Supporting Global Communications of 6G Networks Using AI, Digital Twin, Hybrid and Integrated Networks, and Cloud: Features, Challenges, and Recommendations

Abstract

The commercial deployment of fifth generation (5G) mobile communication networks has begun, bringing with it novel offerings, improved user activities, and a variety of opportunities for different types of organizations. However, there still exist several challenges to implementing 5G technology. Sixth generation (6G) wireless communication technology development has begun on a worldwide scale in response to these challenges. Even though there have been many discussions on this topic in the past, many questions remain unanswered in the present literature. The article provides a comprehensive overview of 6G, including the common understanding of the concept, as well as its technical requirements and potential applications. A comprehensive analysis of the 6G network design, potential uses, and key elements are covered. This research article delineates future study topics and unresolved challenges to stimulate an ongoing global discourse. This analysis and content of this study supports the use of different applications and services that will benefit the community in the near future using the 6G technology. Subsequently, recommendations for each problem are provided, offering solutions to unresolved difficulties where functionalities are anticipated to improve, hence enhancing the overall user experience.

1. Introduction

Due to the swift advancement of communication applications, communication technologies are experiencing successive revolutionary transformations. So far, the progress of cellular mobile communication systems has experienced five generations of development. Each subsequent generation of wireless communication systems, starting from the first generation (1G) analogue systems to the 5G online systems, progressively includes greater frequencies, wider bandwidths, and faster data rates.

The 5G base units are designed to utilize advanced technologies such as massive multiple-input multiple-output (MIMO) [1], mmWave, and ultra-dense networking (UDN) [2]. These innovations allow the ground stations to accommodate up to 64 transceiver links with a greater number of antenna components. At present, there are fully developed commercialized 5G infrastructure solutions equipped having 128 antennas [3]. HUAWEI has taken the lead in introducing a huge MIMO base station with 384 antennas. Furthermore, 5G can facilitate augmented reality (AR) and virtual reality (VR). Despite the substantial advancements it brings beyond 4G connectivity technology, 5G possesses certain limitations. Presently, there exist apps and solutions that demand superior communication reliability surpassing the potential of 5G. These include worldwide coverage, transfer of data at extremely high rates, minimal delay, densely connected networks, precise positioning, highly trustworthy and secure connections, low power consumption, high energy effectiveness, and pervasive creativity. Therefore, the limitations and challenges of the 5G technology are part of the motivation behind the need for higher and faster level of connectivity such as the 6G.

For low transmission speed, the end-to-end (E2E) latency must be below 1 millisecond, and for high speed, the latency ought to be in the microsecond range. Furthermore, to achieve high-precision placement, the exactness of the placement must be enhanced to attain centimetre-level accuracy in outdoor environments and sub-centimetre-level accuracy in indoor environments. Greater dependability is needed in a variety of innovative scenarios, including vehicle-to-everything (V2X) and mobile data centres. Energy consumption is a significant concern for many different applications; hence, it is necessary to decrease power spending and boost network energy efficiency by a factor of 100. Furthermore, a significant multitude of sophisticated applications necessitate communication systems to possess a heightened level of intellect.

The progression of wireless communication systems follows a recurring pattern with a duration of around ten years. Currently, the study on 6G is at its initial phase of investigation, and multiple nations and standards organizations worldwide have declared their intentions to perform an investigation on 6G such as Finland [4], China [5], Japan [6], Germany [7], USA [8], Korea [9], and the EU [10].

The world’s first major 6G research initiative was initiated in 2018 by the Finnish government. In the U.S., 6G development was suggested in 2019, and the terahertz (THz) spectrum was made available for research by the Federal Communications Commission (FCC). The 6G vision for Japan was unveiled in April 2020 by the Ministry of Internal Affairs and Telecommunications. The 6G timetable for South Korea was announced in January 2020, with commercialization anticipated by 2028. The European Union has established the 6G Smart Networks and Solutions Business Organization to promote and develop 6G networks and products.

All of the aforementioned plans include vigorous 6G preparatory work. For instance, in September 2019, an initial 6G white paper [4] was published by Finland’s 6G flagship agency. The article outlined the concept of “ubiquitous wireless intelligence” and focused on the main drivers, obstacles, and associated scientific concerns of 6G. Additionally, following that, a number of white papers were published [11] that addressed many facets of 6G, including networking, ML uses, enterprise, edge intellect, protection, and more. In a white paper published in [12], the authors examined the transition from 5G to 6G and predicted the essential components that will make up 6G. Furthermore, a number of companies have conducted preliminary 6G investigations, including UNISOC [13], China Unicom [14], HUAWEI [15], and China Mobile [16].

In a study conducted by NTT DOCOMO [17], they explored the future development of 5G and the necessary components, applications, and innovations for 6G. They highlighted that the portable communication system undergoes significant shifts every decade, and the process of generating new market worth in the mobile communication industry takes approximately 20 years. In a statement made in [18], Samsung predicted that 6G technology would offer customers an unparalleled experience by enabling hyper-connectivity between people and all other entities. In addition, other alliances within the communication industry have also undertaken preliminary efforts in the development of 6G technology. The Next Generation Mobile Networks alliance conducted an analysis of the key elements that influence the development of 6G. These variables include social aspirations, market desires, and priorities. The alliance also explored the transition from 5G to 6G, focusing on the viewpoint of a 6G vision [19].

There are technical objectives can be obtained by using the 6G system, summarized as follows:

(a)

Data transfer rates of one petabit per second or more, or one hundred times faster than 5G. We anticipate peak data rates of up to 10 Tb/s for some unique use cases, including THz wireless backhaul and fronthaul (x-haul);

(b)

One gigabit per second (Gb/s) as experienced by users, ten times faster than 5G. In certain cases, such indoor hotspots, it is anticipated to offer a data rate of up to 10 Gb/s as perceived by the user;

(c)

A delay of 10–100 µs across the air and elevated mobility exceeding 1000 km/h. This will ensure a satisfactory Quality of Experience for scenarios such as hyper-high-speed rail and aviation systems;

(d)

Tenfold the connectivity frequency of 5G. This will accommodate up to 10 units per km² and an area traffic capacity of up to 1 Gb/s/m² for situations such as hotspots;

(e)

A spectrum effectiveness of 5–10 times and an energy effectiveness of 10–100 times compared to 5G.

Figure 1 shows some of the capabilities of the 6G networks.

Figure 1. Sixth generation key capabilities.

1.1. Summary

Table 1 provides a summary of the shortcomings and constraints of 5G, which are expected to be addressed and handled in coming 6G communication networks. This analysis is based on an examination of the shortcomings between the features of 5G and the anticipated needs of the next decade.

Table 1. Concise constraints and difficulties encountered in the 5G technology.

	Industrial Application	Requirement	Performance Indicator	Expected Challenges
Coverage	Marine and satellite communications	Cover sea and remote area including rural	Ocean coverage: 5% Land coverage: 20%	Coverage: space-air-ground-sea
Data transmission	Ultra-high-definition video	Very high-speed transmission	Rate: <20 Gbps User data rate: ~100 Mbps	Tbps level peak data rate. User date rate: 1–10 Gbps
Low latency	Auto drive, high production level	High speed, and low latency	Delay: <1 ms but low speed	Sub-second (<1 ms)
Solid connection	Crowded malls, auto production lines, stations	Super dense equipment and population	106 devices/km2	Density: 108 device/km2
High placement (accuracy)	UMV positioning and navigation, indoor positioning	Positioning of indoor or outdoor	Indoor: ~1 m Outdoor: ~10 m	Indoor centimetre level positioning. Outdoor metre level positioning
Reliable/safe	V2X, wireless data, telemedicine, internet	Super safe	99%	99%
Power use/energy	Intermediate altitude communications, internet of things	Reduce the consumption, increase efficiency	Energy efficiency: 107 bit/J	Network energy enhancement: 109 bit/J
Universal intelligent	AI applications, sensing technologies, digital twins	Support various applications	Low	High

As seen in the table above, the use of 6G networks will provide a wide coverage that could reach far areas and cities including rural areas. Given the swift progress of the wireless communications sector, it is imperative to address the constraints of 5G and advance the invention of 6G. Utilizing a collection of current proactive 6G studies, we present a thorough analysis and overview of 6G. This study seeks to provide a precise description of 6G, encompassing the prevailing knowledge of 6G, and to thoroughly examine the latest advancements in 6G challenges in different aspects are discussed and explored. Recommendations for different issues are presented in detail. Related literature is presented showing the existing studies and their publication data that hold the necessary knowledge that could support the 6G systems.

1.2. Recent Studies and Insights

Supporting technologies including artificial intelligence (AI), cloud, and/or the digital twin technology (DT) are expected to contribute to the development of the coming 6G networks. Table 2 shows recent studies regarding the use of these technologies.

Table 2. Recent studies showing the use of supporting technologies for the 5G/6G networks.

Technology	Ref.	Source (Journal)	Year	Contribution
Digital twin	[20]	IEEE Internet of Things	2024	The aim of this study is to improve computational performance and network security in 6G Internet of Vehicles (IoV) environments. To attain this, this study proposes an AI enhanced DT framework incorporating an advanced feature engineering module comprising stacked sparse autoencoders (ssAE) for dimension reduction wherein the AutoFS and AutoCM specialize in an online learning module for accurate attack detection. With the proposed DT framework, the performance of attack detection increases significantly, while simultaneously reducing system latency, energy consumption and RAM usage while increasing packet delivery rates in dynamic 6G IoV networks.
	[21]	Sensors	2022	The objective of this study is to develop a DT for NB-IoT wireless communication in an industrial indoor environment, and to analyze its performance when operating within real world conditions. To achieve this, simulations are conducted using Wireless InSite software to model and analyze critical communication metrics like throughput, SNR and received power. This study revealed that industrial environments reduce NB-IoT performance, with throughput 21% to 32% below peak rates. The results provide insights for improving communication efficiency in such settings.
	[22]	IEEE Access	2023	This study seeks to formulate a DT model aimed at unobtrusive patient respiration monitoring in healthcare 4.0. For this purpose, Wi-Fi Carrier State Information (CSI) is used and signal processing approaches like Elliptic Filters, Principal Component Analysis (PCA) to denoise raw data of the signal as well as Machine Learning (ML) algorithms for binary and multi-class classification such as Neural-Networks. To improve patient monitoring and decision support in healthcare settings, the study contributes to improving data processing and classification accuracy.
	[23]	IEEE Wireless Communications	2024	This study focused on the generative AI-enabled DT for wireless networks and propose a hierarchical design approach based on DT considering generative AI to handle challenges in 6G network architecture. This is accomplished using generative AI models such as Transformers and GANs to create message-level and policy-level DTs of the core network control plane. Through admission control and resource allocation, the study improves the stability and reliability of the network and proposes a scalable 6G network DTs-based framework.
	[24]	IEEE Conference	2024	The purpose of this work is to develop a wireless multi-hop network management system that continuously examines the network context in order to obtain network performance insight and suggest reconfiguration. To achieve this, the system combines simulations within the Click Router architecture, graph-based algorithms using the LEMON C++ graph library, and a theoretical link model to construct a Digital Twin Network (DTN) for optimizing Bluetooth Mesh networks. The DTN based system is demonstrated using testbeds which results in improved network resilience and management, especially for traffic scheduling in IoT constrained devices.
	[25]	Sensors	2022	The objective of this study is to enhance the positioning accuracy of UWB technology in indoor environments by developing a DT based indoor positioning system that addresses challenges posed by NLOS and LOS indoor obstructions. This is achieved using Slime Mould Algorithm (SMA) for optimizing position of anchor point. Employing a simulated model of anchor placement with Slime Mold Algorithm and addresses every network’s error utilization through the implementation of neural networks. This approach enhances UWB indoor positioning systems accuracy by improving anchor point positioning and handling errors in LOS and NLOS situations.
AI	[26]	IEEE Internet of Things	2022	The objective of this study is Reducing total MAC layer latency and improving reliability in Beyond 5G (B5G) networks. To achieve this, a Reinforcement Learning (RL)-enabled MAC scheduler is applied, using the UCB1 algorithm. The study successfully lowers MAC layer latency and improves reliability in B5G networks by optimizing resource allocation and queuing for heterogeneous traffic.
	[27]	IEEE BITS the Information Theory Magazine	2022	This objective of this study is to demonstrate how ML can be used to bypass traditional channel modelling to optimize wireless communication system design. This is achieved through the use of Deep Neural Networks (DNNs) and end-to-end training to directly optimize system-level objectives. The study enhances system performance by optimizing RIS reflection coefficients, distributing source coding for massive MIMO and mmWave initial alignment, and solving data-driven optimization problems.
	[28]	IEEE Communications Standards Magazine	2023	This paper attempts to solve bandwidth, privacy, data security, and inconsistent internet speed for Federated Learning (FL) for the Internet of Things (IoT). This is achieved by introducing a new framework based on 6G technology called super wireless over-the-air FL. Using interference resistant radio waves, the framework provides privacy protection for enhancing AI performance, ensuring that IoT system is smarter, faster and safer.
	[29]	IEEE Access	2024	The aim of this work is to determine the optimal beamforming angles for transmitters and receivers in 5G and 6G wireless communication systems while reducing computation complexity. To achieve this RL algorithms were applied such as Q Learning, SARSA, Double Q Learning, and Expected SARSA. With enhanced data throughput and avoiding interference, the study effectively reduces complexity and improves channel capacity.
	[30]	IEEE Networks	2022	The objective of this study is improving the quality of service (QoS) decision-making by preventing traffic congestion in Mobile Networks (MNs). This is achieved by applying Convolutional Neural Networks (CNN) which directly maps the condition of the MN to optimized admission control strategies. The model uses real world data from a telecom operator to enhance decision-making processes to help network operators make better informed more intelligent and timely decision to reduce congestion and improve QoS of user.
	[31]	MDPI Electronics	2023	The aim of this study is to enhance 6G native AI training latency and task accuracy by enhancing the quality of AI services (QoAIS) in 6G native AI wireless networks. To achieve this, the study utilize G-TSRA and NSG-TSRA heuristic algorithms. This improved accuracy and reduction in latency in 6G native AI wireless networks by resolving the issue of optimizing task scheduling and resource allocation in AI training services.
Cloud	[32]	IEEE	2023	The objective of this work is to improve the flexibility in resource management and adaptability to the changing service requirements in the 5G/6G networks. Integration of Software Defined Networking (SDN) and cloud virtualization approaches in the context of Multi-access Edge Computing (MEC) architecture is used to achieve this. Thus, allowing rapid adaptation to critical user requirements and ensure service continuity for the expected user mobility in MEC environments enables service delivery and low latency in time critical services.
	[33]	CMES	2024	The objective of this study is to optimize the performance of FL in 5G/6G networks by considering issues related to non-IID data, low client engagement and heterogeneous device clients. This is achieved by applying the adaptive server selection FedAdaSS algorithm, which utilize cloud computing resources. The study effectively reduces communication overhead by dynamically selecting the best server for each round of training, and random reshuffling on client sides to mitigate performance loss due to low participation in FL processes.
	[34]	Future Generation Computer Systems	2023	This study aims at reducing service request delays and achieving low cost in 6G cloud-edge systems. To achieve this, the Service Deployment with Service Rise and Fall (SD-SRF) algorithm is developed for offline service arrangement based on a greedy approach, and the Proximal Policy Optimization for Multi-layer Service Deployment (PPO-MSD) algorithm is used on an online service deployment model as Markov Decision Process (MDP). The study effectively reduces service delays and improves cost efficiency in 6G networks by optimizing service deployment in multi-layer edge networks.
	[35]	ICTACT Journal on Microelectronics	2022	The aim of this study is to eliminate the interference in 5G ultra-wideband communication antennas and enhance performance in the cloud computing networks. This is achieved by proposing a novel Ultra-Wideband Communication Antenna (UWBCA) design with a ‘double square’ girder structure. This results in reducing latency, and improving spectral efficiency, data rates and connection density over existing antenna models.
	[36]	Sensors	2023	This study aims to improve eHealth IoT system performance by integrating 5G network slicing, cloud computing, and fog computing. This is achieved using an IoT–fog–cloud architecture queuing model analysis employing performance metrics. Through resource allocation optimization with virtual BBU and fog nodes, the study improves throughput, QoS, and minimizes latency in 5G eHealth applications.
	[37]	Sensors	2023	This study aims to enhance energy efficiency and data management for 6G surveillance systems. This is accomplished by utilizing a clustering algorithm with ZigBee energy optimization to minimize power usage and extend the lifetime of WSN. The reduction in network consumption by integration with the AODVjr algorithm, as well as embedded cloud computing providing data processing improvement through reduced latency and better storage, are accomplished. The proposed solution increases system reliability, enhances data security and effectiveness of multi-scale monitoring in 6G environment.

1.3. Paper Organization

The rest of the study is organized as follows: Section 2 presents the goals of the 6G globally with future applications and components. In Section 3, open issues in some applications are discussed and analyzed then supported with recommendations for solving each challenge that could enhance the functionalities of the system and overcome certain issues. Finally, conclusions are written in Section 4.

2. Worldwide 6G Goals and Expectations

Research manuscripts reporting large data sets that are deposited in a publicly available database should specify where the data have been deposited and provide the relevant accession numbers. If the accession numbers have not yet been obtained at the time of submission, please state that they will be provided during review. They must be provided prior to publication.

It is anticipated that 6G will be distinct from 5G in a number of relevant respects. When it comes to the demands of applications, 5G expanded the “Mobile Internet” that was available in 4G to the “Internet of Everything”. After 5G, 6G will strive to improve mobile Internet and Internet of Everything (IoE), but it will also deeply combine these technologies with artificial intelligence and big data in order to achieve smart IoE.

In terms of technological specifications, 6G will aim for broader coverage, enhanced data speeds, increased connectivity, ultra-low latency, exceptional location precision, merging of communication and detecting, greater intellect, improved security, and superior replacement compared to 5G. It is important to recognize that 6G will transcend mere communication. The practical use and technological demands prompt us to conceptualize the nature of 6G.

The 6G concept has been presented in several publications [38,39,40,41,42,43,44,45,46,47]. Specifically, in [38], the authors lay out the cutting-edge 5G tech and point out why 6G research is essential. They expected mobile ultra-broadband, super Internet of Things (IoT) [48], and AI to be three main components of 6G, which will take into account the present and future growth of wireless communications. They then go into the essential technologies needed to make each part a reality. Machine learning methods [49] show great promise as AI candidates, symbiotic radio and satellite-assisted communications can be utilized to accomplish super IoT, and THz communications can be utilized to provide mobile ultra-broadband.

The usages of 6G can be integrated with many technologies including AI, IoT, and mobile broadband, Figure 2. Wireless transmissions of terabits per second are possible with mobile ultra-broadband, which can support far and rural areas including villages and small communities. With super IoT, the present Internet of Things can have better connectivity capabilities and coverage. AI has the potential to intelligently design and improve wireless networks in the years to come.

Figure 2. Expected uses of 6G in the community within various areas: AI, IoT, and mobile ultra-broadband.

The writers devoted their attention on studying intelligent 6G wireless networks and multi-Tb/s in [39], with an eye toward the years 2030 and beyond. They laid out a 6G vision, talked about potential uses, and highlighted 6G’s main features. The study [40] examined the primary factors influencing 6G, which arise from the problems and operational constraints of 5G, with the technology-driven paradigm shift and growth of wireless networks.

As the global installation of 5G systems accelerates, it is imperative to anticipate the evolution towards above 5G systems. This shift is primarily motivated by arising societal trends that demand entirely automated and smart services, bolstered by extra reality and tangible communications. To meet the rigorous demands of these prospective applications, characterized by data-driven needs and necessitating highly low delay, ultra-reliable, rapid, and seamless wireless connectivity, research initiatives are presently concentrating on an evolving roadmap for sixth generation (6G) networks, anticipated to induce revolutionary shifts in this domain.

Reference [41] elucidates several important enabling innovations for 6G, anticipated to transform the foundational architectures of cellular networks and deliver a variety of homogeneous artificial intelligence-enhanced services, encompassing spread communications, management, computing, detecting, and energy, from the centre to the end nodes. The study [42] articulated a comprehensive, future-oriented perspective that delineates the principles of a 6G infrastructure. We believe that 6G will not only include the search of additional spectrum in high-frequency groups but will instead represent a convergence of emerging technology trends propelled by innovative underlying services. Figure 3 presents an overview of the Essential foundations for 6G.

Figure 3. Essential foundations for 6G.

Establishing an idea of potential communications is crucial to guide that research. In [43], scientists endeavoured to delineate the comprehensive landscape of communication requirements and technology within the context of 6G. The 6G will be refined and economically adapted for the novel applications presented in 5G, facilitating their widespread deployment. Simultaneously, it will facilitate novel applications that we are currently unable to conceive or articulate comprehensively. The proliferation of mobile cellular into various sectors, initiated by the advent of affordable IoT technologies in 4G and ultra-reliable low latency Industrial Internet of Things (IIoT) in 5G, will persist, expanding in both scope and depth in 6G.

Figure 4 Present the Corresponding Analytical Instruments for 6G and concepts pertinent to many open topics. Using different tools for the 6G system will definitely support the community and provide higher quality to all users.

Figure 4. Corresponding Analytical Instruments for 6G.

The swift progression of AI/ML innovation and its efficacy in addressing challenges across various sectors indicates a 6G system that will fundamentally leverage these new capabilities to enhance productivity by more effectively responding to the working context. Emerging topics are expected to influence the needs and advancements of 6G systems, including:

(a)

Novel man–machine interactions established by a consortium of various local equipment functioning collaboratively;

(b)

Pervasive global computation dispersed across numerous local devices and the cloud;

(c)

Integration of multi-sensory information to generate multi-verse maps and innovative mixed-reality experiences; and

(d)

Accurate sensing and actuation for the manipulation of the tangible environment.

The above points are summarized in Figure 5, which shows expectations and components. As shown, many devices will be able to function in higher performance and without interruption. This means the operation of services in the community will be enhanced.

Figure 5. Components and expectations for using smartphones and tablets with man–machine interfaces in the community.

Within the next 5 to 10 years, we will initiate the employment of gadgets that individuals can employ to connect to the network. Although smartphones and tablets will persist, we are likely to encounter novel human–machine designs that will significantly enhance our ability to access and manage information [43,47]. Limited assumptions may be established regarding the existing enabling capabilities associated with 6G innovations, described as follows:

(a)

Portable gadgets, including earphones and those integrated into attire, will become prevalent, while skin overlays and bio-implants could also gain acceptance. We may grow dependent on novel cognitive detectors to operate machines. We will possess numerous electronic devices that will function harmoniously together, offering smooth and user-friendly layouts.

(b)

Touchscreen input will become obsolete. Instead, we will use gestures and verbal communication with our devices as standard practice and possibly use AI personal assistant for it.

(c)

The equipment we utilize will be entirely context-aware, and the network will progressively enhance its ability to anticipate our requirements. The integration of context awareness with advanced human–machine gateways will enhance our interactions with both the tangible and virtual domains, rendering them less confusing and more productive.

The computational requirements for these electronics will probably not be entirely contained within the gadgets alone due to constraints related to their size and battery life [43]. Instead, they may need to depend on locally accessible computer assets to do tasks, beyond the edge-cloud [41]. Telecommunications are going to play a crucial part in the future man–machine interface [50,51]. As clients, we may anticipate that the following may occur:

(a)

Contemporary self-driving idea vehicles will be accessible to the general public by the 2030s. They will predominantly operate autonomously, but a remote driver or rider may still be required to assume control under specific circumstances. This will significantly augment the amount of time given for us to access data from the internet in the forms of enhanced amusement, enriched communications, or schooling. Automobiles will substantially increase data consumption: (i) real-time uploads of automobile sensor information to the network, (ii) downloads of high-resolution visualizations, and (iii) immediate inter-vehicle connectivity will occur.

(b)

A substantial deployment of wireless cameras as sensors will occur. Advancements in AI and machine vision, along with its ability to identify individuals and things, will render the camera a ubiquitous sensor applicable in all contexts. Concerns about privacy will be mitigated by restricting access to data and masking material. Additionally, radio and other sensing approaches, such as acoustics, will be employed to collect environmental data.

(c)

Sophisticated methodologies will be employed in security screening protocols to eradicate protection queues. A multifaceted approach utilizing many sensing modalities will be employed to check individuals as they navigate through congested regions, rather than solely at entry points. Radio sensing will be a crucial element in accomplishing this, bolstered by upcoming communication technologies.

(d)

Digital currencies and credentials may emerge as standard, with transactions in both physical and digital realms executed via the multitude of gadgets at our disposal. The forthcoming network must ensure the security and privacy essential for this shift.A multitude of household utility robots will augment the vacuum cleaners and lawnmowers currently in use. These may manifest as a collective of mini robots collaborating to do duties. The robots will be outfitted with video cameras transmitting to a nearby computing servers for immediate execution. Consequently, we will observe a rise in the quantity of products and augmented resource demands within our home networks.

(e)

Healthcare will undergo significant transformation, featuring continuous observation of critical metrics for both healthy individuals and patients via various portable gadgets. Health surveillance will encompass internal gadgets that interact with external wearables, which subsequently transmit information to the web.

Figure 6 shows the network capabilities and possibilities for the man–machine applications including applications and components that support different purposes in the community. Some applications that will be supported by the 6G system are maps and localizations, which means the rural areas will benefit from the services with the wide and strong transmission within the right settings. Other wireless technologies can be used together with the 6G network as hybrid systems including WiFi, light fidelity (LiFi), and/or other broadband systems.

Figure 6. Network support for the man–machine applications and services for the community.

The transition to Industry 4.0 and the initial phase of wireless-enabled automation will occur prior to the 2030s. Fifth-generation systems offering ultra-reliable minimal latency connections will enable real-time computing in the cloud computing, including task scheduling in cloud environment [52]. Nevertheless, applications in industry necessitating significantly more strict demands for wireless connectivity will need 6G. For example, holographic virtual presence is going to be standard for professional and everyday interactions. It will be feasible to create the illusion of being in a specific position while actually residing in another, such as looking to be in the office while being situated in a vehicle. We will develop technologies that integrate current face expressions with an avatar that appears in the virtual illustration of any genuine environment. There will be extensive utilization of mobile robot groups and drones [53,54] across multiple sectors, including tourism, healthcare, warehousing, and parcel delivery. Active digital twins in the digital age, featuring progressively precise and synchronized refreshes of the physical environment, will serve as a crucial platform for enhancing cognitivity among humans. From the aforementioned view of the future, we may derive different key use cases, including augmented reality, telepresence, security monitoring and surveillance, and more. This encompasses the capabilities that 5G will facilitate, alongside widespread adoption anticipated within the 6G period, utilizing novel technology, in addition to a range of new use cases made possible by these 6G innovations.

In 2030 as well as thereafter, the rapid advancement of informatization and digitalization will create a new realm of digital twins [55], where each tangible object possesses an online counterpart, as seen in Figure 7. The virtual realm creates a conduit for the effortless exchange of information and intelligence among individuals, between humans and objects, and even among objects themselves within the living creature. The digital realm simulates and forecasts the live entity, precisely mirroring and anticipating its actual condition. The modelling and anticipation of biological entities from the virtual realm may also be employed to pre-emptively intervene in the operations of a living entity or the physical world, thereby mitigating risks and calamities in preparation.

Figure 7. Illustration of the digital twin.

Detailed explanations of anticipated needs pertaining to the vision will be provided in the paragraphs that follow. First, 6G should be a fully merged network that covers more ground and has better reception in more places. This includes both land and satellite connectivity as well as short-range communication among devices.

Second, 6G is anticipated to operate at greater frequencies in order to attain a broader bandwidth, including mmWave, Terahertz, visible light, and others. With a maximum transfer rate of 1 Tb/s and a user encounter data rate of 10 Gb/s, 6G may achieve a data rate that is 10 to 100 times higher than 5G. Further improvement of reusing frequency productivity is possible with 6G’s dynamic frequency-sharing solution. Third, 6G is a smart, tailored network. When 6G and AI come together, personal mobile communication will become a reality. Instead of a standard function-centralized system, there will be a new three-centralized model: user-centred, data-centred, and completely content-centred. Fourth, the 6G system will incorporate operational security into the architecture or use an internal security mechanism. Cyberspace safety can be achieved with the support of 6G’s self-awareness, real-time flexible assessment, and evolved threat and trust assessment, which are made possible by trust and safety systems. Fifthly, interactions, directions, and detection will all be rolled into 6G. Satellite positioning and localization systems, radar detection systems, and global telecommunication systems are all part of 6G. With its software-defined aerial access networks and backbone systems, 6G will have an accessible design that allows for quick and sophisticated construction as well as flexible activation of network functionalities. Lastly, 6G has the potential to produce vast amounts of data via the Internet of Everything. It may additionally integrate with cutting-edge technologies like edge computing, cloud computation, artificial intelligence, blockchain, and more. With 6G, we can finally achieve entire genius and cluster intellect, which is a kind of group collective intelligence. We can also finally establish a pervasive smarter mobile Community.

3. Challenges and Related Recommendation

More investigation into numerous topics and avenues of study is necessary before the 6G goal of global reach, complete applications, solid security, all perceptions, and all electronics can be realized. Furthermore, there is an immediate need for innovations in 6G telecommunication concepts. In this section, many challenges will be discussed and analyzed. Furthermore, recommendations and features will be suggested for the discussed challenges.

3.1. Fundamental Theories Challenges

Few aspects are included in this section including channel research, information theory of the electromagnetic, and baseband processing including the unified, iterative, hardware generation, and architecture. When discussing Unique Network Research, there are typically four stages to conventional channel research: measuring the channel, estimating its parameters, analyzing its characteristics, and finally, modelling it. Several restrictions are imposed by this passive method of channel recognition. It takes an extensive amount of time, assets, and effort to perform the channel assessment. Also, channel assessments will never be able to capture every possible frequency range or circumstance in the real world. Channel parameter evaluation is complicated and plagued by massive amounts of data. The channel characteristics can only be examined at predetermined frequencies and in predetermined circumstances; thus, the complex link among novel qualities and frequencies/scenarios cannot be thoroughly explored. Last but not least, conventional channel modelling cannot foretell the future channel properties of frequency ranges or scenarios for which no data are now available.

Various channel models, such as 6G established pervasive channel demonstration [56], AI-based forecasting 6G channel demonstrating [57,58], scenario flexible channel modelling, and RIS medium modelling, will need to switch from passively acknowledging channels to proactively controlling them for 6G [59]. Various methods for optimizing antenna–channel interaction are shown in Figure 8. Two methodologies have been employed for channel interaction, distinguished by their level of sophistication. The primary research avenues for AI-based strategies in optimizing antenna and channel compounds involve either the straightforward modification of certain high-level station attributes, such as antenna/beam/tilt choice, or the additional immediate alteration of antenna arrangements according to channel information. AI is anticipated to be especially beneficial for the ideal pattern creation technique illustrated in Figure 8, since a flexible construction of the radiating architecture entails a substantial improvement search space.

Figure 8. Various methods for optimizing antenna–channel interaction. The grey arrows represent the complexity direction (increasing), while the orange arrow represents a two-way direction between the antenna and the channel.

When it comes to ML-based channel description and forecasting, there are two main areas where the techniques shine. In the first, it is commonly used to find the locating function among the surroundings and the channel characteristics or to immediately mimic the assessment data (Figure 9a,b), respectively. In the second, it is more common to acquire knowledge about the past behaviour of channels and use that information to forecast or extrapolate future channels (Figure 9c).

Figure 9. Fundamental concepts of ML-based channel (a) steps included in a simulation, (b) modelling of prediction using ML, and (c) forecasting.

One might come across it in the most recent research: most ML-based channel forecasting techniques are built on top of artificial neural networks (ANNs), which have varied network topologies but both offline training and online executions with minimal computing overhead. As a result, creating effective channel prediction algorithms presents several formidable obstacles related to the type of the ANN, and the input/output type that should be used for training.

Conversely, the conceptual underpinnings of wireless communications rest on the theory of information and electromagnetic (EM) theory. The (EM theory) describes how electromagnetic waves are created and how they travel through space. When dealing with constrained resources like power and bandwidth, the information theory lays out a framework for precise and effective data transmission.

The continuous-space EM field distribution can be derived from the EM theory, which is based on uninterrupted space and time. It is not possible to determine the channel capability only from electromagnetic theory. The channel capability in finite space can be determined using information theory; nevertheless, in an ongoing space, it is not possible. Transmitter theory and wireless transmission channel modelling theory link the electromagnetic theory and information theory. Specifically, the wireless transmission channel connects electromagnetic theory and information theory; it starts at antennas and conveys data via electromagnetic pulses.

Keep in mind that as 6G core innovations advance, there will be issues that none of the aforementioned theories can fully address, creating an opportunity and a threat for a joint application of the four theories. The quantity of users, base stations, relays, and/or RISs, for instance, keeps going up as 6G wireless communication networks grow from covering only a local area on land to covering the entire planet through space, air, ground, and sea interconnected networks.

In mobile networks like low earth orbit (LEO) satellite, UAV, and vehicle networks, the locations of both base stations and users are always shifting. In order to aid in system design, 6G wireless communication networks are evolving from discontinuous to ongoing spaces, necessitating the collecting of channel strength information (CStrI) and its calculation regardless of location in the continuous area. The aforementioned four theories will inevitably be integrated in this scenario. Furthermore, as the dimensions and elements of transmitters in ultra-massive MIMO expand, the Tx/Rx transmitters become increasingly interconnected with their surroundings. Therefore, it is essential to integrate the theories of transmitters and wireless transmission channels in order to describe Tx/Rx antennas and the underlying wireless transmission channel. There are additional requirements for channel characterization, antenna architecture, and continuous-space channel potential computation due to the transformation of antenna grids from isolated to ongoing windows and the decreasing antenna unit separation in holographic MIMO.

Once again, all four of these theories must be considered together. In conclusion, 6G wireless networks offer fresh uses and practical specifications that surpass the reach of any one theory when compared to 5G. Traditional theoretical frameworks have hit roadblocks in their advancement. To lay the theoretical groundwork for 6G wireless communication networks and facilitate fresh findings, it is crucial to investigate the merging of electromagnetic theory and information theory, namely, EM information theory [60,61].

In terms of baseband processing, two distinct varieties can be described: unified and iterative. In communication systems, channel coding has always been crucial for the unifying thanks to its error-correcting potential. Sixth-generation communication systems must include unified implementations of improved coding schemes and simplified decoding algorithms to accommodate dynamic and varied circumstances. In order to satisfy the strict criteria of 6G, it is often necessary to look at the architecture and execution of the entire baseband processor, rather than just refining one channel coding unit. In contrast, iterative receptions have a longer history of being considered to have better system throughput and connection dependability than the current state of distinct baseband processing techniques [62]. The channel estimation, MIMO detection system, network decoder, NOMA detector, and origin decoder perform recurrent soft information swaps with each other [63,64,65,66].

The heightened utilization of every module due to repetitions may impose significant delay and computational overhead on the system. Since the majority of modules focus on addressing the greatest a posteriori estimate problem, they can be integrated into a cohesive Bayesian system by leveraging their identical features, hence enhancing the efficiency of the iterative receiver. Given the serial nature of the signal circulation, one alternative method is to consider many modules collectively, treating the end result as the target for joint processing, including joint detection and decoding [64].

Figure 10 illustrates an example [65] of the instance of sparse code multiple access (SCMA) encoding for J = 6 and K = 4, employing quadrature phase shift keying (QPSK) modulation, with a modulation magnitude of 2 for QPSK. Every user is allocated two assets based on the codebook selected. The J codewords are, thereafter, transferred using K supplies. Each SCMA codeword takes only a portion of the supply. The whole configuration of the SCMA might be depicted by a K × J factor graph matrix. A component that is greater than zero in the k-th line and j-th field of the factor graph matrix indicates that the asset k is allocated to user j. Every asset is utilized by N_u users, and the communication from each user is conveyed by N_r assets.

Figure 10. Example of SCMA encoding procedure.

Reliable unified equipment design and accompanying implementations are crucial for meeting many realistic purposes based on cyclic or cooperative baseband processing. In pre-6G telecommunication networks, the layout and construction of each module are segregated, and the absence of system-level design considerations leads to increased hardware asset expenditures. To enhance the adaptability, compatibility, and hardware effectiveness of 6G baseband handling executions, the VLSI-DSP technique [67], which includes retiming, bending, and unfolding, may be employed to create a solid design featuring fixed handling aspects, improved network interconnections, and iterative timing timetables.

The intricate and varied uses of 6G networks of communication necessitate specialized electronic architecture automated solutions for baseband handling, which can reduce the entry obstacles for circuit designs across numerous scenarios and fulfil client requirements. An effectively developed automation tool can autonomously produce the circuit architecture according to the customer’s quality specifications. Utilizing the unified Bayesian system, the majority of the computation modules may be defined to enhance the auto-generation technique, hence establishing a design space for optimizing hardware attributes [68,69]. Figure 11 illustrates the automated development of hardware architectures, it shows a hardware architectures auto generation where the F represents the latency and throughput while I, II, and III refer to a different type of the polar encoders. Standardization of polar codes for use in improved mobile broadband management channels has piqued the interest of other potential users. For the purpose of automating the generation of Verilog HDL files for the hardware implementation of F_I(N,M), F_II(N,M), and F_III(N,H), a Python compiler called the Polar Compiler was employed in the study [68]. The compiler files are available on GitHub [70]. AI can assist in the discovery of design spaces with complex needs. It can also assist in identifying the compensatory parameters induced by estimation, quantization methodologies, and iterative techniques in the automated hardware design process [71].

Figure 11. Auto-generation of hardware architectures for target polar encoders [68].

3.2. Future Research Challenges

6G networks approaching challenges by using hybrid RF–Optical networks are discussed in this section in addition to space–air–ground–sear (SAGS) networks, integrated sensing and computing networks, digital twins’ networks, and AI-enabled networks.

3.2.1. Hybrid RF–Optical Networks

One exciting new direction for 6G is the integration of systems operating on many frequency channels—including RF and the entire optical wireless bands—to create diverse and mixed networks capable of supporting all of these bands. This will help bring about the entire spectrum vision of 6G. Indoor, outdoor V2V, open space, and submarine settings are only a few examples of the many uses for RF–optical diverse mixed networks [72,73].

However, there are a lot of obstacles to overcome on the road to RF–optical mixed networks and systems. First, RF–optical mixed-signal networks share many of the same difficulties as other types of mixed networks, such as those that span space, air, ground, and sea. These include but are not limited to the following: the development of transfer protocols for the network, traffic balancing [74], synchronization of mixed networks, distribution of resources and enhancement of energy efficiency [75], distribution of spectrum [76], and distribution of power and connection access points [77].

Because of its low energy usage and high attainable data speeds, white LEDs have increased the energy efficiency of data transmission and are now used for both lighting and communications. Consequently, LiFi approaches have a wide range of possible uses, including as anti-jamming in military settings, vulnerable civilian contexts like hospitals and airplanes, and broadband indoor communication. One way to increase the system’s performance is by allowing wireless networks with various technologies to collaborate, which is called a mixed wireless network. The variations in fading channels, transmission losses, and accessible resources among networks allow for these improvements. Such improvements, however, are contingent upon successfully completing the enormous task of creating resource distribution algorithms that distribute bandwidth and power among diverse networks in order to meet varying operation demands. One way to improve the energy efficiency of RF communication systems is to use numerous radio gateways.

6G networks can be utilized for the support of smart buildings together with other technologies. Hybrid LiFi/WiFi networks have been used in smart buildings to assist social distancing [78,79], where vehicle access is managed and controlled after data exchange at the entrance at the gate of the building using the mixed wireless network. The presented approach can be assisted and improved using 6G networks, especially when using other services such as cloud servers [52], and other solutions. However, the challenge will be providing the proper infrastructure and power for big buildings and high-density environments. When designing RF and optical communication networks, it is important to think about transceivers with unique characteristics. For example, users face significant difficulties with frequent handovers (HOs) [80,81,82,83] due to the diversity of access points when they roam. Several studies have proposed different methods for HO for hybrid LiFi networks. However, since LiFi networks used to provide higher data rates than WiFi, when mixing 6G networks with other networks such as LiFi or any optical network, new challenges will occur including the mobility and access point assignment due to the high data rates not only from the LiFi network but also from the 6G networks, which makes the criteria of data rates, capacity, or network speed are not the main aspects of allocating the network to the user. This remains unknown since LiFi and 6G networks are still under development and are not yet popular for the end-user.

Additionally, uplink OWCs still have significant limits owing to the low energy of terminal equipment, the significant impact of mobility on them, and the interference that these uplink OWCs cause with lighting. Lastly, it is impossible to overlook the mixed systems’ impact on system security [84].

Other problems that were studied and investigated in hybrid LiFi networks include optimization of network throughput and data rate provided to the user mainly in load balancing and selecting the best possible access point [74,85,86,87]. On the other hand, machine learning has been used in such networks for problem solving and network optimization [29,88,89,90]. Various simulation results were obtained by the existing studies including increasing the network throughput and/or data rate, reducing the switching/outage probability and complexity. Optimizing LiFi hybrid network performance is essential for user quality in indoor scenarios and in other applications such as vehicle communication.

3.2.2. Space–Air–Ground–Sear (SAGS) Networks and Integrated Sensing and Computing Networks

Realizing the dream of 6G worldwide coverage will require the SAGS integrated network, which is both a trend and a critical technology. There are numerous technical and theoretical concerns with network design, building, preservation, and optimization that need to be addressed before the SAGS integrated network can be considered fully operational. Developing and researching communication systems and networks relies on first establishing communication channels. A prevalent channel structure [73] must take into account how to incorporate distinct channel properties in satellite [91], unmanned aerial vehicle (UAV) [92], ocean [93,94], and ground scenarios, among others, due to the significant difficulties posed by different frequency bands and scenarios when measuring and predicting channels [95].

Satellite networking stands apart due to its extensive support coverage qualities and the fact that it is a proven and operational technology with a lengthy history. Several fields have begun to rely on communications via satellite in recent years, including the broadcasting industry, directions, and earth monitoring. There is a lot to explore in using next-gen domestic wireless communication infrastructure, which means that satellite networks have to meet greater standards for things like bandwidth, QoS, and ubiquitous connection if integrated with 6G networks. Predicting the fundamental genuine transmission channel properties requires a generic, precise, and low-complexity approach, which is essential for the development and efficiency assessment of communications satellites [96].

In addition to dependable control and non-payload communication links, several emerging UAV uses, like distant sensing and aerial snapshots, necessitate a significant transmission data rate for payload connectivity [97]. The vast range and enormous capacity of existing and upcoming 6G networks make it a viable option to use future cellular networks to meet UAV communication requirements. Visions of future 6G wireless communication network deployments point to UAVs acting as base stations for mobile devices as a key component in expanding network reach [39]. Figure 12 shows some tools that enables the UAV system.

Figure 12. UAV communication system tools.

An attractive option for providing ubiquitous access from above to ground user equipment in some situations and locations (such as hotspot regions, big public spaces, and isolated places without network service) is the installation of aerial BSs [39]. The worldwide Internet of Things or global reach via 6G SAGS connected networks can be realized by integrating UAVs with other critical innovations pertaining to satellite, terrestrial, and ocean telecommunications [98]. Multipath elements (MPEs) produced by ground-level landscapes and structures are the root source of small-scale channel fading.

Figure 13a shows a conventional UAV-to-ground channel situation that incorporates all conceivable forms of MPCs. Two possible paths for the broadcast data to reach are the LoS route and the multibounce route. To improve the accuracy of the UAV-to-ground channel simulation while decreasing computational complexity. In Figure 13b, we can see a simplified representation of the network structure for UAV-to-ship connectivity in maritime settings. A transceiver is attached to the UAV side, while a receiver is attached to the ship side, where SB refers to a single bounce, while MB is a multi-bounce. It is possible to place several antennas on each side.

Figure 13. Wireless and broadband applications in different scenarios. (a) UAV-to-ground communication in urban and rural areas, illustrating line-of-sight (LOS) and multi-bounce rays with a UAV trajectory over diverse building environments, (b) UAV-based communication over marine areas, depicting single-cluster and twin-cluster models, with evaporation duct effects and various signal propagation paths, and (c) an overview of various 6G transmission and reception scenarios, including UAV-to-UAV, UAV-to-ship, ship-to-land, underwater acoustic communication, vehicle-to-vehicle (V2V), ultra-massive MIMO, mmWave/THz/optical wireless, and IoT applications for industry and indoor hotspots.

While sub-6 GHz is the most common frequency for UAV-to-ship communication, mmWave bands are also an option for seamless integration with future 6G networks [99]. Ocean surface roughness increases reflections and causes fading of multiple paths as a result of water fluctuation [100]. Figure 13c shows that there are numerous possible uses for UAVs and 6G wireless channels across various frequency bands. There is a striking disparity between the sounders and features of each channel.

In addition, there is an immediate need for creativity in the following areas of the SAGS unified network: structure layout, movement administration, network procedure, resource placement, routing tactics, EE improvement, and beyond. To ensure that users have access to secure and dependable communication solutions, it is important to plan a network layout that is both effective and secure [101]. It is essential that the interconnected networks have efficient, secure, and transparent authentication mechanisms [102]. Advanced mobility control systems are necessary to ensure smooth mobility control between diverse and homogeneous networks, taking into account various forms of movement in different settings [46]. When planning network protocols, resource allocation, and routing techniques, it is important to keep in mind the SAGS combined network’s flexible architecture and elevated movement features [103]. Preservation and improvement of SAGS-based networks necessitates taking into account the power provision of the connectivity base as well as the burden of UAVs and other equipment, all of which place high demands on the total EE of the system.

Simultaneously, it is essential to evaluate how to achieve a balance between enhancing network efficiency and controlling expenses [104]. Furthermore, the application of AI, deep learning, and other smart technologies to improve network structure and enhance entire network reliability [105] is a significant contemporary concern. The current network designs are inadequate for delivering customized network solutions tailored to various task kinds in SAGS. Furthermore, they are unable to effectively address numerous issues in SAGS, including diverse network separation, elevated network latency, sporadic interruptions, and imbalanced network traffic. Edge-cloud processing technology and networks might be employed to resolve the aforementioned issues, facilitating smart networking, and minimizing latencies. Various tasks operating within SAGS and their scheduling [52] may possess distinct network performance needs. SAGS cannot uniformly fulfil the requirements of all tasks. The user quality may be compromised, resulting in an imbalanced network load while addressing a more critical demand.

Moreover, equipment roaming, and an inadequate connection setting may result in sporadic network outages and extended network reaction delay. The current network topologies are inadequate for SAGS in this complex environment. The conventional TCP/IP design is incapable of addressing link interruptions and exhibits drawbacks such as suboptimal bandwidth usage and inadequate security for the network. The information hub system [106], constrained by the extent and volume of data within SAGS, will result in a mass of material namespaces. This will result in significant latency cost for material extraction, hence extending network latency. Moreover, current network topologies do not address the issue of mixed network cooperation, resulting in suboptimal resource use, and proactive propagation is frequently neglected.

3.2.3. Cloud-Edge Computing

Edge and cloud computing can be utilized to improve, secure, the advances the SAGS systems. The advancement of the web, IoT, and cloud computing has enabled diverse innovative wireless gateway devices to connect to the network, offering sophisticated and practical solutions for individuals. The notion of mobile-edge computing (MEC) was established. MEC units are to be deployed at the network’s edge to service individuals, with edges referring to random places along the path between the data provider and the cloud computing core [107]. Every edge-computing component possesses computational, networking, and storage assets that can aid the cloud centre in managing activities. To enhance service efficiency, it is essential to develop an innovative network structure to accommodate the demanding SAGS.

There are numerous tasks in SAGS. Adhering solely to the conventional cloud-computing model will eventually result in significant strain on the cloud-computing core, possibly leading to network congestion and paralysis, so impairing the general efficiency of the entire network. The process of collecting, interpreting, analyzing, and mining data is a collaborative effort between edge computing and conventional cloud computing. Therefore, merging edge computing will add significant advantages to the system. The edge-computing terminals are multi-purpose and can store user data among their many other functions. In order to expedite the reaction time to the demand, the working result and the information to be forwarded can be saved locally in the information storage table. To some degree, the network’s accessibility is enhanced since data packets can be swiftly routed in the event of a disruption or malfunction. They are able to compute and extract the needs for task-oriented networking. Not only do edge-computing components manage the network and provide interface services, but they are also tasked with communicating with cloud-computing canters and other realms of the network. It receives feedback data and tasks from users and transfers it to a cloud-based data centre. Latency decrease, balance of load, and enhanced network security are some of the benefits that can be achieved through the use of edge computing [108].

As seen in Figure 14 [109], the four sector networks are represented by four separate but interconnected processors that make up the cloud computing core. The central data centre in the cloud is able to exert control over SAGS on a worldwide scale by collecting data from the edge computing terminals in every network region. The use of SAGS supports far and wide connections including rural, maritime, and highlands.

Figure 14. Illustration of cloud and edge computing in a SAGS system.

Through high-level convergence and collaborative augmentation, the 6G network will make possible the combined use of mobile communication, smart awareness, and processing power activities. The innovative extensive network known as the interconnected network of communication, detecting, and processing was created by combining all three of these technologies. Waveform layout and processing of signals [110], sensory wireless systems [111], integrated sensing and communication (ISAC) network-based detecting [112], calculating power offloading [113], partnership of multi-layer calculating power supplies [113,114], and cloud-edge gateway asset distribution for the multi-layer widespread computing system are some of the primary areas of present investigation in the field of information system architecture and computer networks.

Cloud/fog computing have made application offloading a popular subject. The decision-making process can be centralized (as is the case in major existing works) or decentralized, with a single or multiple user devices (UEs), and every app can be loaded in three different ways: coarse-grained application level, fine-grained task level, or percentage. Then, it can be loaded into the cloud/fog [115]. It is common practice to think about task allocation and segmentation in a single-UE scenario, with each job having to decide whether to offload based on specific requirements. For optimal productivity in a multi-UE environment, it is important to distribute assets (such as power and bandwidth) that are divided among the fog cluster and the UEs.

By implementing computing solutions close to mobile consumers through the fog computing model, the end-to-end latency that users encounter can be greatly minimized. This lowered latency is helpful for fresh applications that are sensitive to delays, like electronic healthcare, real-time management, and automobile communication [116], which can only withstand latencies of several milliseconds [117]. Limitations on future networks are largely due to their substantial use of energy. Energy is a major factor in increasing running costs and negatively impacting the environment. Due to the fact that base stations consume the vast majority of energy in cellular systems, implementing smart small base station (SBS) execution techniques can greatly decrease the consumption of energy [118].

There are two main approaches to managing SBS operations: centralization and distribution. Within the former, an SBS functions independently by virtue of its intelligent self-organizing characteristics. On the other hand, in a centralized management system, data are pooled in the centralized baseband unit (BBU) for optimized network-aware treatment from BSs and other assisting components. Centralized BBU pools have the capacity to make decisions that affect the entire network and are responsible for managing all of its components. Centralized management has a larger cost than dispersed management, but it improves system performance due to its judgments being more informed and certain. Hence, network units must be in perfect sync to keep QoS adequate before starting the On/Off and traffic offloading procedures [119].

According to the required quality of service demands, such as power and delay, processing workloads can be offloaded from edge units to the cloud or vice versa, much as traffic offloading in cellular networks [120]. That is to say, computing duties can be handled locally by the edge equipment or remotely by the cloud via the MBS’s back-haul connections [121]. Nevertheless, the load on servers in the cloud, communication supplies, and backhaul lines will naturally grow when jobs are pushed to the main cloud. Planned task loading in stacked cloud-fog architectures can also add significant latency due to increased communication latency [122]. Hence, the effects of job loading on the computational and communication nodes must be considered.

From a computational standpoint, executing complicated activities in the central cloud represents a significant advancement in enhancing computing possibilities for humans and computers [123]. Nonetheless, the continually rising quantity of connected equipment within the realms of the IoTs, intelligent homes, and driverless driving will ultimately overwhelm or potentially incapacitate servers in the cloud. Therefore, it is imperative to filter data to alleviate the strain on cloud and network assets and to enhance the quality of user experience, particularly concerning end-to-end latency [124]. Numerous obstacles remain for the combined networks for connectivity, detecting, and computing, particularly in the advancement of ISAC innovations, connected sensing technologies, and processing power network solutions facilitated by the combination of these disciplines. The accessibility and main networks must be able to communicate, sense, and compute in order for the linked sensing with additional computation, sensing, and connectivity to be realized. As a result, specifications for the processing and transceiver parts, network architecture, media access control, navigation procedure, and placement of resources are proposed.

Collaboration sensing across multiple nodes in a network and the need for individualized data collection are further factors to think about. Lastly, research into methods of providing extensive advance knowledge for the best scheduling selection of decentralized computational power via the network sensing feature is still required based on the multi-layer ubiquitous computing system. As a result, using exchanged global computing resources to conduct tailored extraction of features and data synthesis on sensor data is difficult. In addition, it would be fascinating to investigate in the future ways to increase the power of ubiquitous computing by improving communication quality.

3.2.4. Digital Twins Networks

The implementation of 6G’s all-digital concept relies heavily on digital twin modern technology. One positive aspect is that new digital twin applications, such as digital twin body area network [46] and digital twin city [125], will be made possible by the 6G network, which will be an improvement over 5G in terms of performance. By contrast, a more protected, productive, smart, and visually represented 6G network can be realized more quickly with the help of digital twin innovation applied to communication networks. This is achieved via the use of real-time linking and interactivity between the two networks.

Significant network necessities, such as holo-graphic system virtual-real communication visualization, entire life cycle administration, and real-time closed-loop management [126], are inherent to the network virtual twin concept but also provide a number of challenges when trying to put it into practice. Initially, the 6G network will have extensive network components interconnected through intricate network architectures. Consequently, modelling a genuine actual network in real time presents a significant problem.

The wireless route digital twin is an essential component of the system digital twin. A precise, real-time scenario-adaptive channel framework must effectively define the transmission setting in immediate terms and forecast upcoming alterations in wireless channels. Furthermore, due to the discrepancies in the practical execution and functionalities of hardware from various manufacturers inside the system, it is imperative to focus on the suitability of network machinery during data gathering, handling, and modelling [127]. Moreover, the extensive 6G network presents significant issues with gathering information, retrieval, administration, and analysis. It is essential to investigate the computational construction of complex network structure links in extensive systems [127].

3.2.5. Al-Enabled Networks

Network AI is a crucial enabling technology in the 6G era for AI-enabled smart wireless telecommunication networks. It offers a whole AI ecosystem within the network itself, including AI facilities AI process logic, data and design solutions, etc. [128]. The field of network AI remains in its infancy, which means that there is a great deal of unanswered questions. This is despite the fact that AI innovations are advancing at a quick pace and that the previously described fresh approaches may make an impact. To begin, studies on artificial intelligence in wireless networks rely heavily on data. Consequently, the most pressing issue is figuring out methods to set up a shared data set for investigations as well as how to gather and utilize data in wireless networks. Paying close attention to the functioning of AI products is also essential when implementing AI to the network. Improved AI functionality may be obtained by carefully selecting AI algorithms and allocating network assets.

An internet-based network design method that adjusts to network fluctuations and modelling the association across AI outcomes and network setup are the biggest obstacles [129]. It is critical to investigate ways to enhance the EE and decrease energy usage as well as expenses prior to the widespread implementation of AI services [130]. This is because more sophisticated and precise AI systems tend to use a lot of resources and have substantial negative environmental effects.

If present trends continue, the increasing computing load of deep learning may soon limit its use in various fields, rendering the attainment of crucial benchmark goals unreachable. Deep learning becoming computationally limited would not be unprecedented in history. Productivity was constrained by the processing resources accessible even when Frank Rosenblatt created the initial neural networks. Thanks to acceleration from switching to specially designed equipment and a desire to dedicate additional assets to enhance productivity, many functional limitations have been loosened in the last decade. As the amount of data sets increases, the amount of deep learning parameters also has to expand, which can lead to overparameterization. Computational demands increase by a minimum the square of the data sets in the overparameterized case, while the expense of training a deep learning structure grows with a combination of the total amount of variables and the quantity of data sets. When it comes to deep learning systems, this exponential expansion is not nearly quick enough to boost ability. In order to see a linear improvement in achievement, the volume of data used for training typically needs to rise at a faster-than-linear rate. Linear regression provides a good example of a well-developed mathematical acquiring concept (and, by extension, a 1-layer by layer neural system with linear activations) that can be used to comprehend the connection among deep learning system variables, data, and computational demands.

One possible approach to reducing the computational load on deep learning systems is enhancing data curation. More specifically, instead of using more data, models might be trained with improved data to decrease processing requirements [131]. Although higher-quality data would ideally reduce computing expenses, it may still have an unclear impact on overall prices due to the high expense of data curation (e.g., higher human annotator expenses).

One alternative to deal with deep learning’s increasing computing load is to switch to different forms of machine learning, some of which may be just as unexplored or undervalued. Even though “expert” approaches are often cheaper in terms of computing, their efficiency will eventually halt if the experts involved are unable to identify all of the elements that a model with greater flexibility could test. Awareness of known items (e.g., automobiles) [132] and techniques encouraged by biology (e.g., for learning neural controller structures [133]) are two areas in which such approaches are currently outperforming deep learning models.

By locating calculations in energy-efficient data canters, we may lessen the environmental effect of every calculation. Many factors contribute to the efficiency of such data canters, including (i) their location in regions with typical cooling and (ii) their architecture, which allows for more efficient use of the cooling systems that are required. Methods like smart load allocation, flexible voltage expanding, and server downsizing are all examples of how server administration can improve data centre efficiency [134].

The carbon footprint of deep learning systems is further affected by the energy that powers servers. For instance, according to Google’s 2021 sustainability reveal, the tech giant bought sufficient clean energy to power all of its workplaces and data canters [135]. This indicates that carbon production can vary substantially depending on the region of hosting [136].

All the challenges that were explained in this section are summarized in Table 3.

Table 3. Summary of future research challenges of supporting technologies for the 6G networks.

Hybrid RF–Optical Networks	Space–Air–Ground–Sear (SAGS) Networks	Integrated Sensing and Computing Networks	Cloud-Edge Computing	Digital Twins Networks	Al-Enabled Networks
mixed-signal and interference, transfer protocols development, traffic and load balancing, data synchronization, distribution of resources, energy efficiency enhancement, distribution of power, connection access points, access point selection, handover and blockages, infrastructure and power for buildings, mobility and user speed.	network design, building and development, preservation and maintenance, optimization and upgrading, channel structure and topology, coverage and hardware capabilities, environment and weather, adaptability and scalability, installation of aerial BSs, accuracy of the UAV-to-ground channel, UAV-to-ship connectivity settings, mobility, single and multi-bounce, reflections and fading of multiple paths, structure layout, network procedure, resource placement and routing tactics, elevated network latency, sporadic interruptions, imbalanced network traffic, information hub and data privacy, data collection.	computational complexity, transmission data rate requirements, range of transmission, base stations and network reach, LOS route and multi-bounce route, hardware requirements, authentication mechanisms, user data and privacy, components management, data collection and management, communication, detecting, and processing, waveform layout and processing of signals, integrated sensing and communication detection, power supplies, systems’ architectures.	computing core location, data provider reliability, security and data retrieval, storage and capacity, network structure, cloud-computing model, network congestion and delay, collecting, interpreting, analyzing, and mining data, user data and privacy, network speed, gateway asset distribution, decision-making process, task allocation, assets distribution, and segmentation, use of energy, servers and load distribution, task loading and scheduling, media access control.	computational complexity, user data and privacy, power supplies, running costs, systems’ architectures, user interaction modelling, network speed, media access control, body area network requirements, performance optimization, visualization, closed-loop management, life cycle administration, practical execution and functionalities of hardware, data gathering, handling, and modelling, information analysis.	resource usage, processing techniques, AI model, sensory wireless systems, decision-making process, running costs, systems’ architectures, transmission setting and network setup, information analysis, shared data set policies, AI algorithms, assets allocation, negative environmental effects, computing load of deep learning, computational demands, new model training, high quantity of data sets, data types, data curation expenses, energy-efficient in data centres.

Intelligent slice management solutions for 5G and beyond networks have been the subject of few research that have taken AI’s potential into account. The concept of network slicing allows for the coexistence of numerous separate virtual networks, or slices, on top of a single physical network. Network slicing has many benefits [137]. First, network slicing allows for multiple tenancy, or the sharing of physical network infrastructure, via multiplexing virtual networks. Second, network slicing allows for the creation of customized slices for different service types with different quality of service (QoS) requirements. This allows for service differentiation and guarantees service level agreements (SLAs) for each service type. Third, network slicing increases the flexibility and adaptability in network management. It reduces capital expense in network deployment and operation. In order to keep network operation performance high, fulfil new key performance indicators (KPIs), such as dependability, energy consumption, and data quality, and support various and heterogeneous services, the study [138] suggested an efficient, intelligent network slicing framework.

While achieving high-speed communication technologies, the goal of constructing green, low-carbon 6G networks is to minimize environmental effect. Sixth-generation networks may greatly mitigate their impact on climate change by lowering their energy usage and carbon emissions. Maximizing the usage of renewable energy in network base stations and data centres is a primary goal in the design and implementation of 6G network equipment and infrastructure. This will improve energy efficiency and decrease waste. The research [139] offered a model for a 6G wireless environmentally friendly network design. This design’s standout qualities are its ability to schedule tasks in an energy-efficient manner, its intelligent communication nodes, and the thorough separation of the user and control planes. Ensures energy efficiency and sustainability in 6G networks by discussing the difficulties and opportunities in [140].

4. Conclusions

6G networks will enhance worldwide communication solutions by introducing novel application cases, offering innovative technology knowledge, and fostering growth in the economy. We have rigorously evaluated the most recent approaches presented in a substantial corpus of pertinent literature, emphasizing the related advancements and obstacles. We have deliberated on the corresponding concept for 6G. The fundamental difficulties confronting 6G research and the relevant research paths have been examined via the lenses of basic research, sustainable networks, and the essential technologies produced to promote the envisioned 6G framework.

In summary, 6G studies and regulation continue to encounter several unresolved problems. This study has elucidated the anticipated attractive characteristics of 6G, and it is anticipated that it has offered renewed inspiration and encouragement for the community’s 6G development efforts. The discussion and analysis that were presented in this study is expected to benefit the community in the near future especially the developers and researchers.

The following contributions have been achieved in this study:

• Overview of the 6G networks, its features, characteristics, goals and objectives;
• Recent studies and insights of the supporting technologies including AI, digital twin, hybrid and integrated networks, and cloud;
• Review and discussion of the challenges of the supporting technologies;
• Challenges and recommendations for the challenges have been presented and discussed;
• Future research challenges have been presented and summarized.

Future directions for the 6G networks are expected to strengthen the technology and provide new solutions, including integrating 6G network with the supporting technologies in various environments and applications. Another promising avenue for future research in 6G networks lies in AI-driven studies, encompassing the development of advanced models and innovative applications tailored to enhance network intelligence, efficiency, and adaptability. AI algorithm performance comparison is also another future direction that will provide further insights.