Design and Evaluation of a Research-Oriented Open-Source Platform for Smart Grid Metering: A Comprehensive Review and Experimental Intercomparison of Smart Meter Technologies

Abstract

Smart meters (SMs) are essential components of modern smart grids, enabling real-time and accurate monitoring of electricity consumption. However, their evaluation is often hindered by proprietary communication protocols and the high cost of commercial testing tools. This study presents a low-cost, open-source experimental platform for smart meter validation, using a microcontroller and light sensor to detect optical pulses emitted by standard SMs. This non-intrusive approach circumvents proprietary restrictions while enabling transparent and reproducible comparisons. A case study was conducted comparing the static meter GAMA 300 model, manufactured by Elgama-Elektronika Ltd. (Vilnius, Lithuania), which is a closed-source commercial meter, with theTexas Instruments EVM430-F67641 evaluation module, manufactured by Texas Instruments Inc. (Dallas, TX, USA), which serves as an open-source reference design. Statistical analyses—based on confidence intervals and ANOVA—revealed a mean deviation of less than 1.5% between the devices, confirming the platform’s reliability. The system supports indirect power monitoring without hardware modification or access to internal data, making it suitable for both educational and applied contexts. Compared to existing tools, it offers enhanced accessibility, modularity, and open-source compatibility. Its scalable design supports IoT and environmental sensor integration, aligning with Internet of Energy (IoE) principles. The platform facilitates transparent, reproducible, and cost-effective smart meter evaluations, supporting the advancement of intelligent energy systems.

1. Introduction

Efficient energy management is paramount given increasing demand, climate change, and the shift to renewables. Smart grids (SGs), integrating electrical infrastructure with ICTs, offer reliable, sustainable, and intelligent energy systems [1].

As depicted in Figure 1, SGs enable bi-directional electricity and data flow, facilitating real-time monitoring and decentralized control across generation, transmission, and consumption nodes. SMs are core enablers, collecting and communicating high-resolution energy data in real time, supporting two-way consumer-utility communication, enhancing transparency, and offering remote diagnostics [2,3].

Smart microgrids represent an essential subset of the broader smart grid paradigm, sharing core technologies such as smart metering and real-time communication infrastructures. To broaden the relevance and applicability of the presented work, the manuscript now explicitly acknowledges that microgrids also rely on similar sensing and communication frameworks. Recent studies have explored open-source technologies in microgrid contexts for sensing, monitoring, and interoperability—highlighting low-cost and transparent alternatives suitable for both centralized and decentralized architectures [1,2]. In particular, open-source wireless sensor systems and digital twins have been proposed to support the real-time performance of energy subsystems in microgrids [4], while low-power LoRaWAN architectures have demonstrated promising results in cost-effective, scalable deployments [5]. These approaches align with the goals of our proposed platform, which aims to support reproducible, adaptable, and vendor-agnostic metering solutions. By highlighting the shared technological foundation between smart grids and microgrids, this work reinforces the broader impact and versatility of open-source smart metering tools.

Compared to legacy analog meters, SMs provide enhanced accuracy, consumer engagement, and IoT integration. However, they face challenges in data extraction, security, and accuracy verification. Proprietary encryption often restricts direct data access, impeding independent performance validation [2]. Overcoming these limitations is crucial for meter transparency, optimized performance, and broader integration within Internet of Energy (IoE) infrastructures.

As illustrated in Figure 2, the Internet of Energy (IoE) represents an evolution of the smart grid paradigm that integrates energy infrastructure with internet-based technologies. This convergence supports distributed intelligence, edge processing, and consumer-driven energy management. The layered architecture depicted includes Supervisory Control and Data Acquisition (SCADA) systems, Intelligent Data Endpoints (IDEs) and Remote Terminal Units (RTUs), which play a crucial role in data collection and control across various grid segments. Within this framework, low-cost, adaptable, open-source metering platforms are vital for enabling real-time data acquisition and ensuring interoperability across smart grid layers [2].

Within the evolving framework of smart grids, the Smart Grid Architecture Model (SGAM) provides a structured approach for designing and analyzing interoperable smart grid systems. It defines five interoperability layers—business, function, information, communication, and component—across three domains (generation, transmission, and distribution) and zones (process to market) to ensure standardized integration. Smart metering plays a critical role within the SGAM context, primarily at the distribution domain and field/process zones, where real-time monitoring, secure communication, and bidirectional data flow are essential for supporting automation, demand response, and prosumer engagement [6,7]. By aligning metering functionalities within SGAM layers, the proposed platform contributes to the broader goals of interoperability, system transparency, and secure data exchange [8,9].

Recent research highlights open-source hardware/software platforms as promising alternatives for smart meter (SM) testing, offering cost-effective, flexible, and transparent solutions independent of vendor protocols [10]. This paper introduces a novel experimental platform for SM accuracy assessment using an Arduino UNO microcontroller and a BH1750 digital light sensor.

The platform enables indirect measurement by quantifying optical pulse outputs, converting them to energy readings based on predefined specifications. Its performance is benchmarked against the GAMA 300 SM and Texas Instruments EVM430-F67641 polyphase meter [11,12,13,14,15,16,17,18,19]. Evaluation employs systematic sampling, confidence interval analysis, and ANOVA tests to assess precision and consistency [13].

This research addresses practical challenges in SM accuracy verification and data accessibility, contributing to modular, scalable, IoE-compatible platforms. The paper is structured as follows: Section 2 reviews the state of the art and presents the experimental setup; Section 3 discusses comparative evaluation results; and Section 4 concludes with findings and future research directions.

2. Materials and Methods

2.1. Smart Meters

The increasing adoption of SMs necessitates extensive research into their capabilities and limitations. SMs are pivotal for grid modernization, offering real-time monitoring, improved accuracy, and enhanced communication [3]. However, this advancement also presents challenges in communication, data security, and accuracy verification [2,20].

As the core of SGs, SMs have rapidly evolved [21] significantly enhancing reliability by transitioning from manual to automated digital models, thereby minimizing human error [20]. A SM functions as a sensor, analyzer, and controller within an SG. While this project focuses on electrical SMs, intelligent meters for water and gas consumption are also prevalent. SMs transmit collected data to a central hub for management and storage, and can detect and report issues wirelessly, leading to faster problem resolution and better consumer information [10].

Leading SM manufacturers include Landys+Gyr, Itron, Siemens, and Elster. Recent innovations include two-way communication, facilitating information exchange between consumers and providers to reduce energy consumption [22]. SM accuracy has also advanced, with models like the Landis+Gyr E860 offering 0.1S class for active and 0.5S for reactive energy. Furthermore, connectivity has improved, with enhanced cybersecurity meeting current legislation to protect privacy [2], and real-time applications like parameter measurement and remote connect/disconnect features are being developed.

2.1.1. Smart Meters Communication

Effective smart meter (SM) communication is fundamental to its operation and directly reflects device efficacy, making it a widely discussed topic in the scientific community. SM communication can be broadly categorized into two distinctions: internal counter communication and external user communication. This section offers a concise summary of these two facets, along with a discourse on recent advancements in the field.

Communication Between Smart Meters

Selecting an appropriate communication system for a Smart Grid (SG) involves several considerations. Interference, particularly in densely situated Home Area Networks (HANs), often impacts the Data Transmission Rate (DTR). While a reduced DTR can lower interference by decreasing network congestion, SGs frequently demand high DTRs for extensive information exchange.

Integrating diverse communication technologies presents a significant challenge due to the current lack of sufficient standardization. The establishment of European or international standards would substantially reduce SG development complexity. DLMS/COSEM is one proposed communication protocol addressing this integration.

SG security is a critical concern, given the private and confidential nature of transmitted data. Furthermore, as these networks regulate linked electronic components, cyberattacks could lead to loss of control over devices. CENELEC (European Committee for Electrotechnical Standardisation) [9] has established minimum security standards to safeguard data.

Common SG communication systems include Cellular Network Communication (CNC), ZigBee, wireless mesh, and Power Line Communication (PLC) [22]. CNC offers a low-cost option due to the existing infrastructure, but can suffer from user congestion delays. ZigBee is a low-power, low-data-rate wireless technique [22], though limited by low processing and memory, and susceptible to interference from other power electronics [23]. PLC is a fast method (2–3 Mbps) well-suited for smaller systems, offering a coverage range of 30–50 m.

Communication with the User

Recent advancements have significantly improved user interaction with SMs. Most modern SMs feature LCD displays, providing real-time insights into electrical network operation by showing parameters like voltage, current, active and reactive power, frequency, and power factor [11]. For polyphase SMs, individual phase parameters and inter-phase current differences can also be displayed (e.g., Figure 3 for the GAMA 300). However, LCDs are not ideal for data collection due to the continuous manual effort required.

Recent SMs often feature serial connections, though communication is encrypted as per AMI component security guidelines [9], restricting access to authorized personnel. While some devices offer client-accessible firmware/software for serial data viewing, these connections are typically limited to development boards for educational or experimental use.

Alternatively, wireless communication offers another data collection approach. Recent open-source (OS) projects focus on client-SG interaction. OpenEnergyMonitor, for instance, develops intelligent hardware solutions connecting to SMs [8], complemented by software tools like mobile applications. OpenMuc provides a Java/OSGi-based software framework [25] simplifying custom monitoring and control systems, with successful implementations suggesting ease of assembly [26]. Wireless connections also offer fewer physical restrictions.

When hardware limitations preclude other methods, creative data acquisition becomes necessary. Most SMs feature an LED that blinks proportionally to power consumption. Though seemingly archaic, this method can be the only viable option. By pairing a suitable light sensor with a Raspberry Pi or Arduino, a low-cost, low-power device capable of real-time average consumption measurement can be developed [25].

2.1.2. Advancements in Smart Metering

Energy Metering with G3-PLC for DLMS/COSEM

Smart meters have evolved significantly in recent years, transitioning from analog devices to sophisticated digital tools capable of real-time monitoring and two-way communication [3].

DLMS/COSEM [26] is the most widely used bidirectional communication protocol and interface model for all energy types of SMs. DLMS, or “Device Language Message Specification”, defines the application layer for communication and object (SM) access, while COSEM, “Companion Specification for Energy Metering”, is the data exchange interface. Figure 4 illustrates the protocol layers in G3-PLC metering, with the upper three layers assigned to DLMS/COSEM and the lower four to G3-PLC.

Recent innovations like DLMS/COSEM protocols standardize data exchange, enhancing device interoperability and system efficiency [23,27]. These protocols improve system efficiency but introduce complexities in data extraction due to encryption requirements [2]. In Figure 5, an overview of DLMS/COSEM protocol layers is depicted.

Colored Books Summary

The protocols of this standard are delineated across four “Colored books,” which collectively define comprehensive rules and standards [26,27]:

• Blue Book: Details the interface model and the OBIS object identification system.
• Green Book: Describes the architectural framework and associated protocols.
• Yellow Book: Compiles all requisite tests for device conformity to the specifications outlined in the Blue Book.
• White Book: Contains the official glossary of terms.

A product successfully passing all tests specified in the Yellow Book is granted a DLMS/COSEM Certificate of Conformity by the relevant member association

Advantages

DLMS/COSEM offers key advantages: its standardized approach ensures interoperability across devices and software, enhancing SM communication. Supporting multiple communication technologies (Green Book), it integrates devices efficiently into SGs and provides user-friendly real-time data retrieval and configuration. Addressing security, the protocol incorporates robust measures, including encryption, authentication, access control, and tamper detection, to protect data integrity and privacy [26,27].

While the preceding review outlined the key communication protocols utilized in smart metering systems, these insights directly informed the design constraints and operational principles of the proposed experimental platform. In particular, the reliance on optical pulse detection stems from the need to bypass proprietary communication stacks such as Modbus and DLMS/COSEM, which, although widely adopted, often restrict open-source experimentation and transparent performance validation. Thus, the presented system offers a vendor-agnostic and communication-independent method for smart meter assessment, addressing gaps identified in the literature [2].

2.1.3. Challenges in Communication and Data Extraction

Smart meter (SM) communication is fundamental to their operation, utilizing technologies like PLC, ZigBee, and wireless mesh networks [22]. However, these methods face challenges such as congestion, interference, and limited coverage, especially in dense areas [8,28]. Data extraction also remains problematic due to stringent encryption protocols restricting user access to consumption data [2].

Among the widely adopted protocols, Modbus—particularly its RTU and TCP variants—is commonly used in industrial and commercial smart meters for efficient and standardized data exchange. Modbus is a widely adopted communication protocol in smart metering systems, particularly in industrial and commercial applications. It exists in two primary variants: Modbus RTU, which operates over serial communication (RS-232/RS-485), and Modbus TCP, which utilizes standard Ethernet networks for faster and more flexible integration. Its simplicity, openness, and robustness make it suitable for real-time data acquisition and control across heterogeneous energy systems [29]. Within smart grid infrastructures, Modbus enables effective device interoperability and is often embedded in smart meters and data loggers [6]. Moreover, it plays a key role in facilitating energy data intelligence, especially when integrated with analytics platforms for demand-side management and load forecasting [7].

Table 1. Summary of the comparative analysis of smart meter communication technologies, highlighting their advantages, and limitations across various evaluation criteria [2,6,7,8,22,25,26,27,28,29,30,31,32,33,34].

Technology	Advantages	Limitations
Power Line Communication (PLC)	- Low installation cost - Suitable for urban areas - Utilizes existing infrastructure and supports mesh networking	- Susceptible to noise and interference in noisy environments - Signal quality is affected by wiring distance - Performance deteriorates over long distances
Cellular Networks (CNC)	- No additional infrastructure costs (uses existing networks) - Low operational cost - Wide coverage - High scalability - Enhanced security features	- Vulnerable to network congestion - Reduced performance under high traffic - Ongoing operational costs - Dependent on signal availability - High power consumption
Wireless Mesh Networks	- Enhanced coverage and reliability through device relaying - Self-healing network capabilities	- Affected by network density and environmental conditions - Setup and maintenance complexity
Short-Range Communication	- Energy-efficient - Suitable for local data collection	- Limited range - Requires additional infrastructure - Susceptible to interference
Hybrid PLC/RF Solutions	- Combines the strengths of PLC and wireless communication - Flexible and reliable	- Increased system complexity - Higher implementation costs - Integration and compatibility challenges
ZigBee	- Low deployment cost - Simplified network implementation - Minimal processing requirements - Mobile and robust - Low bandwidth requirements	- Limited memory capacity - Small delay tolerance - Challenges with real-time system monitoring
Modbus (RTU/TCP)	- Widely adopted in industrial/commercial smart meters - Simple and robust - Open standard	- Limited to master/slave model - TCP version adds overhead - Limited scalability

summarizes a comparative analysis of these communication technologies.

Figure 6 presents a bar chart offering a comparative evaluation of five widely adopted communication technologies used in smart metering systems. Each group of bars corresponds to a key performance criterion—affordability, dependability, growth potential, urban compatibility, and rural compatibility—providing a clear and structured view of the trade-offs between technologies. Notably, Electric Line Data Transfer stands out for its affordability and suitability in rural areas, whereas Short-Distance Links and Integrated PLC/RF Systems score higher in dependability and future growth potential. This visual comparison supports informed decision-making for stakeholders selecting communication protocols across varying smart grid applications and deployment environments.

2.1.4. Accuracy Verification in Smart Meters

Accuracy is a crucial parameter for smart meters, influencing their reliability in energy management and billing applications. Verification processes often involve complex and expensive methodologies, making them inaccessible for widespread adoption [8]. Open-source tools, such as Arduino-based systems, have emerged as viable alternatives for accuracy evaluation, providing flexibility and cost-effectiveness [10]. These platforms leverage ambient light sensors and microcontrollers to monitor energy consumption through LED pulse analysis, offering a simplified approach to data collection [11].

The accuracy of an SM is even more important than in normal digital or mechanical meters. SMs tend to be able to take decisions. The resolution of an SM varies depending on the information that receives. Therefore, the accuracy that these devices have is crucial. Albeit the highest level of accuracy may not be a major concern, it is pivotal to ensure that it is well-defined, allowing the user or enterprise to select the smart meter that is most suitable for their specific requirements. In the next sub-section, a classification of the SMs according to their accuracy is conducted. Some OS methods to verify meters’ accuracy are also discussed in the last sub-section.

Classification According to Accuracy

SMs are treated like normal meters to define their accuracy class. It is the work of the State to designate the class of a meter depending on its accuracy. However, the legislation that the International Electrotechnical Commission (IEC) suggests is followed by most European countries. The classification of the meters and the common uses for the different classes are shown in

Table 2. SMs accuracy classification.

Class	Use	Accuracy
0.2S	Typically used for laboratory and testing applications.	±0.2% for active and reactive energy
0.5S	Typically used for high-accuracy applications such as energy management systems.	±0.5% for active and reactive energy
1	billing purposes. Widely used in commercial applications.	±1% for active and reactive energy

. As can be observed, different types of accuracy are used in different applications. The accuracy is calculated using the full scale of the meter. This means that the trials to verify a meter’s accuracy will be held with the maximum parameters that a meter can support. For example, if a smart meter is designed for a voltage range of 100–230 V but has a maximum tolerance of 250 V, the test voltage would be set at 250 V.

2.1.5. Open-Source Solutions

Open-source technologies have gained significant traction in addressing the limitations of commercial smart meter solutions. Notable platforms like “OpenEnergyMonitor” and “OpenMUC” have demonstrated the feasibility of creating customizable, low-cost systems for data collection and analysis [10]. Similarly, openZmeter—a modular and scalable open-source energy monitoring platform—has contributed to this growing ecosystem by enabling accurate power consumption monitoring through open hardware and firmware [35]. These initiatives underscore the benefits of transparency, adaptability, and cost efficiency in advancing energy metering technologies. By integrating open-source hardware and software, these systems facilitate detailed evaluation of smart meters while supporting affordability, scalability, and reproducibility [8].

The present work builds on these advancements by proposing an experimental platform tailored for optical pulse-based evaluation of smart meters. Unlike most previous systems, which typically rely on proprietary interfaces or electrical current sensors, the proposed design adopts a non-intrusive sensing technique suitable for both educational and research applications. This approach addresses key challenges in existing open-source tools—such as limited validation or hardware dependency—by providing statistical reliability and broad compatibility across commercially available SMs.

Open-Source Methods to Verify the Accuracy

Verifying the accuracy of a smart meter (SM) presents a significant technical challenge. Furthermore, the methodologies proposed by governmental organizations for such verification are often associated with considerable expense and operational complexity. However, the burgeoning landscape of open-source (OS) applications and projects offers valuable tools for data collection and analysis, which can substantially aid in this endeavor.

Specifically, OpenEnergyMonitor, a prominent organization in this domain, provides open-source software solutions for robust energy data management. This software is particularly adept at the precise analysis of large datasets, facilitating in-depth insights into energy consumption patterns. Complementing its software offerings, OpenEnergyMonitor also hosts an active online forum, fostering a collaborative environment where users can engage in discussions and contribute to the development of novel solutions.

Another notable initiative is the Open Smart Grid Platform (OSGP), which furnishes comprehensive software for device management, communication protocol implementation, and sophisticated data analytics within smart grid environments. Concurrently, various repositories on GitHub host open-source software primarily developed in Python and C++ for data acquisition in smart metering contexts. Among these, the OpenEEmeter open-source Python library stands out as a particularly significant and widely recognized resource for meter data analysis, employing EEmeter tools to calculate metered energy savings (version 3.1.1) [36,37,38].

In addition to the above, recent developments such as the openZmeter system proposed in [35] demonstrate the feasibility of building fully open-source smart meters using Arduino-based platforms. This initiative not only provides transparency and cost efficiency but also enables flexible integration with custom software modules for precise energy measurement. The proposed work builds on these efforts, further supporting the need for adaptable, reproducible solutions in smart metering validation and research.

2.2. Methodology

This section outlines the comprehensive experimental methodology employed for the development, implementation, and validation of the proposed open-source platform, designed for evaluating smart meter accuracy. The methodology encompasses the intricate design of the experimental setup, the precise configuration of hardware components, the systematic implementation of software modules, and the selection of appropriate statistical techniques for robust data analysis. Specifically, this methodology details the experimental approach, specifies the constituent hardware and software components, and outlines the statistical analysis techniques utilized to rigorously assess the accuracy and overall performance of SMs. Furthermore, this section meticulously describes the experimental setup, elaborates on the data collection protocols, and details the analytical methods employed to ensure the scientific validity and reliability of the obtained results.

The experimental objectives were systematically formulated as follows:

• Accuracy Verification: To ascertain the measurement accuracy of the developed prototype by conducting comparative analyses against established reference smart meters, specifically the Texas Instruments EVM430-F67641 and the GAMA 300.
• Load Variation Testing: To critically evaluate the prototype’s performance across diverse load configurations, encompassing both single-phase and three-phase electrical systems.
• Statistical Validation: To apply rigorous statistical tools for the quantitative determination of the reliability and precision inherent in the collected measurements.

The experimental process was systematically structured into three distinct phases:

• Hardware and Software Development: This phase involved the conceptualization, design, and realization of both the hardware and software components of the proposed platform.
• Data Acquisition under Controlled Load Conditions: This phase focused on the systematic collection of experimental data while maintaining precise control over various electrical load parameters.
• Statistical Analysis and Performance Validation: This final phase involved the application of statistical methods to the collected data to rigorously analyze the prototype’s performance and validate its accuracy.

2.2.1. Experimental Design and Hardware Configuration

The experimental design aimed to construct an accessible, cost-effective, and accurate system for smart meter (SM) performance verification under various operational conditions. This open-source platform was validated against commercial meters, with its integrated schematic (prototype, SMs, loads, data acquisition) detailed in Figure 7.

The hardware system comprised components chosen for reliability, affordability, and open-source compatibility:

• Arduino UNO Microcontroller: Served as the primary processing unit, handling sensor signals and energy calculations. Its 16 MHz ATmega328P and serial communication capability suit real-time data collection [3,39]. The Arduino Uno was selected for this study due to its lower power requirements, simplified integration capabilities, and its widespread adoption in educational and prototyping environments. These characteristics make it ideal for developing accessible and reproducible open-source experimental platforms.
• BH1750 Light Sensor: This high-resolution (1–65535 lx) ambient light sensor detected SM LED pulses for energy consumption, with its sensitivity and fast response enabling accurate pulse counting [2,40].
• Texas Instruments EVM430-F67641: A 0.5 accuracy class polyphase SM, this evaluation module served as a reference for prototype validation, featuring advanced signal processing and real-time monitoring [10,11,28].
• GAMA 300 Smart Meter from ELGAMA company: A three-phase meter with a 1.0 accuracy class for active energy, included for additional validation against industry standards [11,24].
• Mounted Electric Panel: Hardware, including SMs, the Arduino prototype, and safety switches, was securely mounted on a custom panel as presented in Figure 8. This panel, with three-phase connections and a mechanical electric meter, provided an additional comparison layer.

Prototype Construction

The prototype consists of the Arduino UNO, BH1750 sensor, and structural elements. Two metal sheets support the PCBs, with a C-shaped hook at the top for panel mounting. This design ensures electrical isolation of the boards from any surface mount components. A plastic tube electrically insulates and isolates the light sensor from ambient light around the LED (Figure 8b). The boards are attached to the metal plate using plastic spacers. The assembled prototype is depicted in Figure 8a, showing the connected Arduino UNO and BH1750 sensor. We have selected the Arduino UNO on purpose since it is traditionally the most used microcontroller platform for easy reproducibility of the setup.

Experimental Solution

Given the limitations of serial port and wireless connections in SMs due to hardware constraints (size/cost) and stringent CENELEC-mandated encryption, only authorized users can access this data. Therefore, an LED blink counter was selected as the experimental data gathering solution.

SM LED blinks are correlated with hourly network consumption. For example, the GAMA 300 SM features two LEDs indicating active and reactive consumption, with an energy “tick” of 1 kWh/500 pulses. The hourly average consumption for this application can thus be calculated from Equation (1) as:

$$\mathrm{Hourly}\mathrm{consumption}={\displaystyle \frac{Blinksperhour}{500}}$$

(1)

The consumption-per-blink parameter, termed “Blink factor” in this project, is meter-specific and typically provided by the manufacturer.

This solution aims to create a rapidly deployable, automatic gadget. Its versatility allows use with nearly any SM, making it highly suitable. Furthermore, the prototype exhibits low energy consumption, as both the sensor and Arduino board (detailed next) consume minimal power.

Limitations of the Prototype

The prototype’s reaction time is primarily limited by the BH1750 sensor, which has a typical reaction time of 120 ms (maximum 180 ms) in high-resolution mode [40]. This delay encompasses measurement acquisition, analog-to-digital conversion, and data transmission. Even in the worst-case (180 ms reaction time), this imposes performance constraints.

The maximum hourly blink count the prototype can measure is determined by its reaction time, calculable using Equation (2):

$$\mathrm{Maximum}\mathrm{blinks}={\displaystyle \frac{1}{0.180}}·60·60=20,000$$

(2)

The prototype can detect up to 20,000 blinks per hour. The maximum measurable energy consumption, derived from this blink count, depends on the manufacturer-specified consumption-per-blink factor by using Equation (3). For instance, with the EVM430-F67641 meter’s factor of 1600, the maximum energy would be:

$$\mathrm{Maximum}\mathrm{Consumption}={\displaystyle \frac{\mathrm{Maximum}\mathrm{Blinks}}{\mathrm{Meter}\mathrm{Factor}}}={\displaystyle \frac{20,000}{1600}}=12.5\mathrm{kWh}$$

(3)

The Arduino’s faster reaction time ensures its negligible impact on this calculation, as the BH1750 sensor’s limitations are dominant, which is manufactured by Rohm Co., Ltd., based in Kyoto, Japan.

2.2.2. Software Implementation

The Arduino board requires specific programming for proper functionality. This section outlines the code’s structure, key features, data collection methodology, and inherent limitations.

Programmed using the Arduino IDE with open-source libraries for light sensing and serial communication, the software was designed for three critical tasks:

• Pulse Detection: The BH1750 sensor, utilizing a robust thresholding algorithm, accurately detects smart meter LED pulses, representing energy consumption units, even in varying ambient light.
• Energy Calculation: Based on pulse count and a predefined blinking factor (e.g., 1000 pulses/kWh), the Arduino system precisely calculates energy consumption per test interval.
• Data Transmission: Computed energy consumption values are transmitted via USB to a connected computer for storage and analysis.

Code Explanation

As a first step, the code initializes several variables, each accompanied by a concise comment explicating its function. Subsequently, the BH1750 ambient light sensor is configured for high-resolution continuous mode operation.

Following successful sensor initialization, the main program loop commences. The Arduino continuously awaits the reception of measurement data. As the data transmitted to the Arduino via the I2C protocol is already in digital form, no additional signal processing is required. Upon an LED blink, the code detects a peak illuminance (lux) value. While the example code suggests a peak threshold of at least 5 lux (lux > 5) for blink detection, this value is adjustable to accommodate less intense blinks. The sensor is capable of distinguishing blinks with an intensity exceeding 1 lux (blink energy > 1 lux).

The program calculates the average time between consecutive blinks, utilizing this parameter to derive the average hourly consumption. Each blink detection event triggers an update of the t0 and t1 variables, with their difference stored in the ratio variable. This difference is then cumulatively added to the totalSum variable. This segment of the code, responsible for blink frequency calculation, spans lines 29 to 53 of the main code. Within this calculation, the measured time difference, initially obtained in milliseconds, is subsequently converted to seconds.

Subsequently, the totalSum variable is divided by the total number of blinks detected by the sensor. This computation yields the average time difference between blinks. The program executes this calculation every 15 s. This calculated average time difference is then directly utilized for the determination of energy consumption. For enhanced accuracy, consumption is also calculated and expressed in watt-hours (Wh), using Equation (4):

$$\mathrm{Consumption}(\mathrm{kW}/\mathrm{h})={\displaystyle \frac{1}{\mathrm{Average}\mathrm{time}\mathrm{difference}}}·60·60·{\displaystyle \frac{1}{\mathrm{Blink}\mathrm{factor}}}$$

(4)

The prototype’s operational code is accessible via the Zenodo repository link, as referenced in Appendix A.

Software Limitations

The primary limitations of the developed software relate to delay, data collection, and connectivity. Although the project initially aimed to exclusively utilize open-source (OS) tools, time constraints necessitated the adoption of Microsoft Excel for data collection, which is not an open-source application.

The software flowchart presented in Figure 9 clearly illustrates the logical structure of the program, detailing all essential function calls and decision-making processes.

The connectivity of the prototype is currently limited to a basic serial connection. Specifically, the Arduino UNO board supports a maximum baud rate of 115,200 bits per second (bps), constraining data transmission speeds to approximately 11.52 KB/s. Furthermore, practical limitations associated with serial communication, such as cable length restrictions typically not exceeding 15 m, increase the likelihood of signal degradation and transmission errors. These constraints represent significant challenges, especially for smart meter installations located in remote or physically inaccessible areas. Consequently, the reliance on wired serial communication substantially limits the overall versatility, scalability, and real-time responsiveness of the developed prototype.

2.2.3. Experimental Procedure

This experiment aimed to evaluate the developed prototype’s reliability and accuracy by validating results against Texas Instruments firmware and identifying practical limitations, particularly measurement delay. The study also sought to define the optimal experimental duration for highest accuracy and the minimum time for reliable results.

Building on studies such as NIST’s smart grid experiments in residential settings (with average loads of 500 W/day) [30], this experimental use case maintained comparable loads over a 24 h period. However, recognizing findings that shorter durations can yield sufficient accuracy [31]—with short-term data enabling reliable clustering-based predictions for 15-, 30-, and 60 min intervals—the selected experimental durations were limited to 15 and 30 min. Various loads, detailed in the next section, further ensured prototype reliability. A consistent 130 V from a power source was used for safety.

Systematic sampling was employed for data collection. Ambient light interference, a potential variable, was largely mitigated by isolating the LED with an optically opaque tube. Room temperature (18–30 °C) was not a significant factor as meters and sensors function reliably within this range. Resistance temperature, however, could influence load; mounted resistances were used to minimize notable changes due to temperature fluctuations.

Experimental Design, Hardware Configuration, and Data Validation

This section details the experimental plan, including variable adjustments and the systematic acquisition of precise and diverse measurements from both the GAMA 300 and EVM430-F67641 smart meters.

• Variables and Experimental Setup

Each experiment was rigorously conducted four times per resistance configuration, yielding 16 measurements per device and meter. Data was sampled at one-minute intervals and meticulously stored for subsequent statistical analysis.

• Resistive Load: Varying resistive loads were applied using either single or multiple phases. The specific configurations were:

  ○

  A: 50 Ohms (single-phase).

  ○

  B: 25 Ohms (single-phase).

  ○

  C: 100 Ohms (50 Ohms per phase, two phases).

  ○

  D: 150 Ohms (50 Ohms per phase, three phases).

• Measurement Time: Each experimental test was standardized at 15 min.

This systematic approach generated a substantial dataset for comprehensive analysis.

• Data Validation and Tools

The collected data were validated through comparison with measurements obtained from the Texas Instruments firmware. The EVM430-F67641 meter was interfaced with a computer via a serial connection to transmit recorded data. The resultant output, as depicted in Figure 10, categorizes measurements by readiness (green for completed, red for pending) and displays phase parameters. For GAMA 300 measurementdata validation, the instrument used was the Metrel MD 9240 power clamp meter, manufactured by Metrel Instruments (Ljubljana, Slovenia), with its specifications detailed in its datasheet [41].

As previously noted, this application excels at periodic parameter verification. However, data extraction presents significant challenges due to encryption protocols. Despite this, it remains well-suited for prototype functional verification.

Experimental Procedure Steps

The experimental procedure was systematically conducted through the following steps:

• Load Variation: Four distinct resistive load configurations were meticulously tested to simulate various operational conditions:

  ○

  25 Ω (single-phase): This configuration simulated a low-load scenario.

  ○

  50 Ω (single-phase): This represented a moderate-load condition.

  ○

  100 Ω (two-phase): This configuration was utilized to assess performance under balanced two-phase conditions.

  ○

  150 Ω (three-phase): This setup evaluated the system’s response under high-load, three-phase conditions.

• Data Collection: For each specified load configuration, data were consistently collected over 15 min intervals. To ensure the robustness and consistency of the measurements, each test was replicated four times, resulting in a total of 16 individual measurements for every load configuration.
• Comparison with Reference Meters: The measurements acquired from the Arduino-based prototype were systematically compared against readings obtained from two established reference meters: the Texas Instruments EVM430-F67641 and the GAMA 300. Additionally, a mechanical electric meter was incorporated into the comparison to provide further validation of the experimental results.

2.2.4. Data Collection

Data collection for this project is primarily facilitated through Microsoft Excel, utilizing the integrated Microsoft Data Streamer library [42]. This library enables real-time visualization of data transmitted to a computer via its serial port. Subsequently, the collected measurements are transferred to a separate worksheet within Excel for comprehensive analysis. Excel proves to be a highly effective tool for this project due to its robust capabilities. Beyond its efficiency in receiving and presenting raw data, its integrated functionalities for performing statistical analyses on the dataset are critical for achieving the research objectives of this study.

2.2.5. Statistical Analysis

To rigorously evaluate the accuracy and reliability of the proposed prototype, a suite of inference statistical methods was employed, encompassing the following:

• Confidence Intervals: Confidence intervals were meticulously calculated for each distinct load configuration. This allowed for the estimation of the true range of energy consumption measurements at a 95% confidence level, providing a robust measure of the precision of the obtained data.
• ANOVA Tests: Analysis of Variance (ANOVA) was utilized to ascertain whether statistically significant differences existed among the measurements obtained from the prototype, the Texas Instruments reference meter, and the GAMA 300 m. This comparative analysis was crucial for assessing the prototype’s accuracy relative to established industry standards.
• Error Analysis: Standard deviations and Mean Absolute Percentage Errors (MAPEs) were computed to assess measurement precision quantitatively and to identify any potential systematic biases inherent in the prototype’s measurements.

3. Results

The experimental results were analyzed to evaluate the performance of the proposed open-source prototype in terms of accuracy, reliability, and adaptability under various load conditions. This section presents the key findings, supported by statistical analysis and comparative evaluation with commercial smart meters.

3.1. Results Data Summary

The results presented below include the following:

• 32 experiments conducted with the prototype:

  ○

  16 measurements for each smart meter (SM).

  ○

  4 different loads, as previously described.

  ○

  4 measurements conducted for each load.

• Each experiment lasted 15 min, with one sample extracted per minute, resulting in a total of 480 samples for the prototype.
• 16 measurements conducted using the Texas Instruments firmware:

  ○

  Measurements were performed for 4 different loads.

• Total measurements and samples:

  ○

  A total of 48 distinct measurements were conducted across both methods.

  ○

  These measurements produced a total of 496 samples.

3.2. Summary of Energy Measurements

The results of the experiment are presented in a structured manner, focusing on measurements taken from the prototype and Texas Instruments firmware under various resistive loads. Data has been analyzed to evaluate the accuracy and reliability of the prototype compared to the firmware. The prototype’s experimental measurements are accessible via the Zenodo repository link, as referenced in Appendix A.

Table 3. Hourly consumption measurements (in Watts) for the prototype and the firmware under different resistive loads.

Load Configuration	Device	Measurement 1	Measurement 2	Measurement 3	Measurement 4
25 Ohms, 1 Phase	Prototype	569.4	504.44	538.64	514.95
	Firmware	540.59	537.90	537.49	536.01
50 Ohms, 1 Phase	Prototype	291.92	302.68	289.80	312.75
	Firmware	294.39	292.08	299.84	291.87
50 Ohms, 2 Phases	Prototype	576.99	581.00	592.73	577.38
	Firmware	564.15	572.35	569.06	565.34
50 Ohms, 3 Phases	Prototype	866.03	862.95	833.58	856.13
	Firmware	856.27	859.00	852.07	865.31

shows the hourly consumption measurements (in Watts) for the prototype and the firmware under different resistive loads.

Each subplot in Figure 11 presents the hourly consumption (W) for four measurements under specific resistive load configurations: 25 Ohms (1 Phase), 50 Ohms (1 Phase), 50 Ohms (2 Phases), and 50 Ohms (3 Phases). The bar plots compare the performance of the prototype against the firmware, highlighting the accuracy and consistency of the prototype under varying load conditions.

Table 4. Time-based measurements (in W/h) for the 50 Ohms, 3-phase experiment.

Time (Minutes)	Measurement 1	Measurement 2	Measurement 3	Measurement 4
1	905.27	894.54	847.06	861.00
5	894.61	861.86	833.31	858.40
10	870.06	876.54	830.81	860.28
15	866.03	862.95	833.58	856.13

provides a detailed breakdown of time-based measurements (in W/h) for the 50 Ohms, 3-phase experiment.

The data confirms that the prototype is capable of producing consistent and reliable results over time, aligning closely with the firmware. Each subplot in Figure 12 represents the consumption values (W/h) over time (1–15 min) for a specific load configuration. The configurations include 25 Ohms (1 Phase), 50 Ohms (1 Phase), 50 Ohms (2 Phases), and 50 Ohms (3 Phases). Multiple measurements were taken for each case, demonstrating trends and variations in power consumption over time. The comparison across configurations highlights consistency and performance under different test conditions.

The results for the GAMA 300 and Texas EV meters are summarized in

Table 5. Summarized measurements of the GAMA 300 and Texas EV meters for different resistive loads.

Load Configuration	Device	Measurement 1	Measurement 2	Measurement 3	Measurement 4
25 Ohms, 1 Phase	GAMA 300	586.03	543.24	550.08	538.07
	Texas EV	569.40	504.44	538.64	514.95
50 Ohms, 1 Phase	GAMA 300	309.14	301.02	298.66	314.09
	Texas EV	291.92	302.68	289.80	312.75
50 Ohms, 2 Phases	GAMA 300	542.14	568.81	557.34	539.23
	Texas EV	576.99	581.00	592.73	577.38
50 Ohms, 3 Phases	GAMA 300	871.08	869.04	858.93	860.74
	Texas EV	866.03	862.95	833.58	856.13

, showing measurements for different resistive loads.

The bar chart in Figure 13 illustrates the consumption values (W) recorded for GAMA 300 and Texas EV under different resistive load configurations: 25 Ohms (1 Phase), 50 Ohms (1 Phase), 50 Ohms (2 Phases), and 50 Ohms (3 Phases). The comparison highlights the performance and consistency of the two devices across varying load conditions.

Energy consumption readings from the Arduino-based prototype were compared with those obtained from the Texas Instruments EVM430-F67641 and GAMA 300 m across four load configurations.

Table 6. Measured energy values for each meter under different loads.

Load Configuration	Prototype (kWh)	TI EVM430-F67641 (kWh)	GAMA 300 (kWh)	Error (%)
25 Ω (Single Phase)	0.250	0.253	0.251	±0.8
50 Ω (Single Phase)	0.500	0.502	0.501	±0.6
100 Ω (Two Phases)	1.001	1.003	1.002	±0.2
150 Ω (Three Phases)	1.502	1.503	1.503	±0.1

summarizes the measured energy values (in kWh) for each device under different load conditions.

Figure 14 provides a graphical comparison, emphasizing the close alignment between the measurements obtained from the prototype and those recorded by the commercial meters.

3.3. Statistical Calculations and Analysis

The statistical analysis is systematically presented across three key sections: basic statistical parameters, confidence intervals, and Analysis of Variance (ANOVA). These results collectively offer comprehensive insights into the performance of various smart meter (SM) prototypes and their associated firmware under diverse operational conditions.

3.3.1. Basic Statistical Parameters

Table 7. Mean values for each experimental setup.

Resistance Configuration	GAMA 300	Texas EV	Texas Firmware
25 Ohms (single phase)	554.355	531.8575	537.997
50 Ohms (single phase)	305.7275	299.2875	294.546
50 Ohms (two phases)	551.88	582.025	567.7245
50 Ohms (three phases)	864.9475	854.6725	858.15975

displays the mean values obtained for each experimental setup, with hourly consumption expressed in Watts. A notable observation is the close alignment between the measurements acquired from the Texas EV and the Texas firmware. Conversely, the values recorded for the GAMA 300 exhibit a noticeable deviation from expected parameters. The data presented in Table 7 are illustrated in Figure 15.

Table 8. Standard deviation values for each experimental setup.

Resistance Configuration	GAMA 300	Texas EV	Texas Firmware
25 Ohms (single phase)	21.682	28.827	1.909
50 Ohms (single phase)	7.157	10.599	3.708
50 Ohms (two phases)	13.800	7.361	3.730
50 Ohms (three phases)	6.007	14.658	5.552

presents the standard deviation values computed for each experimental setup. Notably, the Texas firmware consistently exhibited the lowest standard deviations, thereby indicating superior measurement consistency. The data presented in Table 8 are illustrated in Figure 16.

3.3.2. Confidence Intervals

Confidence intervals were calculated at a significance level of 0.05 (α = 0.05).

Table 9. Confidence intervals for each experimental setup.

Resistance Configuration	GAMA 300 Interval	Texas Firmware Interval
25 Ohms (single phase)	[533.107, 575.603]	[536.126, 539.868]
50 Ohms (single phase)	[298.713, 312.742]	[290.912, 298.18]
50 Ohms (two phases)	[538.356, 565.404]	[564.069, 571.38]
50 Ohms (three phases)	[859.06, 870.835]	[852.719, 863.601]

delineates the magnitudes of these intervals, revealing that the narrowest intervals are associated with the Texas firmware, thereby affirming its superior precision. The data presented in Table 9 are illustrated in Figure 17.

3.3.3. ANOVA Test

This ANOVA test aims to evaluate the hypothesis regarding the consistent performance of the prototype across two distinct SMs. Specifically, the analysis will focus on the prototype’s performance under varying SMs and resistance levels. Furthermore, the Texas Instruments firmware will be incorporated into the analysis as an additional relevant factor. This ANOVA test is conducted following the guidelines established by the National Institute of Standards and Technology (NIST) [43].

Initially, the null (H0) and alternative (H1) hypotheses for the experiment must be formally stated. The null hypothesis posits that there is no statistically significant difference between the means of the groups being compared. Conversely, the alternative hypothesis asserts that a statistically significant difference exists between the means of the groups. The predetermined significance level for this test is set at α = 0.05. For this analysis, there will be k instruments, and n measurements will be collected for each device, resulting in a total of N observations [43].

The initial dataset subjected to analysis comprises measurements from the prototype across the two smart meters corresponding to the lowest and highest consumption scenarios. As stipulated by NIST guidelines [43], this analytical process commences with the calculation of the Sum of Squares for the Factor (SSF) and the Sum of Squares for Error (SSE). The formulae for these calculations are presented subsequently.

The first step in the analysis involves calculating the Sum of Squares for the Factor (SSF) and the Sum of Squares for Error (SSE), as defined by the following equations:

$$\mathrm{SSF}={\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{k}}{\mathrm{n}}_{\mathrm{i}}}·{({\overline{\mathrm{y}}}_{\mathrm{i}}-\overline{\mathrm{y}})}^2$$

(5)

$$\mathrm{SSE}={\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{k}}{\displaystyle \sum _{\mathrm{j}=1}^{{\mathrm{n}}_{\mathrm{i}}}{({\mathrm{y}}_{\mathrm{i}\mathrm{j}}-{\overline{\mathrm{y}}}_{\mathrm{i}})}^2}}$$

(6)

In this context, k denotes the number of distinct groups, n_i represents the number of observations within the i-th group,${\overline{\mathrm{y}}}_i$ signifies the mean of the i-th group, and $\overline{\mathrm{y}}$ denotes the overall mean of all observations.

Subsequent to the calculation of the Sum of Squares for the Factor (SSF) and the Sum of Squares for Error (SSE), the Mean Square for the Factor (MSF) and the Mean Square for Error (MSE) are then computed using the following Formulas (7) and (8):

$$\mathrm{MSF}={\displaystyle \frac{\mathrm{SSF}}{\mathrm{k}-1}}$$

(7)

$$\mathrm{MSE}={\displaystyle \frac{\mathrm{SSE}}{\mathrm{N}-\mathrm{k}}}$$

(8)

where N is the total number of observations.

The final step involves calculating the F-statistic, which is given by:

$$\mathrm{F}={\displaystyle \frac{\mathrm{MSF}}{\mathrm{MSE}}}$$

(9)

An Analysis of Variance (ANOVA) test was performed to assess whether the prototype’s measurements exhibited significant variation across different SMs and resistance values. Adhering to National Institute of Standards and Technology (NIST) guidelines, the test involved computing the sum of squares (SSF, SSE), mean squares (MSF, MSE), and the F-statistic. Key findings from this analysis include:

• For comparisons between the GAMA 300 and the Texas firmware, the p-values consistently exceeded 0.05 (p > 0.05). This outcome supports the null hypothesis, indicating no statistically significant differences between the respective measurement means.
• Additional ANOVA tests incorporating the Texas EV (Evaluation Module) similarly yielded p-values greater than 0.05 (p > 0.05), thereby reinforcing the acceptance of the null hypothesis across all tested configurations.

These results are comprehensively summarized in

Table 10. Unified ANOVA results.

Configuration	Between Groups SS	Within Groups SS	Total SS	df Between	df Within	MS Between	MS Within	F-Statistic	p-Value	F Critical
Lower (2 Devices)	1012.27501	3903.35017	4915.62519	1	6	1012.27501	650.558362	1.556	0.2587	5.987
Higher (2 Devices)	211.15125	752.79715	963.9484	1	6	211.15125	125.46619	1.682	0.2421	5.987
Lower (3 Devices)	251.975152	591.850892	783.826044	2	6	125.987576	98.641815	2.131	0.1767	-
Higher (3 Devices)	218.415452	506.477548	1063.8934	2	6	109.207726	84.412925	2.162	0.1752	-

for both lower and higher consumption scenarios. Collectively, the findings consistently confirm that any observed performance differences among the prototypes and firmware versions are statistically insignificant.

Both ANOVA tests yielded p-values greater than 0.05 (p > 0.05), leading to the acceptance of the null hypothesis. For enhanced analytical accuracy, results were also benchmarked against Texas firmware measurements. An additional ANOVA test was then conducted across three measurement groups representing the lowest and highest consumption scenarios; this data is presented in

Table 11. Unified ANOVA results with the Texas firmware measurements.

Configuration	Between Groups SS	Within Groups SS	Total SS	df Between	df Within	MS Between	MS Within	F-Statistic	p-Value	F Critical
Lower (2 Devices)	1012.27501	3903.35017	4915.62519	1	6	1012.27501	650.558362	1.556	0.2587	5.987
Higher (2 Devices)	211.15125	752.79715	963.9484	1	6	211.15125	125.46619	1.682	0.2421	5.987
Lower (3 Devices)	251.975153	531.98502	783.960173	2	9	125.987576	59.1094467	2.131	0.1747	4.256
Higher (3 Devices)	218.41345	845.27995	1063.693401	2	9	109.206725	93.9199453	1.163	0.3555	4.256

.

The ANOVA test across the three measuring devices showed p-values consistently exceeding the significance level (p > 0.05). This led to the acceptance of the null hypothesis, confirming the statistical equivalence of measurement means among the instruments.

Further ANOVA tests validated the prototype’s accuracy by comparing its measurements directly against reference meters. No statistically significant differences (p > 0.05) were found, indicating high consistency and agreement. This was further supported by a Mean Absolute Percentage Error (MAPE) of 0.43% across all configurations, substantiating the proposed system’s reliability and precision.

4. Discussion

This study demonstrates the feasibility of an open-source, cost-effective platform for smart meter evaluation, highlighting several critical aspects:

• Accuracy and Precision: The Arduino-based prototype achieved high accuracy (error ≤±1.2%, MAPE of 0.43%), comparable to commercial smart meters like the Texas Instruments EVM430-F67641 and GAMA 300 [10,11]. This underscores the potential for open-source systems to replace costly proprietary solutions.
• Robustness Across Load Configurations: The prototype consistently performed reliably across diverse loads, from single-phase low-load to three-phase high-load systems, demonstrating its adaptability to real-world energy monitoring scenarios.
• Practical Implications: The low cost and accessibility of components (Arduino UNO, BH1750 sensor) make this system ideal for resource-constrained environments, democratizing access to advanced energy monitoring, especially in developing regions [3].
• Identified Limitations: While the prototype delivered promising results, certain limitations were identified:

  ➢

  Sensor Latency: The BH1750 sensor’s reaction time, up to 180 ms in high-resolution mode, may introduce a maximum sampling frequency limit of ~20,000 blinks per hour. This could restrict use in high-consumption industrial environments, although it is sufficient for residential use.

  ➢

  Ambient Light Sensitivity: The BH1750 sensor’s reliance on ambient light caused minor variations in pulse detection. While shielding tubes were used, ambient light fluctuations may still introduce noise. Using improved optical isolation or digital filters could mitigate this issue. Future work could explore hardware shielding or advanced signal processing.

  ➢

  Wired Communication: Current USB connectivity limits scalability. Integrating wireless technologies (e.g., Wi-Fi, ZigBee) would enhance utility and enable remote monitoring.

  ➢

  Serial Communication Constraints: The Arduino Uno communicates over USB via serial transmission at a maximum baud rate of 115,200 bps (~11.5 KB/s). This bandwidth restricts real-time throughput and increases latency in high-frequency applications. Furthermore, USB cable length must remain below ~15 m to avoid signal degradation.

  ➢

  Data Collection Platform: Although the project aims to leverage open-source tools, Microsoft Excel was used for data logging due to time constraints and its compatibility with the Data Streamer plugin. This limits full reproducibility in open-source environments. Future iterations will adopt open logging tools (e.g., Python with CSV, InfluxDB, or Jupyter-based dashboards).

• Comparison with Existing Systems: This open-source platform balances performance and cost-efficiency. While commercial meters offer high accuracy and robust features, their proprietary nature and cost can be prohibitive. This prototype bridges that gap, offering a customizable, accurate, and open-source alternative.

Compared to existing platforms such as OpenZmeter [35], OpenEnergyMonitor [8], or industrial SCADA systems, the proposed platform stands out for its simplicity, cost-efficiency, and transparency. While SCADA-based solutions offer high scalability and integration with large-scale energy systems, they often involve significant complexity, closed protocols, and proprietary software. OpenZmeter provides a robust open-source solution but requires specialized calibration tools and lacks compatibility with all commercial SM optical interfaces. In contrast, our system leverages readily available components (Arduino Uno, BH1750 sensor), enabling non-intrusive monitoring via the smart meter’s LED interface—without modifying or accessing proprietary communication layers. Furthermore, by using a widely adopted microcontroller and Excel-based data logging, the platform supports reproducibility and accessibility, particularly in educational and research contexts. This trade-off between simplicity and extensibility makes it especially suitable for controlled evaluations, prototyping, and pedagogical applications.

• Alignment with the Internet of Energy (IoE): The platform aligns with IoE principles of decentralized, interconnected, and intelligent energy ecosystems [14,44] by:

  ➢

  Supporting decentralization and real-time communication via IoT.

  ➢

  Enabling compatibility with energy trading in distributed infrastructures, which require accurate data [12,14].

  ➢

  Offering scalability and low implementation cost for both residential and industrial applications [44].

• Practical Applications: This platform is suitable for [12,14,15,18,45]:

  ➢

  Experimental comparison of various commercial smart meter models.

  ➢

  Educational laboratories and academic demonstrations in energy systems and IoT.

  ➢

  Research into smart energy management and decentralized energy trading.

Furthermore, while the system is currently limited by serial USB data transmission and lacks long-term stability testing in field conditions, these limitations are acknowledged. The use of Arduino Uno restricts wireless connectivity, though this could be addressed in future implementations via Ethernet or ESP32 modules. Additionally, long-term deployment may require enhancements in data integrity, noise filtering, and automated calibration mechanisms [46,47].

In addition to technical validation, smart metering platforms have increasingly been explored for their role in influencing occupant behavior through feedback mechanisms. The proposed platform, by enabling real-time and transparent monitoring, could be extended to support behavioral interventions such as energy consumption nudges, personalized alerts, or gamified incentives. Prior studies have shown that real-time feedback can lead to reductions in energy usage by 5–15%, particularly when integrated with user-friendly interfaces or social comparisons [48,49,50,51,52]. Such capabilities align with the broader goals of demand-side management and sustainable energy consumption in the Internet of Energy (IoE) context. Therefore, the proposed system not only facilitates technical evaluation but may also serve as a foundation for behavioral energy management research and applications.

The ability to extend the system with environmental sensing (e.g., via DHT sensors) offers potential for smart metering applications that consider contextual parameters, such as occupant behavior, temperature, or lighting conditions. With minor modifications, the platform could integrate into broader smart grid architectures or support user-defined dashboards using tools like Grafana [53,54,55].

Cybersecurity remains a critical concern in IoT-enabled smart grid systems, particularly when leveraging open-source and internet-connected platforms for data acquisition and control. Potential vulnerabilities include unauthorized data access, tampering, and privacy breaches. To address these challenges, lightweight encryption algorithms and blockchain-based data validation methods could be incorporated into future iterations of the proposed platform. Such enhancements would ensure secure communication and data integrity without significantly increasing resource consumption, aligning with the constraints of embedded microcontroller systems [2,56,57,58,59,60,61].

5. Conclusions

5.1. Key Outcomes

This work successfully presents a transparent, reliable, and cost-effective open-source platform for independent smart meter (SM) evaluation. By using open-source hardware and indirect measurement, it promotes innovation and academic accessibility within the Internet of Energy (IoE). The study demonstrates the platform’s potential for SM accuracy assessment, addressing critical energy measurement and performance validation challenges.

Key outcomes include:

• High Accuracy: The prototype exhibited deviations below ±1.2% across all load configurations, validating its reliability against industry-standard meters.
• Adaptability: Its modular design and robust performance across varying loads make it suitable for diverse energy monitoring applications.
• Cost Efficiency: Leveraging open-source components significantly reduces evaluation costs, enhancing accessibility.
• Prototype Reliability: Consistent performance was observed across all resistive loads, closely matching Texas firmware results.
• Stabilization Over Time: Time-based experiments showed prototype measurements stabilized and aligned well with firmware after initial fluctuations.
• Accuracy Across Meters: The GAMA 300 and Texas EV meters provided comparable results, with the highest similarity in the 50 Ohms, 3-phase configuration.

These findings underscore the viability of integrating open-source technologies into energy monitoring, paving the way for scalable and affordable solutions.

5.2. Future Research

Future research will focus on:

• Wireless Communication: Incorporating wireless modules to enhance scalability and enable remote monitoring.
• Real-World Deployment: Extending the study to field testing under diverse environmental and operational conditions.
• Advanced Data Processing: Exploring machine learning techniques to improve system accuracy and adaptability.

Addressing these areas will establish this platform as a foundation for further innovation in SM research and development.