diff --git "a/HtFJT4oBgHgl3EQfFSwh/content/tmp_files/2301.11441v1.pdf.txt" "b/HtFJT4oBgHgl3EQfFSwh/content/tmp_files/2301.11441v1.pdf.txt" new file mode 100644--- /dev/null +++ "b/HtFJT4oBgHgl3EQfFSwh/content/tmp_files/2301.11441v1.pdf.txt" @@ -0,0 +1,3479 @@ +1 +Metaverse for Wireless Systems: Architecture, +Advances, Standardization, and Open Challenges +Latif U. Khan, Mohsen Guizani, Fellow, IEEE, Dusit Niyato, Fellow, IEEE, Ala Al-Fuqaha, Senior +Member, IEEE, M´erouane Debbah, Fellow, IEEE +Abstract—The growing landscape of emerging wireless ap- +plications is a key driver toward the development of novel +wireless system designs. Such a design can be based on the +metaverse that uses a virtual model of the physical world systems +along with other schemes/technologies (e.g., optimization theory, +machine learning, and blockchain). A metaverse using a virtual +model performs proactive intelligent analytics prior to a user +request for efficient management of the wireless system resources. +Additionally, a metaverse will enable self-sustainability to operate +wireless systems with the least possible intervention from network +operators. Although the metaverse can offer many benefits, it +faces some challenges as well. Therefore, in this tutorial, we +discuss the role of a metaverse in enabling wireless applications. +We present an overview, key enablers, design aspects (i.e., meta- +verse for wireless and wireless for metaverse), and a novel high- +level architecture of metaverse-based wireless systems. We discuss +metaverse management, reliability, and security of the metaverse- +based system. Furthermore, we discuss recent advances and +standardization of metaverse-enabled wireless system. Finally, we +outline open challenges and present possible solutions. +Index Terms—Virtual reality, mixed reality, augmented reality, +digital twins, and metaverse. +I. INTRODUCTION +The landscape of wireless systems incurred significant +growth during the last few decades. Emerging wireless system +applications have diverse requirements. These diverse require- +ments are in terms of user-defined metrics (e.g., quality of +physical experience) and quality of service (QoS) requirements +(e.g., strict latency and ultra-high reliability). Fulfilling these +diverse requirements is difficult for existing wireless system +infrastructures (e.g., 5G). Many recent works proposed the +use of 6G for such applications [1]–[3]. 6G is still in its +infancy, and many milestones are needed to realize its true +implementation. The work in [3] proposes a digital twin-based +architecture for 6G that consists of three layers: the physical +interaction layer, the twin layer, and the service layer. A twin- +based architecture tries to follow the trends of proactive-online +learning-based systems and self-sustaining wireless systems. +In the case of self-sustainability, there is a minimal possible +L. U. Khan and M. Guizani are with the Department of Machine Learning, +Mohamed Bin Zayed University of Artificial Intelligence, United Arab Emi- +rates. +D. Niyato is with the School of Computer Science and Engineering, +Nanyang Technological University, Singapore. +A. Al-Fuqaha is with the College of Science and Engineering, Hamad Bin +Khalifa University, Qatar. +M. Debbah is with the Technology Innovation Institute, United Arab +Emirates, and also with the Department of Machine Learning, Mohamed Bin +Zayed University of Artificial Intelligence, United Arab Emirates. +Big +Data +Cloud +SBS +with MEC +Edge-based meta space +Physical space +Smart +industry +Autonomous +avators +Mobility +x km/h +East +Legends +Interactive experience +technologies information +Twins +Smart +Hospital +OPD and emergency +Twin of +BS +Twin of +hospital +Twin of smart +factory +Avatars +Avatars +Fig. 1: Example of a metaverse-based wireless system. +intervention requirement from operators, whereas proactive +online learning is needed for analysis of the system before +deployment. To manage wireless and computing resources for +wireless applications with strict latency, there is a need to +proactively analyze the wireless system. +Although a digital twin-based system can offer benefits, +it seems difficult to truly meet the diverse requirements of +wireless systems [1], [3]. For instance, digital twin does not +effectively consider the users/devices mobility which signifi- +cantly affects the performance of wireless systems. Consider +a terahertz (THz) communication system that is significantly +affected (e.g., loss in line of sight (LOS) path) by the human +body. Similarly, the mobility of devices/users significantly +affects the performance of wireless systems. Therefore, we +must effectively take into account the effect of user/device +mobility. To do so, the work in [4] introduced the concept +of metaverse that uses digital avatars to account for mobile +devices/users. A metaverse-based system can be divided into +two spaces, such as (a) meta space and (b) physical space [5]. +One can implement a meta space (i.e., virtual model) using +edge or cloud. On the other hand, all the physical entities (e.g., +edge/cloud servers and devices) that are required for wireless +systems are included in the physical space. Fig. 1 illustrates the +example of wireless system using metaverse. The static entities +arXiv:2301.11441v1 [cs.NI] 26 Jan 2023 + +2 +Design Trends +True self-sustaining wireless systems +True proactive-online learning-based wireless systems +Wireless control center as a black box +These +applications +have +diverse requirements that +need novel wireless system +design. These requirements +are in terms of traditional +quality of service metrics +and user-defined metrics +To +address +diverse +requirements of emerging +applications, we need to +follow these design trends +Driving applications +Key enablers +Main key enablers of +metaverse are avatars, +twins, and interactive +experience technologies +which +are +in +turn +enabled +by +other +technologies/schemes +(e.g., digital modeling +schemes, +artificial +intelligence, +edge +computing, +terahertz +communication, natural +language +processing, +expert systems) +Sensing and +localization +technologies +Digital modeling +technologies +3D modeling +Mathematical +modeling +Data driven +modeling +Experimental +modeling +Terahertz +sensing +Passive sensing +using radar +Channel charting +Context-aware +localization +Communication +technologies +Terahertz +communication +Millimeter-wave +communication +Visible light +communication +Interactive +experience +technologies +Virtual reality +Augmented +reality +Mixed reality +Extended +reality +Artificial +intelligence +Robotics +Expert systems +Computer vision +Natural language +processing +Machine +learning +Computing & +distributed +ledger +technologies +Cloud +computing +Edge computing +Quantum +computing +Blockchain +Extended reality +Brain-computer +interaction +Autonomous +connected vehicles +Smart industry +Intelligent transportation +system +Haptics +applications +xxx +Digital twin +VR/AR/MR/ +XR +Avatars +Fig. 2: Metaverse for wireless systems: Applications, design trends, and key enablers. +are represented by twins in the meta space, whereas the mobile +entities are represented by digital avatars. We will discuss +more regarding the architecture of a metaverse-based wireless +system in Section II-C. An overview of emerging applications, +design trends, and key enablers is given Fig. 2. Emerging +applications are characterized by diverse requirements that +must be fulfilled. These requirements can be fulfilled by +following the design trends of self-sustainability and proactive +online learning analytics. These design trends are met using +metaverse-enabled design that predominantly uses avatars, +digital twins, and interactive experience technologies. Next, +we discuss the research statistics and research trends of the +metaverse and Internet of Everything (IoE). +A. Research Trends and Statistics +According to statistics of [10], the market value of +metaverse in 2021 was USD 51.69 billion and it is expected +to increase at a compound annual growth rate (CAGR) of +44.5% to reach USD 1.3 trillion by 2030. In 2021, the market +share of North America was 46% which made it the highest +contributor among all regions in the world. Among regions, +Asia Pacific will expect the highest growth among all regions. +On the other hand, Qualcomm Technologies, Inc., Roblox Cor- +poration, Decentraland, The Sandbox, Snap Inc., Nextech AR +Solutions Inc., NVIDIA Corporation, Epic Games, Microsoft, +and META will be major players in the metaverse market. +Among applications (e.g., gaming, social media, and virtual +reality), the gaming sector seems to have the highest share in +2021. +Other than the metaverse, the IoE market (i.e., it will +account for emerging wireless applications) share is expected +to reach USD 3,335.2 Billion, globally, by 2027, at 15.1% +CAGR [11]. The key drivers of this increase are the in- +creased implementation of M2M systems and the emergence +of numerous disruptive technologies. On the other hand, the +key players are Cisco System Inc. (US), Nokia Corporation +(Finland), Samsung (South Korea), Huawei Technologies Co +Ltd. (China), Amazon Web Services (US), Qualcomm (US), +AT& T Inc. (US), Koninklijke Philips (Netherlands), Mesh +Systems LLC (US), and Robert Bosch AG (Germany). Addi- +tionally, among the regions, North America is the dominant +region. From the aforementioned discussion, it is clear that the +metaverse and IoE will be key research technologies in the +foreseeable future due to their increasing demand and market +shares. +B. Existing Surveys and Tutorials +Various works [4]–[9] considered metaverse. The work in +[6] surveyed metaverse applications and their recent advances. +Moreover, they studied various initiatives for realizing the +metaverse in various countries. Another work [7] discussed the +metaverse fundamentals and security aspects. Specifically, the + +3 +TABLE I: Summary of existing surveys and tutorials and their primary focus +Reference +Wireless +for +metaverse +Metaverse +for wireless +Recent ad- +vances +StandardizationFocus +Remark +Ning et al. [6] + + + + +Surveyed the concept +and current activities in +various countries for re- +alizing metaverse. +N/A +Wang et al. [7] + + + +(only +for +commu- +nication +between +virtual +and +real world) +Surveyed concept, se- +curity, and privacy of +metaverse. +Our work will present standard- +ization for ML-enable wireless +metaverse +Gadekallu +et +al. [8] + + + + +Surveyed the role of +blockchain +for +meta- +verse +N/A +Khan et al. [4] + + + + +Presented +vision +of +metaverse for wireless +networks +N/A +Khan et al. [5] + + + + +Presented role of ML +in enabling metaverse- +based wireless system +N/A +Xu et al. [9] +(considered +network +edge +for +enabling +metaverse) + + + +Surveyed +concept, +enablers, +computing, +and communication for +edge-based metaverse. +Our work present a novel ar- +chitecture with meta space (i.e., +twins and digital avatars based +on virtual machines) and physical +space. We will discuss how to +deploy them using edge, cloud, +and devices. Moreover, we will +present novel challenges com- +pared to existing surveys +Our Tutorial + + + + + +N/A +authors discussed the security threats and solutions to various +components (e.g., physical space and data management) of the +metaverse. The work in [8] surveyed the role of blockchain in +metaverse. The authors in [4] presented the vision of metaverse +for enabling wireless applications. Specifically, they presented +key requirements, general architecture, and open challenges. +Another work [5] discussed the role of machine learning in +enabling metaverse-enabled wireless systems. The work in +[9] surveyed key enablers, computing, and communication +technologies for the metaverse. Different from the existing +works [4]–[9], our work (as given in Table I) presents the fun- +damentals, key enablers, and recent advances. Additionally, we +present the metaverse management, security, and reliability of +meta space and physical space. Finally, present standardization +of the machine learning-enabled metaverse. +C. Our Tutorial +Our tutorial is different from the existing works [4]–[9] +and will present an overview, design aspects (i.e., metaverse +for wireless and wireless for metaverse), and an architecture +consisting of meta space and physical space along with inter- +faces. We also review the networking, reliability, and security +and present the novel aspects compared to existing works. Fur- +thermore, we discuss the standardization of machine learning- +based metaverse and present recent advances of enabling +wireless applications by metaverse. Finally, we present open +challenges. Specifically, our tutorial will answer the questions +as shown in Fig. 3. Our contributions are summarized as +follows. +• We present an overview, key design aspects, key enablers, +and novel architecture for the metaverse of wireless +systems. Our architecture consists of two spaces: meta +space and physical space that will interact using wireless +interfaces. +• We discuss networking management, reliability, and se- +curity for enabling meta space and physical space. +• We present recent of metaverse towards enabling emerg- +ing IoE applications. Additionally, we discuss standard- +ization of machine learning-based metaverse. +• Finally, novel open research challenges with suggestions +are presented. +II. FOUNDATIONS OF METAVERSE-BASED WIRELESS +SYSTEMS +A. Design Aspects +A metaverse in the perspective of wireless network can +be used to represent various network entities, as shown in +Fig. 4. With the metaverse and wireless systems, we can +have two possible design aspects, as shown in Fig. 5. For +every aspect, the role is shown for meta space, interfaces, +and physical space. Note that a more detailed discussion +about the architecture will be given in Section II-C. The + +4 +Enablers +(Digital twins, avatars, and interactive experience +technologies) + How can we use digital twin and avatars to +model meta space? + How can we use augmented reality, virtual +reality, mixed reality, and extended reality to +help implementation of meta space? +Standardization +(Interface standardization, meta space standardization, AI- +enabled metaverse standardization) + How do we standardize machine learning-based +metaverse? +Deployment +(Edge and cloud based deployment) + How can we use edge, cloud, and devices for +deployment of meta space on a single device/ +entity? + How can we efficient deploy and manage meta +space using multiple devices based on edge, +cloud, and end-devices? +Resource management +(Radio access network, core network, and computing +resources) + How can we efficiently manage communication +resources (i.e., wireless and wired)? + How can we efficiently manage computing +resources (end-devices and edge/cloud servers)? +Security and Privacy +(Physical space, meta space, and interface security) + How can we make wireless interface secure? + How does one enable secure and privacy-aware +training of meta space models? + How do we enable secure meta space and +physical space? +Business models +(Non-fungible tokens and blockchain) + How does one use blockchain for secure +business models for a metaverse-based wireless +system? + How do we use non-fungible tokens for trading +of metaverse-based wireless systems? +Sections II +Virtual model +Avatars +AR/VR/MR/ +XR +Digital twin +Section V +Sections III and IV +Sections III and IV +Sections III and IV +Sections VI +Fig. 3: Overview of research questions and relevant sections. +design aspects are metaverse for the wireless and wireless +for metaverse. A metaverse for wireless systems deals with +resource optimization of computing and communication re- +sources for effectively enabling various wireless applications. +Wireless for metaverse deals with carrying out signaling for +metaverse-enabled wireless system operation. Such a signaling +will be used for the efficient placement of meta space objects +(i.e., twins and digital avatars). For instance, consider the +deployment of meta space using multiple edge and cloud +servers (a more detailed discussion about the deployment will +be given in Section III-A1). To do so, there is a need to +efficiently deploy meta space in such a way as to minimize the +cost. Such a cost can be possibly a transmission latency and +energy consumption. To minimize this, one must choose a set +of edge servers that will result in low latency and low trans- +mission energy. Additionally, to enable proactive analytics, we +must perform training to obtain pre-trained models. Such pre- +trained models can be based on either centralized learning +or distributed learning. Centralized learning will require the +migration of device data to the cloud for training; however, +it has an inherent issue of privacy leakage. To address this, +one can use distributed learning that is based on iterative +interaction between the devices and edge/cloud. To enable +such interaction for getting pre-trained models, there is a +need for effective computing and communication resources +management. Other than wireless for metaverse, metaverse +for wireless will require efficient management of resources +for various applications. For instance, consider infotainment +in autonomous driving cars that require computing resources +at both cars and edge servers installed at the roadside units +(RSUs). Due to the presence of many computing tasks in +autonomous cars, there is a need for offloading these tasks +to the RSUs. How to perform such offloading and manage- +Single component +(e.g., end-device and +autonomous car) +End-device functions +modeling (e.g., edge +resource management) +Modeling of new +devices (e.g., novel +XR headsets for +human-computer +interaction) +Complete service (e.g., +Intelligent +transportation system +and digital healthcare) +Modeling of new +services (e.g., Brain- +computer interaction in +healthcare) +Resource management +for existing services +(e.g., Autonomous +driving cars) +Metaverse for +wireless systems +Fig. 4: Possible uses of metaverse in wireless systems. +ment of computing as well as communication resources? A +metaverse will effectively enable such offloading by making +offloading decisions and management of computing as well as +communication resources. +B. Key Enablers +1) Digital Twins: A digital twin is a virtual model of the +physical wireless system (example scenario is shown in Fig. 1) +[5]. Such a virtual model will be used for analysis and help +in controlling network components/devices. One can model a + +5 +Meta space +Physical space +Virtual model +Avatars +AR/VR/MR/ +XR +Digital twin +Metaverse for wireless +Wireless for metaverse +Enabling IoE applications +using metaverse +Resource management for +metaverse signaling + Seamless interaction of +heterogeneous +IoE +entities + Computing +resource +management + Avatars +and +twin +objects migration due to +mobility of end-devices + Modeling of twins and +avatars + Efficient decoupling of +meta space from the +physical space using +network slicing + Wireless +resource +management +for +interfaces + Efficient +design +of +wireless +interfacing +devices for metaverse +signaling + Encryption/decryption, +channel coding schemes for +metaverse signaling as well +as data +Meta space to physical +space interface +Physical space to meta +space interface +Fig. 5: Metaverse and wireless system design aspects. +TABLE II: Focus of existing surveys and tutorials for meta- +verse key enablers +Existing +sur- +veys/tutorial +Mentioned enabler +Focus +Khan +et +al. [4] +• Digital twins +• Avatars +• Interactive +experience +technologies +This work introduced the +key enablers along with +role of ML in modeling +them without primarily fo- +cusing on mobility mod- +eling of avatars in meta +space. +Khan +et +al. [5] +• Digital twins +• Avatars +• Interactive +experience +technologies +This work focused on role +of ML for modeling twins +and avatars without dis- +cussing mobility as well +as +pose +estimation +for +modeling avatars. +Xu et al. [9] +• Digital twins +• Avatars +• Interactive +experience +technologies +The authors discussed in +detail +about +interactive +experience +technologies +without +providing +in +depth information about +modeling of avatars/twins. +Our work +• Digital twins +• Avatars +• Interactive +experience +technologies +Our work discusses in de- +tail about the twins/avatars +modeling as well as in- +teractive experience tech- +nologies. Additionally, we +discuss in detail about mo- +bility modeling and pose +estimation for modeling +avatars in meta space. +digital twin using experimental modeling, data driven model- +ing, and mathematical modeling. In mathematical modeling, +various assumptions are made during modeling of the real +world systems. For instance, non-linear functions of robotic +systems are generally modeled using linear assumptions, and +thus might not reflect more effective modeling. To handle +this issue, we can employ experimental modeling. A series +of experiments is needed for modeling a physical system +in the case of experimental modeling. Experimental errors +(i.e., human errors and machine errors) are the limitations +of the experimental modeling. One can address the issues +related to experimental and mathematical modeling by using +data-driven modeling that involves using the generated data. +Such a generated data can train machine learning models +using either decentralized training or centralized training. A +centralized training-based machine learning trains a model at +a centralized location. It requires moving device data to the +centralized location, and thus might have privacy loss due +to malicious attack at the cloud. To avoid such an attack, +we can deploy distributed learning (i.e., federated learning). +Federated learning for modeling twins in the metaverse will +involve frequent interaction between end-devices and twins +deployed at edge/cloud. Although FL enables on-devices +machine learning, there are many scenarios where there are +significant limitations on the available computing at end- +devices, and thus they might not be able to compute their +local models within the deadline. To address this issue, a few +works [12]–[14] proposed split FL (SFL) that is based on +computing partial local models by the end-devices and the +remaining at the edge/cloud servers. Different from traditional +FL, SFL needs to offload partial local model computing tasks +to the edge/cloud servers, therefore, there must be an effective +computing resource allocation scheme for SFL. +2) Digital Avatars: A digital avatar is a digital replica of +the humans/ mobile devices controlled by humans in the phys- +ical interaction world. Note that the key enablers presented +here are covering different aspects compared to existing works, +shown in Fig. II. One can use the digital twin to represent +the virtual model of humans in a meta space. Additionally, +interactive experience technologies (e.g., XR, MR, AR, and +VR) can also represent humans in a virtual world. However, +both interactive experience technologies and digital twins +might not effectively represent humans in a meta space. A +human body in physical wireless systems significantly impacts +the quality of service (QoS). Different from the traditional +metaverse, where avatars can be two-dimensional or three- +dimensional (3D) [15], metaverse for a wireless system should +consider 3D avatars. Such a 3D avatar effectively represents +the behavior of humans (e.g., effect of 3D human body on +signal attenuation) in the metaverse, as shown in Fig. 7. From +onward, we will use the digital twin avatars for 3D avatar. To +model 3D avatars for the metaverse, one can propose novel +tools. Existing software tools for 3D modeling of humans are +MakeHuman, Daz Studio, iClone, and Mixamo [16]. These +tools are used for illustration, animation, cinema, and video +games. Although these tools can effectively model humans +in computer simulations, animations, etc., there is a need to +propose a novel design for a metaverse-based wireless system. +The nature of a 3D model of a human in a metaverse- +based wireless system will be different compared to other +applications (e.g., animations). The difference lies in the +incorporation of wireless system features in the avatar of +the metaverse-based wireless system. For instance, Terahertz +(THz) communication is significantly affected by the human +body. A LOS path might be affected by humans. Additionally, +THz communication is significantly affected by the concen- +tration of red blood cells (RBCs) in the human blood. Other + +6 +Sub-global +model 1 +Sub-global +model 2 +Sub-global +model 3 +Transfer of sub- +global models +using wireless +channel +Global +model +Global +model +Global +model +Cluster of devices +with similar +mobility +Fig. 6: Dispersed federated learning for training meta space +models [17]. +applications (e.g., 3D printing, human-computer interaction) +needs accurate avatar modeling. An overview of avatars in +a metaverse-based wireless system is shown in Fig. 1. First, +there is a need to effectively create a 3D model of a human. +Next, we should add effect (e.g., loss in LOS path for THz +communication) of a wireless system in the avatars. We should +also effectively model an avatar’s mobility that can effectively +follow human mobility patterns. Such mobility patterns can +be modeled using various techniques (e.g., optimization the- +ory and deep learning). Mobility management is of signif- +icant importance in traditional wireless networks and many +works proposed various schemes. However, here, a metaverse- +enabled wireless system is different compared to traditional +wireless networks. A meta space deployed at the network +edge must be migrated to the new edge depending on the +device’s mobility. Such migration of meta space can be either +live or non-live. More detailed discussion about migration +schemes will be provided in Sections IV-A2 and III-A1. For +mobility management, one can use prediction schemes based +on deep learning for predicting the device’s mobility. Based +on the predicted mobility, one can perform migration of meta +space. On the other hand, in a metaverse-enabled wireless +system, one must manage the mobility of devices during the +training of meta space models using distributed learning. It +is desirable for devices to remain a range of edge servers +performing aggregation for fast convergence. Therefore, one +should manage device mobility during the training process of +distributed learning models for meta space. Mostly, devices +are mobile, therefore, one must manage such mobility. We can +use dispersed federated learning (i.e., shown in Fig. 6) that is +based on the clustering of devices. Note that devices that will +remain within the vicinity of each other will be placed in a +cluster and a sub-global model is learned, as shown in Fig 6. +Next, to train sub-global models, one can share the sub-global +models among different clusters to yield a global model. +To enable interaction between avatars and the digital twin +objects in a meta space for accurate operation and analysis +prior to deployment, there is a need for efficient computer +vision techniques, such as human-pose tracking, emotion, +(a) Kinematic model +(b) Planar model +(c) Volumetric model +Fig. 7: Categories of human modeling for pose detection. +and expression recognition, and gesture recognition, among +others [18]. Human pose tracking enables the estimation of +multi-person human geometric and motion information. Such +an estimation of human key points-trajectories is necessary +for the performance optimization of metaverse-based wireless +systems. For instance, the exact location and movement of +human body parts can be used for accurate channel estimation +of wireless signals. For instance, meta space having avatars +and twins can be used for the analysis of wireless systems. +As meta space is running a virtual world, therefore, accurate +pose estimation of avatar is necessary for better analysis. +Also, if the meta space interacts with physical space during +the run-time control of physical objects, there is a need to +accurately estimate human positions. Additionally, a human +pose estimation will have a significant impact on various other +applications (e.g., human-computer interaction, healthcare ap- +plications, AR, and VR). Therefore, human pose tracking in +a metaverse-based wireless system has significant importance +for wireless systems. To do pose estimation, the first step is to +model humans. There are three categories, such as kinematic, +planar, and volumetric, of human models shown in Fig 7. +A human body has limbs and joints as well as it has body +shape information and contains body kinematic structure. The +kinematic model is based on a representation of the human +body using limb orientation and joint positions. To represent +humans in more detail, one can use planar models that use +rectangle-type representations to show different parts of the +human body. Although we can represent humans using a +kinematic model and planar model, there is a need for a more +detailed model of humans, such as a volumetric model (i.e., +3D human reconstruction) for the metaverse. +In addition to human modeling, there is a need for +efficient human pose estimation. Human pose estimation can +be of two types: two-dimensional (2D) and three-dimensional +(3D), as shown in Figs. 8 and 9 [19]. In 2D pose estimation, +poses are estimated using images (in terms of pixel values), +whereas 3D pose estimation involves estimation results in +three dimensional spatial arrangement of the human body. +Recent works considered deep learning for estimating human +poses and have shown promising results [20]–[22]. Most +of these works are focused on transforming low-resolution +images into high resolution images for accurately estimating +human poses. Sun et al. in [22] proposed an architecture for +2D estimation, namely, HighResolution Net for maintaining + +7 +Human +detector +2D pose +network +Input +image +Detected +humans +Single +person pose +Output 2D multi- +person poses +(a) Top down approach +2D pose +detector +Body part +association +Input +image +Output 2D multi- +person poses +(a) Top down approach +Body part candidate +detection +Fig. 8: Example of 2D pose estimation [19]. +high resolution during the whole process for estimating 2D +human pose using movements of joints, as shown in Fig. 7c. +Although 2D pose estimation of humans can offer benefits, it +has a few limitations. 2D pose estimation can estimate only +joint movements, not exact human models, and thus might +not be more desirable for metaverse-based systems. Therefore, +there is a need for 3D pose tracking of humans while modeling +digital avatars. In [23], the authors proposed a WiFi-based +IoT-enabled human pose estimation system, namely, MetaFi +for digital avatars. Specifically, the authors used Wi-Fi signals +to estimate the human pose for the metaverse as motivated +by the use of Wi-Fi for human activity recognition. The +MetaFi system comprises two COTS WiFi routers (i.e., TP- +Link N750) acting as a receiver and transmitter. Such data +is sent to the server for AI model inference. Although MetFi +AI model can be easily used for human pose estimation, it +might not perform well in all scenarios. To do so, there is a +need for considering large data sets from a wide variety of +users. A certain group of people in an institution might not +want to share such data outside their institution. To address +this issue, one can use cross-silo federated learning that can +train a human pose estimation model within one institution +and then share only the learning model updates with the other +institution. +3) Interactive Experience Technologies: A trend of a +virtuality-reality continuum is followed by interactive expe- +rience technologies. VR is based on synthetic views along +with additional information and is on the virtuality end, +whereas, AR lies at the reality end. AR is based on enhancing +physical view by using additional information. In a mixed +reality (MR), the virtual world and physical words are merged +and they interact in real time. Extended reality (XR) merges +all three interactive experience technologies, such as MR, +VR, and AR. XR will enable fine-grained human-specific +information perception. Therefore, one can say that XR head- +End to end +network +(a) Skeleton only methods- Direct estimation approach +Off the shelf- +2D HPE +network +(b) Skeleton only methods- 2D to 3D lifting approach +3D pose +network +Off the shelf- +2D HPE +network +(c) Human mesh recovery method +3D pose +and shape +network +Model +regressor +Input image +Output 3D pose +Input image +2D pose +Output 3D pose +Input image +2D joints +Output 3D mesh +Fig. 9: Example of 3D pose estimation [19]. +mounted display, sensors, and embedded systems are the main +source of entrance to metaverse-enabled wireless systems [7]. +Although interactive experience technologies will effectively +enable metaverse, there is a need for efficient design based on +edge computing. In a metaverse-enabled wireless system, there +will be a massive number of running interactive experience +technologies. Such devices will require on-demand computing +resources with low latency. To do so, one must efficiently +manage edge computing resources for various metaverse de- +vices. Note that interactive experience technologies are well +studied in the literature, still, there is a need for more research. +The behavior of interactive experience in a metaverse will be +different and more complex compared to general applications +[4]. Therefore, there is a need for careful design consid- +erations regarding the integration of interactive experience +technologies in metaverse-enabled wireless systems. There +can be many cases where the need for interactive experience +technologies in the metaverse will be crucial [24]. Example +use cases are metaverse-assisted remote expert, metaverse- +based real-time collaboration, and metaverse-based industrial +maintenance, among others. Consider a remote expert system +for industrial maintenance based on metaverse. Cameras and +sensors installed near the industrial machine can take images +and add annotations using interactive experience technologies. +These images are sent to the remote expert using emerging +communication technologies. The remote expert after adding +annotations and suggestions will be shared with the industrial +machine operators for providing guidance to remove faults. +Meanwhile, the metaverse can use the data of the faults to +train/further train machine learning meta space models. +C. High-Level Architecture +First, we discuss the general architecture of a metaverse- +based wireless system, as shown in Fig. 10 [4], [5]. The ar- +chitecture mainly consists of three spaces: physical interaction + +8 +Meta space +Physical space +Services space +User request is translated +using semantic reasoning +Intelligent +transportation +Industry 4.0 +Smart survelliance +Extended reality +Smart farming +Smart buildings + Wireless and +computing resource +management + Devices mobility +management + Edge/cloud server for +deployment of meta +space players (e.g., +twins and avatars) +Metaverse management + Channel coding +schemes (e.g., Turbo +codes) + Homomorphic +encryption +Reliability & Security + Muti-connectivity +and packet +duplication + Efficient deployment +of twins and avatars + Twins and avatars +migration + Low Latency +Consensus +Algorithms for +Blockchain +Metaverse management + Secure implementation +of twins and avatars +using virtualization +schemes +Reliability & Security + Twins and avatars +isolation +User-defined metrics (e.g, quality of +physical experience) +Conventional metrics (e.g, Latency) +Instructions +Machine Learning +Models +Optimization +Schemes +Game Theoretic +Schemes +Virtual model +Avatars +AR/VR/MR/XR +Digital twin +Meta space +implementation +Physical to meta space interface +Meta space to physical interface +Physical entities +User requests +Offline training of twin space +models +Application translation using +semantic reasoning +Authentication +Service request (e.g., healthcare +checkup) +Deployment of meta space +using edge/cloud servers +∑ +Partial local +models +training +Partial local +models +training +Edge aggregation +Learning updates +Fig. 10: An overview of general architecture for metaverse. +space, meta space, and services space. A physical interac- +tion space has all the physical devices, humans, edge/cloud +servers and other network switches necessary for establishing +a wireless system. A meta space is a logical space that handles +the interaction of a digital twin, digital avatars, and interac- +tive experience technologies to analyze/control the physical +wireless system. On the other hand, services space enables +users to request services from a metaverse-enabled wireless +system. To deploy meta space, there is a need to first model +digital twins and avatars. One can use various ways to model +them, such as data-driven modeling, experimental modeling, +and mathematical modeling. In mathematical modeling, we +made a series of assumptions (e.g., linear approximation to +non-linear model). Coping with this limitation, one can use +experimental modeling. Experimental modeling consists of +a series of experiments that may suffer from experimen- +tal errors and equipment malfunctioning. To address these +limitations, one can use data-driven modeling. Data-driven +modeling uses data generated by wireless applications to train +machine learning models. An example of a metaverse based +wireless system is shown in Fig. 1 in Section I. Fig. 1 shows +the role of digital twins, avatars, and interactive experience +technologies in wireless systems. Digital twins in wireless +systems can be used to model the static entities in wireless +systems. These static entities are buildings, base stations, +and mountains, etc. Digital avatars refer to mobile devices +and users. For instance, a user sitting inside an autonomous +car can be modeled using an avatar for the autonomous +car. Similarly, humans with wearables can be modeled using +avatars. Modeling digital avatars in meta space might be +more challenging compared to digital twins. Digital twins +of static entities have no mobility, and thus easier to model +compared to mobile avatars. Modeling the exact mobility of +avatars is difficult. Additionally, mobile users has a significant +impact on wireless system performance. Therefore, there is +a need for effective modeling (e.g., attenuation of signals, +reflection/refraction in wireless signals, and wireless signal +energy absorption) of effect of humans on wireless signals in +a meta space. For instance, Terahertz (THz) communication +suffers from many effects due to humans (i.e., loss in LOS +communication). Additionally, THz communication is affected +by the concentration of red blood cells (RBCs). The molecular +noise and path-loss decreases with a raise in RBCs, and vice +versa. Therefore, there is a need to effectively model avatars +in meta space. +In a metaverse operation, we have two main phases: +offline training and operation. Offline training is used for the +training of meta space models. Such training can be performed +either using centralized machine learning or distributed ma- +chine learning. Although centralized learning converges fast, +but result in loss of users’ privacy due to the transfer of data +from devices/end-users to a centralized location for training. +To remedy this, distributed learning can be used that will +better preserve users’ privacy. On the other hand, during the +operation phase, the devices will request services from the +meta space. The meta space will in turn perform resource + +9 +optimization to instantly serve the end-users using pre-trained +models, optimization theory, and game theory. For both phases +(i.e., offline training and operation), there is a need for wireless +and computing resource optimization. Later, in a tutorial, +we will discuss how can we perform resource optimization +of computing and wireless resources. Additionally, we will +discuss interfaces for communication, deployment of meta +space, and meta space design later on in detail in this tutorial. +III. META SPACE +A. Metaverse Management +1) Efficient Deployment of Twins and Avatars: There can +be two main deployment trends, such as single device-based +deployment (e.g., edge server) and multiple devices-based +deployment (e.g., multiple edge servers). We first discuss the +deployment using single device. To deploy meta space, one can +use edge servers located in close vicinity to the end-devices +[25]. Such deployment of meta space will result in low latency, +however, with low storage and computing power. Generally, +we have less storage capacity and less computing power at +the network edge compared to the remote cloud [26], [27]. +Therefore, one can deploy the meta space at a remote cloud +when the latency requirements are not strict and high storage +as well computing power is needed. Meta space based on +edge servers has more context-awareness compared to cloud- +based meta space [28]. The reason for this context-awareness +(e.g., devices location and mobility pattern) is due to the fact +that edge servers are located close to the devices, and thus +likely have more information about the network devices. Also, +mobility management of devices for edge-based meta space is +easier compared to cloud-based meta space because of the fact +that edge servers are more close the devices, and thus more +readily available knowledge about the devices’ position and +mobility patterns. +Although the aforementioned discussion for implement- +ing meta space on a single device can offer many benefits, it +has a few limitations. The prominent one is scalability which +is typically low for a single device-based implementation. A +meta space located at a centralized location will suffer from +high control signaling overhead, and thus suffer from high +latency. Such a high latency might not be desirable for many +strict latency applications (e.g., digital healthcare). To address +this limitation, one can use meta space implementation in +a hierarchical fashion using multiple devices [29]–[31]. One +can have a root meta space and secondary meta spaces. The +secondary meta spaces can be deployed to serve a part of +whole devices. The primary meta space will coordinate among +all the secondary meta spaces. The secondary meta space +will handle all the signaling required for serving the devices +within its vicinity. Such an approach of deploying meta spaces +in a hierarchical fashion will enable scalable operation by +serving a large number of users without significantly adding +latency to the system. Other than scalability, robustness is +also important. A meta space implemented on an edge server +might stop working due to a malfunction or a security attack. +Implementing meta space on distributed devices can offer +more robustness compared to a single device implementation. +However, this will be at the cost of management complexity. +Therefore, one must make a tradeoff between robustness, +latency, and management complexity during the deployment +of meta space for a metaverse based wireless system. +2) Twins and Avatars Migration: Twins and avatars are +deployed in the meta space upon the end-user request to +serve them. To deploy, there is a need for using resources +on-demand and then releasing these resources after use. To +implement meta space (i.e., twins and avatars) at edge/cloud, +one can use the concept of virtual machines and containers. +We can use virtual machines/containers to create on-demand +meta space and then release the computing resources after +using it. Virtual machines are implemented using the vir- +tualization of layers including hardware, whereas containers +are implemented using software layers, as shown in Fig. 11. +Implementation of containers makes them easy because they +only involve high software layers. This is the main reason +why containers are lightweight and easy to modify for reuse in +future metaverse applications. On the other hand, containers- +based meta space will have low robustness due to having +less isolation compared to virtual machines-based meta space +that is implemented using virtualization of both hardware and +software. On the other hand, virtual machines-based meta +space will offer more isolation and robustness, but at the +cost of high weight compared to containers-based meta space +implementation. Therefore, one must make a tradeoff between +robustness, re usability, and security. +Although virtual machine and containers based twins +and avatars in a meta space can be used to serve end-users, +mobility of devices will cause many challenges in deployment. +For instance, a meta space serving autonomous cars requires +seamless communication during the serving period. However, +autonomous cars have mobility, and thus they may go out +of the range of their meta space deployed to serve them. To +resolve this issue, there is a need for efficient migration of meta +space to account for mobility. Other than mobility, hardware +failure and imbalance loads can be tackled using meta space +migration. Mainly, we can have two main types of meta space +migration, such as live migration and non-live migration [32]. +In live migration, the meta space will be migrated towards the +other supporting devices (i.e., edge or cloud server) without +shutting down, whereas non-live migration first shut down or +suspend before migrating the meta space to another facility. +In non-real-time applications (e.g., training metaverse-based +smart keyword suggestion in keyboard), one can use a non- +live migration, and the states of meta space (based on vir- +tual machines/containers) are transferred to the new running +facility after suspending. Additionally, there is no need for +transfer of the meta space states in case of shutting down. For +real-time applications (e.g., infotainment and remote patient +monitoring), there is a need for seamless service. Therefore, +for such services, one should preferably use live migration. +Although live migration can offer the benefit of the seamless +running of applications, it has challenges in memory data +migration and network connectivity. To tackle these issues, +there is a need for managing the mobility of the devices. +Based on the predicted mobility of devices, one can proactively +live migrate the meta space to the new facility. Such kind + +10 +Virtual machine-based meta space +Metaverse +application +Bins/Libs +Metaverse +application +Bins/Libs +Host operating system +Infrastructure +Container engine +Container-based meta space +Metaverse +application +Bins/Libs +Guest OS +Metaverse +application +Bins/Libs +Guest OS +Hypervisor +Infrastructure +Virtual machine +Virtual machine +Conatiner +Conatiner +Non-live migration + +How to suspend service and migrate +the system states and other resources? + +Can be used for non real time +metaverse applications +Live migration + +How to efficiently and seamlessly +manage storage resources and network +state migration? + +More suitable for real time metaverse +applications +Pros for +metaverse +Cons for +metaverse +Better isolation +More Robustness +Better security +Non light weight +Latency +Pros for +metaverse +Cons for +metaverse +Light weight and +easy to modify for +ruse in new +metaverse +applications +Low robustness +Low isolation +Less secure + +Difficult to migrate VM due to non +light weight nature + +Difficult to allocate resource due to +real time nature +VM/C +VM/C +Mem +Host +Host +Mem +VM/C +VM/C +Host +Host + +Migration is easier due to the delay +tolerant nature of non real time +applications. + +Easy to allocate resource due to non +real time nature +Fig. 11: Overview of virtual machines, containers, and their migration schemes. +of mobility can be predicted using various techniques. Most +prominent is ML-based mobility prediction [33]. Although the +mobility management scheme of [33] can be used for getting +a mobility prediction model, it has privacy leakage issues. +Users from some of the areas might not want to share their +data with a centralized server. To address this one can use +federated learning that will enable sending of only learning +model updates instead of the whole data. +Both for live and non-live migration schemes, there is a +need for efficient resource allocation to carry out the migration +process [34]–[37]. Migration can be performed either using +a wired or wireless network. A wired network has sufficient +bandwidth, whereas a wireless network requires careful de- +sign due to communication resources (i.e., resource blocks) +constraints. For instance, a meta space running on an edge +server needs to migrate to the other edge server if the end- +user moves to coverage of the new base station running that +edge server. Such a migration can be performed wireless +which will require efficient resource allocation with a variety +of constraints. For instance, reusing wireless resource blocks +of existing devices needs a resource allocation scheme such +that interference caused due to reuse of wireless resources, to +the existing devices should not exceed the maximum allowed +limit. Other factors that should be taken are the efficient +allocation of transmit power. Additionally, the migration delay +should not exceed the maximum allowed latency. Therefore, +one must perform resource allocation in such a way as to +fulfill the latency constraints. Wireless resource allocation +can be performed using various schemes. These schemes +can be heuristic schemes, decomposition-relaxation schemes, +game theory, matching theory, and convex optimization-based +schemes [38], [39]. +3) Low Latency Consensus Algorithms for Blockchain: +In a metaverse-enabled wireless system, there will be a variety +of decentralized and distributed datasets. These datasets will be +used for various purposes, such as training machine learning +models and operations (e.g., cached data and security-related +information). To enable these datasets in a transparent, effi- +cient, and immutable manner, one can use blockchain [40]. +One of the main advantages of blockchain is that no node +can change the data without collusion. Note that blockchain +can update the distributed datasets after running the consensus +algorithm. A consensus algorithm is a fault-tolerant mech- +anism that enables an agreement on a set of rules agreed +by decentralized nodes in contrast to a centralized authority. +Therefore, one can use blockchain for various purposes in a +metaverse-based wireless system. In a metaverse for a wireless +system, a set of wireless devices used to train distributed +learning leveraged blockchain to avoid a single point of failure +issue [41]. Similarly, we can consider wireless miners that +communicate with each other. Every miner can have a block +with two parts: body and header. The block body can carry +information about metaverse applications (e.g., control data +and meta space pre-trained models). If some new update is +to be added to the blockchain network, a miner generates +a hash value after running the consensus algorithm. If the +hash value is less than the target value, then the miner is +allowed to update the distributed ledger by its generated block. +The generated block is transmitted to all the miners. Some +of the receiving nodes might be successful in solving the +problem and broadcast their own block in the network before +receiving the generated block of the other node. This event +is called forking. There must be a measure to avoid this +forking event by controlling the block generation rate. Another +factor is the efficient allocation of wireless resources that can +minimize the transmission latency, and thus forking. On the +other hand, the consensus algorithm must be scalable with low +latency. Additionally, the consensus algorithms must consider +the privacy leakage issue as well because of their distributed +nature. In a metaverse-based wireless system, there will be a + +11 +massive number of nodes. To handle distributed datasets using +blockchain, we must propose consensus algorithms that can +work well without adding a significant delay to the system. +Other than scalability, latency is another issue with running +blockchain consensus algorithms. Therefore, we must propose +novel blockchain consensus algorithms that are scalable and +offer low latency along with better privacy preservation. +B. Reliability and Security +Similar to network slicing, there are two ways to im- +plement meta space for various applications: (a) dedicated +physical space hardware and (b) shared physical space hard- +ware [42]–[44]. In the case of dedicated physical space hard- +ware, one can deploy meta space to serve users. Such an +approach will have to advantage of easier management and +better performance but will cost high and is practically not +feasible. To overcome this high issue, one can use shared +physical space hardware that allows multiple meta spaces to +operate. Although using shared physical space hardware for +multiple meta spaces will be a good and feasible solution, +it has a few implementation challenges, such as resource +allocation, reliability, security, and isolation [45]. For instance, +meta spaces deployed to service intelligent transportation and +healthcare at the same network edge might suffer from security +concerns if one of them is being attacked by a malicious +user due to sharing of the same physical space hardware. +Therefore, there is a need for efficient isolation of meta spaces. +Note that isolation for meta spaces will be at various levels: +access network isolation, computing resource isolation, and +core network isolation [46]. Effective isolation will result in +a secure and reliable operation. For a dedicated model, an +access network slice for meta space has a dedicated user +and control plane traffic, spectrum, and MAC scheduler [42]. +This approach will ensure low latency, isolated, secure, and +reliable operation but will cost high and not allow elastic +operation. Every meta space slice will have access to its own +medium access control, radio link control, and radio resource +control instances along with resource blocks. On the other +hand, using dedicated physical space hardware, there is a need +for sharing of spectrum, MAC scheduler, and control plane. +Specifically, the resource blocks for serving different meta +spaces will be managed by a single scheduler and thus, will +face many management and isolation challenges. To resolve +these challenges, one can modify the medium access control +scheduler that can use resource blocks of different network +operators and allocate them to various meta spaces in an effi- +cient way. To do so, one can have an objective function that is +based on maximizing the utility (i.e., overall throughput) while +fulfilling the meta space user requirements (e.g., reliability and +latency). On the other hand, for such an interaction between +meta space scheduler, network operations, and end-users, there +must be some efficient incentive mechanism. Such an incentive +mechanism will enable buying of resources from network +operators and selling them to meta space users with the aim to +maximize the profit while improve users performance. Similar +to access network, one must propose novel scheduling schemes +for sharing of computing as well as core network resources. +All of the above schemes will use optimization theory, game +theory, deep reinforcement learning, and graph theory, among +others [42], [47], [48]. On the other hand, there are various +ways of implementation of meta space. These possible ways +can be meta space implementation using an edge server, cloud, +edge-cloud, or devices [4]. Implementing meta space using +device can have easier management but at the cost of low +reliability and security. An edge server running meta space +might suffer from malfunction either because of security attack +or physical damage. To resolve this, one can deploy meta +space using multiple edge/cloud servers. This approach will +lead to more computing power and storage capacity as well +as security and reliability but at the cost of management +complexity. Based on the aforementioned facts, one can say +that careful attention must be given to the implementation of +meta spaces. Other than this, there must be secure interfaces +for communication between meta space and physical space. +For such security, one can use encryption/decryption schemes. +C. Summary: Lessons learned and Insights +This sections described how can we design and de- +ploy meta space over the physical infrastructure. Specifically, +efficient deployment of meta space, meta space migration, +reliability, and security are discussed. Several lessons learned +are as follows. +• We learned that is a need for efficient deployment of +meta space using edge and cloud. Deployment of the +meta space requires storage and computing resources. +For edge, there will be computing resources limitations, +whereas, for cloud, the latency is the problem. There- +fore, deployment of meta space should be efficiently +performed. Additionally, running the meta spaces for +different applications on the same edge requires careful +design for optimally allocating computing and storage +resources. +• Deployment of meta space (i.e., avatars and twins) on +edge/cloud server must be performed intelligently and +on-demand either using virtual machines or containers +depending on the specifications. For instance, virtual +machines are implemented using virtualization of layers +including hardware as well as software layers, and thus +gives better isolation and security. However, these features +are at the cost of non-light weight nature compared to +a container. Therefore, we must wisely choose contain- +ers and virtual machines for the implementation of on- +demand meta space at edge/cloud. +• Mostly, the emerging wireless applications are real-time, +therefore, we should use live migration of meta space +from one edge to another depending on the mobility of +end-devices. To do so, there is a need for an effective +ML-based scheme for the prediction of device mobility. +Based on the predicted mobility of devices, one can +proactively start the migration of meta space to avoid +latency in the service. For such kinds of predictions, +one must propose effective algorithms based on emerging +schemes of ML. To do so, one can use distributed learning +with better convergence. Normally, distributed learning + +12 +Fig. 12: Overview of computing resource and communication resource management tasks for physical space. +has a low convergence rate due to devices and statistical +heterogeneity as well as fairness issues. Therefore, there +is a need for efficient novel distributed learning schemes +for predicting the mobility of devices. +• To effectively isolate the meta space of one application +from others, there is a need for novel isolation schemes +that allow the operation of various twin spaces for dif- +ferent applications on the shared physical infrastructure. +For wireless resources, one can use the concept of vir- +tualization which can be achieved using a modification +of the existing resource schedulers at the medium access +control layer. To do so, there should be an efficient novel +algorithm based on either contract theory, Stackelberg +game, or matching theory that will enable buying of +wireless resources from various network operators and +selling them to the different meta spaces to increase an +overall utility (i.e., that accounts for network operators +profits and meta space users performance). +IV. PHYSICAL SPACE +A. Metaverse Management +1) Resource Management: The physical space of the +metaverse has a wide variety of players, such as edge/cloud +servers, base stations, autonomous cars, moving devices, and +unmanned aerial vehicles, among others [4]. For the successful +enabling of metaverse-based wireless systems, there is a need +for seamless interaction among devices. Additionally, the key +element of the metaverse, namely, meta space will be run +by the wireless system hardware. To do so, there is a need +for efficient allocation of wireless and computing resources. +Overview of computing and wireless resource optimization +schemes are given in Fig. 12. Typically, wireless resources +of access networks need careful design of resource alloca- +tion schemes. The design of the wireless resource allocation +scheme depends mainly on the access scheme (e.g., orthogonal +multiple access (OFDMA) and non-orthogonal multiple access +(NOMA) [49]–[51]. Typically, devices in wireless systems +have computing tasks and they do not have sufficient re- +sources. Therefore, they compute a part of their task and +send the remaining to the nearby base stations enabled by +Decomposition- +relaxation-enabled +schemes +Limitation of +approximation error +for combinatorial +problems +Optimization of +computing and +communication +resources for +metaverse +Matching theory- +enabled schemes +Can not be directly +used for continuous +time problems (e.g., +transmit power +optimization) +Heuristic schemes +Suffers from high +computing +complexity +Deep reinforcement +learning schemes +Might have high +computing +complexity in +convergence for +some scenarios +Convex +optimization +Cannot be used for +solving +combinatorial +problems +Fig. 13: Overview of resource optimization schemes. +edge servers. Additionally, edge and cloud servers should +support virtual machines and containers running meta space. +For interaction among devices and edge servers, one must +efficiently allocate computing and wireless servers. Also, +for efficient implementation of meta space, the computing +resources must be efficiently managed. The works in [52]–[56] +considered the association and allocation of wireless resources +in a cellular network. In [52], the authors considered an +association-interference problem. They formulated a problem +to maximize the network utility. Due to the non-convex nature +of the formulated problem, a dual decomposition is used. In +another work, [53], Gao et al. proposed an energy-efficient +scheme for user association and on/off of the base stations. The +problem is formulated as a non-convex nonlinear programming +problem which is decomposed into two sub-problems for +an easier solution. The work in [54] proposed a joint user +association and resource allocation in heterogeneous cellular +networks. Similarly, the works in [55] and [56] discussed +resource allocation and association. +Other works [57]–[60] considered task offloading in edge +computing. In [57], the authors surveyed different techniques + +Computing +Virtual model +Management tasks + Meta space +resource +for running + meta space +. Local computing +resource management + Wireless resource +AR/VR/MR Digial twin +Computing +Wireless +Avatars +XR + Edge computing +management +resource for +resource + Management tasks +Edge computing +running + management for + for offloaded task +blockhain +blockhain +Big +resource management for +Local computing +Edge computing +Aggregation for training +Data +offloaded task +consensus +wireless miners +Network +resource +for offloaded task +meta space models +Computing tasks for +management +offloaded tasks at cloud +Core +Big +. +Wireless resource +Computing task for +Network +Data +management +running meta space +Local +.Edge computing +based on virtual +learning +resource +machines and containers +mode +management for +Computing and wireless +Wireless res ource + computation +offloaded task +resource management for +blocks +. +Computing tasks at +Blockchain mining +Wireless res ource +cloud + Mobility management +blocks +Partial local task + Partial local task +Mobility +Miner +. +Partial local task +Partial local task +for meta space migration + computation +computation +management + computation +computation +(a) Overview of traditional wireless networks +(b) Overview of metaverse-based wireless networks13 +used for offloading in edge and cloud computing. specifically, +they studied various applications based on edge computing +and then challenges related to edge offloading. Another work +[58] proposed a decentralized scheme for task offloading in +an edge computing system. The authors in [59] proposed a +game theoretic scheme for enabling efficient task offloading +between multiple users and multiple base stations. The works +in [57]–[59] mainly performed association of devices to base +stations; however, they did not consider the edge comput- +ing and offloading of resource-constrained end-devices. In a +metaverse-based wireless system, there will be a need for local +computational task offloading as well optimization of edge +servers computing resources for performing various tasks, such +as running of meta space, offloaded computing task, aggrega- +tion of meta space models for distributed learning, computing +partial local learning models for split distributed learning, and +running blockchain miners, among others. On the other hand, +the work in [60] considered both tasks offloading and resource +allocation for edge computing. Similarly, other works [61], +[62] considered joint computing and wireless resources (i.e., +transmit power allocation and resource block allocation). In +[61], the authors proposed a game theoretic scheme for joint +computational offloading and resource allocation in mobile +edge computing. They formulated an objective function that +accounts for energy consumption and monetary cost. Due to +the NP-hard nature of the formulated problem, a joint offload- +ing and resource allocation optimization game was proposed to +solve the formulated problem. Another work [62] considered +a system of multiple edge servers and users. They formulated +an objective for minimizing the cost that considers the time +and energy consumption of devices. To solve the formulated +problem, the authors proposed a two-stage algorithm using +alternating optimization and one-dimensional search. One- +dimensional search performs offloading decisions, whereas +the second stage, alternating optimization, performs resource +optimization. Although the works in [60]–[62] can be used +for allocation for computing, offloading of tasks, and wireless +resources in a typical wireless system, they will not perform +well for a metaverse-based wireless system. In a metaverse- +based wireless system (shown in Fig. 13), the scenario is +different and there are a wide variety of players involved +in resource management, such as end-devices computing re- +source, wireless resource blocks, offloaded task computation +at the edge servers, edge computing resource for running meta +space, and storage resource for blockchain miners. Note that +for different services, one can deploy different meta spaces. +To meet the aforementioned challenges of a metaverse-based +wireless system, there is a need for novel frameworks that +will joint perform computing resources (i.e., local devices +computing resources for local model computing and local task +computing, whereas edge computing resources for meta space +running, computing offloaded tasks, computing partial local +meta models for split learning case) and communication re- +source (i.e., for transmission of learning updates, user requests, +offloaded task, and mining information). +2) Devices Mobility Management: The mobility of de- +vices poses different challenges in a metaverse compared to +traditional wireless networks. For instance, a mobile device +connected to one base station can easily be handed over to +the new base station if it enters its coverage area. However, +the case is different in the metaverse where simply handover +will not work. There must be different and novel schemes +to address the mobility of users. There are two main phases +in the metaverse: (a) offline training of meta space models +and (b) online operation [4], [5]. For training, one can use +various schemes, such as centralized ML or distributed learn- +ing [63]. Distributed learning can offer many benefits over +centralized ML. For distributed learning, frequent communi- +cation takes place between the meta space deployed at the +network edge/cloud and end-devices. For such interaction, +there must be seamless communication between devices and +the edge/cloud server. It is desirable that the devices should +remain in the coverage area of the edge-based base station. +such a fashion will generally result in a faster convergence +[17], [28]. Frequent changes in the devices for a typical edge +server in case of multiple edge servers will result in changes in +local datasets (i.e., of devices), and thus will suffer from a slow +convergence rate. To resolve this issue, one can use a clustering +approach that should be based on the clustering of devices +that have more probability to remain within the coverage +area of each other with one of the nodes as a central node +acting for aggregation [17]. On the other hand, during serving +the end-devices by a meta space deployed at edge/cloud, we +should also tackle mobility. For serving the devices, the meta +space will enable them with efficient resource management +that will require seamless communication among devices and +meta space. Therefore, during the training phase and operation +phase, there is a need for efficient management of device +mobility. +Mobility management of devices in wireless systems +is considered by various works [64]–[68]. Broadly, one can +divide the management schemes into categories: (a) within a +network of one network operator and (b) between different +network operators. For instance, a device connected to meta +space deployed on edge supported by one network opera- +tor can go under the coverage area. In this case, mobility +management will be easy and will generally require less +signaling information compared to the case when the device +moves to the coverage of new network operators. Therefore, +there is a need for novel schemes that can efficiently handle +the mobility of devices served by meta space. To continue +seamless operation, the meta space should also be migrated +based on the mobility of devices. In [64], the authors proposed +a spectrum-aware mobility management scheme. Specifically, +they presented an architecture for mitigating heterogeneous +spectrum availability. Using this architecture, a unified mo- +bility management framework is presented to cope with the +issue of mobility events. Moreover, the authors proposed +inter-cell resource allocation. Other works [65]–[68] surveyed +and presented schemes for mobility management of devices +in wireless networks. However, note that the nature of a +metaverse-based wireless system is different. Along with the +mobility of devices, there is a need to migrate corresponding +meta space (i.e., those serving the mobile devices) as well. +Therefore, traditional mobility management schemes will not +work well for a metaverse-based wireless system, and we + +14 +Big +Data +Core Network +(a) Hybrid design of edge and cloud for meta space +Virtual model +Avatars +AR/VR/MR/ +XR +Digital twin +Virtual model +Avatars +AR/VR/MR/ +XR +Digital twin +Meta space +deployed at edge +Meta space +deployed at cloud + More available +computing power + Complex +management + High latency for +meta space +component deployed +at cloud + Low latency service +from meta space +component deployed +at edge + Meta space can have +local as well global +context-awareness +Key features +Big +Data +Virtual model +Avatars +AR/VR/MR/ +XR +Digital twin +Virtual model +Avatars +AR/VR/MR/ +XR +Digital twin +Meta space +deployed at edge +Meta space deployed at core +network or macro base station + Sufficient available +computing power + More complex +management + Low latency for +meta space + Meta space can have +local context- +awareness + More scalable +Key features +Virtual model +Avatars +AR/VR/MR/ +XR +Digital twin +Meta space +deployed at edge +Core Network +(b) Hierarchical placement edge servers for meta space +Small cell +base station +Small cell +base station +First tier meta +spaces +Second tier +meta spaces +Fig. 14: Overview of edge and cloud deployment for running meta space. +should propose novel schemes. +3) Edge and Cloud Deployment: To deploy edge and +cloud servers for serving wireless system users, there is a +need for efficient deployment of edge and cloud servers [69]. +Every edge and cloud server requires sufficient backup power +for operation. They will require cooling especially for cloud +servers [70]–[72]. Additionally, edge servers have limited com- +puting power, and thus they should be deployed intelligently. +Therefore, there is a need for efficient deployment of edge and +cloud servers. Such efficient placement of edge/cloud servers +is necessary for efficient running meta space based on virtual +machines and containers. Various works [73]–[75] considered +efficient deployment of edge servers. In [73], Vitello et al. +proposed the efficient placement of edge data centers in urban +environments. They focused on using user mobility along with +spatial deployments to assist in the efficient deployment of +data centers. Another work [74], studied three key issues, +such as capacity at edge locations, user association, and +edge location, related to edge servers deployment. Cong et +al. in [75] proposed a scheme for cost-efficient deployment +for cooperative edge computing. Specifically, the idea of the +authors was to share the edge servers (i.e., overlapped) during +the time of peak load in a cooperative manner. The advantage +of this approach is to avoid using a large number of edge +servers to fulfill the demands during peak hours. +On the other hand, there are computing limitations on +the edge server running meta space/s. To address this, one +can have a hybrid placement of meta space/s. Such a hybrid +placement will allow us to use both edge and cloud for +placing meta space. This approach will enable to use of edge +computing resources by the meta space first and then the +cloud computing resource if needed. Although this approach +of hybrid deployment can offer the benefits of high computing +power, performing a task by meta space deployed in the +cloud will suffer from a high latency that is undesirable. To +resolve this issue, one can use the concept of hierarchical edge +deployment of meta space/s, as shown in Fig. 14. Although +hierarchical fashion Fig. 14b can significantly improve the +performance of a metaverse, it has a few limitations. Its context +awareness (i.e., information about the surrounding network +nodes) is local. The reason for this is the low converge area +associated with the small cell base stations. On the other +hand, context-awareness is global for hybrid due to the fact +that the cloud is associated with a large coverage area, and +thus might have information about the nodes located over a +large geographical area. The hierarchical fashion of deploying +edge servers for enabling a meta space can offer scalability as +well. The reason for this is the number of devices associated +with a meta space deployed at the edge will be less. Only +the top tier will provide service if the computing power +available in the bottom tier is insufficient. Additionally, a +top-tier meta space will control the bottom-tier meta spaces. +This approach will enable more scalable operations. Note +that we can have multiple meta spaces for enabling a single +service/application (e.g., infotainment in autonomous cars). +For enabling infotainment, one should perform caching in +addition to other schemes. Such caching based on meta space +can be performed either at meta spaces deployed at the first +tier and also at the second tier. Such a fashion of hierarchical +caching has been considered by many works [76]. Therefore, +we must efficiently deploy meta space/s for metaverse-based +wireless systems. +B. Reliability and Security +In the physical space of the metaverse-based wireless +system, there are a wide variety of entities, such as blockchain +miners, end-devices, software-defined networking switches, +edge/cloud servers, and unmanned aerial vehicles, among +others [4]. Devices’ physical access from a malicious user +is very difficult because of their distributed nature. Therefore, +effective authentication schemes should be used for avoiding +the attacks due to malicious users. One must propose efficient +and lightweight authentication schemes. Such a scheme can be +based on tokens generated by a server (e.g., Auth2 protocol) +or a non-tokens-based scheme that uses the user name and +password [77]. Authentication can be performed using vari- +ous ways: three-way authentication, two-way authentication, +and one-way authentication. One-way authentication involves +authenticating only one party (e.g., client) without considering +the other (e.g., server). This approach might have low com- +plexity but might suffer from inefficiency in the case of the ma- +licious second user for which authentication is not required. To + +15 +address this limitation, one can have two-way authentication, +both parties agree to authenticate with each other. To make the +system more secure, one can use three-way authentication that +involves a third party authenticating the two parties. Other than +authentication schemes, there must be some mechanism for +secure wireless communication. A malicious user might access +the wireless signal and cause leakage/alteration of sensitive +information. Additionally, during training of the meta space +model using distributed learning model, a malicious user can +access the wireless local learning model and infer the device- +sensitive information [77]. Therefore, there is a need for good +encryption schemes before transmitting wireless signals for a +metaverse. A data encryption scheme transforms plaintext data +into encoded data, namely, ciphertext to avoid the man-in-the- +middle attacks that can result in the leakage of devices’ sensi- +tive data. One can use various encryption/decryption schemes. +One of the popular ones is homomorphic encryption [78]. An +advantage of using homomorphic encryption is that there is +no need of sharing a key between the two parties involved +in communication to avoid privacy concerns. In homomorphic +encryption, the receiving party can operate on the data without +the need for decryption. For instance, training a meta space +learning model using distributed learning, devices send their +locally trained models to the meta space where aggregation +has to take place. However, distributed learning still suffers +from privacy attacks (i.e., inference attacks at aggregation +server). Therefore, we can use homomorphic encryption to +encrypt the local model and at the aggregation server, one can +perform aggregation without the need for decryption [79]– +[81]. Homomorphic encryption can be divided ointo three +types: (a) partial homomorphic encryption, (b) somewhat ho- +momorphic encryption, and (c) fully homomorphic encryption. +Note that homomorphic encryption enables effective security, +there is a need for efficient wireless resource allocation as it +results in a significant overhead, especially fully homomorphic +encryption. Therefore, there must be a tradeoff while selecting +a homomorphic encryption scheme. +Other than security, there must be reliable communication +between the devices of physical space. For reliable commu- +nication, one can use effective channel coding that enables +encoding the input bits into a coded sequence of bits to make +the system robust against channel errors. One can use Turbo +codes, convolutional codes, and linear block codes [82]–[84]. +Although linear block codes have low computing complexity, +they might not perform well in all scenarios. To overcome +this, one can use convolution codes that may perform well +but will be generally at the cost of the increase in computing +and communication costs. Turbo codes can be employed +that are based on either parallel or series concatenation of +linear block codes or convolutional codes. Generally, Turbo +codes can outperform all other schemes, but they have high +computing and communication cost. Therefore, one must make +a tradeoff between performance and cost. For Ultra-Reliable +Low Latency Communications, recent works proposed the +use of Short Block-Length Codes [85], [86]. Turbo codes, +convolutional codes, Low-density parity-check (LDPC) codes, +and Bose, Chaudhuri, and Hocquenghem (BCH), can be used +for URLLC. Similarly, one can use these codes for requesting +devices in a metaverse-based wireless system. BCH codes +have shown good reliability under optimal decoding conditions +among various codes (e.g., polar codes and convolutional +codes) [85]. From the aforementioned discussion, one can say +that we must properly select a code with low overhead for +metaverse-based wireless systems. +C. Summary: Lessons Learned and Insights +In this section, we discussed various management func- +tions (e.g., resource management and deployment of edge and +cloud servers) of the physical space. Moreover, we discussed +reliability and security of physical space. Several lessons +learned from this section are as follows. +• There must be efficient joint computing and wireless +resource allocation schemes for a metaverse-based wire- +less system. Such a resource allocation in a metaverse- +enabled wireless system is different compared to tradi- +tional resource allocation problems due to the presence +of a wide variety of players. Such a problem will be a +kind of mixed integer non-linear programming problem +(MINLP) along with numerous constraints. To solve such +kind of problem, there is a need for novel solutions +based on decomposition-relaxation, game theory, deep +reinforcement learning, and graph theory [28]. +• It is evident that the deployment of edge servers for +running meta spaces must be performed intelligently. One +can deploy meta space at the network edge or cloud or +both edge and cloud. Deployment at the network edge +will result in more context awareness compared to cloud- +based meta space but at the cost of low computing and +storage resources. More context awareness (e.g., device +location) will result in better mobility management and +vice versa. On the other hand, one can use both cloud and +edge for the deployment of meta space to offer benefits of +both edges (i.e., low latency and more context-awareness) +and cloud (i.e., more storage and computing power). +• Novel low overhead channel coding schemes should be +proposed for a metaverse-enabled wireless system. These +low overhead channel coding schemes can be comprised +of the existing schemes (e.g., Turbo codes and linear +block codes) or modified version for further reducing the +overhead while fulfilling the bit error rate requirements +of the applications as well metaverse signalling. +• Mobility of the devices must be given proper attention as +it will significantly affect the performance of a metaverse- +enabled wireless system. Both during training of meta +space models and service request/operation, there is a +need for effective mobility management. For mobility +management, one can use novel schemes based on deep +reinforcement learning or federated learning. +V. STATE-OF-THE-ART AND STANDARDIZATION +A. Advances +In this section, we discuss various recent advances [87]– +[91], [91]–[95] towards enabling wireless system by a meta- +verse. As the metaverse is still in its infancy, only a few works + +16 +Key generation +center +Data owner +Private +server +Public +server +Data user +S +1 +TK +RK +2 +Fig. 15: Encryption of data for metaverse [87]. +Capturing and rendering +Transmission +Display +0.3 in. x +0.3 in. +0.3 in. x +0.3 in. +Server +Vin1 +Vin2 +Vout1 +Vout1 +Vout2 +Vout2 +Driver +Light field camera +(11.82 Gbps) +Switch +Server +SLM +driver +Projector +C +Fig. 16: 3D holographic communication framework [88]. +presented architectures/frameworks for enabling emerging ap- +plications using the metaverse. In [87], the authors proposed +a metaverse-enabled healthcare framework that diagnoses a +patient. Meanwhile, there is healthcare data that is used by a +metaverse and stored on edge servers. To ensure the privacy of +such metaverse data, they propose the use of attribute-based +encryption. The system consists of a data user, private server, +public server, data owner, and key generation center. The data +user submits the attribution set for registration to the key +generation center that issues reclaiming key and transformation +key for the data owner. The intermediate cipher texts are given +to the owner of data, as shown in Fig. 15. Then, the cipher +texts are fed to the private and public servers. Finally, the +transformation keys are shared with servers when the data is +required to be downloaded by a user. The proposal of [87] +can be used for ensuring the privacy and security of data in a +metaverse architecture presented in Section II-C (i.e., Fig. 10). +He et al. in [88] proposed a three-dimensional holographic +communication system for the metaverse. Their system has +four components, such as display, transmission, hologram +generation, and capture, as shown in Fig. 16. To support 3D +communication, one must use 3D display and imaging tech- +nologies, such as light field (LF) display, volume display, and +binocular vision display [96]–[100]. To capture images, light +field and structured light cameras are used for an object for +Extended reality in metaverse +Application +domains +XR can overcome +limitations of space +and time +XR enables people +to experience things +impossible in real +world +XR enables to +experience various +perspectives +Presence +Agency +Embodiment +XR affordances +Threat appraisal +For +myself +For +others +Coping threat +For +myself +For +others +Behavior change +facilitators +Increasing healthy behavior and health communication +challenges +Fig. 17: A theoretical framework for metaverse using +extended reality [89]. +capturing dynamic 3D models and objects that change slow, +respectively. Next to capturing 3D images, computer-generated +holograms (CGHs) are used to denote 3D intensity patterns in +computer holography under coherent illumination. The phase- +only CGHs are computed by the capturing and rendering part +using the layer-based angular-spectrum method (ASM). The +layer-based ASM used shading images and depth images. +Next, the CGHs are transmitted over a wireless channel using +some communication technology (e.g., 5G). At the receiver +side, a 3D video is generated using a holographic optical +display system. Although the proposed 3D communication +system offers many benefits, there are many challenges that +need to be addressed. The first one is communication resource +management. For a massive number of applications based on +3D holographic communication, we must propose efficient +resource management schemes that increase the throughput of +the overall system. Other than this issue, mobility management +is necessary for such systems. For instance, a user might +move outside the coverage area of one capturing device during +the mid of capturing phase, and thus the capturing device +will not get complete information. To resolve this, one can +predict the mobility of the devices and based on the predicted +mobility, one can better associate the user with a better +image-capturing device. On the other hand, there must be +novel encryption schemes for 3D holographic communication +systems. A malicious user might access the wireless signal and +thus, causes privacy leakage or alter important information. +Therefore, there is a need for efficient and effective encryption +schemes for 3D holographic communication systems. +Plechata et al. in [89] highlighted the role of extended +reality in enabling metaverse for healthcare applications. A +metaverse-enabled architecture for disease prevention and +health promotion was considered that is based mainly on two +phases: threat appraisal and coping appraisal. Threat appraisal +refers to vulnerability and threat severity (i.e., level of damage +to health), whereas coping appraisal refers to self-efficacy and +response efficacy. In response to efficacy, an individual belief’s +whether the measures of coping will minimize the health threat + +17 +Fig. 18: MeTAI ecosystem using AI and metaverse [90]. +or not. On the other hand, self-efficacy refers to individual con- +fidence in the ability for performing behavior recommended +by the architecture. Similar to many existing applications, +extended reality can play a crucial role in healthcare communi- +cations. To do so, one can use metaverse using extended reality +to support patient support groups as well as expert-moderated +health communities. Specifically, the metaverse using extended +reality will provide presence, agency, and embodiment. Based +on these extended reality affordances, the architecture can bet- +ter enable the behavior change facilitators, as shown in Fig. 17. +The framework proposed by the authors can help in improv- +ing healthcare services using metaverse, but it needs further +efforts. To implement the theoretical framework, in reality, +there is a need t resolve many challenges. These challenges +are sensing, adding healthcare annotations using extended +reality, and communication of sensory data (e.g., human body +temperature and 3D images of body parts). Therefore, there is +a need for modifications in the framework of [89] to enable +healthcare services. Wang et al. in [90] proposed the use +of metaverse for healthcare. They presented an architecture, +namely, MeTAI ecosystem, for enabling intelligent healthcare +based on the metaverse. The MeTAI ecosystem shown in +Fig. 18, has four applications: (a) virtual cooperative scanning, +(b) raw data sharing, (c) augmented regulatory science, and +(d) metaversed medical intervention. The purpose of virtual +cooperative scanning is to find suitable scanning technology +for healthcare diseases. The digital twin scanners are installed +to take scans of digital avatars. The architecture also provides +ubiquitous and secure medical data access to various patients +for using it by healthcare personnel and experts. Although +the framework presented in [90] can be applied to many +healthcare applications, it needs much effort to apply in reality. +For instance, to immerse interactive experience technologies +with medical imaging, there is a need to design various +schemes depending on the nature of the disease. Additionally, +there is a need for effective three-dimensional (3D) computed +tomography (CT) of human models for use in the analysis. +On the other hand, there are many challenges that needs to +be resolve prior to using MeTAI system. These challenges +are privacy, security, management, and disparity reduction. +As MeTAI system can be deployed commercially on a large +scale, therefore, there must be some set of laws to ensure +the privacy of users (e.g., Health Insurance Portability and +Accountability Act (HIPAA) in the United States). In addition +to laws, one must use modern security-related technologies, +such as blockchain and privacy-aware distributed learning. +Other than security and privacy, there must be an efficient +mechanism for the management of such a complex system. +Lim et al. in [91] proposed an edge intelligence-based +architecture for realizing the metaverse. They focused on +infrastructure, the metaverse engine, the virtual world, and +the physical world. They identified the key requirements +for enabling metaverse. Additionally, they discussed various +interfaces for communication among various players of the +metaverse architecture. They also presented a case study of the +edge-based metaverse and finally, they presented open research +challenges. The authors in [91] considered the aspect wireless +for the metaverse. On the other hand, one can use metaverse +to fulfill the diverse requirements of various applications (e.g., +intelligent transportation systems). For doing so, one can +deploy a metaverse that uses digital twins, avatars, and other +schemes for efficient and effective management of various +resources. In another work [92], the authors presented vision, +applications, and technologies of enabling vehicular networks +by metaverse and they named it vetaverse. Their identified key +technologies are artificial intelligence, speech understanding, +human motion detection, physiological parameters monitor- +ing, and emotion recognition. They also gave an architecture +for vetaverse. Finally, they presented open challenges with +suggestions. In [93], Alpala et al. presented an experimental +framework for enabling collaboration between virtual environ- + +Virtual +comparative +Cloudlet +scanning +Digital twin scans +Fog +Analysis for the +X +Cloud +Node +best reconstruction +of twins +MEC Server +B +X +Patient +Raw data sharing +Real patient +Patient that requires +scan +imaging +for +the +suspected disease +3D phantom +Remote access +printed scan +Doctors, Engineers, +Scan with realistic +and researchers, etc +features for quality +Phantom +control and model +validation +Phy sical phantom +3D printed form of +the digital twin +Avatar +Testing and learning +Human and model federated learning +Digital twin model ++ +observer studies +with + and +without +D +avatars +Metavers ed +Augmented +medical +regulatory +intervention +science18 +Meta space +Physical space +Machine +Learning Models +Optimization +Schemes +Game Theoretic +Schemes +Virtual model +Avatars +AR/VR/MR/ +XR +Digital twin +Services layer +User request is translated +using semantic reasoning +Instructions to manage +end-devices/learning +model updates +Learning model updates/ +feedback of end-devices +Intelligent +transportation +Industry 4.0 +Smart survelliance +Extended reality +Smart farming +Smart buildings + Authentication schemes + Mobility management + Intelligent transceivers + Intelligent devices design (i.e., Intelligent hardware- +software co-design)Traffic prediction (i.e., proactive +analytics) + Intelligent wireless resource management + Estimation of wireless channel + Wireless signal transmit power control + Deep learning-based channel coding + Avatars and twins design + Efficient placement of avatars and twins + Twins and avatars migration + Twin and avatar deployment (i.e., edge, cloud, or both +edge and cloud) + AR/VR/XR integration with avatars and twins + Computing resource management (i.e., for training +meta space models and blockchain mining etc) + Authentication + Translation schemes + Semantic reasoning schemes + Service layer to meta space interface +(a) Layered architecture +(b) Role of machine learning +Fig. 19: Role of ML for metaverse. +Meta space +Physical space +Machine +Learning Models +Optimization +Schemes +Game Theoretic +Schemes +Virtual model +Avatars +AR/VR/MR/ +XR +Digital twin +Services layer +User request is translated +using semantic reasoning +Instructions to manage +end-devices/learning +model updates +Learning model updates/ +feedback of end-devices +Intelligent +transportation +Industry 4.0 +Smart survelliance +Extended reality +(a) Layered architecture +Interface A +Interface B +Interface C +Interface D +Actuator +API +Actuator +data +Actuator +command +IEEE 2888.2 +Sensory +API +Sensor +capacity +Sensory +data +IEEE 2888.1 +ML manager +IEEE 2888.3 +IEEE 802.1AR- +2018 +Blockchain standards +ERC-20 +ERC-721 +Fig. 20: Standardization of ML-enabled metaverse. +ments using a virtual reality-based metaverse. Their system +consists of an online multi-user system, interfaces, object- +oriented configurations, and other functional components. As +a case study, they presented a metaverse-based digital factory +and presented experimental results. Although the framework +of [93] can be used for smart factories, the authors did not +consider a few important aspects, such as security, privacy, and +resource management. Another work [94] proposed the use +of metaverse to enable smart cities. Specifically, the authors +presented a high-level architecture element of a metaverse to +enable smart applications. Additionally, it helps in providing +guidelines for using emerging technologies in the metaverse. +They also presented the key projects for real-time implementa- +tion of metaverse. Although the authors discussed various key +enablers and use cases of metaverse, they did not provide more +concrete implementation of use cases of the metaverse. In +[95], Du et al. proposed an attention-aware network resource +allocation scheme for a metaverse. The key idea is to allocate +more resources to the virtual objects that are more important +to users. Specifically, they discussed the key requirements (i.e., +remote rendering, eye-tracking, and QoE analysis) of enabling +of metaverse. Then, using the existing user-object-attention +level, an attention-aware network resource allocation algorithm +that has two steps (i.e., QoE maximization and attention +prediction is proposed. Their proposal allocates resources (i.e., +edge devices rendering capacity) based on the predicted user +object-attention values and shows promising results. Finally, +the authors provided future directions. +B. Standardization +In this section, we will discuss the standardization of a +metaverse-based wireless system. Prior to discussing the stan- +dardization of the metaverse, there is a need to highlight the +role of many emerging technologies in enabling the metaverse. +First of all, we will highlight the role of machine learning in +enabling metaverse for wireless systems. At the physical layer, +one can use machine learning to enable efficient authentication +schemes to avoid unauthorized users to access distributed +devices. As the devices are distributed in physical space, +therefore, enabling security to avoid unauthorized access can +be performed using effective authentication schemes. Such +authentication schemes can be based on machine learning +[101]–[104]. Other than authentication, one can use machine +learning for the mobility of the management of devices. Such +mobility management schemes can use prediction based on +various machine learning schemes (e.g., convolutional neural +networks). Based on the predicted outcomes, one can better +manage the mobility of devices [105]–[108]. Furthermore, one +can use machine learning for enabling intelligent transceivers. +These transceivers will use machine learning for intelligent + +19 +resource allocation, intelligent channel estimation, and in- +telligent transmit power control, among others [109]–[112]. +Also, one can use machine learning to design efficient devices +using hardware-software co-design [3], [28]. Generally, train- +ing of local models for training a distributed learning meta +space model consumes a significant amount of computation +resources which in turn will consume significant power/energy. +Therefore, one must use neural architecture search (NAS) +that tries various architectures of machine learning models in +order to select optimal architecture for a particular dataset and +task [113]–[115]. NAS is a sub-field of automated machine +learning that enables one to find a suitable design for a given +design. Although NAS enables efficient software design, there +is a need for software-hardware co-design that consider both +hardware and software during the design of end-devices. Such +designs can be based on machine learning [28]. Therefore, +there is a need to propose standardization schemes for machine +learning-enabled metaverse, as shown in Fig. 19. +In 2019, IEEE 2888 project was launched to standardize +interfaces between the cyber and physical worlds, as shown +in Fig. 20. One can use these interfaces along with other +interfaces in metaverse. IEEE 2888.1 and IEEE 2888.2 in- +terfaces can be used for moving sensory information from +physical space to meta space and actuator controls from meta +space to physical space, respectively. On the other hand, +IEEE 2888.3 standard can be used for the definition of +digital things [7]. Additionally, for efficient communication +between meta space entities (e.g., avatars and twins), there +is a need for novel interfaces based on novel standards. +Due to the important role of machine learning in enabling +metaverse systems, there is a need for a standardized ML +manager that can control various interfaces, such as interface +A, interface B, interface C, and interface D. Interface A +will deal with the efficient deployment and resources (i.e., +computing and communication) management of the physical +space using machine learning. Interface B will control commu- +nication between the meta space and physical space by mod- +ifying/assisting the existing IEEE 2888.1, IEEE 2888.2, and +IEEE 2888.3 standardized interfaces using machine learning +schemes. Interface C will use machine learning to deploy meta +space using the physical space infrastructure (e.g., edge/cloud +servers and unmanned aerial vehicles). Such a deployment +will include virtual machines/containers-based design. Addi- +tionally, the deployment of these containers/virtual machines +on single/multiple hardware devices. Such kind of opera- +tions/functions will be performed by interface C. To handle the +meta space data, one can use blockchain. ERC-20 helps in the +implementation of standard APIs tokens. Additionally, ERC- +20 supports basic functionality for transferring tokens [116]. +On the other hand, ERC-721 helps in the implementation of a +standard programming interface for non-fungible tokens within +a smart contract [117]. Other than ERC-20 and ERC-721, +there is a need for other standards using machine learning to +enable consensus among blockchain nodes with less latency +and energy consumption. Finally, interface D will perform +secure authentication using existing/modified schemes. IEEE +802.1AR-2018 (IEEE 802.1AR-2009 suspended) can provide +unique per-device identifiers (DevID) as well as cryptographic +binding of identifiers with devices [118]. +C. Summary: Lessons Learned and Insights +In this section, we discussed recent advances of meta- +verse. Moreover, we identified their design aspect along with +primary focus. Several lessons learned from this sections are +as follows: +• There is a need for wireless channel models for holo- +graphic communications. One can use 3D holographic +communication for the transmission of 3D human images. +To efficiently perform this communication, there is a +need for wireless channel models similar to existing +channel models (e.g., Stanford university interim (SUI) +Channel models, such as SUI-1, SUI-2, SUI-3, SUI- +4, SUI-5, and SUI-6) [119], [120]. Such a specialized +channel model will be used for the analysis of holographic +communication systems [121]. Additionally, to cope with +fading effects of a wireless channel, one should design +an efficient and effective channel estimator. To do so, the +channel model can help to analyze the performance of the +various channel estimators prior to actual implementation +for a metaverse-based wireless system. +• To enable various emerging applications using metaverse, +there is a need to resolve the issue of interoperability due +to the presence of many players (e.g., edge/cloud servers, +devices, blockchain miners, and unmanned aerial vehi- +cles). Therefore, enabling a seamless interaction among +these players is challenging due to their different under- +lying technologies. To do so, one can propose a general +interface that will allow us to efficiently and seamlessly +communicate. +• Most of the existing works presented theoretical frame- +works for metaverse-based wireless system applications. +These theoretical frameworks (e.g., autonomous driving +cars) can be extended to many specific applications (e.g., +lane change assistance) by medications. Although theo- +retical frameworks offer many benefits, there is a need +for mathematical models related to specific applications. +• Due to the important role of machine learning in en- +abling a metaverse-based wireless system, there is a need +for effective standardization of machine learning for a +metaverse-enabled wireless system. To do so, one can +propose a machine learning manager that can control +various, diverse players of the metaverse using differ- +ent interfaces. Meanwhile, the machine learning-based +metaverse system can use existing standards in addition +to novel standards for effectively enabling a metaverse- +enabled wireless system. +VI. OPEN CHALLENGES +In this section, we present open research challenges. +Existing tutorials and surveys on metaverse considered in- +teraction problem, computation issues, ethical issues, privacy +issues, compatibility, endogenous security, empowered meta- +verse, cloud-edge-end orchestrated secure metaverse, cross- +chain interoperable and regulatory metaverse, energy-efficient + +20 +and green metaverse, content-centric and human-centric meta- +verse, resource optimization, blockchain-based data manage- +ment, incentive mechanism, prototyping, training fashion, +standardization of ML-based metaverse, blockchain for se- +cure ML-enabled metaverse-based wireless systems, advanced +multiple access for immersive streaming, multi-sensory multi- +media networks, multimodal semantic/goal-aware communi- +cation, integrated sensing and communication, digital edge +twin networks, edge intelligence and intelligent edge, sus- +tainable resource allocation, avatars (Digital Humans), the +industrial/vehicular metaverse, quality of experience, market +and mechanism design for metaverse services. In contrast, +we consider interoperable meta spaces, non-fungible tokens +for metaverse trading, personalized distributed learning-based +avatars modeling, isolation of meta spaces, machine learning- +enabled semantic communication for metaverse, and zero- +touch networking for metaverse. +A. Interoperable Meta Spaces +How do we enable seamless interaction between the +avatars and twins modeled for different meta spaces? In a +metaverse, the concept of interoperability is different com- +pared to existing wireless systems. In a traditional wireless +system, the goal of interoperability is to enable seamless +interaction between a wide variety of players (e.g., devices +and edge/cloud servers). In contrast, here, the metaverse has +two main aspects: (a) wireless devices and (b) meta spaces. To +enable interoperability between various wireless devices, there +is a need for the design of general interfaces that can enable +seamless communication. However, different devices have +different structures. Therefore, we must define novel interfaces +for a metaverse-based wireless system. On the other hand, +meta space mainly constituted by digital avatars and twins +must be interoperable (i.e., one virtual machine-based meta +space (as explained already in Section III) must work on the +new edge/cloud servers as well due to meta space migration). +For instance, meta space based on virtual machine might not +work on containers deployed at the network edge. Addition- +ally, within a single design (i.e., container-based or virtual +machine-based), the meta space might not be compatible. +Therefore, for efficient deployment of wireless system, one +must propose interoperable meta spaces. For virtual machines- +based meta space, one can have three levels of interoper- +ability [122]. The first one (i.e., level 1) involves running +of virtual machine-based meta space on a virtual hardware +selection/ CPU architecture, and or particular virtualization +product. Level-1 migration is equivalent to suspend at source +and resume at destination. Additionally, one can live migrate +meta space based on level-1, it faces some limitations. The +prominent one is preservation of IP addresses. In other level- +2, virtual machine-based meta space will run on a specific +family of hardware and works by shutting down in the current +edge and rebooting at the destination edge. On the other hand, +level-3 has more freedom of running meta space on multiple +hardware, and thus gives more flexible operation with better +interoperability. +B. Non-Fungible Tokens for Metaverse Trading +How does one use non-fungible tokens for the trading +of metaverse entities (e.g., digital avatars and twins) among +various players? Enabling ownership of digital items (e.g., in- +game items, collectibles, videos, art, and music) in a metaverse +is challenging and needs careful design. In a metaverse, to +uniquely represent the digital assets, non-fungible tokens are +used that are a unit of data stored on a blockchain. Alternately, +non-fungible tokens serve as a certificate of authenticity in +a metaverse-enabled wireless system. One can also say that +non-fungible tokens form a link between physical world items +and metaverse virtual items. A unique value is associated +with a non-fungible token in the metaverse that is used for +permanently storing them in a blockchain network. One of +the recent events of non-fungible tokens was the selling of +digital work created by Beeple [9]. Although non-fungible +tokens can be effectively used for representing digital as- +sets in the metaverse, their many challenges that must be +resolved. The first one is how to use non-fungible tokens +for the representation of digital assets in a wireless system. +For instance, in a metaverse, how do we use a non-fungible +token to define an entity? The entity can be an end-device, +a system made of many devices, or a complete application +(e.g., autonomous cars) made of many systems. There should +be a proper and worldwide acceptable framework for assigning +non-fungible tokens to wireless systems. Also, the unique +numbering of non-fungible tokens must be done in an efficient +way to effectively cover all massive numbers of entities in a +metaverse. +C. Personalized Distributed Learning-based Avatars Model- +ing +How do we enable efficient modeling of avatars using +personalized, privacy-aware distributed learning schemes? To +model avatars, one can use distributed learning. However, get- +ting a generalized global model using distributed for modeling +avatars might not effectively model them. Therefore, there +is a need for modified distributed learning modeling. One +can use personalized distributed learning to model avatars. +For instance, consider a metaverse-based vehicular network, +there is a wide variety of vehicles, such as cars, trucks, +and bikes, among others. If we want to model mobility and +driving assistance using distributed learning in a metaverse- +based intelligent transportation system, there is a need for +more personalized models. Such modeling will be based +on the training of a general global model and then further +training of local data to make it more personalized. Although +such an approach will enable efficient modeling, it may face +challenges. The local data might not be sufficient to well train +the personalized model. Additionally, the local data may have +noise. Therefore, we must effectively take into account all the +factors while using personalized distributed learning for the +modeling of avatars in a metaverse. On the other hand, there +might be very less local data associated with some of the +devices. To address this challenge, one can use a clustering +approach that will be based on the clustering of devices with +similar data distribution. In each cluster, after getting a global + +21 +model, a local model will be trained that will be used by the +associated devices. +D. Isolation of Meta Spaces +How does one enable isolated operation of meta spaces +using shared hardware without affecting the performance of +each other? To deploy meta spaces (i.e., twins and avatars +along with computing storage), there are two main ways: ded- +icated hardware and shared hardware. Dedicated hardware will +result in a good performance, but it comes with a high cost that +is not practically feasible. There is a need for shared hardware +usage for various meta spaces associated with various applica- +tions/functions. To do so, there is a need for isolation at various +levels (e.g., access network and core network). For an access +network, one can use the concept of virtualization which will +consist of buying network resources from the operators and +selling them to metaverse users. For such a design, one can +define a utility that will jointly maximize the profit of network +operators and metaverse users. For such maximization, one +can use mathematical optimization, matching theory, and game +theory, among others. On the other hand, computing resources +must be efficiently managed in such a way as to run multiple +meta spaces on computing hardware (i.e., edge/cloud server) +without affecting the performance of other metaverse users. +E. Zero-Touch Networking for Metaverse +How does one use zero-touch networking to enable effec- +tive self-sustaining metaverse-enabled wireless applications? +Deploying metaverse for emerging applications to service a +massive number of users requires seamless metaverse sig- +naling. Such signaling must be done in a way that requires +less intervention from end users and operators. To do so, one +can use zero-touch networking (i.e., autonomous networking) +for metaverse signaling. For the efficient realization of zero- +touch networking for the metaverse, one can use various +schemes/technologies, such as network slicing, machine learn- +ing, and optimization theory. Note that there are two aspects: +zero-touch networking for metaverse and metaverse for zero- +touch networking. Metaverse for zero-touch networking re- +quires training of meta space models using emerging machine +learning schemes and mathematical tools that can assist the +network operation with the lowest possible intervention of net- +work operators. On the other hand, zero-touch networking for +the metaverse deals with the efficient signaling of a metaverse +using emerging schemes to enable various metaverse-based ap- +plications. One can use machine learning schemes (specifically +distributed learning) to train various models for performing +metaverse signaling. Such models will perform optimization +of computing and communication resources for performing +signaling. Additionally, the interruption in metaverse services +due to faults or security attacks must be addressed using zero- +touch networking models. +F. Machine Learning-enabled Semantic Communication for +Metaverse +How do we enable applications using metaverse and ma- +chine learning while performing service-level optimization and +service diversity? Enabling the metaverse for a massive num- +ber of devices requires service-level optimization and service +diversity for cost-efficient operation. In contrast to traditional +data-oriented wireless systems that require a channel with an +infinite capacity for real-time applications, there is a need +to combine reasoning tools and knowledge representation in +training machine learning tools for the metaverse. Traditional +data-oriented wireless systems represent information simply as +bits that are not sufficient. Therefore, semantic communication +combines reasoning tools and knowledge representation along +with machine learning tools for communication in a metaverse. +Semantic communication only sends important information in +contrast to traditional data-oriented communication systems, +and thus improves the system’s efficiency. The key com- +ponents of semantic communication in a metaverse will be +a semantic encoder, semantic decoder, and semantic noise +interferes. The purpose of a semantic encoder is to detect +semantic information out of all available information. On +the receiving end, the semantic decoder decodes the rele- +vant information from the received information. In [123], a +semantic communication scheme based on auto-encoder over +a Rayleigh channel was proposed. The purpose of the auto- +encoder is to encode and decode the information in semantic +communication. Another work [124] proposed a deep learning- +enabled semantic communication. Specifically, a DeepSC, +using a transformer encoder and decoder for text transmission +was proposed. Based on the aforementioned facts, there is a +need to propose a novel distributed learning schemes for a +privacy-aware semantic communication system. +G. Hybrid Modeling for Meta Space +How do we effectively model meta space that truly reflects +the actual entities in the physical space? Modeling of twins +and avatars can be performed using various techniques, such +as machine learning, experimental, and mathematical. Every +technique has pros and cons, therefore, it might not be more +suitable to model twins and avatars using a single technique. +For instance, machine learning-enabled might not converge +well, and thus fail to effectively model twins and avatars. +Similarly, mathematical modeling also has limitations due to +various assumptions required for modeling. Moreover, experi- +mental modeling also has experimental errors. Keeping in view +the aforementioned facts, one can conclude that there is a +need for hybrid modeling based simultaneously on different +techniques. For instance, consider a wireless system that has +a variety of players. For mobility modeling, one can use deep +learning (i.e., machine learning-enabled modeling). For some +entities (e.g., resource block allocation), one can use mathe- +matical modeling (e.g., optimization theory, game theory, and +graph theory). For 3D modeling of mobile devices/humans, +one can use experimental modeling to effectively model the +effect on wireless communication (e.g., the effect on THz +communication). Therefore, there is a need to propose hybrid +models for the effective modeling of meta space. +VII. CONCLUSION +In this tutorial, we have presented a detailed overview +of the fundamentals of the metaverse for wireless systems. + +22 +Specifically, we presented design aspects, key enablers, and +general architecture. As a part of the general architecture, +we studied the network management, reliability, and security +of both meta space and physical space. We also outlined +the recent advances and evaluated them. Furthermore, we +presented the standardization of the machine learning-enabled +metaverse. Finally, open challenges are presented with possible +guidelines. +REFERENCES +[1] L. U. Khan et al, “6G wireless systems: A vision, architectural +elements, and future directions,” IEEE Access, vol. 8, pp. 147 029– +147 044, August 2020. +[2] L. U. Khan, Z. Han, D. Niyato, and C. S. Hong, “Socially-aware- +clustering-enabled federated learning for edge networks,” IEEE Trans- +actions on Network and Service Management, vol. 18, no. 3, pp. 2641– +2658, 2021. +[3] L. U. Khan, Z. Han, W. Saad, E. Hossain, M. Guizani, and C. S. Hong, +“Digital twin of wireless systems: Overview, taxonomy, challenges, and +opportunities,” arXiv preprint arXiv:2202.02559, 2022. +[4] L. U. Khan, Z. Han, D. Niyato, E. Hossain, and C. S. Hong, “Meta- +verse for wireless systems: Vision, enablers, architecture, and future +directions,” arXiv preprint arXiv:2207.00413, 2022. +[5] L. U. Khan, I. Yaqoob, K. Salah, C. S. Hong, D. Niyato, Z. Han, +and M. Guizani, “Machine learning for metaverse-enabled wire- +less systems: Vision, requirements, and challenges,” arXiv preprint +arXiv:2211.03703, 2022. +[6] H. Ning, H. Wang, Y. Lin, W. Wang, S. Dhelim, F. Farha, J. Ding, and +M. Daneshmand, “A survey on metaverse: the state-of-the-art, technolo- +gies, applications, and challenges,” arXiv preprint arXiv:2111.09673, +2021. +[7] Y. Wang, Z. Su, N. Zhang, R. Xing, D. Liu, T. H. Luan, and X. Shen, +“A survey on metaverse: Fundamentals, security, and privacy,” IEEE +Communications Surveys & Tutorials, 2022. +[8] T. R. Gadekallu, T. Huynh-The, W. Wang, G. Yenduri, P. Ranaweera, +Q.-V. Pham, D. B. da Costa, and M. Liyanage, “Blockchain for the +metaverse: A review,” arXiv preprint arXiv:2203.09738, 2022. +[9] M. Xu, W. C. Ng, W. Y. B. Lim, J. Kang, Z. Xiong, D. Niyato, +Q. Yang, X. S. Shen, and C. Miao, “A full dive into realizing the edge- +enabled metaverse: Visions, enabling technologies, and challenges,” +IEEE Communications Surveys & Tutorials, 2022. +[10] “Precedence research,” https://www.precedenceresearch.com/metaverse- +market, [Online; accessed Oct. 16, 2022]. +[11] “Internet +of +everything +(ioe) +market,” +https://www.marketresearchfuture.com/reports/internet-of-everything- +market-6841, [Online; accessed Oct. 16, 2022]. +[12] C. Thapa, P. C. M. Arachchige, S. Camtepe, and L. Sun, “Splitfed: +When federated learning meets split learning,” in Proceedings of the +AAAI Conference on Artificial Intelligence, vol. 36, no. 8, 2022, pp. +8485–8493. +[13] A. Singh, P. Vepakomma, O. Gupta, and R. Raskar, “Detailed com- +parison of communication efficiency of split learning and federated +learning,” arXiv preprint arXiv:1909.09145, 2019. +[14] Y. Gao, M. Kim, S. Abuadbba, Y. Kim, C. Thapa, K. Kim, S. A. +Camtepe, H. Kim, and S. Nepal, “End-to-end evaluation of federated +learning and split learning for internet of things,” arXiv preprint +arXiv:2003.13376, 2020. +[15] “Metaverse +avatar +guide; +embody +yourself +in +the +metaverse,” +https://metamandrill.com/metaverse-avatar/, [Online; accessed June. +16, 2022]. +[16] “Chara design in 2022: Top software to design your own 3d +character,” +https://www.sculpteo.com/en/3d-learning-hub/3d-printing- +software/chara-design-2/, [Online; accessed June. 16, 2022]. +[17] L. U. Khan et al, “Dispersed federated learning: Vision, taxonomy, and +future directions,” arXiv preprint arXiv:2008.05189, 2020. +[18] “The +expanding +frontiers +of +computer +vision +in +the +meta- +verse,” https://www.telusinternational.com/articles/computer-vision-in- +the-metaverse, [Online; accessed Oct. 16, 2022]. +[19] Q. Dang, J. Yin, B. Wang, and W. Zheng, “Deep learning based 2d +human pose estimation: A survey,” Tsinghua Science and Technology, +vol. 24, no. 6, pp. 663–676, 2019. +[20] A. Newell, K. Yang, and J. Deng, “Stacked hourglass networks for +human pose estimation,” in European conference on computer vision. +Springer, 2016, pp. 483–499. +[21] B. Xiao, H. Wu, and Y. Wei, “Simple baselines for human pose +estimation and tracking,” in Proceedings of the European conference +on computer vision (ECCV), 2018, pp. 466–481. +[22] K. Sun, B. Xiao, D. Liu, and J. Wang, “Deep high-resolution rep- +resentation learning for human pose estimation,” in Proceedings of +the IEEE/CVF conference on computer vision and pattern recognition, +2019, pp. 5693–5703. +[23] J. Yang, Y. Zhou, H. Huang, H. Zou, and L. Xie, “Metafi: Device-free +pose estimation via commodity wifi for metaverse avatar simulation,” +arXiv preprint arXiv:2208.10414, 2022. +[24] “Why does ar/vr need 5g?” https:https://stlpartners.com/articles/edge- +computing/how-5g-and-edge-computing-will-transform-ar-vr-use- +cases/, [Online; accessed Oct. 18, 2022]. +[25] T. Taleb, S. Dutta, A. Ksentini, M. Iqbal, and H. Flinck, “Mobile edge +computing potential in making cities smarter,” IEEE Communications +Magazine, vol. 55, no. 3, pp. 38–43, 2017. +[26] H. Zahmatkesh and F. Al-Turjman, “Fog computing for sustain- +able smart cities in the iot era: Caching techniques and enabling +technologies-an overview,” Sustainable Cities and Society, vol. 59, p. +102139, 2020. +[27] Z. Ning, J. Huang, and X. Wang, “Vehicular fog computing: Enabling +real-time traffic management for smart cities,” IEEE Wireless Commu- +nications, vol. 26, no. 1, pp. 87–93, 2019. +[28] L. U. Khan, S. R. Pandey, N. H. Tran, W. Saad, Z. Han, M. N. Nguyen, +and C. S. Hong, “Federated learning for edge networks: Resource opti- +mization and incentive mechanism,” IEEE Communications Magazine, +vol. 58, no. 10, pp. 88–93, 2020. +[29] Y. Peng, A. Tan, J. Wu, and Y. Bi, “Hierarchical edge computing: A +novel multi-source multi-dimensional data anomaly detection scheme +for industrial internet of things,” IEEE Access, vol. 7, pp. 111 257– +111 270, 2019. +[30] R. Smeliansky, “Hierarchical edge computing,” in 2018 International +Scientific and Technical Conference Modern Computer Network Tech- +nologies (MoNeTeC). +IEEE, 2018, pp. 1–11. +[31] L. Tong, Y. Li, and W. Gao, “A hierarchical edge cloud architecture for +mobile computing,” in IEEE INFOCOM 2016-The 35th Annual IEEE +International Conference on Computer Communications. IEEE, 2016, +pp. 1–9. +[32] F. Zhang, G. Liu, X. Fu, and R. Yahyapour, “A survey on virtual +machine migration: Challenges, techniques, and open issues,” IEEE +Communications Surveys & Tutorials, vol. 20, no. 2, pp. 1206–1243, +2018. +[33] Y. Tang, N. Cheng, W. Wu, M. Wang, Y. Dai, and X. Shen, “Delay- +minimization routing for heterogeneous vanets with machine learning +based mobility prediction,” IEEE Transactions on Vehicular Technol- +ogy, vol. 68, no. 4, pp. 3967–3979, 2019. +[34] T. Sharma, M. Singh, S. Selvan, D. Krah et al., “Energy-efficient +resource allocation and migration in private cloud data centre,” Wireless +Communications and Mobile Computing, vol. 2022, 2022. +[35] Z. Xiao, W. Song, and Q. Chen, “Dynamic resource allocation using +virtual machines for cloud computing environment,” IEEE transactions +on parallel and distributed systems, vol. 24, no. 6, pp. 1107–1117, +2012. +[36] L. Yang, D. Yang, J. Cao, Y. Sahni, and X. Xu, “Qos guaranteed +resource allocation for live virtual machine migration in edge clouds,” +IEEE Access, vol. 8, pp. 78 441–78 451, 2020. +[37] X. Ruan, H. Chen, Y. Tian, and S. Yin, “Virtual machine allocation +and migration based on performance-to-power ratio in energy-efficient +clouds,” Future Generation Computer Systems, vol. 100, pp. 380–394, +2019. +[38] L. U. Khan, Y. K. Tun, M. Alsenwi, M. Imran, Z. Han, and C. S. Hong, +“A dispersed federated learning framework for 6g-enabled autonomous +driving cars,” IEEE Transactions on Network Science and Engineering, +2022. +[39] L. U. Khan, U. Majeed, and C. S. Hong, “Federated learning for cellular +networks: Joint user association and resource allocation.” in APNOMS, +2020, pp. 405–408. +[40] J. Wang, X. Ling, Y. Le, Y. Huang, and X. You, “Blockchain-enabled +wireless communications: A new paradigm towards 6g,” National +Science Review, vol. 8, no. 9, p. nwab069, 2021. +[41] H. Kim, J. Park, M. Bennis, and S.-L. Kim, “Blockchained on-device +federated learning,” IEEE Communications Letters, vol. 24, no. 6, pp. +1279–1283, 2019. + +23 +[42] S. A. Kazmi, L. U. Khan, N. H. Tran, and C. S. Hong, Network slicing +for 5G and beyond networks. +Springer, 2019, vol. 1. +[43] S. Zhang, “An overview of network slicing for 5g,” IEEE Wireless +Communications, vol. 26, no. 3, pp. 111–117, 2019. +[44] X. Foukas, G. Patounas, A. Elmokashfi, and M. K. Marina, “Network +slicing in 5g: Survey and challenges,” IEEE communications magazine, +vol. 55, no. 5, pp. 94–100, 2017. +[45] H. Zhang, N. Liu, X. Chu, K. Long, A.-H. Aghvami, and V. C. Leung, +“Network slicing based 5g and future mobile networks: mobility, re- +source management, and challenges,” IEEE communications magazine, +vol. 55, no. 8, pp. 138–145, 2017. +[46] J. Ordonez-Lucena, P. Ameigeiras, D. Lopez, J. J. Ramos-Munoz, +J. Lorca, and J. Folgueira, “Network slicing for 5g with sdn/nfv: Con- +cepts, architectures, and challenges,” IEEE Communications Magazine, +vol. 55, no. 5, pp. 80–87, 2017. +[47] M. Leconte, G. S. Paschos, P. Mertikopoulos, and U. C. Kozat, “A re- +source allocation framework for network slicing,” in IEEE INFOCOM +2018-IEEE Conference on Computer Communications. +IEEE, 2018, +pp. 2177–2185. +[48] L. U. Khan, I. Yaqoob, N. H. Tran, Z. Han, and C. S. Hong, “Network +slicing: Recent advances, taxonomy, requirements, and open research +challenges,” IEEE Access, vol. 8, pp. 36 009–36 028, 2020. +[49] L. Dai, B. Wang, Y. Yuan, S. Han, I. Chih-Lin, and Z. Wang, “Non- +orthogonal multiple access for 5g: solutions, challenges, opportunities, +and future research trends,” IEEE Communications Magazine, vol. 53, +no. 9, pp. 74–81, 2015. +[50] Z. Ding, X. Lei, G. K. Karagiannidis, R. Schober, J. Yuan, and +V. K. Bhargava, “A survey on non-orthogonal multiple access for 5g +networks: Research challenges and future trends,” IEEE Journal on +Selected Areas in Communications, vol. 35, no. 10, pp. 2181–2195, +2017. +[51] Z. Ding, F. Adachi, and H. V. Poor, “The application of mimo +to non-orthogonal multiple access,” IEEE Transactions on Wireless +Communications, vol. 15, no. 1, pp. 537–552, 2015. +[52] C. Kim, R. Ford, and S. Rangan, “Joint interference and user associa- +tion optimization in cellular wireless networks,” in 2014 48th Asilomar +Conference on Signals, Systems and Computers. IEEE, 2014, pp. 511– +515. +[53] J. Gao, Q. Ren, P. S. Gu, and X. Song, “User association and small- +cell base station on/off strategies for energy efficiency of ultradense +networks,” Mobile Information Systems, vol. 2019, 2019. +[54] D. Fooladivanda and C. Rosenberg, “Joint user association and resource +allocation in heterogeneous cellular networks: Comparison of two +modeling approaches,” in 2019 31st International Teletraffic Congress +(ITC 31). +IEEE, 2019, pp. 66–74. +[55] N. Trabelsi, C. S. Chen, R. El Azouzi, L. Roullet, and E. Altman, +“User association and resource allocation optimization in lte cellular +networks,” IEEE Transactions on Network and Service Management, +vol. 14, no. 2, pp. 429–440, 2017. +[56] M. Yemini and A. J. Goldsmith, “Optimal resource allocation for +cellular networks with virtual cell joint decoding,” in 2019 IEEE +International Symposium on Information Theory (ISIT). +IEEE, 2019, +pp. 2519–2523. +[57] F. Saeik, M. Avgeris, D. Spatharakis, N. Santi, D. Dechouniotis, +J. Violos, A. Leivadeas, N. Athanasopoulos, N. Mitton, and S. Pa- +pavassiliou, “Task offloading in edge and cloud computing: A survey +on mathematical, artificial intelligence and control theory solutions,” +Computer Networks, vol. 195, p. 108177, 2021. +[58] X. Wang, J. Ye, and J. C. Lui, “Decentralized task offloading in +edge computing: a multi-user multi-armed bandit approach,” in IEEE +INFOCOM 2022-IEEE Conference on Computer Communications. +IEEE, 2022, pp. 1199–1208. +[59] J. Luo, Q. Qian, L. Yin, and Y. Qiao, “A game-theoretical approach +for task offloading in edge computing,” in 2020 16th International +Conference on Mobility, Sensing and Networking (MSN). IEEE, 2020, +pp. 756–761. +[60] T.-Y. Kan, Y. Chiang, and H.-Y. Wei, “Task offloading and resource +allocation in mobile-edge computing system,” in 2018 27th wireless +and optical communication conference (WOCC). +IEEE, 2018, pp. +1–4. +[61] J. Zhang, W. Xia, Y. Zhang, Q. Zou, B. Huang, F. Yan, and L. Shen, +“Joint offloading and resource allocation optimization for mobile edge +computing,” in GLOBECOM 2017-2017 IEEE Global Communications +Conference. +IEEE, 2017, pp. 1–6. +[62] K. Zhang, X. Gui, and D. Ren, “Joint optimization on computation +offloading and resource allocation in mobile edge computing,” in 2019 +IEEE Wireless Communications and Networking Conference (WCNC). +IEEE, 2019, pp. 1–6. +[63] L. U. Khan, E. Mustafa, J. Shuja, F. Rehman, K. Bilal, Z. Han, and C. S. +Hong, “Federated learning for digital twin-based vehicular networks: +Architecture and challenges,” arXiv preprint arXiv:2208.05558, 2022. +[64] W.-Y. Lee and I. F. Akyildiz, “Spectrum-aware mobility management +in cognitive radio cellular networks,” IEEE Transactions on Mobile +Computing, vol. 11, no. 4, pp. 529–542, 2011. +[65] S. Smys and J. S. Raj, “A self-organized structure for mobility man- +agement in wireless networks,” Computers & Electrical Engineering, +vol. 48, pp. 153–163, 2015. +[66] S. Fernandes and A. Karmouch, “Vertical mobility management ar- +chitectures in wireless networks: A comprehensive survey and future +directions,” IEEE Communications Surveys & Tutorials, vol. 14, no. 1, +pp. 45–63, 2010. +[67] F. Siddiqui and S. Zeadally, “Mobility management across hybrid +wireless networks: Trends and challenges,” Computer Communications, +vol. 29, no. 9, pp. 1363–1385, 2006. +[68] A. Achour, L. Deru, and J. C. Deprez, “Mobility management for wire- +less sensor networks a state-of-the-art,” Procedia Computer Science, +vol. 52, pp. 1101–1107, 2015. +[69] L. U. Khan et al, “Edge-computing-enabled smart cities: A compre- +hensive survey,” IEEE Internet of Things Journal, vol. 7, no. 10, pp. +10 200–10 232, October 2020. +[70] C. D. Patel, C. E. Bash, R. Sharma, M. Beitelmal, and R. Friedrich, +“Smart cooling of data centers,” in International Electronic Packaging +Technical Conference and Exhibition, vol. 36908, 2003, pp. 129–137. +[71] Z. Liu, Y. Chen, C. Bash, A. Wierman, D. Gmach, Z. Wang, +M. Marwah, and C. Hyser, “Renewable and cooling aware workload +management for sustainable data centers,” in Proceedings of the 12th +ACM SIGMETRICS/PERFORMANCE joint international conference +on Measurement and Modeling of Computer Systems, 2012, pp. 175– +186. +[72] T. Evans, “The different technologies for cooling data centers,” APC +white paper, vol. 59, 2012. +[73] P. Vitello, A. Capponi, C. Fiandrino, G. Cantelmo, and D. Kliazovich, +“Mobility-driven and energy-efficient deployment of edge data centers +in urban environments,” IEEE Transactions on Sustainable Computing, +2021. +[74] Y. Shao, Z. Shen, S. Gong, and H. Huang, “Cost-aware placement +optimization of edge servers for iot services in wireless metropolitan +area networks,” Wireless Communications and Mobile Computing, vol. +2022, 2022. +[75] R. Cong, Z. Zhao, L. Zhang, and G. Min, “Coopedge: Cost-effective +server deployment for cooperative multi-access edge computing,” in +2022 19th Annual IEEE International Conference on Sensing, Com- +munication, and Networking (SECON). +IEEE, 2022, pp. 208–216. +[76] S. Ghosh and D. P. Agrawal, “A high performance hierarchical caching +framework for mobile edge computing environments,” in 2021 IEEE +Wireless Communications and Networking Conference (WCNC), 2021, +pp. 1–6. +[77] M. El-Hajj, A. Fadlallah, M. Chamoun, and A. Serhrouchni, “A survey +of internet of things (iot) authentication schemes,” Sensors, vol. 19, +no. 5, p. 1141, 2019. +[78] A. Acar, H. Aksu, A. S. Uluagac, and M. Conti, “A survey on +homomorphic encryption schemes: Theory and implementation,” ACM +Computing Surveys (Csur), vol. 51, no. 4, pp. 1–35, 2018. +[79] L. U. Khan, W. Saad, Z. Han, E. Hossain, and C. S. Hong, “Federated +learning for internet of things: Recent advances, taxonomy, and open +challenges,” IEEE Communications Surveys & Tutorials, 2021. +[80] H. Fang and Q. Qian, “Privacy preserving machine learning with +homomorphic encryption and federated learning,” Future Internet, +vol. 13, no. 4, p. 94, 2021. +[81] J. Ma, S.-A. Naas, S. Sigg, and X. Lyu, “Privacy-preserving federated +learning based on multi-key homomorphic encryption,” International +Journal of Intelligent Systems, 2022. +[82] M. Zhan, Z. Pang, D. Dzung, and M. Xiao, “Channel coding for +high performance wireless control in critical applications: Survey and +analysis,” IEEE Access, vol. 6, pp. 29 648–29 664, 2018. +[83] K. Arora, J. Singh, and Y. S. Randhawa, “A survey on channel coding +techniques for 5g wireless networks,” Telecommunication Systems, +vol. 73, no. 4, pp. 637–663, 2020. +[84] E. Biglieri, G. Caire, and G. Taricco, “Coding for the fading channel: +a survey,” Signal processing, vol. 80, no. 7, pp. 1135–1148, 2000. +[85] M. Shirvanimoghaddam, M. S. Mohammadi, R. Abbas, A. Minja, +C. Yue, B. Matuz, G. Han, Z. Lin, W. Liu, Y. Li et al., “Short block- + +24 +length codes for ultra-reliable low latency communications,” IEEE +Communications Magazine, vol. 57, no. 2, pp. 130–137, 2018. +[86] M. Sybis, K. Wesolowski, K. Jayasinghe, V. Venkatasubramanian, and +V. Vukadinovic, “Channel coding for ultra-reliable low-latency com- +munication in 5g systems,” in 2016 IEEE 84th vehicular technology +conference (VTC-Fall). +IEEE, 2016, pp. 1–5. +[87] T. Zhang, J. Shen, C.-F. Lai, S. Ji, and Y. Ren, “Multi-server assisted +data sharing supporting secure deduplication for metaverse healthcare +systems,” Future Generation Computer Systems, 2022. +[88] L. He, K. Liu, Z. He, and L. Cao, “Three-dimensional holographic +communication system for the metaverse,” Optics Communications, +vol. 526, p. 128894, 2023. +[89] A. Plechat´a, G. Makransky, and R. B¨ohm, “Can extended reality in the +metaverse revolutionise health communication?” NPJ digital medicine, +vol. 5, no. 1, pp. 1–4, 2022. +[90] G. Wang, A. Badal, X. Jia, J. S. Maltz, K. Mueller, K. J. Myers, +C. Niu, M. Vannier, P. Yan, Z. Yu et al., “Development of metaverse +for intelligent healthcare,” Nature Machine Intelligence, pp. 1–8, 2022. +[91] W. Y. B. Lim, Z. Xiong, D. Niyato, X. Cao, C. Miao, S. Sun, and +Q. Yang, “Realizing the metaverse with edge intelligence: A match +made in heaven,” arXiv preprint arXiv:2201.01634, 2022. +[92] P. Zhou, J. Zhu, Y. Wang, Y. Lu, Z. Wei, H. Shi, Y. Ding, Y. Gao, +Q. Huang, Y. Shi et al., “Vetaverse: Technologies, applications, and +visions toward the intersection of metaverse, vehicles, and transporta- +tion systems,” arXiv preprint arXiv:2210.15109, 2022. +[93] L. O. Alpala, D. J. Quiroga-Parra, J. C. Torres, and D. H. Peluffo- +Ord´o˜nez, “Smart factory using virtual reality and online multi-user: +Towards a metaverse for experimental frameworks,” Applied Sciences, +vol. 12, no. 12, p. 6258, 2022. +[94] Z. Allam, A. Sharifi, S. E. Bibri, D. S. Jones, and J. Krogstie, +“The metaverse as a virtual form of smart cities: Opportunities and +challenges for environmental, economic, and social sustainability in +urban futures,” Smart Cities, vol. 5, no. 3, pp. 771–801, 2022. +[95] H. Du, J. Wang, D. Niyato, J. Kang, Z. Xiong, D. I. Kim et al., +“Exploring attention-aware network resource allocation for customized +metaverse services,” arXiv preprint arXiv:2208.00369, 2022. +[96] B. Xu, Q. Wu, Y. Bao, G. Chen, Y. Wang, and S. Ren, “Time- +multiplexed stereoscopic display with a quantum dot-polymer scanning +backlight,” Applied Optics, vol. 58, no. 16, pp. 4526–4532, 2019. +[97] H. Hiura, K. Komine, J. Arai, and T. Mishina, “Measurement of +static convergence and accommodation responses to images of integral +photography and binocular stereoscopy,” Optics Express, vol. 25, no. 4, +pp. 3454–3468, 2017. +[98] T. North, M. Wagner, S. Bourquin, and L. Kilcher, “Compact and high- +brightness helmet-mounted head-up display system by retinal laser +projection,” Journal of Display Technology, vol. 12, no. 9, pp. 982–985, +2016. +[99] Y. Su, X. Tang, Z. Zhou, Z. Cai, Y. Chen, J. Wu, and W. Wan, +“Binocular dynamic holographic three-dimensional display for optical +see-through augmented reality using two spatial light modulators,” +Optik, vol. 217, p. 164918, 2020. +[100] H. Huang and H. Hua, “Systematic characterization and optimization +of 3d light field displays,” Optics express, vol. 25, no. 16, pp. 18 508– +18 525, 2017. +[101] S. Y. Kung, M.-W. Mak, S.-H. Lin, M. Mak, and S. Lin, Biometric au- +thentication: a machine learning approach. Prentice Hall Professional +Technical Reference New York, 2005. +[102] P. Punithavathi, S. Geetha, M. Karuppiah, S. H. Islam, M. M. Hassan, +and K.-K. R. Choo, “A lightweight machine learning-based authentica- +tion framework for smart iot devices,” Information Sciences, vol. 484, +pp. 255–268, 2019. +[103] Y. Xun, J. Liu, N. Kato, Y. Fang, and Y. Zhang, “Automobile driver +fingerprinting: A new machine learning based authentication scheme,” +IEEE Transactions on Industrial Informatics, vol. 16, no. 2, pp. 1417– +1426, 2019. +[104] M. Hazratifard, F. Gebali, and M. Mamun, “Using machine learning +for dynamic authentication in telehealth: A tutorial,” Sensors, vol. 22, +no. 19, p. 7655, 2022. +[105] M. Simsek, M. Bennis, and I. G¨uvenc¸, “Context-aware mobility man- +agement in hetnets: A reinforcement learning approach,” in 2015 ieee +wireless communications and networking conference (wcnc). +IEEE, +2015, pp. 1536–1541. +[106] H. Zhang, R. Wang, W. Sun, and H. Zhao, “Mobility management for +blockchain-based ultra-dense edge computing: a deep reinforcement +learning approach,” IEEE Transactions on Wireless Communications, +vol. 20, no. 11, pp. 7346–7359, 2021. +[107] M. M. U. Chowdhury, W. Saad, and I. G¨uvenc¸, “Mobility management +for cellular-connected uavs: A learning-based approach,” in 2020 +IEEE international conference on communications workshops (ICC +Workshops). +IEEE, 2020, pp. 1–6. +[108] W. Liu, P. Si, E. Sun, M. Li, C. Fang, and Y. Zhang, “Green +mobility management in uav-assisted iot based on dueling dqn,” in ICC +2019-2019 IEEE International Conference on Communications (ICC). +IEEE, 2019, pp. 1–6. +[109] B. Zheng, C. You, W. Mei, and R. Zhang, “A survey on channel +estimation and practical passive beamforming design for intelligent +reflecting surface aided wireless communications,” IEEE Communica- +tions Surveys & Tutorials, vol. 24, no. 2, pp. 1035–1071, 2022. +[110] B. Zheng and R. Zhang, “Intelligent reflecting surface-enhanced ofdm: +Channel estimation and reflection optimization,” IEEE Wireless Com- +munications Letters, vol. 9, no. 4, pp. 518–522, 2019. +[111] Z. Wang, L. Liu, and S. Cui, “Channel estimation for intelligent reflect- +ing surface assisted multiuser communications: Framework, algorithms, +and analysis,” IEEE Transactions on Wireless Communications, vol. 19, +no. 10, pp. 6607–6620, 2020. +[112] Q.-U.-A. Nadeem, A. Kammoun, A. Chaaban, M. Debbah, and +M.-S. Alouini, “Intelligent reflecting surface assisted wireless com- +munication: +Modeling +and +channel +estimation,” +arXiv +preprint +arXiv:1906.02360, 2019. +[113] B. Zoph and Q. V. Le, “Neural architecture search with reinforcement +learning,” arXiv preprint arXiv:1611.01578, 2016. +[114] T. Elsken, J. H. Metzen, and F. Hutter, “Neural architecture search: A +survey,” The Journal of Machine Learning Research, vol. 20, no. 1, +pp. 1997–2017, 2019. +[115] Y. Jaafra, J. L. Laurent, A. Deruyver, and M. S. Naceur, “Reinforcement +learning for neural architecture search: A review,” Image and Vision +Computing, vol. 89, pp. 57–66, 2019. +[116] S. Casale-Brunet, P. Ribeca, P. Doyle, and M. Mattavelli, “Networks +of ethereum non-fungible tokens: a graph-based analysis of the erc- +721 ecosystem,” in 2021 IEEE International Conference on Blockchain +(Blockchain). +IEEE, 2021, pp. 188–195. +[117] P. Christodoulou and K. Christodoulou, “A decentralized voting mech- +anism: engaging erc-20 token holders in decision-making,” in 2020 +Seventh International Conference on Software Defined Systems (SDS). +IEEE, 2020, pp. 160–164. +[118] “Ieee standard for local and metropolitan area networks - secure device +identity,” IEEE Std 802.1AR-2018 (Revision of IEEE Std 802.1AR- +2009), pp. 1–73, 2018. +[119] L. U. Khan, Z. Sabir, S. A. Mahmud, and G. M. Khan, “Comparison +of three interpolation techniques in comb-type pilot-assisted channel +coded ofdm system,” in 2013 27th International Conference on Ad- +vanced Information Networking and Applications Workshops. +IEEE, +2013, pp. 977–981. +[120] L. U. Khan, M. H. Zafar, M. I. Khattak, and N. Khan, “A novel +channel estimation error minimizing interpolation technique for ofdm +systems,” in 2014 9th International Symposium on Communication +Systems, Networks & Digital Sign (CSNDSP). +IEEE, 2014, pp. 716– +720. +[121] L. Wei, C. Huang, G. C. Alexandropoulos, E. Wei, Z. Zhang, +M. Debbah, and C. Yuen, “Multi-user wireless communications with +holographic mimo surfaces: A convenient channel model and spectral +efficiency analysis,” in 2022 Joint European Conference on Networks +and Communications & 6G Summit (EuCNC/6G Summit). IEEE, 2022, +pp. 488–493. +[122] A. Lenk, G. Katsaros, M. Menzel, J. R. Revelant, R. Skipp, E. C. +Leon, and V. Gopan, “Tiosa: Testing vm interoperability at an os +and application level–a hypervisor testing method and interoperability +survey,” in 2014 IEEE International Conference on Cloud Engineering. +IEEE, 2014, pp. 245–252. +[123] X. Luo, B. Yin, Z. Chen, B. Xia, and J. Wang, “Autoencoder- +based semantic communication systems with relay channels,” in 2022 +IEEE International Conference on Communications Workshops (ICC +Workshops). +IEEE, 2022, pp. 711–716. +[124] H. Xie, Z. Qin, G. Y. Li, and B.-H. Juang, “Deep learning enabled +semantic communication systems,” IEEE Transactions on Signal Pro- +cessing, vol. 69, pp. 2663–2675, 2021. + +25 +Latif U. Khan received his Ph.D. degree in Com- +puter Engineering and M.S. degree in Electrical +Engineering with distinction from Kyung Hee Uni- +versity (KHU), South Korea in 2021 and UET Pe- +shawar in 2017, respectively. He worked as a leading +researcher in the intelligent Networking Laboratory +under a project jointly funded by the prestigious +Brain Korea 21st Century Plus and Ministry of +Science and ICT, South Korea. Prior to joining +the KHU, he has served as a faculty member and +research associate in the UET, Peshawar, Pakistan. +He has published his works in highly reputable conferences and journals. He +is the recipient of KHU best thesis award. He is the author/co-author of two +conference best paper awards. He is also author of two books titled ”Network +Slicing for 5G and Beyond Networks” and ”Federated Learning for Wireless +Networks”. His research interests include analytical techniques of optimization +and game theory to edge computing, end-to-end network slicing, digital twins, +and federated learning for wireless networks. +Mohsen Guizani (S’85, M’89, SM’99, F’09) re- +ceived his B.S. (with distinction) and M.S. degrees +in electrical engineering, and M.S. and Ph.D. degrees +in computer engineering from Syracuse University, +New York, in 1984, 1986, 1987, and 1990, respec- +tively. He is currently a Professor with the Ma- +chine Learning Department, Mohamed Bin Zayed +University of Artificial Intelligence (MBZUAI), Abu +Dhabi, UAE. Previously, he served in different aca- +demic and administrative positions at the University +of Idaho, Western Michigan University, the Univer- +sity of West Florida, the University of Missouri-Kansas City, the University +of Colorado-Boulder, and Syracuse University. His research interests include +wireless communications and mobile computing, computer networks, mobile +cloud computing, security, and smart grid. He was the Editor-in-Chief of +IEEE Network. He serves on the Editorial Boards of several international +technical journals, and is the Founder and Editor-in-Chief of the Wireless +Communications and Mobile Computing journal (Wiley). He is the author +of nine books and more than 500 publications in refereed journals and +conferences. He has guest edited a number of Special Issues in IEEE journals +and magazines. He has also served as a TPC member, Chair, and General +Chair of a number of international conferences. Throughout his career, he +received three teaching awards and four research awards. He also received the +2017 IEEE Communications Society WTC Recognition Award as well as the +2018 AdHoc Technical Committee Recognition Award for his contribution to +outstanding research in wireless communications and ad hoc sensor networks. +He was the Chair of the IEEE Communications Society Wireless Technical +Committee and the Chair of the TAOS Technical Committee. He served as +a IEEE Computer Society Distinguished Speaker and is currently an IEEE +ComSoc Distinguished Lecturer. He is a Senior Member of ACM. +Dusit Niyato (M’09–SM’15–F’17) received the +Ph.D. degree in electrical and computer engineering +from the University of Manitoba, Winnipeg, MB, +Canada, in 2008. He is currently a Professor with +the School of Computer Science and Engineering, +Nanyang Technological University, Singapore. He +has published more than 400 technical articles in +the area of wireless and mobile computing. He +received the Best Young Researcher Award of the +IEEE Communications Society Asia Pacifica and +the 2011 IEEE Communications Society Fred W. +Ellersick Prize Paper Award. He is also serving as a Senior Editor of the IEEE +Wireless Communication Letters, an Area Editor of the IEEE Transactions +on wireless Communications and the IEEE Communications Surveys and +Tutorials, an Editor of the IEEE Transactions on Communications, and +an Associate Editor of the IEEE Transactions on Mobile Computing, the +IEEE Transactions on Vehicular Technology, and the IEEE Transactions on +Cognitive Communications and Networking. He was a Distinguished Lecturer +of the IEEE Communications Society from 2016 to 2017. He was named a +highly cited researcher in computer science. +Ala Al-Fuqaha (Senior Member, IEEE) received +the Ph.D. degree in computer engineering and net- +working from the University of Missouri-Kansas +City, Kansas City, MO, USA, in 2004. He is cur- +rently a Professor at the Information and Computing +Technology Division, College of Science and En- +gineering, Hamad Bin Khalifa University (HBKU), +Doha, Qatar. His research interests include the use +of machine learning in general and deep learning +in particular in support of the data-driven and self- +driven management of large-scale deployments of +the Internet of Things (IoT) and smart city infrastructure and services, wireless +vehicular networks (VANETs), cooperation and spectrum access etiquette in +cognitive radio networks, and management and planning of software-defined +networks (SDNs). +Merouane Debbah is Chief Researcher at the Tech- +nology Innovation Institute in Abu Dhabi. He is a +Professor at Centralesupelec and an Adjunct Pro- +fessor with the Department of Machine Learning +at the Mohamed Bin Zayed University of Artifi- +cial Intelligence. He received the M.Sc. and Ph.D. +degrees from the Ecole Normale Sup´erieure Paris- +Saclay, France. He was with Motorola Labs, Saclay, +France, from 1999 to 2002, and also with the Vienna +Research Center for Telecommunications, Vienna, +Austria, until 2003. From 2003 to 2007, he was +an Assistant Professor with the Mobile Communications Department, Institut +Eurecom, Sophia Antipolis, France. In 2007, he was appointed Full Professor +at CentraleSupelec, Gif-sur-Yvette, France. From 2007 to 2014, he was the +Director of the Alcatel-Lucent Chair on Flexible Radio. From 2014 to 2021, +he was Vice-President of the Huawei France Research Center. He was jointly +the director of the Mathematical and Algorithmic Sciences Lab as well as +the director of the Lagrange Mathematical and Computing Research Center. +Since 2021, he is leading the AI +Digital Science Research centers at the +Technology Innovation Institute. He has managed 8 EU projects and more than +24 national and international projects. His research interests lie in fundamental +mathematics, algorithms, statistics, information, and communication sciences +research. He is an IEEE Fellow, a WWRF Fellow, a Eurasip Fellow, an +AAIA Fellow, an Institut Louis Bachelier Fellow and a Membre emerite +SEE. He was a recipient of the ERC Grant MORE (Advanced Mathematical +Tools for Complex Network Engineering) from 2012 to 2017. He was a +recipient of the Mario Boella Award in 2005, the IEEE Glavieux Prize +Award in 2011, the Qualcomm Innovation Prize Award in 2012, the 2019 +IEEE Radio Communications Committee Technical Recognition Award and +the 2020 SEE Blondel Medal. He received more than 20 best paper awards, +among which the 2007 IEEE GLOBECOM Best Paper Award, the Wi-Opt +2009 Best Paper Award, the 2010 Newcom++ Best Paper Award, the WUN +CogCom Best Paper 2012 and 2013 Award, the 2014 WCNC Best Paper +Award, the 2015 ICC Best Paper Award, the 2015 IEEE Communications +Society Leonard G. Abraham Prize, the 2015 IEEE Communications Society +Fred W. Ellersick Prize, the 2016 IEEE Communications Society Best Tutorial +Paper Award, the 2016 European Wireless Best Paper Award, the 2017 Eurasip +Best Paper Award, the 2018 IEEE Marconi Prize Paper Award, the 2019 IEEE +Communications Society Young Author Best Paper Award, the 2021 Eurasip +Best Paper Award, the 2021 IEEE Marconi Prize Paper Award, the 2022 +IEEE Communications Society Outstanding Paper Award, the 2022 ICC Best +paper Award as well as the Valuetools 2007, Valuetools 2008, CrownCom +2009, Valuetools 2012, SAM 2014, and 2017 IEEE Sweden VT-COM-IT Joint +Chapter best student paper awards. He is an Associate Editor-in-Chief of the +journal Random Matrix: Theory and Applications. He was an Associate Area +Editor and Senior Area Editor of the IEEE TRANSACTIONS ON SIGNAL +PROCESSING from 2011 to 2013 and from 2013 to 2014, respectively. From +2021 to 2022, he serves as an IEEE Signal Processing Society Distinguished +Industry Speaker. + +BGF \ No newline at end of file