Smart Grid Systems: Addressing Privacy Threats, Security Vulnerabilities, and Demand–Supply Balance (A Review)

Abstract

The smart grid (SG) plays a seminal role in the modern energy landscape by integrating digital technologies, the Internet of Things (IoT), and Advanced Metering Infrastructure (AMI) to enable bidirectional energy flow, real-time monitoring, and enhanced operational efficiency. However, these advancements also introduce critical challenges related to data privacy, cybersecurity, and operational balance. This review critically evaluates SG systems, beginning with an analysis of data privacy vulnerabilities, including Man-in-the-Middle (MITM), Denial-of-Service (DoS), and replay attacks, as well as insider threats, exemplified by incidents such as the 2023 Hydro-Québec cyberattack and the 2024 blackout in Spain. The review further details the SG architecture and its key components, including smart meters (SMs), control centers (CCs), aggregators, smart appliances, and renewable energy sources (RESs), while emphasizing essential security requirements such as confidentiality, integrity, availability, secure storage, and scalability. Various privacy preservation techniques are discussed, including cryptographic tools like Homomorphic Encryption, Zero-Knowledge Proofs, and Secure Multiparty Computation, anonymization and aggregation methods such as differential privacy and k-Anonymity, as well as blockchain-based approaches and machine learning solutions. Additionally, the review examines pricing models and their resolution strategies, Demand–Supply Balance Programs (DSBPs) utilizing optimization, game-theoretic, and AI-based approaches, and energy storage systems (ESSs) encompassing lead–acid, lithium-ion, sodium-sulfur, and sodium-ion batteries, highlighting their respective advantages and limitations. By synthesizing these findings, the review identifies existing research gaps and provides guidance for future studies aimed at advancing secure, efficient, and sustainable smart grid implementations.

1. Introduction

The smart grid (SG) technology has sparked a revolutionary shift in the manner in which energy is produced, distributed, and consumed. These modernized grids are an enhancement of traditional grids, empowered by digital technologies, the Internet of Things (IoT), and Advanced Metering Infrastructure (AMI). This transformation has enabled them to bridge the gap between energy efficiency, sustainability, and resource reliability [1]. With the help of SGs, energy and electricity can flow in both directions, allowing consumers to interact with energy providers in real time. This interaction facilitates real-time monitoring of energy consumption, automatic billing, and demand-response systems [2,3].

Nonetheless, like every other digital transformation, the accompanying increase in data usage, communication, andautomation introduces issues of data privacy, security, and cyber threats. Examples of sensitive information circulating within SG networks include patterns of electricity consumption, supply/demand data, and price fluctuations [4]. If such data is intercepted or manipulated by malicious actors, it can reveal users’ personal information, potentially leading to identity theft, energy theft, and interruptions in energy services [5,6]. Therefore, to defend these systems against evolving cybersecurity threats, robust protection mechanisms and strong security measures must be established.

The primary distinction between SGs and traditional electrical grids lies in their bidirectional electricity flow, moving beyond the unidirectional model of the past. SGs integrate real-time data processing and two-way communication [7]. The inclusion of digital technologies allows for more efficient grid management, preventing real-time overloads or shortages in energy demand and supply. This results in improved energy efficiency, reduced operational costs, and enhanced reliability [8]. Furthermore, the penetration of renewable energy forms, such as solar and wind, becomes easier, fostering a more environmentally friendly energy ecosystem.

SGs also enable the development of more sophisticated consumer services. Using smart meters, consumers can make informed decisions regarding their electricity consumption in real time, based on their actual usage. Additionally, dynamic pricing models offer consumers the opportunity to optimize electricity usage when it is most cost-effective, thereby saving money on energy costs and conserving energy during peak hours [9].

Nevertheless, security risks emerge alongside these innovations. Because SGs gather and transmit sensitive consumer information and offer the functionality to operate or command power-related devices remotely, they present various attack vectors that can be exploited by cybercriminals [10]. Beyond information interception, SGs face challenges such as Denial-of-Service (DoS) attacks, among others [11].

IoT technologies have been central to the emergence of SGs. Integrating smart meters (SMs), grid sensors, smart appliances, and electric vehicles (EVs) with intelligent capabilities, IoT facilitates real-time data monitoring, collection, and control of grid operations. These devices capture data and transmit it to control centers (CCs) that analyze and interpret the information to predict demand, modify generation levels, and optimize energy distribution [12].

Advanced Metering Infrastructure (AMI) is a core element of SGs, implementing the use of smart meters to transmit and gather data regarding consumers’ electricity consumption and transfer that information to utility providers. The real-time transfer of this data enables utilities to understand demand fluctuations, energy consumption habits, and efficiency opportunities [13]. Moreover, the integration of renewable energy sources, such as solar panels and wind turbines, into SG systems is achievable with the help of highly advanced data communication and control systems. AMI systems also allow for automated billing, minimizing human error and ensuring accurate charge calculations [14].

Despite these advantages, the advent of IoT devices and AMI brings significant privacy concerns that cannot be neglected. Data stored in these devices, such as electricity usage habits over time, can be used to infer users’ activities, daily lifestyles, and even the presence of individuals in a household. Accessing such information by an unauthorized party can lead to identity theft, surveillance, or physical security breaches [15]. Furthermore, IoT devices are often susceptible to hacking due to their inherent vulnerabilities, exposing them to attackers who can tap into or modify data, leading to severe privacy and security implications.

Secure communication requires participants to share information such as energy demand, supply, prices, and the time of delivery or requirement with market operators [16]. Using this information, market operators can estimate the expected Demand and Supply Ratio, which helps determine the price of energy through uniform pricing or auctions by deploying several techniques such as optimization [17], game-theoretic models [18], and artificial intelligence models [19].

An effective price signal further assists the grid in maintaining a balanced state by distributing surplus energy among buyers. However, the stochastic nature of renewable energy sources creates challenging conditions for keeping the grid in a balanced state [20]. To overcome this problem, the Demand–Supply Balance Program (DSBP) provides further guidelines to participants by deploying several models such as optimization [21], game-theoretic [22], and artificial intelligence [23].

Energy storage systems (ESSs) further improve the grid’s balance by injecting surplus energy during times of high demand [24]. This energy is stored in batteries made from different chemical compositions, such as lead–acid [25], lithium-ion [26], sodium–sulfur [27], and sodium-ion [28].

In this review article, we have deeply explored all these important elements and discussed in detail the methods and techniques that have been extensively used in the past and are currently being deployed in the smart grid. Based on an intensive literature analysis, this review article further assists researchers by highlighting current research gaps and providing a clear path for future research directions.

The rest of this paper is organized as follows: Section 2 and its associated subsections discuss data privacy in the smart grid, along with vulnerabilities and entry points for possible cyberattacks. Furthermore, in this section, several techniques have been explored in detail to analyze their usage in the smart grid. Section 3 and its associated sections deeply analyze pricing models along with their solution techniques. Section 4 highlights the important factors that impact Demand–Supply Balance Programs and how different methods are used to solve them. Section 5 discusses energy storage systems in the smart grid by incorporating several types of batteries and their associated merits and challenges. Section 6 provides existing research gaps and highlights future research directions. Finally, Section 7 presents the conclusion of this work.

2. Data Privacy Threats in Smart Grids

Data privacy is a crucial concern for smart grid (SG) systems, primarily due to the vast amount of sensitive information exchanged across the network. The information gathered by smart meters (SMs) can be processed to reveal specific user behaviors, including their energy usage hours, the nature of appliances they use, and the time they spend at home. When this information becomes available to malicious actors, it can be used to infer an individual’s lifestyle or even anticipate their future actions [29]. Because SGs collect vast amounts of detailed energy consumption data, they create the potential for consumer profiling. Malicious or unauthorized parties could use this data to deduce sensitive information about individuals, such as their daily routines, presence or absence from home, and even the appliances they use [30]. For example, an attacker could infer when people are at home, what devices they are using, or predict their behavior based on consumption patterns.

Malicious entities may alter consumption data to lower electricity bills or access electricity without paying, a practice known as energy theft. Since SG systems rely on continuous data transmission, attackers can manipulate the data exchanged between smart meters and aggregators, leading to revenue loss for utility companies [31]. Additionally, with inadequate encryption or authentication, data being transmitted from SMs to central hubs can be intercepted, revealing confidential consumption patterns or allowing attackers to tamper with the data [32,33]. Furthermore, attackers can link data from different sources to reconstruct a user’s schedule, inferring when they are present, when they are absent, or even when they are engaged in certain activities based on their energy consumption profile [5].

As the collection of sensitive data grows, these privacy concerns become more significant. Without effective privacy-preserving solutions, users may lose confidence in SG systems, which could lead to a reluctance to adopt smart meters and smart appliances. Therefore, privacy preservation is a critical requirement for the successful implementation and widespread acceptance of SGs.

2.1. Overview of Smart Grid Systems

The smart grid (SG) is an advanced, bidirectional electrical power system that integrates traditional grid infrastructure with cutting-edge Information and Communication Technology (ICT). This integration enables SGs to dynamically manage and monitor the flow of energy, offering numerous advantages over traditional power grids. The SG system optimizes electricity generation, distribution, and consumption through the integration of smart meters, sensors, automated control systems, and renewable energy sources [34].

The SG is composed of a series of interrelated subsystems that contribute to the provision of sustainable, reliable, and efficient energy management. These subsystems are as follows:

Smart Management System: This system supports the balance of electricity demand and supply. It operates through control centers (CCs) where smart meters (SMs) and other devices relay information to be analyzed. This analysis helps anticipate demands, thereby enabling optimal control of grid operations. An intelligent management system also facilitates the integration of renewable energy sources into the power grid and the efficient distribution of available resources.

Smart Infrastructure System: This subsystem allows for the bidirectional flow of electricity and information between end-users and utilities. It incorporates smart meters (SMs) that monitor electricity consumption, smart appliances that can manage energy levels in buildings or businesses, and communication systems that facilitate the seamless flow of information among various entities within the grid.

Smart Protection System: This subsystem improves grid reliability by monitoring for faults and preventing permanent equipment damage. It has built-in automated units that can respond promptly to conditions like power failures or equipment malfunctions, thereby reducing repair time and ensuring uninterrupted service.

Smart Grid Communication Network: A communication network is a primary characteristic of the SG because it enables the interchange of information between all elements. The structure of communication systems in SGs is typically hierarchical, with information flowing from SMs to control centers via intermediate aggregators. This architecture allows for the optimization of data transmission, the minimization of delays, and ensures all components collaborate effectively. The communication framework of a smart grid generally consists of the following networks:

• Customer Premises Area Network (HAN/BAN): Data on electricity usage by individual devices is sent to SMs via the Home Area Network (HAN) and Building Area Network (BAN). The communication protocols used by these networks are low-power, such as ZigBee, Wi-Fi, and blacktooth [35,36]. The primary concerns in this layer are privacy and security because personal data regarding consumer actions is collected and distributed.
• Wide Area Network (WAN): The WAN links all regional networks and the CCs. It is responsible for long-distance data transfer, often spanning tens of kilometers. For this purpose, optical communication and cellular networks are widely applied because they facilitate fast data transfer. The WAN is vital for providing a reliable control linkage to the distributed smart grid and its control systems [37].

2.2. Key Components of the Smart Grid

The SG system consists of multiple important elements, which allow an enhanced performance of the grid and its intelligence, as shown in Figure 1.

From Figure 1, it can be seen that the smart grid consists of several components such as the smart meter (SM), control center (CC), aggregators, smart appliances, and renewable energy sources (RESs).

Smart meters are intelligent devices installed in residential areas and businesses to collect electricity usage data and transmit it back to the operations centers [38]. SMs enable real-time monitoring, a factor that is vital for optimizing demand-response programs and implementing dynamic pricing [39].

The CC is the central hub of the SG that receives, transmits, and analyzes data from SMs and other equipment to facilitate real-time decision-making regarding energy supply and demand. This information is used by the CC to optimize grid operation, identify possible failures, and react to unforeseen situations like power cuts or supply shortages [40,41].

Nodes that gather data from various SMs, known as aggregators, also serve as an intermediary between SMs and the CC. They collect usage data and send it to the CC for analysis [42].

Smart appliances are intelligent products that can regulate the amount of energy they use based on directions from the SG. Smart appliances are instrumental in demand-side management as they can reduce energy consumption during peak periods [43].

RESs are a means of energy generation. The SG promotes the inclusion of renewable energy sources such as solar panels and wind turbines. This integration helps in balancing supply and demand, in addition to reducing the carbon footprint of energy production [44].

2.3. Vulnerabilities and Entry Points for Data Leaks in the Smart Grid Network

While SGs offer numerous benefits in terms of efficiency and sustainability, they also present various vulnerabilities that can be exploited by attackers. One of the primary concerns is the reliance on cyber-physical systems, which combine computational systems with physical infrastructure like SMs, aggregators, and CCs. These devices and their interconnected communication channels present numerous points of vulnerability.

The SG system, while revolutionizing the energy sector, has introduced new cybersecurity vulnerabilities due to its reliance on digital communication and data exchange [45]. As shown in Figure 2, cyberattacks pose significant risks, as evidenced by several high-profile incidents that have affected energy infrastructures worldwide. These recent cyberattacks on SG and energy infrastructure highlight therapidly evolving threats and vulnerabilities.

One of the major threats to the SG network is the Man-in-the-Middle (MITM) attack. During this attack, an intruder listens in on and intercepts a legitimate communication between two devices in a network (e.g., between an SM and an aggregator) [46]. This allows the culprit to intercept or manipulate information being exchanged, resulting in data loss, privacy breaches, and interruptions to the energy supply chain. Hackers can also forge data, such as inaccurate electricity usage measurements, to manipulate power distribution [5,47].

The Denial-of-Service (DoS) attack is another type of vulnerability where intruders flood the network with excessive data, thereby slowing down or blocking legitimate service requests [48]. Such attacks can cause a system failure, preventing users from accessing important services like real-time energy consumption details or the ability to adjust their energy usage during peak demand times. A Distributed Denial-of-Service (DDoS) attack is particularly harmful because it uses multiple sources of traffic, making it more difficult to mitigate [49].

Another significant issue in SG networks is the Replay attack. This involves the interception and retransmission of data communicated on the network, causing distorted information to be processed by the receiving system [50]. This can lead to improper load balancing, overcharging of users, and even blackouts if the corrupted data affects grid management decisions.

In SG networks, Theft of Resources and Insider attacks are also highly dangerous. Insider attacks occur when malicious individuals (including disgruntled employees or contractors) gain unauthorized access to system resources, including control systems, databases, or SMs. Such actions can result in the manipulation of data or physical assets, thereby disrupting grid operations and compromising customer privacy [51].

The major attacks and their historical impact are given as follows:

• In 2023, Hydro-Québec, a major grid operator in Canada, experienced a cyberattack that disrupted its outage verification app and website. This incident underscores the increasing cyber threats faced by energy utilities and the need for robust cybersecurity measures to protect critical infrastructure [52,53].
• In 2024, Spain’s National Cybersecurity Institute (Incibe) investigated a massive blackout that disrupted essential services. The focus was on smaller electricity generators, which may have lacked robust cyber defenses, making them potential vulnerabilities in the national power grid [54].
• In May 2023, a critical vulnerability in the MOVEit-managed file transfer software was exploited by the ransomware group CL0P, leading to unauthorized access to sensitive databases. The breach affected over 2,700 organizations across various sectors, including energy, highlighting the systemic risks inherent in the interconnected nature of digital supply chains [54,55].
• From December 2023 through to early 2024, Iranian hacker groups engaged in successful cyberattacks on various sectors, including the energy sector. Such attacks interrupted services and exposed weaknesses in critical infrastructure, pointing to the necessity of more efficient cybersecurity practices to defend against state-sponsored cyber threats [56].

2.4. Security and Privacy Requirements in the Smart Grid Network

The smart grid (SG) network involves a large number of interdependent devices and systems that regularly transfer sensitive information. To maintain an efficient and secure operational network, several key security and privacy requirements must be met. These needs include upholding confidentiality, data integrity, availability, and secure storage, as well as ensuring scalability.

Confidentiality is crucial in SG networks. It refers to the protection of sensitive information from unauthorized access. This is especially necessary in SG systems, where data specific to individuals, such as energy usage trends, is constantly uploaded from smart meters (SMs) to local aggregation nodes [57]. If this data is intercepted, attackers could learn about users’ consumption habits, leading to serious privacy invasions. Furthermore, businesses that optimize operations based on electricity consumption patterns may expose sensitive data like trade secrets or customer behavior analytics [58].

Integrity ensures that information transmitted through the SG network cannot be altered and remains trustworthy. When applied to SGs, a lack of integrity can be a significant issue [59]. For instance, if an attacker gains control over electricity usage data transmitted from a smart meter to the control center, the grid could over- or underestimate demand, which might cause blackouts or inefficient resource utilization. Threats to data integrity, such as Man-in-the-Middle (MITM) attacks and replay attacks, must be addressed and monitored constantly [60].

Availability is a safeguard that ensures services and data within the SG network are accessible when needed. Having real-time energy consumption information is critical for grid operators to make real-time decisions, which helps manage peak demand and implement curtailment strategies. In case of a loss of availability due to a Denial-of-Service (DoS) or Distributed Denial-of-Service (DDoS) attack, grid operators would be unable to control resource distribution, potentially causing widespread outages and operational inefficiencies [61].

The SG network is also affected by major issues in terms of secure storage. Because SG networks contain a large amount of sensitive data, the importance of keeping such data in a secure environment cannot be understated. Without proper encryption and data protection mechanisms, stored data is at risk of theft or manipulation [62]. The risk of data breaches escalates if hackers can access data centers or other storage facilities.

Scalability is another important concern in SG networks. Security solutions must be able to support an expanding network and adapt as more smart devices are deployed, thus handling the increasing data and communication traffic [63]. Traditional security procedures may not be adequate for larger, decentralized systems, necessitating more robust and dynamic security solutions.

To address these security and privacy needs, a layered security approach is essential. This includes features such as advanced encryption, multi-factor authentication, intrusion detection systems, and regular security audits. The SG network can only be safe and consistent by ensuring that all its aspects—confidentiality, integrity, availability, secure storage, and scalability—receive due attention.

2.5. Privacy Preservation Techniques in Smart Grid Network

Since SGs are becoming an essential part of modern-day energy systems, it is of utmost importance to ensure the privacy of highly sensitive data, including the energy consumption patterns of consumers. An SG connects energy producers, consumers, and utility providers, and facilitates real-time communication. Nevertheless, the constant flow of data exposes consumers to vulnerabilities, particularly concerning data confidentiality. Preserving privacy in SG networks is aimed at protecting sensitive information, such as the energy usage of individuals, preventing unauthorized access, and mitigating the potential risks of privacy disclosure that arise from sharing this sensitive information. Various methods have been introduced in the literature, all of which aim to address a particular aspect of privacy and the secure transfer of information via the grid, as shown in Figure 3.

2.5.1. Cryptographic Techniques

Cryptography is a major defense applied to establish the security of SG network data, whether in transit or at rest, and also ensures that sensitive information links are not hacked or altered. Within the dynamics of the smart grid, cryptographic methods are utilized to ensure the privacy and integrity of data while still preserving the functionality of the grid. Various cryptographic mechanisms have been suggested to ensure data privacy.

2.5.2. Homomorphic Encryption (HE)

HE is a cryptographic technique that enables computers to perform operations on encrypted data without first decrypting it. The subsequent output is also encrypted and can only be decoded by the authorized user, which means the data remains confidential throughout processing [64,65]. Such an approach is highly beneficial in SG systems because data is often consolidated and processed on third-party servers. In recent years, a number of researchers have suggested Homomorphic Encryption schemes to implement SG metering systems whose aggregation protocols can collect consumption data without interfering with the privacy of specific users [66,67]. Furthermore, HE can be utilized when collecting data to allow electricity companies to aggregate the information they obtain through smart meters (SMs) without revealing intimate user behaviors [68,69,70].

HE provides the dual advantage of preventing the disclosure of sensitive data while simultaneously enabling utilities to gain insights into energy consumption trends [71]. This is especially important for privacy in SGs, where exposing consumption patterns can lead to security threats or user profiling [72]. Nevertheless, HE is associated with a high computational cost due to its more complex operations on encrypted data compared to conventional encryption or plaintext processing [73]. Research has therefore focused on streamlining these methods to minimize overhead costs and ensure they are viable for handling large-scale SG networks, as can be summarized in

Table 1. Summary of Homomorphic Encryption for privacy preservation in Smart Grids.

Refs.	System Model	Goal	Security Parameters	Performances	Limitations
[67,74]	SG, Data Aggregator, SMs	Aggregate encrypted energy consumption data to preserve user privacy in SG.	Confidentiality, Integrity, and Availability of encrypted data	Efficient data aggregation, reducing exposure of sensitive information.	High computational overhead, especially in large networks.
[75]	Smart Metering System, Cloud-based Framework	Enable secure demand-side energy management while preserving privacy using HE.	Privacy of users’ consumption data, Data confidentiality	Secure privacy-preserving aggregation, effective for dynamic demand-response systems.	Heavy computation, delays in processing large-scale data.
[76,77]	SG, Distributed Energy Systems	Preserve privacy during SG data processing and forecasting.	Confidentiality of real-time data, Robustness to cyberattacks	Secure processing and forecasting for energy data, supports multiparty computation.	High encryption cost, latency in data handling for large datasets.
[78]	SM, Consumer-side Aggregators	Enable privacy-preserving energy consumption analysis for smart grid forecasting.	Secure computation, Data integrity, Privacy of consumers	Improved privacy, secure aggregation of energy data, real-time analysis possible.	Large overhead for dynamic data and issues with real-time processing.
[69]	SG, Centralized Data Servers	Securely aggregate and analyze energy data from multiple SMs while preserving privacy.	Data confidentiality, Authentication, Integrity of aggregated data	Low-risk data breaches, secure load forecasting, scalable implementation.	Performance degradation with large datasets and high computation costs.

.

2.5.3. Zero-Knowledge Proofs (ZKPs)

Zero-Knowledge Proofs (ZKPs) are a cryptographic protocol in which one party can prove the veracity of a claim to another party without revealing any information about the proof itself, other than whether the claim is true or not [79]. ZKPs are especially useful for authorization and identity verification in SG systems, where privacy is a primary concern [80]. Liang et al. [81] explained how blockchain-based ZKPs can enhance the privacy of transactions and smart contract predicates in SG systems. With ZKPs, SG participants can authenticate themselves without revealing sensitive data, such as consumption statistics or personal identifiers [80].

Works by [80,82] highlight that the primary advantage of ZKPs is that they preserve user privacy during the authentication process. For example, a smart meter can prove that it is in proper working order without revealing any specific information on the amount of energy it is consuming, as has been suggested in [83,84]. This capability is vital for addressing the rising concerns regarding individual privacy, as no external party can access data that could then be used to profile a user or perform other nefarious actions.

Moreover, ZKPs are increasingly employed in privacy-preserving applications where proving the correctness of data is required without revealing the underlying data. El-Hajj et al. [85] compared various ZKP protocols, including zk-SNARKs, for privacy-preserving authentication in blockchain-based systems and discussed their applications in billing verification and identity authentication. Similarly, Ref. [85] describes how ZKPs can guarantee that smart grid users receive accurate billing without exposing sensitive usage patterns. However, as pointed out by El-Hajj et al., while ZKPs provide strong security guarantees, they introduce significant computational complexity, making them unsuitable for real-time applications requiring rapid processing, such as fault detection or real-time load balancing [85,86].

One drawback of ZKPs is that they can be computationally intensive, particularly in real-time applications, which may limit their scalability in large SG networks, as shown in

Table 2. Summary of Zero-Knowledge Proof (ZKP) and blockchain-based privacy techniques.

Refs.	System Model	Goal	Security Parameters	Performances	Limitations
[87,88]	Blockchain-based SG	Identity verification, ensuring users’ identity without revealing sensitive data	Identity verification, confidentiality, integrity	Efficient identity verification in decentralized systems	Computationally expensive, scalability issues with large-scale grids
[89]	Permissioned and permissionless blockchain	Privacy for identity, transactions, and smart contracts	Privacy preservation, transaction confidentiality, anonymity	Reduces transaction-related privacy risks	Limited by blockchain size and complexity
[90,91]	Blockchain-based SG	Billing verification and identity authentication	Privacy, non-repudiation, confidentiality	Fast and efficient in verifying billing information and user identity	Complex implementation in large-scale smart grid systems
[92]	Blockchain-based Privacy-Preserving Systems	Comparing ZKP protocols (zk-SNARKs, zk-STARKs) for secure authentication	Secure authentication, privacy preservation, efficiency	Comparison of zk-SNARKs, zk-STARKs, and bulletproof protocols for low-latency systems	Limited application to large-scale networks, high computational load for zk-SNARKs
[76]	IoT-based SG	Implementing ZKP for secure data exchange between consumers and utility providers	Confidentiality, authentication	Enables secure and private communication for energy data	High processing overhead in resource-constrained devices

.

2.5.4. Secure Multiparty Computation (SMPC)

SMPC is a cryptographic method that enables multiple parties to jointly compute a function over their private inputs while keeping those inputs confidential [93]. In SGs, SMPC can be used to enable energy consumers, utility companies, and third-party aggregators to collaborate on energy consumption analysis and demand-response programs without sharing sensitive consumption data [76]. Samuel et al. [75] discussed an SMPC framework for SGs that ensures users’ energy usage data remains private, even during collaborative optimization and demand-response computations. Furthermore, this technique ensures that each party can perform computations on their own encrypted data, with only the final aggregated result being revealed [94].

Additionally, SMPC has the advantage of allowing collaboration among multiple stakeholders in a smart grid without exposing confidential private data [95]. For example, a consumer may want to share their energy usage information with the utility to optimize energy pricing but not reveal detailed personal behavior. SMPC achieves this by ensuring that computations are conducted over encrypted data [96]. However, the main challenge with SMPC is its high computational complexity and communication overhead. Handling large datasets and ensuring that computations scale efficiently are key challenges that ongoing research is trying to address [97]. Recent work has focused on improving the efficiency and practicality of SMPC in real-time smart grid systems, as shown in

Table 3. Summary of Secure Multiparty Computation (SMPC) privacy-preserving techniques in smart grids.

Refs.	System Model	Goal	Security Parameters	Performances	Limitations
[88]	SG with multiple users, demand-response systems	Ensure privacy of users’ energy usage data during collaborative optimization and demand-response computations	Data privacy, integrity, confidentiality	Efficient in computation and communication overhead	Scalability issues with large systems and multiple participants
[75,98]	Cloud-based AI and SG systems	Address data privacy and security concerns during energy consumption optimization	Data confidentiality, encryption, non-repudiation	Enhances security for cloud-based smart grid systems	Increased latency and computation costs due to complex cryptographic operations
[99]	Distributed SG with multiple consumers and grid operators	Secure collaborative computations for demand response without compromising privacy	Privacy, confidentiality, trust, authentication	Highly efficient for small to medium-sized grids	Scalability issues with large datasets and numerous participants

.

2.5.5. Anonymization and Aggregation

Anonymization and aggregation are techniques used to protect personal identities by altering data in such a way that it cannot be traced back to an individual. These methods are often applied to aggregate datasets, which reduces the granularity of the information while maintaining its usefulness for grid management.

• Differential Privacy: The process of anonymization, particularly through differential privacy (DP), has gained significant attention in the smart grid sector for protecting individual data. Differential privacy guarantees that no individual’s data or the data collected on themsignificantly affects the response of a query [100]. Researchers Hu et al. [101] proposed a privacy-preserving data aggregation and management approach for smart grid architecture, highlighting the feasibility of using differential privacy in data aggregation in smart meters. As another example, Rai et al. [102] used differential privacy to provide privacy guarantees to energy consumption data and showed that, under this scheme, individual consumer privacy could be achieved without compromising the utility of the aggregate consumption statistics for optimizing a smart grid. Bourechak et al. [103] demonstrated that a smart grid model based on differential privacy could be constructed, in which noise is added to smart meter data in such a way that the privacy of individual consumers is guaranteed. Among the main benefits of DP is that it provides formal privacy guarantees and is, therefore, a powerful strategy to protect sensitive information in smart grids. The method ensures that no individual’s data can significantly influence query results, even if an adversary has access to auxiliary information [104]. However, a known trade-off of DP is that the addition of noise may slightly reduce the accuracy of aggregated results [105].
  Further research focuses on optimizing the amount of noise added to maintain a balance between privacy and data utility, ensuring that the SG can still function efficiently while maintaining strong privacy guarantees [106,107]. The following Table 4 provides a summary of a few works concerning the deployment of differential privacy in the smart grid.

  Table 4. Summary of differential privacy techniques in smart grids.

  Refs.	System Model	Goal	Security Parameters	Performances	Limitations
  [108]	General privacy-preserving model	Provide privacy guarantees while sharing data from smart meters in smart grids	Privacy, data perturbation, noise addition	Provable privacy guarantees under differential privacy	Increased noise can affect data utility
  [109]	SG with individual consumer data	Protect individual user data while enabling utility companies to aggregate energy consumption	Data privacy, utility preservation, statistical analysis	High privacy preservation with minimal data alteration	Performance degradation in the aggregation of data due to noise
  [107]	SG with multiple consumers and data aggregation points	Enable privacy-preserving data aggregation for demand-response management	Privacy, data aggregation, differential privacy	Ensures that data leaks do not occur, and efficient aggregation is maintained	Limited effectiveness for highly detailed data
  [109]	SG with large-scale data collection	Apply differential privacy to protect energy usage data for large-scale grids	Data perturbation, confidentiality, utility maximization	Enhances privacy without large overhead for large datasets	Requires balancing between privacy and data quality

  Table 4. Summary of differential privacy techniques in smart grids.

  Refs.	System Model	Goal	Security Parameters	Performances	Limitations
  [108]	General privacy-preserving model	Provide privacy guarantees while sharing data from smart meters in smart grids	Privacy, data perturbation, noise addition	Provable privacy guarantees under differential privacy	Increased noise can affect data utility
  [109]	SG with individual consumer data	Protect individual user data while enabling utility companies to aggregate energy consumption	Data privacy, utility preservation, statistical analysis	High privacy preservation with minimal data alteration	Performance degradation in the aggregation of data due to noise
  [107]	SG with multiple consumers and data aggregation points	Enable privacy-preserving data aggregation for demand-response management	Privacy, data aggregation, differential privacy	Ensures that data leaks do not occur, and efficient aggregation is maintained	Limited effectiveness for highly detailed data
  [109]	SG with large-scale data collection	Apply differential privacy to protect energy usage data for large-scale grids	Data perturbation, confidentiality, utility maximization	Enhances privacy without large overhead for large datasets	Requires balancing between privacy and data quality

• k-Anonymity, another widely studied anonymization technique, has been proposed [110] and applied to ensure that energy consumption data cannot be linked to fewer than k-1 other users. Sweeney et al. [111] initially introduced this concept, and it has been applied in various domains, including SGs. In the context of SGs, k-Anonymity helps obscure individual energy usage to prevent privacy violations. However, recent studies, such as the one proposed in [112], indicate that k-Anonymity is vulnerable to attacks when auxiliary information is available, which makes it insufficient for protecting against linkability attacks. Moreover, Sakthivel et al. [113] proposed a k-Anonymity-based model for anonymizing smart meter data, where they ensure that the consumption patterns are generalized so that no individual data point can be traced back to a particular consumer.
  Additionally, the advantage of k-Anonymity is that it is relatively simple to implement and can provide a basic level of privacy by masking individual consumption details [114]. The proposed studies make it a useful tool for SGs where data must be aggregated and shared, but individual privacy must be preserved [115]. However, k-Anonymity can lead to a loss of granularity in the data, which might diminish its usefulness for detailed grid analysis [116]. Additionally, the technique is vulnerable to attacks that use auxiliary data or background knowledge, such as the homogeneity attack or the background knowledge attack, where attackers may still infer sensitive information [117]. A summary of a few other studies based on k-Anonymity is presented in Table 5.

  Table 5. Summary of k-Anonymity and data anonymization techniques.

  Refs.	System Model	Goal	Security Parameters	Performances	Limitations
  [118]	Residential SM data	Protect consumers’ privacy by ensuring that energy consumption patterns are indistinguishable from other consumers	k-Anonymity, data generalization, suppression	Achieves a high level of privacy with minimal loss of data quality	Data utility significantly decreases when increasing k
  [119]	SG with aggregated consumption data	Preserve privacy of consumers’ energy usage data while allowing aggregation for demand response	Data anonymization, k-Anonymity, data perturbation	Effective at protecting privacy without large computational overhead	Performance degradation in data analysis when large k is required
  [120]	SG with individual consumption data from residential users	Protect sensitive data while maintaining usability for grid management tasks	k-Anonymity, noise injection, clustering	Effective for privacy protection in real-time smart grid applications	Increased noise can reduce the accuracy of demand forecasting

  Table 5. Summary of k-Anonymity and data anonymization techniques.

  Refs.	System Model	Goal	Security Parameters	Performances	Limitations
  [118]	Residential SM data	Protect consumers’ privacy by ensuring that energy consumption patterns are indistinguishable from other consumers	k-Anonymity, data generalization, suppression	Achieves a high level of privacy with minimal loss of data quality	Data utility significantly decreases when increasing k
  [119]	SG with aggregated consumption data	Preserve privacy of consumers’ energy usage data while allowing aggregation for demand response	Data anonymization, k-Anonymity, data perturbation	Effective at protecting privacy without large computational overhead	Performance degradation in data analysis when large k is required
  [120]	SG with individual consumption data from residential users	Protect sensitive data while maintaining usability for grid management tasks	k-Anonymity, noise injection, clustering	Effective for privacy protection in real-time smart grid applications	Increased noise can reduce the accuracy of demand forecasting

• Data aggregation models have also been extensively researched as an alternative to individual data storage. Zhan et al. [121] discussed how aggregation at the community level can maintain grid performance while ensuring that individual data is protected. The proposed research in [122] highlights the advantage of aggregating energy usage data, as it reduces the risk of sensitive data leakage while optimizing grid control. Privacy protection through aggregation, however, has limitations because it increases data generalization, which can limit its value for individualized services such as adaptive pricing or demand-response programs [123].
  However, a balance must be maintained between data aggregation and the loss of detail, as privacy protection should not undermine effective grid management [124]. Furthermore, Xu et al. [125] proposed a data aggregation framework for SG systems utilizing Secure Multiparty Computation (SMPC) to aggregate data from multiple smart meters while preserving the privacy of individual consumption information. This approach ensures that only aggregated consumption data is shared with the grid operator, thereby preventing the disclosure of sensitive user information.
  Data aggregation in the smart grid is essential as it enables operators to analyze overall consumption and develop an understanding of future requirements and efficient strategies for resource allocation. However, there are some limitations in achieving both privacy and utility because excessive noise added during the aggregation process may reduce its usefulness for grid optimization [126]. Researchers are working on developing models that aggregate data in a way that maintains privacy while providing accurate and useful insights for grid management.

  Table 6. Summary of privacy-preserving data aggregation techniques.

  Refs.	System Model	Goal	Security Parameters	Performances	Limitations
  [115]	SG with smart meters and aggregators	Aggregate energy consumption data while preserving consumer privacy	Data aggregation, privacy preservation, encryption	Efficient aggregation with reduced communication overhead	Aggregated data may lose individual-level accuracy
  [127]	Residential energy usage data from SMs	Aggregate data in a way that maintains privacy without compromising data utility for analysis	Homomorphic Encryption, data anonymization	Effective at maintaining data utility for grid management	Higher computational complexity due to encryption processes
  [128]	SG with distributed energy resources	Perform secure aggregation of energy consumption data across different smart meters	Data aggregation, privacy preservation, Secure Multiparty Computation	Reduced data leakage with minimal performance impact	Some aggregation models increase latency due to encryption
  [129]	Energy data aggregation in a smart city	Aggregate data from multiple smart meters while maintaining privacy of users’ energy consumption patterns	Data aggregation, differential privacy, noise injection	High accuracy in aggregated results with strong privacy guarantees	Increased noise can reduce the accuracy of the aggregated data

  presents a summary of a few studies with their limitations.

2.5.6. Blockchain-Based Mechanism

Blockchain technology has gained popularity in the smart grid space for its ability to provide decentralized, secure, and transparent data management. By using a distributed ledger system, blockchain enables secure transactions without the need for a central authority [130].

Smart contracts, which are a key part of blockchain technology, are self-executing contracts with the terms of the agreement written directly into code. Saeed et al. [131] proposed using smart contracts in peer-to-peer energy trading platforms to automate transactions between consumers and producers of renewable energy. Moreover, smart contracts ensure transparency, security, and privacy while eliminating the need for intermediaries, thereby reducing the risk of data manipulation or privacy breaches [132]. As noted in the literature, however, blockchain scalability remains a major concern, particularly as the number of participants grows [133].

A privacy-preserving scheme based on blockchain was proposed in a study by Alsheavi et al. [134] to provide privacy in SGs by using blockchain and smart contracts that automate the sharing of energy consumption data only with authorized entities. Smart contracts eliminate the need for a central authority to implement privacy policies, which leads to greater transparency and trust for all participants. Furthermore, to enable all transactions to be registered and stored in the blockchain in an immutable and easily traceable manner, smart contracts can be used to automate important processes in SGs, such as releasing energy data, delivering payments for energy use, and implementing demand-response measures [135].

Nevertheless, blockchain’s slow transaction speed and the computational resources it requires can cause performance bottlenecks in large-scale smart grid applications, as depicted in

Table 7. Summary of smart contract applications in smart grids.

Refs.	System Model	Goal	Security Parameters	Performances	Limitations
[136]	Blockchain-based SG with decentralized control	Automate energy transactions and ensure secure, tamper-proof agreements between consumers and producers	Blockchain, cryptographic signatures, encryption	High transparency and immutability; reduces fraud in transactions	Scalability issues with large-scale deployments
[137]	Peer-to-peer energy trading in SGs	Use smart contracts to facilitate secure and automated energy trading between prosumers	Smart contracts, blockchain, digital signatures	Efficient energy trading with minimal overhead and reduced human error	Potential delays in transaction execution due to blockchain confirmation times
[138]	Energy management system using smart contracts	Automate billing and payment for energy usage while ensuring privacy and transparency	Blockchain, encryption, access control	Provides secure billing and payment system with reduced administrative costs	High computational power required for contract execution
[88]	Decentralized SG system with smart contracts for grid operation	Enable autonomous grid management and decision-making through automated contracts	Decentralized access control, blockchain, cryptography	Improves operational efficiency and reduces human intervention	Limited adoption due to infrastructure requirements and implementation complexity
[139]	Decentralized energy trading platform	Ensure the privacy and integrity of trading data through secure smart contracts	Blockchain, Zero-Knowledge Proofs (ZKPs), cryptographic protocols	Enhances privacy while automating secure energy trading	Complex implementation and maintenance costs in large-scale grids

.

Decentralized access control has been suggested as a way to use blockchain-based systems to secure consumer information in smart grids (SGs) [140]. Hasan et al. [130] discussed the possibility of making access to energy consumption data decentralized with the help of blockchain, where consumers can control access to their data using cryptographic keys. This strategy provides enhanced privacy and security compared to conventional centralized approaches [141]. Researchers like Alsheavi et al. [134] also demonstrated that a well-structured decentralized access control system can be implemented in SGs to regulate access to sensitive information, including consumption statistics and pricing.

The distributed nature of decentralized access control effectively eliminates the threats of a single point of failure and a central authority being corrupted by an adversary. It achieves this by sharing access control decisions across the entire network. The advantage of blockchain-based decentralized access control is that it enhances transparency and accountability, making it easier to track who accessed what data and when [142]. This ensures that access is granted based on pre-defined permissions and that users can verify that their data is not being accessed without authorization [143]. However, one key challenge of using blockchain for access control is ensuring scalability, as the number of transactions in a large SG network can increase substantially, as shown by the comparative analysis of a few studies in

Table 8. Summary of access control mechanisms in smart grids.

Refs.	System Model	Goal	Security Parameters	Performances	Limitations
[140]	Decentralized energy management system	Provide secure and efficient access to energy data and smart grid resources	Blockchain, public-key infrastructure, cryptographic access control	High security and flexibility; allows dynamic access control	High computational cost and complexity in implementation
[144]	Blockchain-based SG with decentralized control	Enable secure, role-based access to SG data and services	Blockchain, multi-signature authentication, access policies	Enhanced privacy and control over energy data sharing	Complexity in managing large numbers of access policies and users
[145]	SG with decentralized access control for energy data sharing	Allow users to control and grant access to their energy consumption data	Smart contracts, encryption, Zero-Knowledge Proofs	Efficient data sharing with enhanced privacy; reduces risks of unauthorized access	Risk of performance bottlenecks in real-time applications

.

Several consensus algorithms have been used in blockchain for deployment in the smart grid.

Table 9. Comparison of consensus mechanisms for smart grid applications.

Consensus Mechanism	Description	Advantages	Limitations	Storage Requirements	Latency	Scalability	Common Use Cases in Smart Grid	Key References
Proof of Work (PoW)	Miners solve cryptographic puzzles to create a new block.	High security, widely tested	High energy consumption, slower transactions	High (full blockchain replication)	High (minutes scale)	Low (scales poorly)	Data aggregation, electricity consumption recording	[146]
Proof of Stake (PoS)	Validators chosen based on coin holdings staked as collateral.	Energy efficient, faster than PoW	Potential centralization risk	Moderate (depends on node role)	Moderate (seconds scale)	Moderate to high	Distributed energy resources management	[147]
Delegated Proof of Stake (DPoS)	Stakeholders elect delegates to validate transactions and create blocks.	Faster consensus, scalable	Trust issues, potential centralization	Moderate	Low (seconds or less)	High	Real-time energy trading, demand response	[147]
Practical Byzantine Fault Tolerance (PBFT)	Nodes agree on transaction order through voting rounds.	Low latency, suitable for permissioned networks	Scalability limits in large networks	Low to moderate	Very low (milliseconds to seconds)	Low to moderate	Consortium blockchain for monitoring, maintenance	[148]
Proof of Task (PoT)	Nodes complete real-time control tasks for consensus based on contribution.	Real-time control, contribution-based	Emerging, under development	Moderate	Very low	Moderate	Real-time regulation and control of renewable energy systems	[147]
Proof of Authority (PoA)	Trusted validators are pre-approved and known entities maintaining the network.	High throughput, low energy use	Centralization risk	Low (permissioned blockchain)	Very low	High	Permissioned smart grid networks	[146]
Proof of Credit Scores (PoCS)	Consensus based on node credit scores reflecting trustworthiness and performance.	Encourages honest participation, efficient	Complexity in credit evaluation	Moderate	Moderate	Moderate	Smart grid power trading systems	[149]
Proof of Importance (PoI)	Nodes gain importance score based on activity and stake influencing block creation rights.	Encourages active participation and stakeholding	Less common, requires additional metrics	Moderate	Moderate	Moderate	Energy trading and decentralized energy markets	[150]
Proof of Elapsed Time (PoET)	Random wait times assigned to nodes; the first to finish produces the block.	Energy efficient, fair	Requires trusted execution environment	Low	Very low	Moderate	Suitable for permissioned smart grids	[151]
Algorand Consensus	Pure PoS variant with verifiable random functions for leader election.	High security, fast finality	Complexity in implementation	Moderate	Low	High	Distributed energy resource coordination	[152]
RAFT Consensus	Leader-based consensus utilized mostly in permissioned blockchains.	Simplicity, low overhead	Not decentralized	Low	Very low	Moderate	Consortium smart grid implementations	[153]

presents a comparative analysis of several consensus mechanisms for smart grid applications. From Table 9, it can be observed that among the various consensus mechanisms, PBFT and PoA are the most practically applicable to large-scale smart grid applications, as they offer low latency rates with high scalability.

2.5.7. Tokenization

Tokenization, another blockchain-based technique, involves replacing sensitive data with tokens that can be securely exchanged or traded. Tokenization has been proposed [154] and applied in energy systems to represent energy credits or trading rights, allowing consumers to engage in energy trading without revealing detailed consumption data. Taherdoost et al. [155] explored tokenization in decentralized energy markets and demonstrated how it can be leveraged to ensure privacy while allowing participants to trade energy efficiently. Moreover, Chen et al. [156] proposed a tokenization-based approach for protecting sensitive data in SGs, where energy usage information is replaced with anonymous tokens during billing processes, ensuring that users’ data cannot be easily traced or identified.

The primary advantage of tokenization is that it reduces the risks of data breaches because tokens cannot be reverse-engineered to rediscover the original data. This makes tokenization a viable solution for SGs whose transactions and sensitive data should be secured. However, managing the tokenization process requires careful design to ensure that tokens cannot be easily misused and that they do not introduce processing delays or errors in energy data handling [157]. As noted in their study, key management is a critical issue in tokenization systems. The security of tokenized data relies on the secure handling of encryption keys, which, if compromised, can expose sensitive information, as shown in

Table 10. Summary of tokenization applications in smart grids.

Refs.	System Model	Goal	Security Parameters	Performances	Limitations
[158]	Tokenization in SG data management	Protect users’ personal energy consumption data while enabling secure transactions	Cryptographic token generation, data masking	Enhanced privacy by converting sensitive data into non-sensitive tokens	Risk of token theft if not properly managed
[159]	SG platform using tokenized identities for energy trading	Enable secure, anonymous energy transactions among prosumers and consumers	Tokenized identities, encryption, blockchain	Secure energy trading with privacy-preserving features	Possible scalability issues when handling a large number of transactions
[159]	Blockchain-based tokenized smart SG system for data access control	Prevent unauthorized access to smart grid data using tokens for verification	Smart contracts, tokenized access control	Fast, secure data access control with reduced risks of unauthorized access	High computational overhead and reliance on blockchain networks

.

2.5.8. Machine Learning Approaches

Machine learning (ML) techniques can also be applied to privacy preservation in SGs. By learning patterns from data without directly accessing sensitive information, ML models can improve system efficiency while maintaining privacy, with methods like federated learning and adversarial learning gaining popularity.

Federated learning is a decentralized approach to training machine learning models where the data remains on local devices, such as SMs, and only the model updates are shared with a central server [160]. Su et al. [161] explored how federated learning could be applied in smart grid load forecasting, where data privacy is preserved by not transmitting sensitive consumer data to the central server. Furthermore, Alshehri et al. [162] discussed a federated learning framework for detecting energy theft in smart grids, where local models are trained at individual SMs and then aggregated to build a global model without revealing the sensitive energy consumption data of each user. The advantage of federated learning in SGs is that it provides a privacy-preserving way to train machine learning models without centralizing sensitive user data [163]. It enables the detection of anomalies in energy consumption patterns without compromising privacy.

However, federated learning requires efficient communication between devices to share model updates, and it can suffer from performance issues when there is significant data heterogeneity across devices [164]. While federated learning reduces the risk of data breaches, it faces challenges in terms of communication overhead and model convergence, especially when the data distribution across devices is heterogeneous, as shown in

Table 11. Summary of federated learning applications in smart grids.

Refs.	System Model	Goal	Security Parameters	Performances	Limitations
[165]	Federated learning for distributed energy resources (DERs)	Train models for optimal energy dispatch across DERs without revealing individual data	Secure aggregation, federated model synchronization	Efficient energy distribution, better user privacy preservation	Data heterogeneity, model imbalance due to unshared local data
[166]	SG demand-response system using federated learning	Optimize energy consumption models while maintaining privacy	Local model training, differential privacy	Enhanced privacy for users, reduced central data processing	Delays in model updates, potential inefficiencies with large-scale systems
[167]	Federated learning for distributed energy resources (DERs)	Train models for optimal energy dispatch across DERs without revealing individual data	Secure aggregation, federated model synchronization	Efficient energy distribution, better user privacy preservation	Data heterogeneity, model imbalance due to unshared local data

.

Adversarial learning, in which synthetic data is generated to simulate real-world data without revealing sensitive information, has been applied to data augmentation in privacy-preserving machine learning models [107]. Adewole et al. [168] pioneered the use of adversarial learning with Generative Adversarial Networks (GANs), which have since been employed for privacy-preserving data generation in SGs. Takiddin et al. [169] applied adversarial learning to SGs for anomaly detection, training models to identify fraudulent activities, such as energy theft, even in the presence of malicious data tampering. By making the models robust to adversarial inputs, SG systems can become more resilient against data manipulation. While adversarial learning can effectively protect data privacy, it incurs computational costs, and ensuring the quality of synthetic data remains a key challenge, as shown in

Table 12. Summary of adversarial machine learning applications in smart grids.

Refs.	System Model	Goal	Security Parameters	Performances	Limitations
[170]	Adversarial machine learning for anomaly detection in smart grids	Detect adversarial attacks in energy consumption data	Adversarial training, perturbation-based attack detection	Improved robustness against data tampering, anomaly detection in real time	Risk of adversarial model manipulation, requires extensive training data
[171]	SG cybersecurity using adversarial learning for load forecasting	Detect adversarial inputs influencing energy consumption predictions	GANs for synthetic attack generation, adversarial robustness	Increased prediction accuracy under adversarial conditions, resilient against data poisoning	Computational cost of adversarial model training, vulnerability to strong attacks
[31]	Adversarial training for secure energy trading in P2P systems	Enhance security of decentralized energy trading systems by learning adversarial behaviors	Generative adversarial networks (GANs), secure training protocols	Enhanced resistance to cyberattacks in peer-to-peer energy transactions	Difficulty in balancing model performance with adversarial robustness

.

The trade-offs of privacy preservation approaches in smart grids vary significantly. For instance, Homomorphic Encryption (HE) and Zero-Knowledge Proofs (ZKPs) provide strong privacy guarantees but entail high computational costs, limiting their scalability in real-time systems. Secure Multiparty Computation (SMPC) enables collaborative data analysis without sharing raw data but introduces substantial communication overhead. In contrast, anonymization and differential privacy (DP) offer lightweight and scalable solutions, albeit with reduced accuracy. Blockchain-based mechanisms provide tamper-proof data management; however, transaction speed and scalability remain concerns.

Table 13. Comparison of privacy-preserving techniques for smart grids.

Method	Key Advantages	Main Limitations	Key Insights and Suitability for Smart Grids	Latency	Computation Cost/Scalability	Refs.
HE	Strong privacy; allows computation over encrypted data	Very high computational overhead; unsuitable for real-time	Suitable for offline analysis or small-scale aggregation	0.5–5 s (per operation/aggregation)	<10 k users (current demos)	[172]
ZKP	Enables authentication without revealing data	High complexity; latency in large-scale systems	Effective for billing/identity verification, not for bulk data	300–600 ms (per proof)	<5 k users	[173]
SMPC	Collaborative computation without sharing raw data	Heavy communication overhead; scales poorly	Useful for cooperative analytics among few entities	200-ms (For Three Parties)	1.5 Mbps Per operation	[173]
DP	Lightweight; scalable; strong formal guarantees	Loss of accuracy due to noise addition	Good for large-scale smart meter data aggregation	<50 ms	>100 k users	[76]
BC	Tamper-proof records; transparency; decentralization	Scalability and transaction delays; energy consumption	Suitable for settlements and secure trading platforms	1–10 s per transaction (PoW); 200–400 ms (PoS/DAG)	<1 k users (PoW); >50 k users (PoS/DAG)	[174,175]
FL	Data never leaves the devices; transparency; decentralization	Communication overhead, heterogeneity	Suitable for offline analysis or small-scale aggregation	100–300 ms per round	>10 k users	[176]

presents a comparative analysis of these privacy-preserving techniques, highlighting their key insights, computational intensity, and suitability for smart grid applications.

Noting the latency and computational scalability from Table 13, it can be observed that traditional privacy-preserving techniques such as Homomorphic Encryption (HE), Zero-Knowledge Proofs (ZKPs), and blockchain (BC) are still immature for deployment on resource-constrained IoT devices. Therefore, future research should focus on modifying these techniques to make them suitable for such devices. For instance, a lightweight privacy-preserving model has been proposed in [177], which is based on proxy re-encryption and asymmetric scalar product preserving encryption. In this model, the encryption process involves only addition and multiplication operations along with pairing operations rather than more computationally intensive individual operations, thus offering reduced computational overhead while maintaining the privacy of IoT devices.

Another approach addressing computational challenges in ZKPs was proposed in [178], where the authors utilized a Merkle tree architecture to aggregate user attributes and construct commitments for Zero-Knowledge Proof verification. This modification ensures that user attributes and access policies remain confidential while reducing computational overhead.

Regarding the computational intensity of blockchain, a scalable blockchain model has been proposed in [179]. In this model, the authors employed physically unclonable functions (PUFs), one-way hash functions, and bitwise XOR operations to achieve lightweight mutual authentication between smart meters and regional gateways, improving efficiency while maintaining security.

2.6. Trade-Offs Between Privacy Guarantees and Data Utility

Privacy-preserving mechanisms employed in smart grids span a spectrum from cryptographic techniques (Homomorphic Encryption (HE), Zero-Knowledge Proofs (ZKP), Secure Multiparty Computation (SMPC)) to statistical/anonymization approaches (differential privacy (DP), k-Anonymity, aggregation), blockchain/tokenization, and machine learning methods (federated and adversarial learning). Each class delivers a different point on the privacy–utility–performance frontier. Cryptographic tools provide very strong confidentiality; for example, HE enables computation over encrypted measurements but incurs substantial computation and latency that often makes it impractical for hard real-time use cases (high computation/latency, limited scalability). ZKPs provide strong verification without disclosure but add proof-generation latency and implementation complexity. SMPC offers collaborative analytics without revealing raw inputs but typically introduces large communication overheads and scaling difficulties. By contrast, DP and aggregation are lightweight and highly scalable and provide formal privacy guarantees, yet their injected noise or data generalization reduces accuracy for downstream tasks such as short-term load forecasting or fine-grained demand response. Blockchain and tokenization improve tamper resistance and auditability but suffer transaction speed and storage costs that reduce their applicability in very high-throughput measurement scenarios. Federated and adversarial learning keep raw data local and preserve privacy to varying degrees, but they face communication costs, data heterogeneity, and potential leakage through model updates. These qualitative trade-offs are summarized in Table 13.

To make these trade-offs concrete, systems researchers should move beyond qualitative statements and report a small set of standardized metrics for each use case:

• A privacy-level parameter (e.g., DP’s $\epsilon$, key sizes or proof sizes for cryptographic techniques, attacker success rate/re-identification probability);
• Utility metrics tied to the application (e.g., RMSE/MAE for forecasting, billing error for billing tasks, detection F1/AUC for theft/anomaly detection);
• Operational costs (CPU cycles, latency, memory, communication bytes).

Reporting curves that sweep the privacy parameter and plot utility and cost (e.g., utility vs. $\epsilon$, latency vs. message size, re-identification risk vs. k for k-Anonymity) reveals practical operating points. For instance, DP implementations require choosing $\epsilon$ to balance forecasting error and privacy: very small $\epsilon$ (strong privacy) implies larger injected noise and degraded forecasting/billing accuracy, while larger $\epsilon$ reduces noise but weakens formal guarantees. Similarly, HE/ZKP/SMPC variants differ in proof size/operation time and hence in feasibility for real-time grid control. The literature repeatedly emphasizes these trade-offs and the need to optimize noise/complexity for smart grid constraints, as shown in Table 9, Table 10, and Table 12.

Therefore, future research should focus on hybrid architectures: use lightweight DP or aggregation for continuous high-frequency monitoring (to preserve utility at scale), and reserve heavier cryptographic proofs (HE, ZKP, SMPC) for periodic audits, billing settlement, or offline analytics where latency is acceptable. Federated learning can complement this approach by keeping model training local and sharing only encrypted/DP-protected updates for centralized model aggregation. Such a mixed strategy leverages the strengths of each technique while controlling utility loss and performance overhead.

3. Pricing Mechanisms for Energy Trading in Smart Grid

Secure information motivates smart grid energy market participants to share their data with confidence, without fear of tampering. Such secure information enables regulatory authorities to determine fair energy trading prices. Market price determination also involves various pricing mechanisms, which depend on supply and demand, encouraging efficient use of energy resources and stabilizing the market.

Depending on the trading process, energy pricing in peer-to-peer (P2P) energy trading can be categorized into synchronous and asynchronous techniques, assuming P2P energy trading occurs within a two-sided market among multiple prosumers and consumers exchanging electrical energy [16]. This categorization provides a foundational understanding of energy prices by highlighting the trading process and the peculiarities of each approach.

A synchronous pricing mechanism is one where market players submit bids, and the energy price and trading volume are determined using market objectives, which may include optimizing social welfare or minimizing production costs. This system-centric approach to P2P energy trading resembles traditional wholesale electricity markets [17,180]. In this process, a forward energy trading market is initiated on a platform at a specified time. Prosumers and consumers submit sealed bids indicating the price and quantity of energy they wish to trade during this window. After the bidding period ends, the platform reveals the results and clears the market. The resulting energy price controls the welfare of all market participants.

Synchronous energy trading can enhance market efficiency compared to asynchronous pricing, particularly if trading partners directly manage transaction parameters. However, its effectiveness may diminish if a participant manipulates the merit order to exercise market power [180].

Figure 4 illustrates the types of synchronous and asynchronous pricing mechanisms.

Different synchronous pricing models, as depicted in Figure 4, share a common objective function outlined in

Table 14. Objective functions for synchronous energy pricing mechanisms.

Pricing Mechanism	Objective Function
Uniform Pricing	Maximize social welfare: $$W=\sum _{i=1}^n\left(U_i\left(P\right)-C_i\left(P\right)\right)$$ where W is total social welfare, $U_i\left(P\right)$ is the utility function for participant i at price P, and $C_i\left(P\right)$ is the cost for participant i at price P.
Discriminatory Pricing	Maximize individual profits or market participation: $${\Pi }_i=(P_i·q_i)-C_i\left(q_i\right)$$ where ${\Pi }_i$ is the profit for participant i, $P_i$ is the price set by participant i, $q_i$ is the energy traded, and $C_i\left(q_i\right)$ is the cost for participant i.
Locational Marginal Pricing (LMP)	Minimize total system cost subject to power flow and grid constraints: $$\sum _{i=1}^n{C}_i\left(P_i\right)+\sum _{j=1}^m{\lambda }_j\left(P_j-P_{\mathrm{demand}}\right)$$ where $C_i\left(P_i\right)$ is the cost of generation, ${\lambda }_j$ is the shadow price at node j, and $P_{\mathrm{demand}}$ is the demand.
SDR-based Pricing	Maximize market efficiency and fairness by balancing supply and demand: $$\mathrm{SDR}\mathrm{Price}={\displaystyle \frac{S}{D}}$$ where S is the total supply and D is the total demand.

. While these models offer multiple advantages, they also present operational challenges, summarized in Table 15.

Asynchronous energy pricing enables numerous bilateral agreements between prosumers and consumers during the trading window, analogous to a flea market where participants negotiate energy prices that suit them individually. This methodology aims to achieve a collective agreement among market players in diverse situations. To facilitate optimal outcomes, distributed decision-making procedures have been proposed to support cooperation and coordination among participants [181].

In the asynchronous model, the market platform allows prosumers and consumers to negotiate energy prices and trading volumes directly. The consensus process varies depending on the framework employed. A key feature of asynchronous energy pricing is that trading is participant-driven, which reduces economic and political interventions typical in centralized markets. This decentralized approach allows market players to consider multiple factors, such as energy price, participant reputation, and preferences for renewable energy sources [182].

The objective functions, advantages, and operational challenges for asynchronous pricing models are summarized in Table 16 and Table 17.

Table 15. Advantages, disadvantages, and challenges of synchronous pricing mechanisms in energy markets.

Type	Advantages	Disadvantages	Challenges	Refs.
Uniform Pricing	Simple, transparent, easy to implement.	Does not reflect grid constraints, inefficiencies in large markets.	May lead to price distortion, congestion handling issues.	[18,183,184,185,186]
Discriminatory Pricing	Aligns payment with bidder’s price, promotes competition.	Can lead to inefficiency and mark-ups, potential for strategic bidding.	Complexity in matching bids and price setting, potential for unfair market behavior.	[187,188,189,190,191]
Locational Marginal Pricing (LMP)	Reflects true cost of energy at different locations, efficient, encourages investment in grid capacity.	Computationally complex, exposes participants to locational price risk.	Requires accurate modeling of grid constraints, difficult for dynamic systems with high renewable penetration.	[192,193,194,195,196,197,198,199,200]
SDR-based Pricing	Simple to implement, encourages participation, promotes fairness.	Lack of grid constraint consideration, incentivization issues.	Requires accurate supply and demand data, challenges in fluctuating renewable energy markets.	[201,202,203,204,205,206,207]

Table 15. Advantages, disadvantages, and challenges of synchronous pricing mechanisms in energy markets.

Type	Advantages	Disadvantages	Challenges	Refs.
Uniform Pricing	Simple, transparent, easy to implement.	Does not reflect grid constraints, inefficiencies in large markets.	May lead to price distortion, congestion handling issues.	[18,183,184,185,186]
Discriminatory Pricing	Aligns payment with bidder’s price, promotes competition.	Can lead to inefficiency and mark-ups, potential for strategic bidding.	Complexity in matching bids and price setting, potential for unfair market behavior.	[187,188,189,190,191]
Locational Marginal Pricing (LMP)	Reflects true cost of energy at different locations, efficient, encourages investment in grid capacity.	Computationally complex, exposes participants to locational price risk.	Requires accurate modeling of grid constraints, difficult for dynamic systems with high renewable penetration.	[192,193,194,195,196,197,198,199,200]
SDR-based Pricing	Simple to implement, encourages participation, promotes fairness.	Lack of grid constraint consideration, incentivization issues.	Requires accurate supply and demand data, challenges in fluctuating renewable energy markets.	[201,202,203,204,205,206,207]

Table 16. Advantages, disadvantages, and challenges for asynchronous pricing mechanisms in energy markets.

Pricing Type	Advantages	Disadvantages	Challenges	Refs.
Pay-as-Bid/Vickrey Pricing	Predictable for sellers; transparency in price determination.	Can lead to market inefficiency; high risk for sellers if bids are too low.	Manipulation of bids; price uncertainty.	[208,209,210]
Bilateral Negotiation Mechanisms	Flexibility in terms; enables tailored agreements.	Lack of transparency; can lead to inefficiency if not well-negotiated.	Lack of market transparency; disagreement on terms.	[211,212,213,214,215]
Reserved Pricing	Provides a minimum price guarantee; protects sellers from low market prices.	Can prevent market clearing; prices may be set too high, reducing competition.	Determining an optimal reserve price; potential market inefficiency.	[216,217,218,219,220]
Forward and Futures Contracts	Price certainty for future transactions; mitigates price risk.	Not suitable for all markets; limited flexibility for customization.	Determining an appropriate contract price; market volatility risks.	[221]

Table 16. Advantages, disadvantages, and challenges for asynchronous pricing mechanisms in energy markets.

Pricing Type	Advantages	Disadvantages	Challenges	Refs.
Pay-as-Bid/Vickrey Pricing	Predictable for sellers; transparency in price determination.	Can lead to market inefficiency; high risk for sellers if bids are too low.	Manipulation of bids; price uncertainty.	[208,209,210]
Bilateral Negotiation Mechanisms	Flexibility in terms; enables tailored agreements.	Lack of transparency; can lead to inefficiency if not well-negotiated.	Lack of market transparency; disagreement on terms.	[211,212,213,214,215]
Reserved Pricing	Provides a minimum price guarantee; protects sellers from low market prices.	Can prevent market clearing; prices may be set too high, reducing competition.	Determining an optimal reserve price; potential market inefficiency.	[216,217,218,219,220]
Forward and Futures Contracts	Price certainty for future transactions; mitigates price risk.	Not suitable for all markets; limited flexibility for customization.	Determining an appropriate contract price; market volatility risks.	[221]

Table 17. Objective functions for asynchronous pricing mechanisms.

Pricing Type	Objective Function (Mathematical Formulation)
Pay-as-Bid/Vickrey Pricing	Pay-as-Bid: Minimize total market cost, where each seller receives the price they bid. $$\mathrm{Objective}:\sum _{i=1}^n{P}_i{B}_i$$ where $P_i$ is the price bidder i bids and $B_i$ is the quantity sold by bidder i. Vickrey Pricing (Uniform Pricing): Maximize social welfare by finding the market-clearing price $P^{*}$. $$\mathrm{Objective}:\underset{P}{max}\sum _{i=1}^n(P-C_i)Q_i$$ where $C_i$ is the cost of seller i, $Q_i$ is the quantity, and P is the uniform market price.
Bilateral Negotiation Mechanisms	Maximize the utility of the buyer and seller based on the negotiated price $P^{*}$. $$\mathrm{Objective}:\underset{P^{*}}{max}\left(U_{\mathrm{seller}}\left(P^{*}\right)+U_{\mathrm{buyer}}\left(P^{*}\right)\right)$$ where $U_{\mathrm{seller}}\left(P^{*}\right)$ and $U_{\mathrm{buyer}}\left(P^{*}\right)$ are the utility functions of the seller and buyer, respectively, which depend on the negotiated price $P^{*}$.
Reserved Pricing	Maximize seller’s revenue subject to the reservation price $P_{min}$ being met. $$\mathrm{Objective}:\underset{P\ge P_{min}}{max}P·Q$$ where P is the market price, Q is the quantity, and $P_{min}$ is the minimum price set by the seller.
Forward and Futures Contracts	Maximize the expected profit from a future transaction based on the agreed price $P_f$. $$\mathrm{Objective}:\underset{P_f}{max}E\left[(P_f-C)·Q\right]$$ where $P_f$ is the agreed future price, C is the cost of production, and Q is the quantity to be traded. For futures, the expected value E accounts for uncertainty in market conditions.

Table 17. Objective functions for asynchronous pricing mechanisms.

Pricing Type	Objective Function (Mathematical Formulation)
Pay-as-Bid/Vickrey Pricing	Pay-as-Bid: Minimize total market cost, where each seller receives the price they bid. $$\mathrm{Objective}:\sum _{i=1}^n{P}_i{B}_i$$ where $P_i$ is the price bidder i bids and $B_i$ is the quantity sold by bidder i. Vickrey Pricing (Uniform Pricing): Maximize social welfare by finding the market-clearing price $P^{*}$. $$\mathrm{Objective}:\underset{P}{max}\sum _{i=1}^n(P-C_i)Q_i$$ where $C_i$ is the cost of seller i, $Q_i$ is the quantity, and P is the uniform market price.
Bilateral Negotiation Mechanisms	Maximize the utility of the buyer and seller based on the negotiated price $P^{*}$. $$\mathrm{Objective}:\underset{P^{*}}{max}\left(U_{\mathrm{seller}}\left(P^{*}\right)+U_{\mathrm{buyer}}\left(P^{*}\right)\right)$$ where $U_{\mathrm{seller}}\left(P^{*}\right)$ and $U_{\mathrm{buyer}}\left(P^{*}\right)$ are the utility functions of the seller and buyer, respectively, which depend on the negotiated price $P^{*}$.
Reserved Pricing	Maximize seller’s revenue subject to the reservation price $P_{min}$ being met. $$\mathrm{Objective}:\underset{P\ge P_{min}}{max}P·Q$$ where P is the market price, Q is the quantity, and $P_{min}$ is the minimum price set by the seller.
Forward and Futures Contracts	Maximize the expected profit from a future transaction based on the agreed price $P_f$. $$\mathrm{Objective}:\underset{P_f}{max}E\left[(P_f-C)·Q\right]$$ where $P_f$ is the agreed future price, C is the cost of production, and Q is the quantity to be traded. For futures, the expected value E accounts for uncertainty in market conditions.

3.1. Uniform Pricing Mechanism

The uniform pricing rule is a synchronous market-clearing mechanism in which all matched trades—whether a prosumer’s sale or a consumer’s purchase—are settled at a common price. Typically, this price is determined by aggregating all buy and sell bids in the form of supply and demand curves. The intersection of these curves establishes the clearing price for all transactions, regardless of individual bid values submitted by participants [222,223].

Several researchers have developed models to identify uniform pricing in peer-to-peer (P2P) energy trading markets. For instance, optimization-based models aim to maximize objectives such as social welfare [224,225]. Alternatively, the price may be set through stakeholder agreement. Uniform pricing contributes to reducing voltage imbalances within the distribution network and promoting supply–demand equilibrium [226,227,228].

The market-surplus-maximizing energy price is generally defined as the highest bid at the intersection of the supply and demand curves [229]. In practice, however, market clearing often uses the second-highest bid instead of the absolute highest bid to promote fairness [229,230,231]. Studies such as [232,233] examine uniform pricing in P2P markets, demonstrating that a common clearing price can ensure fairness, even under grid constraints or in 100% renewable electricity markets.

Fair competition among sellers determines the single clearing price, typically selected via a heuristic search over submitted bids to maximize revenue or minimize costs [223,234]. While uniform pricing enhances transparency and provides a common price for all participants, it is susceptible to strategic bid shading, where bidders may inflate offers above their true willingness to pay [235].

3.2. Discriminatory Pricing Mechanism

The discriminatory pricing system settles each approved transaction at the specific price bid by the respective participant. In this mechanism, sellers post their quantity–price combinations, and bids are accepted until demand is fulfilled. Successful sellers are then paid their individual bid price rather than a single uniform price, and the traded quantity corresponds to the minimum of the bid and ask quantities [236,237].

Discriminatory pricing has also been employed in renewable energy markets to provide more accurate price signals and optimize distribution performance. For instance, Ref. [238] demonstrates that discriminatory pricing can enhance the integration of renewable energy sources while maintaining fairness and effectiveness. The mechanism better reflects actual energy costs by allowing prices to vary according to local generation capacity and fluctuations in demand [239].

Pricing strategies under discriminatory pricing are typically aligned with specific market objectives. For example, Ref. [240] links matching rules to pricing approaches that can achieve the Nash equilibrium in simultaneous games [241]. In [242], a distributed approach is proposed to maximize overall energy-user benefit, creating an envy-free and Pareto-optimal outcome. Similarly, a two-stage Stackelberg game model [243] allows small distributed energy sources to trade at higher payments by targeting users willing to pay more, thereby reducing controller costs while maximizing user utility. Overall, discriminatory pricing improves market efficiency by emphasizing price flexibility, while necessitating careful pairing and matching of trading partners.

3.3. Locational Marginal Pricing Mechanism

Building on discriminatory pricing, Locational Marginal Pricing (LMP) incorporates the physical characteristics of the energy network. Unlike standard discriminatory pricing, which relies solely on bid prices, LMP determines a price at each grid node that accounts not only for the energy cost but also for transmission congestion and line losses. This provides a more accurate and location-specific pricing signal, facilitating cost-effective energy delivery across large, complex networks [244].

LMP calculation typically involves three main components:

• The marginal cost of electricity generation at a specific power plant [245].
• Costs associated with failing to clear electricity flows in the transmission network, requiring additional dispatch of generation resources [246].
• Energy losses due to resistance in transmission lines [247].

Research by [248] approximates LMPs using AC and DC Optimal Power Flow (OPF) models, providing a detailed decomposition to understand the pricing mechanism in power markets. The study in [200] examines implementation challenges for LMP in transitioning power systems, emphasizing complexities introduced by decarbonization and heterogeneous participants. Other works, such as [197], highlight LMP’s role in grid optimization and transparent energy pricing, taking into account generation costs, delivery, and transmission constraints at different network locations.

3.4. SDR-Based Pricing Mechanism

Building on discriminatory and Locational Marginal Pricing, the SDR (Supply and Demand Ratio)-based method provides an alternative where the ratio of available energy supply to demand directly influences energy prices. Unlike LMP, which considers grid physical characteristics, SDR focuses on real-time supply versus demand, making it suitable for small or decentralized energy markets [207,249].

The SDR is calculated as the total energy provided divided by the total energy required. This ratio, combined with bidding information, is used to determine energy prices. Studies such as [250] have shown SDR to be effective in renewable energy markets with fluctuating supply. Community-based energy trading scenarios also benefit from SDR, as it promotes fairness and encourages market participation [225,251]. Price determination may involve a convex combination of local purchasing and grid-selling prices [252,253].

3.5. Pay-as-Bid (PAB) and Vickrey Pricing

Pay-as-bid (PAB) and Vickrey auctions are competitive bidding mechanisms for energy markets.

In a PAB auction, participants submit bids indicating their desired selling or buying price. The market clears by accepting bids starting from the most competitive until total demand is met. Each successful seller receives their submitted price, incentivizing competitive bidding to maximize returns [254,255,256,257,258].

Vickrey pricing modifies PAB by having the winning bidder pay the second-highest bid rather than their own. This promotes truth-telling, as bidders reveal their true valuation without overbidding or underbidding, enhancing efficiency [259,260,261,262]. Recent research explores repeated and multi-unit PAB auctions [263] and adaptive bidding strategies [264,265], optimizing utility and market performance.

3.6. Reserved Pricing Mechanism

Reserved pricing sets the minimum acceptable price for energy in an auction. Bids below this threshold are rejected, protecting sellers from losses and ensuring that auction outcomes reflect real market value [266,267].

Reserved pricing is commonly applied in pay-as-bid or Vickrey auctions, influencing bidding behavior and maintaining market stability, especially with high renewable penetration or variable demand [206,268,269]. Research has examined its effect on operating reserves, scarcity pricing, and market incentives [270,271,272,273,274,275].

3.7. Forward and Futures Contracts

Forward and futures contracts are asynchronous mechanisms where prices are agreed upon today for delivery at a future date.

A forward contract is privately negotiated and customizable between two parties, while a futures contract is standardized and traded on an exchange [276,277,278,279,280,281,282]. These contracts allow participants to hedge against price volatility or speculate on future prices.

Forward/futures contracts provide price stability, particularly important in markets with renewable generation or fluctuating supply–demand conditions [221,283,284,285,286]. They also inform price discovery and market decision-making while enabling small-scale generators to secure predictable revenue streams.

3.8. Game-Theoretic Techniques

Game theory is a branch of mathematics that studies strategic interactions among rational decision-makers, where the payoff of each participant depends on the actions of others [19]. In energy markets, game-theoretic methods are used to model interactions among market players—such as buyers, sellers, operators, and regulators—to establish optimal pricing, bidding strategies, and contract negotiations.

Nash equilibrium is a fundamental concept in game theory [19,287], representing a set of outcomes where no player can improve their payoff by unilaterally changing their strategy. In energy markets, Nash equilibrium helps predict participants’ behavior in competitive pricing environments, particularly during auctions.

Several studies have applied Nash equilibrium to optimize energy market prices:

• A Nash Stackelberg game model [288] was proposed to study strategic interactions between distributed energy resource (DER) aggregators and electricity retailers. This hierarchical model captures profit-maximizing behaviors of DER aggregators while accounting for retailers’ pricing strategies, providing insights into optimal bidding strategies in decentralized systems.
• A hybrid approach combining Mixed-Integer Linear Programming (MILP) with game theory [289] has been used to represent P2P energy market settlements.
• Nash equilibrium has been used to identify optimal trading strategies among prosumers in decentralized energy markets [290] and to maximize bids in bilateral electricity markets [291].
• Game-theoretic bargaining solutions have been applied to determine prices for Tradable Green Certificates (TGCs) and other auctioned energy products [292].
• Learning-based approaches, such as the no-regret algorithm, have been integrated with Nash equilibrium to model supplier bidding strategies in forward electricity markets [293].

Cooperative vs. Non-Cooperative Games [294] are also widely applied:

• Cooperative game theory analyzes coalition formation among prosumers in P2P networks and equitable profit distribution using concepts like the Shapley value or the core. For example, online coalitional games have been proposed to compute payoffs in real-time P2P markets [295], ensuring fairness and scalability.
• Distributed negotiation mechanisms facilitate stable bilateral contracts within coalitions, improving participation and user satisfaction [296].
• Hedonic games incorporate social preferences and community relationships, enabling coalition formation that maximizes total energy exchange [297].

Stackelberg models are used in non-cooperative scenarios, where grid operators or aggregators act as leaders, and DER prosumers are followers. These models optimize network costs and prosumer utilities simultaneously [298]. Hybrid models combining cooperative and Stackelberg frameworks balance self-interest with group stability [294,299].

Game theory provides diverse tools for modeling pricing in electricity markets, accommodating both market-like structures and P2P energy trading.

Table 18. Applications of game theory in energy pricing.

Application Area	Technique Used	Description	Refs.
Auction Design (Pay-as-Bid, Vickrey–Clarke–Groves)	Nash Equilibrium/Auction Theory	Game theory models are used to design and analyze auction mechanisms like pay-as-bid and Vickrey auctions to determine optimal bidding strategies in competitive electricity markets.	[208,300]
Bilateral Electricity Markets	Nash Equilibrium	Nash equilibrium models are used to study bidding strategies in bilateral electricity markets, where players (sellers and buyers) interact to optimize their payoffs.	[301,302]
Peer-to-Peer (P2P) Energy Trading	Non-Cooperative Game Theory	Game theory is applied to model the strategic interactions of prosumers and consumers in P2P energy trading systems, optimizing pricing and transaction behavior.	[241,303,304]
Distributed Energy Resource (DER) Aggregation	Nash–Stack-elberg Game	Nash–Stackelberg game theory models are applied to optimize pricing and bidding strategies in DER aggregation by modeling the interactions between aggregators and retailers.	[305,306,307]
Tradable Green Certificates (TGCs)	Nash Bargaining Theory	Nash bargaining models are used for price determination and profit-sharing in energy markets, ensuring mutual benefits between stakeholders.	[308,309,310]
Forward and Futures Markets	Nash Equilibrium + No-Regret Learning	Nash equilibrium and no-regret learning algorithms are used to determine optimal bidding strategies in forward and futures electricity markets, considering market volatility.	[293,311]

summarizes recent applications of game-theoretic methods in energy price determination.

3.9. Optimization Techniques

Optimization methods are widely used for efficient and fair cost determination in energy markets, particularly in market-based electricity systems. These methods are crucial for forecasting prices and improving market mechanisms by considering multiple constraints, including supply–demand balance, transmission limits, and cost minimization. This section reviews recent applications of optimization in energy pricing.

Linear Programming (LP) is employed in centralized market clearing to determine optimal generation dispatch, transmission allocation, and price setting. The objective is typically to minimize total generation costs while satisfying demand and operational constraints. LP models are attractive for their simplicity and computational efficiency. For example, Ref. [312] proposes an LP model that accounts for generation and transmission line constraints to clear the electricity market in advance, minimizing costs and maximizing reliability. Similarly, Ref. [313] applied LP for renewable energy integration, optimizing market prices while ensuring that solar and wind generation meets demand.

Mixed-Integer Linear Programming (MILP) extends LP by incorporating both continuous and integer variables, enabling modeling of discrete decisions such as unit commitment, generation scheduling, and bidding strategies. MILP is particularly useful for markets requiring binary choices, e.g., turning generators on/off [314]. Studies like [315,316] use MILP to minimize operational costs while respecting environmental limits, ensuring generation flexibility, and optimizing contracts between market participants. MILP has also been applied in auction-based electricity markets [316,317,318] to optimize bidding strategies and long-term investment planning.

Convex Optimization has gained popularity in price forecasting and market equilibrium analysis due to its guarantee of global optimality for convex problems. These techniques are used to maximize social welfare or profit while minimizing costs under operational constraints. For instance, Ref. [319] uses convex optimization to forecast electricity prices under renewable energy uncertainty, whereas [320] incorporates demand response and DER integration to enhance price stability.

Stochastic Optimization is essential in markets with uncertain future conditions, such as real-time pricing, futures, and renewable integration. By modeling probabilistic variations in demand, generation, and market factors, stochastic optimization improves pricing under uncertainty. Applications include real-time pricing considering wind and solar forecasting errors [321] and modeling electricity price fluctuations under water supply–demand uncertainty and financial risk [322].

Metaheuristic Optimization Techniques such as Genetic algorithms (GAs), Particle Swarm Optimization (PSO), and Simulated Annealing (SA) are used to solve complex, non-linear problems in bidding strategies and market behavior [312]. These methods are particularly useful for dynamic, real-time price adjustments. For example, Ref. [323] applies GA to optimize bidding strategies considering competitor actions, Ref. [324] uses PSO for real-time auction pricing, and [325] employs SA for calculating real-time P2P energy prices.

Table 19. Applications of optimization techniques in energy pricing.

Objective	Optimization Technique	Description	Refs.
Market Clearing	Linear Programming (LP)	Linear Programming (LP) is widely used for optimizing the market clearing process, minimizing costs while satisfying supply and demand constraints.	[313,326,327,328]
Bidding Strategies in Electricity Auctions	Mixed-Integer Linear Programming (MILP)	MILP is applied to model bidding strategies and generation scheduling, helping market participants determine optimal strategies for both price and quantity.	[329,330]
Price Forecasting in Spot Markets	Convex Optimization	Convex optimization is used for forecasting prices in spot markets, where prices are determined based on supply–demand dynamics and operational constraints.	[319,331]
Real-time Pricing Adjustment	Stochastic Optimization	Stochastic optimization is applied in real-time price adjustment by modeling uncertainty in market conditions and helping determine optimal pricing strategies under uncertainty.	[321,332]
Generation Scheduling and Profit Maximization	Mixed-Integer Non-Linear Programming (MINLP)	MINLP models are applied to optimize generation scheduling while maximizing profit, considering operational and environmental constraints such as emission limits.	[333,334]
Energy Storage Optimization	Particle Swarm Optimization (PSO)	PSO is used to optimize the operation of energy storage systems by minimizing cost and maximizing energy delivery efficiency based on demand forecasts and storage capacity.	[335,336,337]
Bidding in Renewable Energy Markets	Genetic Algorithms (GAs)	Genetic algorithms (GAs) are used to model and optimize bidding strategies in renewable energy markets, considering factors like energy production and market prices.	[338,339]

summarizes key studies that use optimization techniques for energy price determination.

3.10. Numerical Method-Based Techniques

Numerical methods play a critical role in energy pricing, particularly for complex, non-linear problems where analytical solutions are not feasible. They are widely applied in modeling market behavior, handling uncertainty, stochastic processes, price prediction, real-time pricing, and renewable energy integration.

Monte Carlo simulation (MCS) is extensively used to model uncertainty in pricing mechanisms, especially in stochastic environments such as forward contracts, futures markets, and renewable integration. It enables analysis of price variations and forecasts likely outcomes based on random variables. For instance, Ref. [340] investigates the impact of solar energy uncertainty on real-time electricity pricing using Monte Carlo simulation, while [341] uses it to project future prices in electricity futures markets.

Partial Differential Equations (PDEs) arising in price evolution and energy derivatives are often solved using Finite Difference Methods (FDMs). These methods approximate PDEs to simulate dynamic energy pricing, where prices depend on multiple factors such as demand, supply, and external conditions. For example, Ref. [22] models electricity price fluctuations in forward contracts using FDM, enhancing time-dependent price prediction accuracy. Other studies [342,343] apply FDM for non-linear pricing problems under market uncertainties.

Iterative Techniques are used for solving non-linear optimization problems and market-clearing algorithms, especially in decentralized energy markets. Iterative algorithms assist in balancing supply and demand to achieve market equilibrium [344,345].

The Newton–Raphson Method is an iterative approach for solving non-linear equations, often applied in real-time market clearing under complex price relationships [346,347]. Similarly, gradient descent is applied to optimize bidding and pricing strategies iteratively, particularly under competitive market conditions [348,349]. Stochastic gradient descent has been used for price modeling under renewable energy uncertainty and volatile markets [350,351].

3.11. AI-Based Techniques

AI-based approaches, including machine learning (ML), reinforcement learning (RL), and deep learning (DL), are increasingly applied to dynamic and complex pricing problems in energy markets, especially with the integration of renewable energy sources and smart grids.

Machine learning (ML) techniques predict energy prices using historical data, weather conditions, and market factors. Supervised learning models such as regression and ensemble methods (e.g., random forests) are widely used for real-time price forecasting [352,353]. Support Vector Machines (SVMs) have been employed for electricity futures pricing under renewable generation uncertainty [354,355].

Reinforcement learning (RL) is suitable for dynamic pricing, where agents learn optimal pricing strategies through interaction with market conditions. Q-learning has been applied to real-time pricing in smart grids [356,357], while Deep Reinforcement Learning (DRL) models optimize bidding strategies considering market states and competition [358,359].

Deep learning (DL) models, including neural networks (NN), LSTM, and CNN, are effective for complex pattern recognition in price forecasting. LSTM models capture long-range dependencies in electricity price data [360,361], while CNNs characterize price variability under uncertain renewable generation [362].

Hybrid AI Models integrate multiple techniques (e.g., DL, RL, ML) to optimize pricing comprehensively. Examples include CNN-LSTM models for spot price prediction [363], DRL-DL hybrids for dynamic smart grid pricing [364], and DNN-GA hybrids for P2P energy market bidding [365]. These hybrid approaches outperform conventional models in forecasting accuracy and market decision-making.

Table 20. Advantages and disadvantages of the techniques used for price determination.

Technique	Advantages	Disadvantages	Challenges
Game Theory Techniques	Helps model strategic interactions between market participants. Provides insights into optimal bidding strategies. Can handle competitive pricing environments.	Assumes rational behavior, which may not always hold in real-world markets. Models can become complex for large-scale systems. Computational complexity for large datasets.	Uncertainty in behavior modeling. Equilibrium calculation in large markets is difficult. Handling non-cooperative behaviors in dynamic markets.
Optimization Techniques	Offers efficient solutions for complex problems (e.g., market clearing, generation scheduling). Can optimize for cost minimization and profit maximization. Can incorporate uncertainty and constraints effectively.	Limited flexibility in some non-linear or dynamic pricing models. Requires accurate data inputs, which may be difficult to obtain in real-world markets. Scalability issues in larger markets.	Data accuracy issues in forecasting. Requires computational resources for large-scale problems. Difficulty in modeling renewable energy variability in optimization problems.
Numerical-Based Techniques	Can handle uncertainty and complexity in pricing models. Useful for real-time price determination. Can simulate various market scenarios and price behaviors.	May not provide global optimal solutions for non-linear problems. Computationally expensive for large-scale models. Sensitivity to initial conditions can affect results.	High computational costs for real-time price simulation. Difficulty in calibrating models under dynamic market conditions. Handling large data in Monte Carlo simulations or FDM.
AI-Based Techniques	Adaptability to dynamic market conditions. Improved accuracy in price forecasting and real-time pricing. Can handle large datasets and complex patterns.	Requires large amounts of training data. Models can become black-boxes, making it difficult to interpret results. Overfitting can be an issue in some models.	Data quality and availability. Model interpretability can be a concern for decision-makers. Training time for large-scale models and adaptation to market dynamics.

summarizes the advantages, disadvantages, and challenges of the numerical and AI-based pricing techniques discussed.

4. Demand–Supply Balance Program (DSBP)

The integration of distributed energy resources (DERs) into smart grids is facilitated by effective pricing signals and energy conversion technologies. However, DER generation is inherently variable due to changing weather conditions, directly affecting generation capacity. This variability often causes a mismatch between supply and demand in the smart grid.

During periods of favorable weather, DERs, such as photovoltaic (PV) systems, operate near maximum capacity, potentially producing excess supply relative to demand. Conversely, during evenings or cloudy conditions, DER output declines, leading to a supply deficit relative to the energy demands of buyers.

To maintain demand–supply balance, the smart grid often relies on energy imports from or exports to the utility grid, which primarily consists of fossil-fuel-based generation. Such interactions not only increase CO₂ emissions but also impose operational stress on the grid and participants. Importing energy from the utility grid is costlier than P2P energy, reducing potential savings for buyers, while exporting excess energy may reduce profitability for DER sellers.

The Demand–Supply Balance Program (DSBP) addresses these challenges by dynamically balancing generation and consumption while optimizing energy flows and prices. The DSBP can be formulated using various methodologies, including iterative methods, optimization models, game-theoretic frameworks, and AI-based approaches, as illustrated in Figure 5.

Each formulation approach in Figure 5 has unique merits and operational challenges.

Table 21. Comparative analysis of several models used to design DSBP.

Model Type	Main Principle	Typical Use Scenarios	Strengths	Weaknesses	Operational Challenges	Scalability	Data Requirements	Real-Time Capability
Iterative Model	Progressive updates until convergence; decentralized interactions	Decentralized coordination, iterative pricing, DSM	Simple implementation; distributed; scalable	Slow convergence; sensitive to delays; suboptimal solutions	Communication reliability; synchronization	High	Low to Moderate	Moderate
Optimization Model	Mathematical programming (LP/MILP/heuristics)	Cost minimization, load balancing, scheduling	Achieves optimality; constraint flexibility; performance control	Computationally intensive; centralized; complex modeling	Data availability; solving high-dimensional problems	Moderate	High	Challenging
Game Theoretic Model	Strategic decision-making among multiple agents (users, suppliers)	Pricing, market design, incentive mechanism, auctions	Captures strategic intelligence; equilibrium concepts	Equilibrium may not be unique; complexity; data dependency	Ensuring incentive compatibility; privacy; convergence	Moderate	Moderate to High	Moderate
AI-Based Model	Learning from data (ML, deep learning, RL, expert systems, etc.)	Forecasting, anomaly detection, demand prediction	Adapts to uncertainty; predictive; pattern recognition	Needs training data; transparency issues; bias risks	Integration with legacy systems; computation; security risks	High	High	High

provides a comparative overview of these models in terms of methodology, application, advantages, and limitations.

4.1. Demand–Supply Balance Program Solving Models

Several researchers have proposed advanced methods to address the challenges of DSBP. This section explores notable approaches, highlighting their objectives and strengths.

4.1.1. Iterative Models

Iterative models rely on numerical formulations of the DSBP, updating system values repeatedly until a pre-defined convergence threshold is met. These models are widely used in smart grid management due to their simplicity and adaptability.

For instance, Ref. [20] introduced a Mixed-Integer Linear Programming (MILP)-based iterative model to manage real-time energy in an IEEE-33 AC-DC distribution network. Similarly, Ref. [366] proposed an iterative MILP model to optimize household energy consumption while respecting the operative supply envelope.

Another iterative approach using an inclined block tariff with real-time varying prices was developed in [367], enabling the adjustment of home appliances according to real-time energy costs. A demand-side management iterative sizing method was proposed in [368] to improve system efficiency.

Iterative weighting strategies have also been employed. In [369], different weights were assigned to household appliances, and appliance operating cycles were adjusted based on the Demand–Supply Ratio. Another iterative model [370] maintained the balance between requested demand and supply, achieving an average peak-hour load reduction of 52.8%, which is approximately 26.1% more efficient than comparable iterative methods. Additionally, this model demonstrated a 33.9% profit increase for aggregators by dynamically adjusting the demand–supply balance.

Table 22. Summary of selected studies on DSBP based on iterative methods.

Ref.	Model Focus	Data Used	Findings	Strength	Limitation
[371]	Demand peak reduction via incentives	Smart grid agent load data	Novel incentive algorithm learning user response iteratively under budget constraints	Realistic agent modeling, budget aware	Simulator-based results; limited real-field validation
[372]	Demand–supply balancing in power	Generator and consumer data	Stackelberg equilibrium algorithm balancing supply and demand interactions	Strong theoretical foundation	Heavy assumptions on rationality and market structure
[373]	Integration of smart buildings	Building and grid integrated data	Proposes iterative optimization for building-grid demand coordination	Building-grid integration focus	Scale limitations from building to entire grid
[374]	Market clearing and price setting	Electricity market bid data	Fast converging algorithm robust to stochastic bids	Efficient market clearing with robustness	Focus on market, less on grid operation
[375]	Smart Grid Energy Management	Machine learning iterative algorithms	Demand forecasting, failure detection	Smart meters and sensors	Integrates ML for iterative energy load balancing and anomaly detection
[376]	Multi-level supply–demand matching	Cost/performance measures	Reduces mismatch via multi-level iterative solution	Hierarchical structure improves scalability	Solution complexity not fully addressed

presents a comparative analysis of several iterative DSBP models, summarizing their methodologies, objectives, and performance outcomes in maintaining demand–supply equilibrium in smart grids.

Although iterative models are scalable and easy to deploy in smart grids and perform effectively under pre-defined threshold conditions, the solutions they provide do not capture the strategic interactions among participants in DSBP programs.

Strategic interactions are crucial in DSBP design, as participants select their consumption based on private information, such as financial constraints, comfort preferences, and other individual factors. These interactions can be effectively modeled and analyzed using game-theoretic approaches.

4.1.2. Game-Theoretic Models

Game-theoretic models are an effective tool to capture and analyze the strategic behavior of participants by formulating their interactions as cooperative, non-cooperative, or evolutionary games.

In cooperative game models, participants form coalitions to determine optimal demand and supply values collectively. The achieved outcomes are then distributed among participants using allocation schemes such as Shapley values or other fair division models.

In non-cooperative game models, each participant independently selects their strategy, and outcomes are not shared with others. These models allow the analysis of individual optimization behavior in decentralized energy markets.

Several studies have utilized game-theoretic approaches to identify optimal strategies in DSBP programs.

Table 23. Game theory applications in energy pricing and management.

Ref.	Game Type	Objectives	Strengths
[22]	Nash Equilibrium	Optimize energy consumption and pricing strategies between consumers and utilities; balance consumer cost savings with utility revenue while promoting green energy use	Uses smart meter data for informed, strategic decision-making; achieves a mutually beneficial equilibrium where both consumers and utilities maximize payoffs
[377]	Non-cooperative Game	Analyse interactions between distributed generation (DG) units and autonomous demand-response programs in smart distribution grids; reduce total operational costs and power losses	Enhances overall grid performance in terms of power quality and stability
[378]	Nash Bargaining Model	Achieve fair benefit allocation between trading peers while maximizing social welfare	Balances fairness and efficiency, ensuring stable cooperation
[379]	Nash Equilibrium	Optimize energy consumption and pricing strategies between consumers and utilities using green energy sources	Balances consumer and utility interests; reduces costs; promotes green energy adoption
[380]	Nash Equilibrium	Demand side management in smart grids to optimize energy use and costs, maximizing utility and consumer payoffs	Enhances energy efficiency; reduces environmental pollution through green energy use
[381]	Non-cooperative Game	Manage smart distribution grids with distributed generation and autonomous demand response; analyze effects compared to centralized control	Reduces total costs and power losses; improves reactive power support, voltage profiles, and load profile flattening
[382]	Noncooperative Game	Encourage residential consumers to adjust electricity use via dynamic pricing; minimize costs while maintaining comfort	Significant cost savings without sacrificing comfort; optimized appliance scheduling using NSGA-II
[383]	Stackelberg Game	Optimize V2G pricing for aggregators and EV users, balancing benefits while considering charging costs and inconvenience	Models multi-entity interactions with realistic factors; validated through multi-aggregator EV simulations
[229]	Double Auction-Based Stackelberg Game	Enable P2P energy trading among prosumers to maximize participant profits, ensure social welfare, and maintain privacy	Achieves incentive compatibility and individual rationality; supports real-time trading via blockchain; effective under various scenarios
[384]	Concave N-Person Game	Optimize global power consumption scheduling of TCL users while considering individual preferences and REN generation forecasts	Adaptive and flexible pricing mechanism; fast solution via simplified model; smooths tie-line power in microgrids

provides a comparative summary of notable research using game-theoretic models for demand–supply balancing in smart grids.

4.1.3. Optimization Models

Although iterative models are highly scalable and game-theoretic models effectively capture participant interactions in DSBP design, both approaches often overlook critical system constraints, such as network operating capacity, available generation, and supply-to-demand ratios. To address these challenges, optimization models have been widely employed to design effective Demand–Supply Balance Programs in smart grids.

For instance, Zhao et al. [21] proposed a multi-objective optimization model to balance demand and supply while considering both economic and environmental impacts. This model utilized the AUGMECON2 optimization algorithm to obtain optimized solutions under multiple constraints.

An Ant Colony Optimization (ACO)-based model was introduced in [385] to determine optimized demand and supply values. The model was compared with other optimization approaches, such as Evolutionary Algorithm, Moth-Flame Optimization, and Bacterial Foraging Optimization, with simulation results demonstrating superior performance of the ACO model. Similarly, Nanibabu et al. [386] applied the Butterfly Optimization Algorithm, achieving an 18.4% cost reduction during peak hours through effective demand-side management strategies.

Optimization models have also been applied to control appliance operation and social welfare. For example, Ref. [367] employed Optimal Adaptive Wind-Driven Optimization to schedule appliance usage and balance supply and demand efficiently. Particle Swarm Optimization (PSO)-based models have been proposed in [387,388] to maximize social welfare and formulate the DSBP. Furthermore, a two-layered optimization model was introduced in, where the first layer manages appliance operation cycles, and the second layer optimizes consumption costs while maintaining consumer comfort levels.

These optimization models demonstrate the ability to integrate system constraints, operational flexibility, and economic objectives, making them a robust approach for designing effective DSBPs in smart grids.

A bi-level optimization model to maintain the balance between demand and supply in the smart grid was proposed in [389]. Additionally, a study in [390] introduced a model that considers multi-energy sources along with their planning and operational costs. This model employs a hybrid genetic algorithm (GA) and pattern search algorithm to balance energy generation and supply efficiently in the smart grid.

Beyond these studies, several other works deploying optimization models to maintain demand–supply balance are compared in

Table 24. Optimization techniques in energy pricing and management.

Ref.	Optimization Type	Objectives	Strengths
[391]	Multi-objective optimization with real-time supply–demand balance	Minimize average electricity cost for the VPP operation, maximize renewable energy utilization (solar and wind), maintain grid stability while meeting demand	Manages uncertainty in renewable generation; adjusts supply allocation based on real-time market prices; selects optimal mix of multiple energy sources; reduces average electricity cost by 15%; increases renewable utilization by 20%
[392]	Multi-objective, market-based optimization (LP, MILP, Evolutionary Algorithms)	Reduce peak demand, ease power flow congestion, integrate DERs while maintaining grid stability, maximize economic benefits, improve DR participation and scheduling	Coordinates DER integration through AMI; enhances network and market operations; improves decision-making for DER management; increases flexibility and resilience of the power system
[393]	Market-based bilateral bidding model for pre-listing energy consumption rights trading	Connect medium-/long-term and spot markets, enable multi-day rights transfer, ensure fair bidding, align volume and price, improve resource allocation	Allows flexible listing and withdrawal within price tolerance; encourages engagement from supply and demand; improves market clearing with defined computation methods
[394]	DSM optimization with IoT-enabled monitoring and blockchain	Shift load from peak to off-peak, reduce costs, encourage behavioral change, cut emissions, improve grid efficiency	Uses smart meters and IoT for adaptive monitoring; supports competitive pricing; integrates tech and social change; enables secure transactions via blockchain
[395]	Goal-oriented classification and selection optimization for DR schemes	Improve performance and reliability, enhance decision-making, maximize economic benefits, integrate DR into ancillary services	Provides practical classification for DR plan selection; considers benefits and barriers; supports all stakeholders; aligns DR schemes with smart community concepts
[396]	MILP for optimal scheduling of integrated energy systems (CCHP, carbon capture, DR)	Minimize total operating cost (energy purchase, maintenance, carbon, compensation), enable low-carbon operation, enhance user satisfaction, optimize multi-energy flow	Integrates cooling, heating, and electricity DR with carbon trading; enables flexible load shifting; reduces gas purchase; cuts emissions; lowers costs by 5.9% plus 3.1% with DR
[397]	Distributed MPC with forecasting and P2P negotiation	Incorporate forecasting and peer-to-peer negotiation in distributed MPC for meshed grids	Enables coordinated, predictive control with negotiation; improves adaptability in meshed grids
[398]	Optimal decentralized operation model (P2P-enabled)	Achieve optimal decentralized operation of smart distribution grids with P2P trading	Facilitates decentralized decision-making; improves local autonomy and operational efficiency
[399]	Distribution network-constrained optimization for multi-microgrid P2P trading	Optimize P2P energy trading among multiple MGs considering network constraints	Integrates network constraints into market transactions; enhances feasibility and reliability
[400]	Data-driven distributionally robust co-optimization of P2P trading and network operation	Provide robust collaborative optimization for interconnected MGs considering fairness	Uses distributionally robust optimization and ADMM for decentralized, fair, and resilient decisions
[401]	Distributed consensus-based optimization	Reach agreement on energy prices and quantities via iterative local exchanges	Low communication overhead; convergence to global optimality
[402]	Bi-level optimization model	Upper level sets market prices; lower level optimizes prosumer schedules	Captures hierarchical decision-making between market operator and prosumers
[403]	Robust optimization	Account for uncertainty in renewable generation and demand during scheduling	Ensures feasible solutions under worst-case scenarios

.

4.1.4. AI-Based Models

While optimization models provide highly accurate outputs for the present time horizon, their effectiveness can diminish for future horizons due to the need to analyze large volumes of historical data. As the dataset grows, computational time increases, which directly impacts the scalability of these models.

To overcome these limitations and capture the true dynamics of the system, artificial intelligence (AI) models are increasingly employed in designing DSBP programs. AI methods can learn important patterns and events from historical data, enabling better forecasting and adaptive demand–supply management in smart grids.

Table 25. AI models applied in DSBM and smart grid management.

Ref.	AI Model	Objectives	Strengths
[23]	Artificial Neural Networks (ANNs)	Predict energy consumption patterns for improved demand-response performance	Can capture complex, non-linear relationships in energy data, enabling accurate predictions
[404]	Supervised Machine Learning (SML)	Develop predictive models for load forecasting in demand response	Flexible with various algorithms; performance improves with quality feature engineering
[405]	Hybrid AI Approach (ANN + Optimization)	Optimize demand-response scheduling by combining predictive modeling with decision optimization	Integrates prediction accuracy with operational optimization for better scheduling outcomes
[406]	AI-based (ANN, Supervised ML)	Present an overview of AI methods in demand response, identify research gaps, and propose future study directions	ANNs can capture complex non-linear patterns in energy data; Supervised ML provides flexibility with various algorithms but benefits greatly from effective feature engineering
[407]	Deep Learning (LSTM) and Unsupervised Learning (K-means, Hierarchical Clustering, PCA)	Provide insight into demand forecasting in smart grids and explore deep learning techniques for improving forecasting accuracy	LSTM handles temporal dependencies in load profiles; Unsupervised learning reveals hidden patterns and clusters in consumption data
[408]	Reinforcement Learning	Develop price-driven demand-response management to minimize system cost via optimal pricing for prosumers	Handles incomplete information and outperforms other price-based DRM approaches
[409]	Hybrid GRU–TCN with Attention	Improve short-term power load forecasting accuracy and efficiency; enhance energy information management in smart grids	Incorporates uncertainty modeling with stochastic and Monte Carlo methods for robust forecasting
[410]	AI-based diagnostics and prognostics (Machine Learning)	Address challenges in smart grid management and security; develop a comprehensive framework for optimal grid reliability	Enables real-time monitoring and predictive maintenance for improved grid performance
[24]	AI methods (Machine Learning, ANN)	Provide an overview of AI techniques in energy demand response; identify future research directions in demand-side response	Offers diverse modeling capabilities; adaptable to various DR scenarios
[411]	Model-free DDPG on Hyperledger Fabric	Integrate distributed controllable resources for grid services; optimize DCR allocations and maximize prosumer profits	Handles continuous control without system model; on-chain execution and auditability; coordinates many resources securely

summarizes several studies that applied AI models to determine optimized demand–supply values, highlighting the techniques, objectives, and key results.

4.2. AI-Driven Approaches in Pricing and Balance: Economic and Environmental Perspectives

The integration of artificial intelligence (AI) in smart grid (SG) systems, particularly through Demand–Supply Balance Programs (DSBPs) and pricing models, enhances operational efficiency and supports economic viability by leveraging real-time data from Advanced Metering Infrastructure (AMI) and Internet of Things (IoT) devices [2,12,13]. AI models facilitate demand prediction, supply optimization, and dynamic pricing, which contribute directly to reduced operational costs and improved resource reliability [8]. For instance, AI-driven DSBPs enable market operators to estimate demand–supply ratios using optimization and game-theoretic frameworks augmented by AI [17,18], allowing for uniform pricing or auction mechanisms that minimize energy waste during peak hours. This results in economic benefits such as lower electricity bills for consumers through informed usage decisions and dynamic pricing models that incentivize off-peak consumption, thereby conserving energy and reducing costs [9]. Utilities also benefit from automated billing and real-time monitoring, which reduce human error and enhance revenue management by preventing overages or undersupplies [14].

From a financial viability standpoint, AI deployment in SG pricing and balance programs offsets initial implementation costs, such as those for smart meters (SMs), control centers (CCs), and aggregators [38], through long-term gains in grid efficiency and scalability. By integrating renewable energy sources (RESs), such as solar panels and wind turbines, AI supports bidirectional energy flow and demand-response systems that balance supply and demand, reducing the need for expensive fossil fuel backups and fostering a more sustainable energy market [1]. AI approaches, combined with energy storage systems (ESSs), further amplify economic returns by injecting surplus energy during high-demand periods, mitigating the stochastic nature of RESs, and stabilizing prices [20].

Environmentally, AI implementation in the energy sector promotes sustainability by easing the penetration of RESs into SG architectures, which lowers the carbon footprint of energy production [44]. For example, AI-optimized DSBPs and pricing strategies enable efficient distribution of surplus renewable energy, reducing reliance on non-renewable sources and supporting dual energy flows that align with environmental goals [3]. However, challenges such as the energy demands of AI computations and potential vulnerabilities in IoT-integrated systems [12] must be addressed to ensure that these benefits do not inadvertently increase electronic waste or grid instability. Overall, the reviewed paradigms suggest that AI not only bolsters economic resilience through cost savings and reliability [8] but also advances environmental objectives by promoting cleaner, more balanced energy ecosystems [44], warranting further research into scalable, low-impact AI deployments.

4.3. Energy Storage Systems (ESS)

Although the Demand–Supply Balance Program (DSBP) assists the smart grid in maintaining a balanced state, this balance can be further enhanced by incorporating energy storage systems (ESSs). By utilizing ESSs, energy generated from photovoltaic (PV) systems and wind turbines during periods of high generation or favorable weather conditions can be stored in batteries. This stored energy can then be injected back into the system during peak-hour demand, thereby reducing the gap between requested demand and available renewable energy.

In smart grids, energy can be stored using multiple technologies, as illustrated in Figure 6. Among these, batteries are one of the most widely adopted methods due to their flexibility, scalability, and relatively straightforward integration. Batteries come in various chemical compositions, each with specific advantages and limitations, as also shown in Figure 6.

4.3.1. Lead–Acid Batteries

Lead–acid batteries are among the most cost-effective energy storage solutions, offering a low upfront cost and simple control mechanisms with high surge current capability. Due to these attributes, they are frequently employed in smart grid research to address energy storage challenges.

Table 26. Lead–acid battery types, objectives, and strengths.

Ref.	Lead–Acid Type	Objectives	Strengths
[25]	Lead–acid Battery Model with New Control Scheme	Propose a new control scheme for hybrid energy storage systems; introduce a lead–acid battery model considering DoD and temperature effects	Enhances power filtering and voltage limit adherence; improves performance in poor thermal conditions
[412]	Hybrid Battery Configurations	Investigate hybrid energy storage system performance; analyze charge/discharge cycling of battery configurations	Enables round-trip efficiency measurement; provides insight into DoD impact on efficiency
[413]	Battery Energy Storage System (BESS)	Introduce battery energy storage for emergency power supply; improve reliability of separated power networks during outages	Based on real measurement data; enhances network reliability during main line damage or transmission limitations
[414]	Lead–acid Battery Pack + Supercapacitor Pack (HESS)	Manage hybrid energy storage for photovoltaic grid integration; optimize service level and minimize battery degradation	Balances power delivery between battery and supercapacitor; extends battery lifespan while maintaining optimal performance
[415]	Hybrid Electrochemical Energy Storage System (HEESS)	Summarize recent research progress in HEESS development; stimulate innovative thoughts for HEESS applications	Integrates system configuration, DC/DC converter design, and energy management strategy; supports innovative approaches for performance and lifespan improvements
[416]	Lead–Carbon Battery	Optimize positive plate performance and production process; enhance high-current charging and deep discharge capabilities	Improved positive plate structure and lead alloy selection; better deep discharge performance and tolerance to high-current charging
[417]	Hybrid MTM–ANN Method	Control load variability for real-time demand-side management; optimize reserved ESS capacity	Combines Markov Transition Matrix with ANN for improved real-time maximum demand control
[418]	Energy Storage Technology	Identify suitable energy storage devices for grid support applications; evaluate technical, economic, and environmental impacts	Provides a comprehensive comparative analysis of various storage devices for different stationary applications
[419]	Residential Energy Consumption and BESS Analysis	Analyze electrical power consumption patterns in residential areas; assess economic viability of aggregated and distributed battery energy storage systems	Provides insights into consumption patterns and potential business models for BESS deployment
[27]	Battery ESS and Power-to-Gas	Analyze balance and flows of electrical energy in networks; model operational modes incorporating electricity storage systems	Combines short-term and long-term storage solutions for improved grid flexibility

presents a comparative summary of several studies that utilized lead–acid batteries for energy storage in smart grid applications.

Figure 6. Storage categories and types of batteries.

Figure 6. Storage categories and types of batteries.

However, the main operational challenges of lead–acid batteries include their relatively short lifecycle and degraded performance under adverse weather conditions, which limits their effectiveness in long-term or harsh operating environments.

4.3.2. Nickel-Based Batteries

To address the limitations of lead–acid batteries, nickel-based batteries have been proposed for energy storage in smart grids. Compared to lead–acid batteries, nickel-based batteries offer higher tolerance to extreme weather conditions, enhanced reliability, and longer operational lifetimes, making them a more robust option for backup and renewable energy storage.

Table 27. Nickel-based battery types, objectives, and strengths.

Ref.	Nickel-Based Type	Objectives	Strengths
[420]	Nickel–iron battery (NI)	Provide reliable backup power in a PV/battery hybrid system to compensate for main grid outages	Long lifespan, high durability, suitable for harsh environments, low maintenance needs
[421]	Nickel–Cadmium Battery (NiCd)	Durable industrial battery storage, withstands extreme temperatures	High cycle life (1000–1500+), wide temperature range, reliability
[422]	Membraneless Ni–Fe battery with gel electrolyte	Develop a high-performance Ni–Fe battery without membrane, improving redox kinetics and addressing passivation, hydrogen evolution, and self-discharge	Enhanced ion transport via self-assembled nanostructures; simpler design; reduced cost; improved durability
[423]	Nickel-based selenides	Summarize preparation methods (hydrothermal, thermal solvent, thermal decomposition, heat treatment) and discuss performance optimization paths	Provides a foundation for future research by consolidating methods and highlighting electrode–electrolyte interaction considerations
[424]	Ni-based batteries	Review current trends and classifications of electrochemical storage for smart grids and EVs, focusing on nickel- based technologies	Broad comparison with other battery types considering cost, impact, maintenance, advantages, and protection; supports informed selection for future applications
[425]	Ni-based (general)	Sustainable, scalable battery storage in smart grids	Long cycle life, safety, moderate energy density, scalable for grid and industrial use
[419]	Residential BESS Business Model Analysis	Analyze electrical power consumption patterns in residential areas; assess economic viability of aggregated and distributed battery energy storage systems	Provides insight into consumption behavior for optimized storage deployment; supports development of viable business models for nickel-based BESS integration
[426]	Optimization and Benefit Evaluation in High-RES Smart Grids	Summarize challenges of integrating high renewable energy sources in smart grids; analyze optimization planning and benefit evaluation methods for energy storage	Incorporates diverse decision-making and optimization techniques; enables comprehensive assessment of energy storage benefits and planning under complex grid conditions

provides a comparative overview of several studies that have implemented nickel-based batteries in smart grid applications, highlighting their performance, advantages, and deployment scenarios.

4.3.3. Sodium–Sulfur-Based Batteries

One of the primary drawbacks of nickel-based batteries is their high self-discharge rate, which negatively affects the overall efficiency of the storage system. To overcome this limitation, sodium–sulfur (NaS) batteries have been increasingly proposed for energy storage in smart grids. NaS batteries offer several advantages, including high energy density, long operational life, and deep discharge capabilities, making them well-suited for large-scale storage and renewable energy integration.

Table 28. Sodium–sulfur battery research, objectives, and strengths.

Ref.	Battery Type	Objectives	Strengths
[427]	Room-temperature Sodium–Sulfur Battery	Enhance cycling stability; achieve complete conversion of sodium polysulfides in cathodes	Advanced analysis using DFT adsorption energy calculations and in situ synchrotron XRD for species tracking
[428]	Room-temperature Sodium–Sulfur Battery	Clarify operating principles and technical challenges; propose future strategies for practical RT-Na–S batteries	Strategies include regulating electrolyte components, adding additives, developing new electrolytes, and multifunctional separators to address low conductivity, volume expansion, and Na dendrite formation
[428]	Sodium–Sulfur Battery with Covalent Sulfur Confinement	Enhance sodium–sulfur battery performance by confining covalent sulfur and accelerating polysulfide redox kinetics; improve cycling stability	Uses ex situ XPS and theoretical calculations to study covalent sulfur bond breakage; mitigates sluggish reactivity, polysulfide dissolution, and sulfur’s insulating nature
[429]	Room-Temperature Sodium–Sulfur Battery	Summarize working principles of RT-Na–S batteries; address key scientific problems in cathode, anode, electrolyte, and separator design to enhance energy storage performance	Tackles poor safety performance, high cost, and limited lithium resources; provides a pathway for safer, more sustainable, and cost-effective large-scale energy storage
[430]	Room-Temperature Sodium–Sulfur Battery	Enhance reversibility and cyclability by designing an all-fluorinated electrolyte (FDMA, MTFE, FEC) and forming NaF- and Na3N-rich cathode electrolyte interphase; improve compatibility between electrolytes and electrodes	Mitigates polysulfide shuttle, improves electrode–electrolyte compatibility, and boosts long-term cycling stability
[431]	Room-Temperature Sodium–Sulfur Battery	Summarize historical progress toward practical RT-Na/S batteries; promote balanced research trends by addressing advanced sulfur host design, Na metal anode protection, electrolyte optimization, separator modification, and binder engineering	Highlights overlooked components affecting polysulfide migration and reaction kinetics; encourages holistic and unbiased development for broader practical applications
[432]	Room-Temperature Sodium–Sulfur Battery	Develop a stable quasi-solid-state gel polymer electrolyte; enhance performance using in situ preparation and density functional theory for ion interaction analysis	Addresses dendrite formation and polysulfide shuttle issues; offers lower-cost alternative to expensive solid electrolytes in high-temperature Na–S systems
[433]	Room-Temperature Sodium–Sulfur Battery	Improve stability of sodium–sulfur batteries using chemical and spatial dual-confinement and covalent bonding of sulfur to carbon materials; achieve high-capacity retention after multiple cycles	Enhances cathode stability; mitigates sodium polysulfide formation; improves cycling performance and capacity retention
[434]	Sodium–Sulfur Battery	Mitigate polysulfide shuttle via covalent bonding of sulfur to polymeric backbone; optimize electrolyte compositions for long cycle life	Reduces capacity fade; improves stability and cycling performance through tailored cathode–electrolyte interactions
[435]	Room-temperature sodium–sulfur battery with FEC additive and tetraethylene glycol dimethyl ether solvent	Improve lifespan by mitigating shuttle effect in sulfur cathode	Targets key degradation mechanism (“solid–liquid–solid” reaction), enhancing cycle life through electrolyte optimization

summarizes several studies that have deployed sodium–sulfur batteries, highlighting their performance characteristics, deployment strategies, and operational benefits in smart grid applications.

4.3.4. Sodium-Ion Batteries

Although sodium–sulfur batteries provide high energy density and long life, they require regular maintenance, which increases the operational costs of the energy storage system. To mitigate these costs, sodium-ion (Na-ion) batteries have been proposed as an alternative for energy storage in smart grids. Sodium-ion batteries offer reduced maintenance requirements while remaining well-suited for stationary storage applications.

Table 29. Sodium-ion battery research, objectives, and strengths.

Ref.	Battery Type/Focus	Objectives	Strengths
[28]	Sodium-ion battery with varied electrode coatings and materials analyzed using SEM, EDS, EIS, and C-rate testing	Investigate structural and electrochemical characteristics of sodium-ion batteries	Provides detailed performance insights across conditions, supporting a knowledge base for further sodium-ion battery research
[436]	Study of capacity degeneration mechanisms in cathodes and review of effective strategies for long-cycle-life cathodes	Focus on developing long-cycle-life, low-cost cathodes for sodium-ion batteries	Identifies structural/morphology changes and unstable interphases, providing targeted strategies for stable cycling performance
[437]	Review fundamentals and progress of sodium-ion batteries, focusing on novel materials and electrochemistry tuning factors	Provide sustainable, low-cost alternative to lithium-ion batteries and guide future SIB development	Addresses lithium scarcity, promotes material sustainability, and supports long-term battery technology advancement
[438]	Na4MnCr(PO4)3 cathode	Develop high-energy cathode materials for sodium-ion batteries and investigate redox mechanisms and ionic migration	Moderate 7.7% volume change during cycling, enabling better structural stability and performance
[439]	Diglyme-based electrolyte SIBs	Explore sodium-ion batteries for sustainable energy storage systems and discuss advancements in diglyme-based electrolytes	Highlights significant impact of electrolyte selection on electrochemical performance
[440]	Non-aqueous, aqueous, and solid-state SIBs	Review current sodium-ion battery technologies for energy storage and explore sustainable alternatives to lithium-ion batteries	Addresses resource and supply chain limitations, highlighting cost-effective and sustainable solutions
[26]	Electrolyte compositions and electrode materials	Examine advancements in sodium-ion battery technology and discuss future research directions for scalability and commercial viability	Provides insights into material degradation and sodium-ion diffusion challenges for improved performance
[437]	Materials, degradation mechanisms, full-cell design, and electrolyte progress	Assess progress and challenges in sodium-ion battery technology and present a roadmap for future SIB implementation in energy storage	Identifies fundamental degradation mechanisms and enduring challenges, providing a foundation for targeted improvements
[437]	Standardized evaluation criteria and retention index for sodium cathodes	Discuss the role of sodium-ion batteries in energy storage and explore low-cost, performance, and sustainability benefits	Establishes a standardized evaluation framework, enabling consistent comparison and improvement across studies
[441]	One-pot solid-state reaction synthesized biphasic cathodes	Improve sodium-ion battery performance and enhance cycle stability and rate capability	Simple synthesis method with in situ structural analysis for better understanding of performance factors

summarizes several studies that have deployed sodium-ion batteries, highlighting their advantages, deployment strategies, and performance outcomes in smart grid implementations.

4.3.5. Lithium-Ion Batteries

One of the main limitations of sodium-ion batteries is their lower energy density, which restricts their full-scale deployment in smart grid energy storage applications. To address this limitation, lithium-ion (Li-ion) batteries are widely adopted due to their higher energy density, longer cycle life, and better efficiency. Many researchers have incorporated lithium-ion batteries into smart grid energy storage models to enhance performance and reliability.

Table 30. Lithium-ion battery research, objectives, and strengths.

Ref.	Type	Objectives	Strengths
[442]	Li-ion BESS with active network management	Design ANM architecture with Li-ion BESSs and develop adaptive controllers considering battery aging	Incorporates aging effects into control strategy to maintain performance and reliability over time
[443]	Grid-connected Li-ion BESS	Review management approaches for grid-connected Li-ion BESSs and evaluate participation in electricity markets	Provides comprehensive analysis of battery modeling, BMS architecture, and market integration challenges
[444]	Li-ion BESS reliability assessment	Propose a reliability assessment algorithm for BES systems and analyze battery lifetime degradation effects on reliability	Combines universal generating function and weak-link analysis to quantify safety and reliability impacts, offering a systematic evaluation method
[445]	Second-life Li-ion batteries (SLB) for fast-charging	Evaluate second-life batteries for fast-charging energy storage and compare economic and environmental impacts in U.S. cities	Provides detailed comparison of SLB and grid-based configurations, analyzing cost, lifetime, efficiency, and environmental metrics like GWP and CED
[446]	Direct and Model-Based SOH Diagnostics for Battery Reuse	Diagnose state of health for battery repurposing; assess direct, model-based, and data-driven methods for EV and ESS applications	Enables accurate SOH estimation to support battery reuse, promoting sustainability and circular economy goals
[415]	Hybrid Energy Storage System (HEESS) Development	Summarize recent research progress in HEESS; develop system configuration, DC/DC converter design, integrated sizing, and energy management strategies	Integrates battery and other storage elements to improve performance and extend lifetime; encourages innovative applications despite cost and degradation constraints
[447]	Smart Sensing Technologies for Batteries	Highlight advancements in smart sensing technologies for batteries; analyze limitations and challenges of different sensor applications (electrical, thermal, mechanical, acoustic, gas)	Provides comprehensive monitoring for battery health and safety; multi-sensor approach enables early detection of issues and improved battery management despite data collection challenges
[448]	Advanced Battery Management Systems	Improve battery energy density, power density, and cycle life; develop advanced BMS for safety and efficiency through online electrochemical spectroscopy impedance estimation and SOC estimation using adaptive unscented Kalman Filter	Enables accurate state-of-charge estimation and robust thermal management; enhances performance, safety, and lifespan of Li-ion batteries
[449]	Grid-Integrated Battery Storage	Discuss applications and benefits of battery storage in electricity grids; compare advantages and disadvantages of various electrochemical batteries	Provides comprehensive comparison of battery types; highlights suitability of Li-ion and other chemistries for managing renewable intermittency and long-term cost considerations
[413]	Emergency Backup for Isolated Networks	Introduce battery energy storage for emergency power supply; improve reliability of separated power networks during outages	Uses real measurement data for accurate ESS dimensioning; enhances reliability and resilience during main line damage or transmission limitations

provides a comparison of several studies that implemented lithium-ion batteries, highlighting their objectives, deployment strategies, and key strengths in energy storage systems.

4.4. Cost–Performance Trade-Offs Across Battery Technologies in Smart Grid Scenarios

Energy storage systems (ESSs) are integral to smart grids (SGs) for balancing demand and supply, but each battery technology presents distinct cost–performance characteristics. Lead–acid batteries are among the cheapest to deploy, offering simple control and high surge currents, which make them attractive for low-cost stationary storage. However, their shorter lifecycle and poor performance under adverse weather conditions limit their long-term cost-effectiveness [25]. Nickel-based batteries improve durability and environmental tolerance, with high cycle life (1000–1500+ cycles) and low maintenance requirements, but they suffer from higher self-discharge currents, which increase operational inefficiency [420].

Sodium–sulfur (NaS) batteries provide high energy density, long life, and deep discharge capability, offering better performance for large-scale integration. However, they require regular maintenance, which increases operational expenditure over time [427]. Sodium-ion batteries (SIBs) offer a cost-effective and sustainable alternative to lithium-ion by leveraging abundant materials. They are optimal for stationary storage models, providing moderate performance at lower cost, yet their lower energy density restricts deployment in applications requiring high power density [437].

Lithium-ion (Li-ion) batteries, conversely, deliver high energy density and scalability, making them suitable for both grid-scale and distributed applications. Nevertheless, they come with higher upfront costs and face degradation challenges over time [442]. Research on second-life Li-ion batteries has shown that reusing electric vehicle (EV) batteries for grid services can significantly reduce lifecycle costs while maintaining acceptable performance, improving both economic and environmental metrics (e.g., lowering global warming potential and cumulative energy demand compared to new battery packs) [445].

In practical SG scenarios, these cost–performance trade-offs indicate that lead–acid may be chosen for short-duration, cost-constrained backup; nickel-based and sodium–sulfur for environments demanding resilience and deep cycling; sodium-ion for cost-sensitive stationary storage with moderate performance; and lithium-ion or second-life Li-ion where high performance and integration flexibility justify higher upfront cost. Hybrid configurations combining, for instance, Li-ion and supercapacitors or lead–acid with supercapacitors, further optimize both cost and performance by distributing stress across chemistries to extend lifetime and improve round-trip efficiency [414]. A comparative analysis of cost–performance metrics of these storage systems is presented in

Table 31. Cost–performance trade-offs across battery technologies for smart grid (SG) applications.

Battery Type	Cost Level	Performance (Cycle Life/Energy Density)	Maintenance/Degradation	Typical SG Use Cases
Lead–acid	Low (cheapest upfront)	Low–Moderate/Low energy density	Short lifecycle, sensitive to adverse weather	Short-duration backup, cost-constrained stationary storage
Nickel-based (NiCd, NiMH)	Moderate	High cycle life (1000–1500+), good temperature tolerance	Higher self-discharge, moderate maintenance	Resilient stationary storage, moderate-to-high cycling frequency
Sodium–sulfur (Na–S)	Moderate–High	High energy density, long life, deep discharge	Requires regular maintenance, thermal management	Large-scale stationary storage, renewable smoothing
Sodium-ion (SIBs)	Low–Moderate	Moderate energy density, emerging tech	Limited high-power performance, early stage scaling	Sustainable, cost-sensitive stationary storage
Lithium-ion (Li-ion)	High	High cycle life, high energy density, scalable	Degradation over time, higher upfront cost	Distributed and grid-scale storage, fast-response balancing
Second-life Li-ion	Moderate (lower lifecycle cost)	Maintains acceptable performance for SG use	Performance depends on EV battery history	Cost-effective grid services, environmental lifecycle benefits
Hybrid systems (e.g., Li-ion + supercapacitor)	Case-dependent	Optimized performance via complementary chemistries	Reduced stress on individual cells	Applications requiring both high power and high energy capabilities

.

4.5. Role of AI/Privacy-Preserving Techniques in ESS

Energy storage systems (ESSs), particularly their battery management systems (BMSs), introduce unique cybersecurity and privacy challenges. BMS devices control charging, discharging, and state-of-charge estimation, making them attractive targets for adversaries. False data injection into BMS signals can lead to misreporting of storage capacity, resulting in market imbalances or unsafe operating conditions.

Privacy-preserving techniques, such as differential privacy or Secure Multiparty Computation, can be applied to aggregated storage data to hide individual user behavior, while blockchain can provide tamper-proof audit trails of storage transactions. However, these mechanisms must be complemented with system-level intrusion detection and secure firmware, since privacy alone cannot prevent adversaries from disrupting storage operations.

Thus, the interaction between privacy and ESS security is twofold: cryptographic and privacy-preserving methods help protect user and market data, while robust BMS cybersecurity ensures that the underlying physical storage assets remain safe and trustworthy. A summary of privacy-preserving techniques and their interaction with ESS is presented in

Table 32. Privacy-preserving mechanisms for ESS/BMS interaction.

Mechanism	Interaction with ESS/BMS
Homomorphic Encryption	Encrypts storage data before sharing
Zero-Knowledge Proofs	Verifies BMS data authenticity without revealing full details
Differential Privacy	Preserves privacy in aggregated SoC/charging data
Blockchain	Provides immutable record of ESS transactions and logs
IDS/Lightweight Crypto	Detects abnormal SoC/voltage patterns

.

5. Case Studies

This section presents case studies based on pricing models, demand–supply management, privacy preservation, and attack identification in P2P smart grid energy trading, following recent studies [1,450,451].

5.1. Case Study: Uniform Pricing Model

Uniform pricing models bring all participants in the energy market onto a single platform, where they agree on the same price. Several factors, such as the Supply–Demand Ratio, generation costs, transmission, and grid constraints, can influence the determination of a uniform price signal. Reputation is another key factor, which can be used to determine energy prices in P2P markets. Incorporating reputation reduces bias among participants and promotes fairness in energy trading. A reputation-based pricing model has been proposed in the recent study by Amin et al. [450].

The energy trading model proposed by the authors is illustrated in Figure 7.

From Figure 7, it can be seen that the proposed model formulates a market where participants interact with each other to encourage energy trading under the supervision of a virtual market operator (MO), which oversees all trading activities. Sellers are weighted based on factors such as reliability $\left(\alpha \right)$, price dynamism $\left(PD\right)$, presence of participants $\left(\omega \right)$, persistence factor $\left(\beta \right)$, etc. The MO evaluates the reputation $\left(\mathcal{R}\right)$ of each participant and assigns a corresponding score.

At the start of a trading session t, the MO invites participants to share their information. Sellers provide the MO with their offered price $P_{pro}$ and the offered surplus energy $E_{off}$, while buyers in the set B notify the MO of their energy demand $E_{dem}$ and offered price $P_{off}$. The MO keeps this information private and evaluates the reputation of each participant. The reputation evaluation for sellers, according to this study, can be computed using the following equations.

$${\left({\mathrm{Score}}_{s_i,t}\right)}_{\omega }=\left\{\begin{array}{cc}1,\hfill & {\omega }_{i,t}=1,\hfill \\ 0,\hfill & {\omega }_{i,t}=0,\hfill \end{array}\right.$$

(1)

$${\mathrm{Score}}_{s_i,t}^{\beta }={\beta }_{s,t}·\mathbb{I}({\omega }_{i,t}=1)$$

(2)

$${\mathrm{Score}}_{s_i,t}^{\beta }=\left\{\begin{array}{cc}(1-\gamma ){\beta }_{s,t-1}+\gamma ,\hfill & {\omega }_{s_i,t}=1,\hfill \\ (1-\gamma ){\beta }_{s,t-1},\hfill & {\omega }_{s_i,t}=0,\hfill \end{array}\right.$$

(3)

$${\mathrm{Score}}_{i,t}^{\alpha }=\left\{\begin{array}{cc}{\mathrm{Score}}_{i,t}+1,\hfill & {\alpha }_{i,t}=1,\hfill \\ {\mathrm{Score}}_{i,t}-1,\hfill & {\alpha }_{i,t}=0,\hfill \end{array}\right.$$

(4)

$${\mathrm{Score}}_{s_i,t}^{PD}=\left\{\begin{array}{cc}{\mathrm{Score}}_{i,t-1}+1,\hfill & P_{i,t}^{pro}\ne P_{i,t-1}^{pro},\hfill \\ {\mathrm{Score}}_{i,t-1}-1,\hfill & P_{i,t}^{pro}=P_{i,t-1}^{pro},\hfill \end{array}\right.$$

(5)

$${\mathrm{Score}}_{s_i,t}^{BPD}=\sum _{p^{off}\in P^{off}}\left\{\begin{array}{cc}{\mathrm{Score}}_{i,t}+{\displaystyle \frac{p^{off}-P_{i,t}^{pro}}{p^{off}}}\times 100,\hfill & P_{i,t}^{pro}<p^{off},\hfill \\ {\mathrm{Score}}_{i,t}-{\displaystyle \frac{P_{i,t}^{pro}-p^{off}}{p^{off}}}\times 100,\hfill & P_{i,t}^{pro}>p^{off},\hfill \end{array}\right.$$

(6)

$${\left({\mathrm{Score}}_{i,t}\right)}_{E_{off}}=\left\{\begin{array}{cc}{\mathrm{Score}}_{i,t}+2,\hfill & E_{off,i,t}={\Theta }_{sur,t},\hfill \\ {\mathrm{Score}}_{i,t}-\mathrm{Penalty}·{\displaystyle \frac{|E_{off,i,t}-{\Theta }_{sur,t}|}{{\Theta }_{sur,t}}},\hfill & E_{off,i,t}<{\Theta }_{sur,t},\hfill \\ {\mathrm{Score}}_{i,t}+\mathrm{Bonus}·{\displaystyle \frac{E_{off,i,t}-{\Theta }_{sur,t}}{{\Theta }_{sur,t}}},\hfill & E_{off,i,t}>{\Theta }_{sur,t},\hfill \end{array}\right.$$

(7)

$$\begin{array}{cc}\hfill {\mathrm{TotalScore}}_{s_i,t}& =W_{\omega }{\left({\mathrm{Score}}_{s_i,t}\right)}_{\omega }+W_{\beta }{\mathrm{Score}}_{s_i,t}^{\beta }+W_{\alpha }{\mathrm{Score}}_{s_i,t}^{\alpha }\hfill \\ \hfill & +W_{PD}{\mathrm{Score}}_{s_i,t}^{PD}+W_{BPD}{\mathrm{Score}}_{s_i,t}^{BPD}+W_{E_{off}}{\left({\mathrm{Score}}_{i,t}\right)}_{E_{off}}\hfill \end{array}$$

(8)

$$S_{\mathrm{winner}}\left(t\right)=arg\underset{s_i\in S}{max}\left({\mathrm{TotalScore}}_{s_i,t}\right)$$

(9)

Similarly, buyers’ evaluation criteria are presented in equations given as

$${\left({\mathrm{Score}}_{b_i,t}\right)}_{\omega }=\left\{\begin{array}{cc}1,\hfill & {\omega }_{b_i,t}=1,\hfill \\ 0,\hfill & {\omega }_{b_i,t}=0,\hfill \end{array}\right.$$

(10)

$${\mathrm{Score}}_{b_i,t}^{\beta }={\beta }_{b,t}\mathbb{I}({\omega }_{b_i,t}=1)$$

(11)

$${\mathrm{Score}}_{b_i,t}^{\beta }=\left\{\begin{array}{cc}(1-\gamma ){\beta }_{b,t-1}+\gamma ,\hfill & {\omega }_{b_i,t}=1,\hfill \\ (1-\gamma ){\beta }_{b,t-1},\hfill & {\omega }_{b_i,t}=0,\hfill \end{array}\right.$$

(12)

$${\mathrm{Score}}_{b_i,t}^{PD}=\left\{\begin{array}{cc}{\mathrm{Score}}_{i,t-1}+1,\hfill & P_t^{off}\ne P_{t-1}^{off},\hfill \\ {\mathrm{Score}}_{i,t-1}-1,\hfill & P_t^{off}=P_{t-1}^{off},\hfill \end{array}\right.$$

(13)

$${\left({\mathrm{Score}}_{b_i,t}\right)}_{BPD}=\sum _{p^{off}\in P^{off}}\left\{\begin{array}{cc}{\mathrm{Score}}_{i,t}+{\displaystyle \frac{p^{pro}-P^{off}}{p^{pro}}}\times 100,\hfill & P^{off}>p^{pro},\hfill \\ {\mathrm{Score}}_{i,t}-{\displaystyle \frac{p^{pro}-P^{off}}{p^{pro}}}\times 100,\hfill & P^{off}<p^{pro},\hfill \end{array}\right.$$

(14)

$${\left({\mathrm{Score}}_{i,t}\right)}_{E_{dem}}=\left\{\begin{array}{cc}{\mathrm{Score}}_{i,t}+2,\hfill & E_{dem,t}={\Theta }_t^{dem},\hfill \\ {\mathrm{Score}}_{i,t}-\mathrm{Penalty}·{\displaystyle \frac{|E_{dem,t}-{\Theta }_t^{dem}|}{{\Theta }_t^{dem}}},\hfill & E_{dem,t}>{\Theta }_t^{dem},\hfill \\ {\mathrm{Score}}_{i,t}+\mathrm{Bonus}·{\displaystyle \frac{E_{dem,t}-{\Theta }_t^{dem}}{{\Theta }_t^{dem}}},\hfill & E_{dem,t}<{\Theta }_t^{dem},\hfill \end{array}\right.$$

(15)

$${\mathrm{TotalScore}}_{b_i,t}=\sum _k{\mathrm{Weight}}_k·{\mathrm{Score}}_{b_i,t}^{\left(k\right)}$$

(16)

$$B_{\mathrm{winner}}\left(t\right)=arg\underset{b_i\in B}{max}\left({\mathrm{TotalScore}}_{b_i,t}\right)$$

(17)

Once the scores for buyers and sellers have been evaluated, the market operator (MO) calculates the Supply–Demand Ratio (SDR) to determine the possible mode of operation. Based on the identified mode of operation, the final trading price of energy can be determined. The process is summarized as follows:

$$(\mathrm{Sellers}’\mathrm{mode}){\mathrm{FTP}}_t=P_{S_{\mathrm{winner}}}^{pro}$$

(18)

$$(\mathrm{Buyers}’\mathrm{mode}){\mathrm{FTP}}_t=P_{B_{\mathrm{winner}}}^{off}$$

(19)

$$(\mathrm{Equilibrium}\mathrm{mode}){\mathrm{FTP}}_t={\displaystyle \frac{1}{2}}\left(P_{S_{\mathrm{winner}}}^{pro}+P_{B_{\mathrm{winner}}}^{off}\right)$$

(20)

This proposed model exhibits several desirable characteristics, such as non-dictatorship, Pareto efficiency, independence of irrelevant alternatives, and collective rationality, thereby promoting fairness in the energy market while determining the uniform price of energy.

The model was evaluated using a one-year dataset consisting of varying demand and supply values. It was also compared with several state-of-the-art methods, including those proposed in [226,235,294,452]. Additionally, the proposed model considers the grid price as another baseline parameter to assess seller profitability and buyer savings. The experimental results are presented in Figure 8a,b.

The experimental results demonstrate that the proposed reputation-based pricing model can increase the probability of sellers participating in the market by 2.1% to 20.41% compared with other state-of-the-art methods, and improve seller profitability by 22.91% compared to the grid export rate.

Similarly, buyers benefit from energy bill savings ranging from 2.81% to 44.27% compared with other state-of-the-art methods, while the savings relative to the grid import rate can reach approximately 27.78%.

5.2. Case Study: Integrated Demand–Supply Balance, Pricing, and Privacy-Preserving Model

A recent study [1] proposed an integrated model that combines demand–supply balance, pricing, and privacy-preserving techniques to ensure an effective energy trading platform in P2P energy markets. The proposed model is illustrated in Figure 9.

In this model, for each time slot t, each energy consumer (EC) i reports its energy demand $D_i$, while each energy producer (EP) j reports its surplus energy $S_j$. Additionally, each buyer communicates to the system operator (S.O.) their comfort indices and budgetary constraints, denoted as $({\mathbb{C}}_{min},{\mathbb{C}}_{max})$ and $({\mathbb{B}}_{min},{\mathbb{B}}_{max})$, respectively.

The information shared by the participants is secured through a quorum-based architecture that integrates SHA-256 hashing, the Elliptic Curve Digital Signature Algorithm (ECDSA), and Shamir’s Secret Sharing (SSS). Each participant signs and hashes their data as follows:

$$\begin{array}{cc}\hfill {\mathrm{Sig}}_{E{C}_i}& ={\mathrm{Sign}}_{S{K}_i}(D_i\parallel C_{min,i}\parallel C_{max,i}\parallel B_{min,i}\parallel B_{max,i})\hfill \end{array}$$

(21)

$$\begin{array}{cc}\hfill H\left(E{C}_i\right)& =\mathrm{SHA}-256(D_i\parallel C_{min,i}\parallel C_{max,i}\parallel B_{min,i}\parallel B_{max,i})\hfill \end{array}$$

(22)

After verification, the gateway encodes the data using SSS:

$$f\left(x\right)=a_0+a_{1}x+a_2{x}^2+\cdots +a_{\lambda -1}{x}^{\lambda -1}modp$$

(23)

Each gateway receives a share:

$${\mathrm{Share}}_k=f\left(x_k\right)modp,k=1,2,\dots ,t$$

(24)

The S.O reconstructs the secret using Lagrange interpolation:

$$S=f\left(0\right)=\sum _{k=1}^{\lambda }{\mathrm{Share}}_k{L}_{k}modp$$

(25)

Once the information is successfully secured, the S.O determines the market operation mode by evaluating the Supply–Demand Ratio (SDR):

$$\mathrm{SDR}={\displaystyle \frac{{\sum }_{i=1}^N{D}_i}{{\sum }_{j=1}^M{S}_j}}$$

(26)

If $\mathrm{SDR}>1$, the market operates in Sellers’ Mode, otherwise in Buyers’ Mode. At each trading time ‘t’ the ideal trading price (ITP) is determined at the equilibrium point of supply and demand, as presented in Figure 10.

Based on the market mode, the S.O. computes the trading price (TP). For the Sellers’ Mode, the TP is calculated as follows:

$$TP=\left|{\displaystyle \frac{{\sum }_{j=1}^M{S}_j-{\sum }_{i=1}^N{D}_i}{{\sum }_{j=1}^M{S}_j+{\sum }_{i=1}^N{D}_i}}\right|·ITP+ITP$$

(27)

Once the price of energy has been found the next objective for the S.O is to formulate the energy allocation model as a convex optimization problem presented as follows:

$$\begin{array}{c}\hfill min\sum _i{x}_i\end{array}$$

(28)

$$\begin{array}{cc}\hfill \mathrm{s}.\mathrm{t}.B_{min,i}\le TP·x_i\le B_{max,i},& \forall i\hfill \end{array}$$

(29)

$$\begin{array}{cc}\hfill x_i\le D_i,& \forall i\hfill \end{array}$$

(30)

$$\begin{array}{cc}\hfill C_{min,i}\le {\displaystyle \frac{x_i}{D_i}}\le C_{max,i},& \forall i\hfill \end{array}$$

(31)

$$\begin{array}{cc}\hfill x_i,x_j>0,& \forall i,j\hfill \end{array}$$

(32)

For Buyers’ Mode, the objective can be expressed as follows:

$$\begin{array}{c}\hfill max\sum _i{x}_i\end{array}$$

(33)

$$\begin{array}{cc}\hfill \mathrm{s}.\mathrm{t}.TP·x_i\le B_{max,i},& \forall i\hfill \end{array}$$

(34)

$$\begin{array}{cc}\hfill \sum _{i=1}^N{x}_i=\sum _{j=1}^M(S_j-x_{g,j}),& \forall j\hfill \end{array}$$

(35)

The experimental results demonstrate the robustness of the proposed model against data manipulation during transmission. Any tampering results in the failure of the integrity test for affected participants, as illustrated in Figure 11.

Due to the failed integrity test, the S.O. refuses to reconstruct the information of affected participants, as shown in Figure 12.

This mechanism ensures privacy preservation in the P2P energy trading market. Furthermore, the model is computationally efficient; Figure 13 shows that it requires less execution time compared to baseline models.

Finally, evaluation over a one-year dataset demonstrates improved grid efficiency by reducing the grid imbalance state in Buyers’ Mode, as presented in Figure 14.

5.3. Cyberattack Detection in Smart Grid

This case study demonstrates the detection of cyberattacks in smart grids, specifically focusing on False Data Injection Attacks (FDIAs) that can manipulate participants’ shared information and disrupt market operations, based on the recent study [451].

With the rapid integration of renewable energy and advanced communication technologies, smart grids have become crucial for efficient power generation and distribution. However, this interconnectivity exposes grids to cyber threats such as FDIAs, which can compromise state estimation, disrupt grid operations, and potentially cause large-scale blackouts. Traditional machine learning techniques often fail to capture temporal dependencies in smart grid time-series data, reducing detection accuracy. To address this, Ref. [451] proposes a robust detection framework using a Bidirectional Long Short-Term Memory (Bi-LSTM) network integrated with an Attention Mechanism, offering high detection accuracy, computational efficiency, and interpretability.

The architecture diagram of the proposed model is shown in Figure 15.

As illustrated in Figure 15, the proposed model workflow is as follows:

• Data Acquisition: Time-series data from smart meters, including participant information such as energy demand, surplus energy, offered and proposed prices, is collected.
• Data Preprocessing: The raw input data passes through a preprocessing block to clean, normalize, and structure it for use in the Bi-LSTM model.
• Bi-LSTM Network: Captures both forward and backward temporal dependencies, enhancing recognition of subtle patterns associated with potential FDIA.
• Attention Mechanism: Dynamically assigns weights to critical time steps, focusing on significant events while filtering irrelevant information, thereby improving detection accuracy and interpretability.

This integrated Bi-LSTM and Attention framework enables the smart grid system to effectively detect FDIAs, ensuring the integrity of participant data and maintaining stable energy market operations.

The proposed model is evaluated on a real-time smart grid dataset, and the reported results are summarized in

Table 33. Performance metric of the proposed model [451] in comparison with other state-of-the-art baseline models.

Model	Accuracy (%)	F1-Score (Attack)	F1-Score (Natural)	F1-Score (NoEvents)
Proposed Model	92.32	88.17	89.54	99.14
LSTM-CNN	79.01	66.29	73.82	96.92
LSTM-Autoencoder	90.17	86.21	87.80	97.81
Bert	80.00	68.25	74.60	98.51

. From Table 33, it can be observed that the proposed Bi-LSTM with Attention model effectively identifies potential cyberattacks in the form of False Data Injection Attacks (FDIAs), achieving an accuracy of 92.32%, which outperforms other state-of-the-art baseline models.

6. Future Research Directions

This section briefly explores potential future research directions based on the extensive literature reviewed in the preceding sections.

6.1. Privacy-Preserving Future Research Directions

The analysis of privacy-preserving techniques in smart grids reveals several limitations that highlight essential avenues for future research. For cryptographic methods such as Homomorphic Encryption (HE), the high computational overhead and latency in processing large-scale data necessitate the development of streamlined algorithms to reduce costs and enhance feasibility for real-time applications [64,65,73]. Similarly, Zero-Knowledge Proofs (ZKPs) face challenges in computational complexity and scalability, indicating the need for optimized protocols such as zk-SNARKs and zk-STARKs for efficient deployment in large-scale networks [85,86]. Secure Multiparty Computation (SMPC) also requires advancements to mitigate high communication overhead and improve scalability in decentralized systems [93,97].

In anonymization and aggregation techniques, differential privacy (DP) demands improved optimization of noise addition to balance privacy guarantees with data utility, particularly for demand-response management [100,106]. The vulnerabilities of k-Anonymity to auxiliary information attacks call for more robust models incorporating clustering and noise injection to enhance resistance [111,117]. Data aggregation models should aim to minimize generalization losses while still supporting individualized services without compromising overall grid optimization [122,123].

Blockchain-based mechanisms, including smart contracts and decentralized access control, require solutions to scalability and transaction speed limitations to accommodate increasing network traffic [130,142]. Similarly, tokenization approaches demand improved key management to prevent misuse and ensure seamless integration into energy trading systems [156,157].

Machine learning approaches, such as federated learning, must address challenges related to data heterogeneity and communication inefficiencies to achieve better model convergence for anomaly detection [160,164]. Adversarial learning, particularly using GANs, should focus on improving the quality of synthetic data and computational efficiency to develop robust privacy-preserving models [168,169].

Table 34. Future research directions in privacy preservation for smart grids.

Research Area	Future Research Direction	Refs.
Cryptographic Techniques (HE)	Streamline methods to minimize computational overhead and latency for large-scale SG networks.	[64,65,73]
Cryptographic Techniques (ZKPs)	Optimize protocols for reduced complexity and improved scalability in real-time applications.	[85,86]
Cryptographic Techniques (SMPC)	Enhance efficiency to handle high communication overhead in decentralized systems.	[93,97]
Anonymization and Aggregation (DP)	Optimize noise addition for better balance between privacy and data utility.	[100,106]
Anonymization and Aggregation (k-Anonymity)	Develop robust models resistant to auxiliary attacks using clustering and noise.	[111,117]
Anonymization and Aggregation (Data Aggregation)	Minimize data generalization losses for individualized services and grid optimization.	[122,123]
Blockchain-Based Mechanisms (Smart Contracts/Access Control)	Address scalability and transaction speed for large network traffic.	[130,142]
Blockchain-Based Mechanisms (Tokenization)	Improve key management to prevent misuse in energy trading.	[156,157]
Machine Learning Approaches (Federated Learning)	Mitigate data heterogeneity and communication issues for model convergence.	[160,164]
Machine Learning Approaches (Adversarial Learning)	Enhance synthetic data quality and reduce computational costs.	[168,169]

summarizes the key future research directions for privacy-preserving techniques in smart grids.

Although the discussed literature presents remarkable efforts to preserve participants’ privacy and ensure grid stability, their resilience against coordinated, multi-layered state-sponsored attacks as discussed in Section 2.3 remains uncertain and highly dependent on implementation rigor and adherence to established standards.

For instance, while Homomorphic Encryption ensures data confidentiality during processing, its high computational overhead may make it susceptible to resource exhaustion attacks, a common tactic in state-sponsored DDoS campaigns, unless complemented by NIST-recommended resource management and intrusion detection systems [453]. Similarly, Zero-Knowledge Proofs, although effective for authentication without exposing data, face scalability challenges and may not withstand sustained high-volume attacks without IEC 62351-compliant key management and session integrity controls.

Blockchain-based solutions provide tamper-resistant transaction records and decentralized trust; however, their slower transaction speeds and scalability limitations (as highlighted in Table 7 and Table 13) could be exploited in prolonged attacks targeting consensus mechanisms. The International Electrotechnical Commission (IEC) [454] emphasizes the necessity of resilient consensus algorithms and real-time anomaly detection to mitigate such risks.

Although differential privacy and k-Anonymization techniques protect against data linkage attacks, they may fail under advanced adversarial inference leveraging auxiliary data, a known capability of state actors. IEC 62443-3-3 [455] recommends augmenting these techniques with continuous monitoring and adaptive noise injection to maintain utility while resisting sophisticated de-anonymization attempts.

Furthermore, the hierarchical communication architecture of smart grids (HAN/BAN, WAN) introduces multiple attack surfaces. State-sponsored actors often exploit unpatched systems, vulnerable protocols, and insider threats, as illustrated in Figure 2. NIST CSF and IEC 62351 stress the importance of defense-in-depth strategies, including (i) network segmentation and encryption across all layers, (ii) regular security audits and patch management, (iii) multi-factor authentication and strict access controls, and (iv) real-time intrusion detection and response systems.

Based on this discussion, it is recommended that future research adopt a holistic, standards-based approach integrating cryptographic methods with robust network security practices, continuous monitoring, and incident response frameworks to enhance smart grid resilience against advanced persistent threats.

6.2. Future Research Direction: Pricing Models

Future research in energy pricing mechanisms for smart grid P2P trading should focus on addressing the operational challenges and limitations observed in synchronous and asynchronous pricing models. For synchronous pricing, efforts could aim at enhancing market efficiency by mitigating manipulation in the merit order, improving congestion handling in uniform pricing, and developing more accurate grid constraint modeling for Locational Marginal Pricing (LMP) in dynamic systems with high renewable penetration. Similarly, Supply–Demand Ratio (SDR)-based pricing requires advancements to better handle fluctuating renewable energy markets through improved accuracy in supply and demand data.

Asynchronous mechanisms offer opportunities to reduce inefficiencies in pay-as-bid pricing by minimizing bid manipulation and price uncertainty, while also optimizing bilateral negotiations for greater transparency and mutual agreement on contract terms. Reserved pricing could benefit from methods to determine optimal reserve prices without hindering market clearing, and forward/futures contracts require improved models for contract pricing under market volatility.

Regarding price determination techniques, future directions include refining game-theoretic approaches to better model participant uncertainty and equilibrium calculations in large markets, as well as effectively handling non-cooperative behaviors in dynamic environments. Optimization methods should address forecasting data accuracy and computational demands for large-scale problems, particularly in systems with high renewable variability. Numerical methods can evolve to reduce computational costs in real-time simulations and improve model calibration under dynamic conditions. AI-based techniques present significant potential to overcome data quality challenges, enhance model interpretability, and reduce training time for large-scale adaptations to market dynamics.

Table 35. Future research directions in energy pricing.

Research Area	Future Research Direction	Refs.
Synchronous Pricing Mechanisms	Mitigate price distortion and congestion issues in uniform pricing; improve grid constraint modeling for LMP in high renewable systems; enhance data accuracy for SDR in fluctuating markets.	[18,183,184,185,186,192,193,194,195,196,197,198,199,200,201,202,203,204,205,206,207]
Asynchronous Pricing Mechanisms	Reduce bid manipulation and uncertainty in pay-as-bid/Vickrey; improve transparency in bilateral negotiations; optimize reserve prices to avoid market inefficiency; handle volatility in forward/futures contracts.	[208,209,210,211,212,213,214,215,216,217,218,219,220,221]
Game-Theoretic Techniques	Address uncertainty in behavior modeling and equilibrium in large markets; handle non-cooperative behaviors in dynamic settings.	[19,241,287,288,289,293,301,302,303,304,305,306,307,308,309,310,311]
Optimization Techniques	Improve data accuracy in forecasting; reduce computational resources for large problems; model renewable variability effectively.	[313,319,321,326,327,328,329,330,331,332,333,334,335,336,337,338,339]
Numerical Method-Based Techniques	Lower computational costs for real-time simulations; enhance calibration under dynamic market conditions.	[22,340,341,342,343,344,345,346,347,348,349,350,351]
AI-Based Techniques	Overcome data quality and availability issues; improve model interpretability; reduce training time for large-scale models.	[352,353,354,355,356,358,359,360,361,362,363,364,365]

summarizes the future research directions for pricing models in smart grid P2P energy trading.

6.3. Future Research Directions: DSBM

The Demand–Supply Balance Program (DSBP) faces several challenges, including achieving scalability, real-time adaptability, and effective integration of strategic participant interactions. Iterative models, while simple and scalable, often exhibit slow convergence and suboptimal performance in dynamic environments [20,370]. Future research should focus on accelerating convergence through advanced synchronization techniques and enhancing robustness against communication delays.

Optimization models can achieve near-optimal solutions but are computationally intensive and struggle with high-dimensional problems [21,385]. Developing hybrid algorithms that combine heuristic methods with real-time data processing could mitigate these limitations.

Game-theoretic models capture strategic interactions effectively but face challenges in ensuring unique equilibria and preserving participant privacy [22,229]. Future research should explore decentralized game-theoretic frameworks integrated with blockchain to enable secure, privacy-preserving interactions.

AI-based models offer significant potential in managing uncertainty and large datasets but require substantial training data and can face integration challenges with legacy grid infrastructure [23,408,456]. Future efforts should focus on developing lightweight AI models, leveraging transfer learning to reduce data dependency, and ensuring seamless integration with existing systems.

Table 36. Future research directions for Demand–Supply Balance Program (DSBP).

Research Area	Future Research Direction	References
Iterative Models	Enhance convergence speed using advanced synchronization techniques; improve robustness against communication delays	[20,370]
Optimization Models	Develop hybrid optimization algorithms combining heuristics with real-time data processing to reduce computational intensity	[21,385]
Game-Theoretic Models	Explore decentralized frameworks with blockchain for secure, privacy-preserving strategic interactions	[22,229]
AI-Based Models	Develop lightweight AI models for seamless integration with legacy systems; leverage transfer learning to reduce data dependency	[23,408]

summarizes the key future research directions for DSBM.

6.4. Future Research Directions: ESS

Energy storage systems (ESSs) play a crucial role in balancing demand and supply in smart grids, yet each battery technology presents unique challenges. Lead–acid batteries, while low-cost, suffer from short lifecycles and limited performance under adverse conditions [25,418]. Future research should focus on enhancing lifecycle performance through advanced thermal management and hybrid configurations.

Nickel-based batteries offer durability but are constrained by high self-discharge rates [420,424]. Developing low-cost, high-efficiency electrolytes and improved cell designs could mitigate this limitation. Sodium–sulfur batteries provide high energy density but require frequent maintenance [427,428]. Research should explore low-maintenance architectures and novel electrolyte additives to improve operational stability.

Sodium-ion batteries are promising for stationary storage but exhibit lower energy density [28,437]. Future work should investigate advanced cathode materials and optimized cell designs to enhance energy density while maintaining cost-effectiveness. Lithium-ion batteries, with high energy density and scalability, face challenges related to cost and degradation [442,448]. Research should emphasize second-life applications, advanced battery management systems, and recycling strategies to extend lifespan and reduce costs.

Table 37. Future research directions for energy storage systems (ESSs).

Research Area	Future Research Direction	References
Lead–Acid Batteries	Improve lifecycle through advanced thermal management and hybrid configurations	[25,418]
Nickel-Based Batteries	Develop low-cost, high-efficiency electrolytes to reduce self-discharge rates	[420,424]
Sodium–Sulfur Batteries	Explore low-maintenance designs and novel electrolyte additives for enhanced stability	[427,428]
Sodium-Ion Batteries	Investigate advanced cathode materials to improve energy density while maintaining cost-effectiveness	[28,437]
Lithium-Ion Batteries	Focus on second-life applications and advanced battery management systems to extend lifespan and reduce costs	[442,448]

summarizes the key future research directions for ESSs in smart grid applications.

7. Conclusions

The smart grid (SG) represents a transformative platform in energy management, integrating efficiency, sustainability, and reliability through two-way communication, the Internet of Things (IoT), and Advanced Metering Infrastructure (AMI). However, the extensive transfer of sensitive data exposes SGs to privacy and security risks, as evidenced by recent global incidents including data interception, energy theft, and cyberattacks such as MITM, DoS, and replay attacks. This review systematically analyzes the SG architecture and its core components substations, concentrators, aggregators, smart appliances, and renewable energy sources highlighting associated vulnerabilities and fundamental security requirements, including confidentiality, integrity, availability, secure storage, and scalability.

Effective countermeasures include cryptographic protocols (Homomorphic Encryption, Zero-Knowledge Proofs, Secure Multiparty Computation), anonymization techniques (differential privacy, k-Anonymity), and data aggregation methods. Recent studies also demonstrate that machine learning and blockchain-based solutions can complement these defenses. Further analyses of pricing models, Demand–Supply Balance Programs (DSBPs) using optimization and game-theoretic approaches, AI-driven models, and energy storage systems across various battery chemistries underscore their crucial role in maintaining grid stability amid the stochastic nature of renewable energy sources.

Finally, this review identifies persistent research gaps, such as the computational overhead of cryptographic techniques, trade-offs in the utility of anonymized data, and network scalability challenges. Addressing these issues is critical to developing resilient SGs that safeguard user privacy, mitigate cyber threats, and facilitate widespread adoption of sustainable energy technologies.