Keywords

1 Introduction

The 5G wireless cellular system has emerged in response to the need for providing high-speed, reliable wireless connectivity not only to conventional wireless services, such as video streaming, but also to the so-called Internet of Things (IoT) system that interconnects machine-type devices such as sensors and autonomous robots. Indeed, beyond providing high data rates to the so-called enhanced mobile broadband (eMBB) services such as mobile TV, 5G systems are expected to guarantee ultra-reliable, low-latency communication (URLLC) to IoT services. The need for URLLC was hailed as the ultimate game-changer for 5G systems and beyond.

In 5G, URLLC services are essentially viewed as IoT applications that must reliably collect very short packets (few bytes) from small sensors or robots with an uplink over-the-air latency of less than 1 ms. For such IoT services, high reliability is defined by 3GPP as achieving a percentage of successfully delivered packet within the application time deadline constraint within the range 99.9% to 99.999%, depending on the application. Moreover, for such URLLC IoT services, data rate can be completely neglected since the transmitted uplink packets are only a few bytes long.

While such a URLLC design is sound when dealing with IoT sensors networks or factory automation applications, it is questionable whether one can keep restricting URLLC services to uplink short packets in the near future. In particular, the IoT itself is now witnessing an unprecedented revolution that is disrupting its original premise as a mere machine-to-machine communication system and transforming it into a complex Internet of Everything (IoE) system in which machines, people, and complex processes must constantly communicate with one another. As explained in [1], the IoT can be viewed as “the equivalent of a railroad line, including the tracks and the connections, whereas the IoE is all of that and the trains, ticket machines, staff, customers, weather conditions, etc.” and more. Indeed, the IoE will bring forth new wireless services, such as large-scale autonomy (including fully autonomous vehicles, drones, and flying cars), a massive tactile Internet system, wireless brain-computer interfaces, and advanced eXtended reality (XR) applications (encompassing augmented, mixed, and virtual reality (AR/MR/VR)) [2]. These will impose very stringent quality-of-service (QoS) requirements across the rate, reliability, and latency spaces and will blur the boundary between classical 5G URLLC and eMBB services. As such, although 5G may be able to meet the QoS needs of basic XR or autonomous robotics, it will still fall short in meeting the more stringent rate (e.g., above 100 Gbps for some XR applications such as the so-called Ultimate VR class of services), latency (e.g., below 1 ms for wireless brain-computer interfaces), and high reliability (near-zero packet errors at low latency, i.e., extreme reliability [3]) needs of tomorrow’s IoE applications.

To overcome the challenges of emerging IoE services, there is a need to develop a novel sixth generation (6G) wireless system, whose design is inherently tailored to the need for highly reliable, low-latency, and high-rate services. 6G will be a byproduct of traditional trends in communication technologies (e.g., densification, higher rates, and massive antennas) coupled with recent services and technological advances that include new wireless devices (e.g., body implants, XR apparatus, etc.), emerging artificial intelligence (AI) paradigms [4], and the need for supporting multiple functions that range from imaging to sensing and control. Beyond supporting new IoE services, the road toward 6G must also be able to overcome some of the limitations of 5G that were identified in early rollouts of the system, including:

  1. (1)

    High Frequency, High Rate, High Mobility: At its early stages, 5G was envisioned to be a system that operates almost exclusively at high-frequency millimeter wave (mmWave) bands that can deliver the promised data rates. However, the early deployment of 5G systems is still primarily relying on sub-6 GHz spectrum bands, particularly for highly mobile scenarios. Indeed, thus far, mmWave has been deployed by a handful of operators and only for fixed wireless access. Therefore, the road to beyond 5G systems must revisit the problem of delivering high-speed wireless access at high frequencies for highly mobile environments.

  2. (2)

    Elusive Reliability: Although early deployments of 5G have shown very promising performance in terms of data rate and low latency, reliability targets remain elusive. To date, outside of laboratory trials, there has been no fully fledged deployment that achieved the reliability targets set forth when 5G was conceived, i.e., a five nine (99.999%) reliability at low latency. This shortcoming in terms of reliability can be partially attributed to the lack of URLLC fundamentals, as outlined in [3]. The challenges of reliability will be also further exacerbated when looking at higher frequency such as mmWave bands that will be a hallmark of beyond 5G systems.

  3. (3)

    Coverage in Extreme Conditions: Despite the growing success of 5G communication systems, conservative estimates show that more than half of the global population, mostly in extreme conditions, such as rural areas and disaster-affected areas, will still live in “wireless darkness” post the 5G era. Indeed, providing wireless coverage under extreme conditions has remained a major problem facing wireless systems, ever since the inception of the successful 2G cellular network. Examples of this lack of coverage are abundant. For example, in 2016, the FCC estimated that over 10% of all Americans and 40% of the US rural population did not have access to high-speed wireless connectivity [5]. Meanwhile, in the aftermath of hurricane Harvey, about 95% of cell sites in Houston stopped working, and wireless networks along the Texas coast suffered significant outages. As we enter the era of smart cities, the persistence of such a lack of connectivity will have adverse economic and societal consequences, and, thus, providing connectivity to rural and disaster-affected areas must become a priority for beyond 5G systems.

  4. (4)

    Spectral and Energy Efficiency: Although 5G will provide significant spectrum efficiency gains compared to 4G, as it stands, the originally sought target of a three-times increase in spectrum efficiency has not been met yet. Moreover, early deployments of 5G show that the system is less energy-efficient than 4G which once again motivates more fundamental research in spectrum and energy efficiency whose targets will be even higher for beyond 5G systems.

Motivated by these limitations and the challenges of future IoE services, in this chapter, we provide a holistic overview on how 6G wireless systems will entail. In particular, we develop a bold new vision of 6G systems (detailed in Fig. 7.1) that uncovers the key drivers of 6G across applications, trends, metrics, and technologies. We then present new 6G services and provide a prospective road map to accelerate the leap from 5G toward 6G while going by a “beyond 5G” milestone.

Fig. 7.1
figure 1

An overview on our broad vision for 6G: Applications, trends, and technologies

Full size image

2 6G Driving Applications, Metrics, and New Service Classes

As has been the case in every past generation of cellular systems, every new “G” is often motivated by a plethora of emerging wireless applications that bring forth new performance requirements and designs. 6G will be no exception: It will be driven by an unparalleled emergence of exciting new applications, ranging from XR to haptics, robotics and autonomous systems, and brain-computer interface over wireless networks, that will constitute the heart of the IoE and smart cities. These new services will require establishing new target performance metrics for 6G while radically redefining standardized 5G application types such as URLLC, eMBB, and massive machine-type communication (mMTC) as will be evident from Table 7.1. In this section, we first introduce the main driving applications of 6G cellular systems, and, then, based on those applications, we present some desirable metrics and performance requirements.

Table 7.1 Requirements of 5G vs. Beyond 5G vs. 6G
Full size table

2.1 6G: Driving Applications and Their Performance Requirements

Even though classical wireless applications, such as video services, live mobile streaming, and voice/messaging services, will still be essential in 6G systems, the performance of 6G will, however, be determined by four new classes of IoE applications that include as follows:

2.1.1 Multi-sensory XR Applications

6G must support several XR applications within the broad AR/MR/VR spectrum. Although 5G will surely support basic VR services, it will not be able to deliver a full immersive XR experience for more advanced AR/MR/VR applications (including holographic teleportation) that require capturing all sensory inputs from the users, thus imposing more stringent latency and reliability constraints than those possible with 5G. Moreover, many XR applications will require highly reliable, ultra-low-latency, and high-rate communications with requirements that cut across traditional eMBB and URLLC services. In addition, providing a realistic and immersive XR user experience requires a joint system design that integrates not only engineering (communications, storage, networking, computing) requirements but also perceptual requirements related to human cognition, senses, and physiology. Indeed, for future XR services, minimal and maximal human perceptual requirements and limits must be integrated within the communication, computing, and processing functions of the system. In other words, physical and cognitive human perceptions will now be a key determinant of QoS in XR applications. For this purpose, we envision defining a new notion of quality-of-physical-experience (QoPE) metric that combines physical/cognitive factors from the human user itself with the more common QoS measures (e.g., delay and rate) and conventional quality-of-experience (QoE) concepts such as the mean-opinion score. This concept of QoPE will be affected by various human-centric factors such as brain cognition/capabilities, body physiology, gestures, and even age. For instance, in our recent work in [7], we have demonstrated that the human brain fails to distinguish between different latency measures, particularly when operating within the URLLC regime. As a result, a network can save its communication resources (particularly energy resources to enhance energy efficiency) by being aware of the human-in-the-loop when performing resource management and network optimization. Meanwhile, in [8], we characterized the impact of visual and haptic perceptions on wireless performance using the just-noticeable difference (JND), an established measure from psychophysics. We have shown how using notions such as JND is necessary to truly understand the performance limits of wireless communication for XR-type services. In summary, the requirements of XR services for 6G will essentially be a combination of classical URLLC and eMBB with the need to integrate perceptual factors.

2.1.2 Connected Robotics and Autonomous Systems (CRAS)

A major driver for 6G is the anticipated proliferation of CRAS applications including drone systems, self-driving cars and platoons, flying vehicles, swarms of drones, and autonomous robotics. Deploying CRAS over cellular systems will not be a simple case of “an additional uplink IoT devices with short packets.” On the one hand, next-generation CRAS will require exchange of high-rate data such as high-definition (HD) maps that can serve for navigation purposes and require eMBB-level communication with high reliability. On the other hand, CRAS will have complex control systems that will dictate the latency and reliability requirements for the system. This, in turn, motivates a cyber-physical design to the wireless system in which physical constraints, extracted from the control system requirements and the physical world, must be incorporated into traditional, cyber-only QoS metrics. Indeed, here, we can once again redefine the QoPE notion while substituting the human factors with the physical control system factors. In addition, CRAS will undoubtedly open up the door for an unimaginably rich set of applications that we cannot even foresee at this point. Hence, CRAS can be seen as a prime use case for beyond 5G systems that will require not only stringent requirements across the reliability-rate-latency spectrum – a balance not yet available in 5G – but also a joint cyber-physical system design that merges control with communications and even navigation/localization.

2.1.3 Wireless Brain-Computer Interactions (BCI)

Traditionally, BCI applications were limited to healthcare scenarios involving electroencephalography (EEG) devices that allow humans to control prosthetic limbs or interact with nearby devices using brain implants. However, the ongoing revolution in the field of wireless brain to computer interfaces will inevitably lead to novel BCI paradigms (e.g., multi-brain-controlled cinema [9]) that will rely on 6G connectivity. Most recently, Neuralink announced that it has successfully built special microchips and flexible fiber electrodes that can be implanted into a human brain enabling it to seamlessly connect to computing systems or neighboring machines. Clearly, these tremendous advances in the field of BCI will soon lead to a broad range of brain implants that can wireless connect to the entire IoE system. In such applications, there is a seamless integration between human cognition and the wireless network, a feature that is not seen in 5G applications. Using BCI, one can envision transmitting brain signals over the wireless channel for various purposes such as communicating with nearby devices or even remotely controlling CRAS devices. As discussed in [6, 10], such wireless cognition can strain the capacity of existing 5G systems and will require very strict rate, latency, and reliability guarantees. Indeed, once implants and BCI devices become more pervasive, we envision a world in which, instead of conventional smartphones, individuals will interact with their environment and other individuals using a range of worn, embedded, and implanted devices. Such interactions will enable control of environments using gestures and communication with others using haptic messages. This, in turn, yields many interesting questions such as what happens when humans communicate through touch instead of words? Meanwhile, the emerging notion of “emotion-driven devices” also called affective computing that helps devices understand our mood and then make suggestions to match it (e.g., rescheduling a meeting due to a mood swing or tiredness) will also leverage BCI devices. Such affective, emphatic, and haptic communications will potentially become major use cases for 6G. Compared to the main 5G services, wireless BCI application will need fundamentally different performance indicators. Analogous to XR, wireless BCI applications will require high reliability, high data rates, and extremely low latency. However, compared to XR, BCI applications will be more sensitive to physical perceptions thus requiring performance guarantees in terms of QoPE. In addition, wireless BCI opens the door for new research problems at the intersection of neuroscience and wireless communications.

2.1.4 Blockchain and Distributed Ledger Technologies (DLT)

Many next-generation IoE services will rely on blockchain and DLT ideas in order to provide security and distributed operation. Although the notions of blockchain and DLT are not directly tied to communication, their operations will require a reliable communication infrastructure. In particular, massive machine-type communications (mMTC) services in tomorrow’s IoE will largely integrate blockchain and DLT ideas. This, in turn, motivates a need for understanding the synergies between communication and blockchain. For example, it has been recently shown in [11] that wireless network errors can impact the probability of forking in a blockchain. This, in turn, motivates revisiting the fundamentals of URLLC and mMTC services in order to understand how blockchain and DLT applications will impact the requirements. In many ways, blockchain and DLT applications can be seen as massive, distributed IoE sensing applications that need a wireless connection that can provide a synergistic mix of massive machine-type communications (mMTC) and URLLC in order to maintain reliable connectivity, low latency, as well as scalability.

2.2 6G: Key Trends and Metrics

Clearly, the driving applications of 6G outlined in Sect. 7.2.1 will yield several novel cross-system trends that will, in turn, determine the objectives that 6G must be able to meet. These trends can be summarized into seven key groups:

  • Trend 1 – More Bits, More Spectrum, More Reliability: A majority of the applications outlined in Sect. 7.2.1 will require higher data rates and higher spectrum efficiency than what 5G can deliver. For example, we anticipate yet another need for a 1000× increase in data rates, yielding a target of around 1 terabit/second, in order to meet the QoS requirements of next-generation IoE services such as XR or even wireless BCI. This, in turn, once again requires the exploration of more spectrum resources that go beyond sub-6 GHz and mmWave frequency bands. Moreover, much of the aforementioned 6G services will require an even more stringent reliability than basic IoT sensor systems. Therefore, delivering higher reliability is a major challenge for all wireless systems starting from 5G. This challenge is particularly exacerbated by the fact that going higher in frequencies negatively impacts reliability. As such, one key driving trend behind 6G will be the need for higher reliability at higher frequency bands.

  • Trend 2 – From Areal to Volumetric Spectral and Energy Efficiency: Drones and flying vehicles will become a major component of beyond 5G systems. Indeed, 6G must serve both ground and flying users, encompassing smartphones, XR devices, and BCI implants, along with aerial vehicles and drones. Hence, we are now witnessing a transformation from traditional, two-dimensional wireless systems into fully fledged three-dimensional (3D) wireless systems [12]. This 3D nature of 6G mandates an evolution toward defining volumetric spectrum and energy efficiency, rather than the traditional areal definition. We particularly anticipate that 6G systems must deliver very high spectral and energy efficiency (SEE) requirements quantified in bps/Hz/m3/Joules. This is aligned with the evolution that we have seen starting from 2G systems (bps) to 3G systems (bps/Hz), then 4G systems (bps/Hz/m2) to 5G systems (bps/Hz/m2/Joules).

  • Trend 3 – Emergence of Smart Surfaces and Environments: All current generations of wireless cellular systems used base stations (BSs), of different sizes, forms, and number of antennas, to transmit data to their users. However, satisfying the very stringent rate, coverage, spectrum efficiency, and delay requirements of the 6G applications discussed in Sect. 7.2.1 through the traditional trends of designing better BS-centric transceivers or using more antennas at BS towers will no longer be possible. One promising solution is to exploit the tremendous recent advances in metamaterials that allows one to transform man-made structures such as buildings, walls, and roads into electromagnetically active metasurfaces with radio frequency (RF) capabilities. This transformation, exemplified by the Berkeley ewallpaper project,Footnote 1 will allow future wireless systems to use such smart metasurfaces as large, reconfigurable intelligent surfaces (RISs) and environments to provide pervasive, high-speed wireless communications. This trend that shifts from traditional tower-mounted BSs toward RISs will be a major force behind the 6G architectural evolution.

  • Trend 4 – Massive Availability of Small Data: AI is witnessing a radical departure from traditional centralized “big data” cloud architectures toward a distributed AI paradigm in which massive, “small” data is dispersed across multiple edge devices and must be processed in a distributed manner using on-device machine learning. This paradigm shift will be further fueled by the emerging IoE discussed in Sect. 7.2.1 in which sending large data volumes to a cloud faces major communication, privacy, and scalability challenges. In consequence, 6G systems are expected to leverage both big and small datasets that are dispersed across the system so as to improve network operation and deliver new services. In consequence, it is necessary to investigate new AI and machine learning techniques that go beyond classical big data analytics so as to address the aforementioned challenges with effective distributed AI techniques that can exploit local, on-device edge data processing tailored to the real-time, private, and mission-critical nature of the IoE services.

  • Trend 5 – From Self-Organizing Networks (SONs) to Self-Sustaining Networks: Classical or legacy cellular trends, such as SONs, will witness yet another evolution in 6G systems. For instance, SON functions, which have remained elusive or scarcely integrated into 4G/5G systems, become a necessity for 6G given the highly distributed nature of driving applications such as CRAS and DLT technologies. Therefore, 6G must be able to deliver intelligent SON functions that can be used to manage network resources and operations, as well as system optimization. In fact, the deployment of CRAS and DLT applications calls for a paradigm shift from traditional SON, using which the cellular system simply adapts its functions to some states of the environment, to a self-sustaining network (SSN) that is able to guarantee its key performance indicators (KPIs), in near-perpetuity, under largely complex and dynamic environments that will result from the diverse domain of 6G applications and services. SSNs should be capable of not only adapting their network functions but also sustaining their system resource usage and management (e.g., by properly exploiting spectrum resources and potentially harvesting energy) to autonomously maintain stringent, long-term KPIs. Naturally, the deployment of SSN functions will make use of the recent advances in the AI domain. Ultimately, 6G could potentially incorporate AI-powered SSN algorithms.

  • Trend 6 – Convergence of Communications, Computing, Control, Localization, and Sensing (3CLS): To date, all existing wireless cellular systems were designed with one exclusive purpose: providing wireless connectivity. However, it is expected that 6G will be able to deliver services beyond communications. In particular, 6G will mark a convergence of diverse functions that include communications, control, computing [13], localization, and sensing. We view 6G as a multi-purpose and versatile system that can offer a diverse set of 3CLS applications which are particularly suitable and arguably necessary for services like CRAS, XR, and DLT where tracking, control, localization (e.g., for navigation), and computing are an inherent feature. In addition, by leveraging the use of sensing functions, 6G can build a 3D mapping of the radio environment across multiple frequency bands. This mapping can then be used to assist in network functions and user operation. In a nutshell, 6G systems must seamlessly integrate and manage a broad range of 3CLS functions.

  • Trend 7 – End of the Smartphone Era: Smartphones were the driving force behind the wireless revolution from 3G all the way up to 5G. However, as exemplified by the advances in the BCI and XR fields, the next decade will experience an exponential rise in the number of embedded wearable devices and implants whose functionalities will gradually start replacing those of smartphones. For instance, XR and BCI devices that include smart wearables, integrated headsets, and advanced body implants that can take direct sensory inputs from human senses can potentially bring an end to the smartphones era and constitute major 6G use case scenarios. For instance, we may see a shift from traditional BS-to-smartphone communication links, which are integral to all current and previous cellular systems, toward RIS-to-implant communication links in 6G and beyond.

From Table 7.1, we can observe that the aforementioned trends will collectively lead to new desirable performance targets and requirements that will be met in two stages of cellular system evolution: (a) a first (evolutionary) stage that we can call a “beyond 5G” stage and (b) a second, revolutionary 6G stage.

2.3 New Service Classes for 6G

In addition to introducing new performance metrics and targets, the aforementioned technological trends will necessitate redefining the different application and service types in 5G by enhancing and potentially combining conventional URLLC, eMBB, and mMTC services while also leading to new types of services (summarized in Table 7.2), as discussed next:

Table 7.2 Summary of 6G service classes, their performance indicators, and example applications
Full size table

2.3.1 Mobile Broadband Reliable Low-Latency Communication

From the discussion in Sect. 7.2.2, we can easily observe that the existing distinction between eMBB and URLLC is not sustainable to support tomorrow’s IoE applications such as advanced XR, wireless BCI, CRAS, or even smart city services. This is due to the fact these applications will need not only very low latency and high reliability but also high data rates (at the level of eMBB services). To cater for these requirements, we can introduce a new cellular service class that we dub mobile broadband reliable low-latency communication (MBRLLC) which will allow 6G to deliver any desirable performance target that lies in the rate-reliability-latency dimensions. In general, MBRLLC can be seen as a generalization of classical URLLC and eMBB services. In addition, energy efficiency will be a major design challenge for MBRLLC not only because of its effect on reliability and data rate but also because devices in 6G will continuously shrink in their size and increase in their functions, thus requiring highly energy-efficient designs.

2.3.2 Massive URLLC

In 5G, URLLC services pertain to guaranteeing reliability and low latency for well-defined uplink IoE applications such as IoT sensors. The fundamentals of URLLC have already been widely explored in the literature (e.g., see [14]). However, in 6G, there is a need to scale traditional URLLC across the device dimension. This, in turn, can yield a new massive URLLC (mURLLC) service that combines 5G URLLC with conventional mMTC. mURLLC exhibits a trade-off in reliability-latency-scalability, and, thus, it requires a departure from average-based system designs (e.g., based on average data rate or average latency). Instead, a principled and scalable framework which accounts for latency, reliability, packet size, network architecture, topology (across edge, access, and core), and decision-making under uncertainty is needed [15]. Moreover, in mURLLC one must also deal with very extreme networking conditions as outlined in [3].

2.3.3 Human-Centric Services

6G will have to deal with human-centric services (HCS), a new type of service class that imposes QoPE performance targets (tightly integrated with the human users and their body/physiology, as discussed in Sect. 7.2.1) instead of raw rate-reliability-latency metrics. A prime example of HCS would be wireless BCI applications in which service performance is determined by the cognition, actions, and even physiology of human users. For HCS, a new set of QoPE performance indicators should be defined and quantified as function of traditional (raw) QoE and QoS performance metrics.

2.3.4 Multi-purpose 3CLS and Energy Services

6G systems must jointly deliver 3CLS services and their derivatives. For example, 6G systems can provide navigational and localization inputs to CRAS devices. In addition, using new advances in wireless energy transfer, 6G systems can potentially provide energy to recharge small devices such as IoE sensors. Hence, we anticipate that 6G will have to define a new class of service that goes beyond communication. These multi-purpose 3CLS and energy services (MPS) will be of central importance for CRAS and wireless BCI applications, among others. For MPS, there is a necessity for (a) joint uplink-downlink designs and (b) meeting desirable performance targets for the control (e.g., in terms of control stability), computing (e.g., computing delay), energy (e.g., amount of energy to transfer), localization (e.g., precision of localization), as well as mapping and sensing functions (e.g., accuracy of a mapped environment). MPS services are also suitable to operate cyber-physical systems over the wireless infrastructure.

3 6G: Enabling Technologies

To enable all the foreseen 6G services and meet their required QoS and QoPE performance, a broad range of new, disruptive technologies must be integrated into 6G systems. These technologies, their challenges, and their 6G integration are explained next.

3.1 6G at Above 6 GHz: From Small Cells Toward Tiny Cells

From Trends 1 and 2, we can see that, in 6G, higher data rates and SEE will be needed anywhere, anytime. This, in turn, motivates the exploration of higher-frequency bands beyond sub-6 GHz which had already started with 5G and mmWave. First, and foremost, as previously mentioned, one of the key limitations of 5G systems is the lack of high-speed wireless access at high frequencies for highly mobile environments. Therefore, a first step in this area requires developing new fundamental science to understand how one can make mobile mmWave a reality in early 6G systems or even at the beyond 5G step. The enabling technologies needed to realize the vision of mobile mmWave communications can include a combination of sub-6 GHz and mmWave bands, as well as the use of caching to minimize handover failures as proposed in [16]. As 6G progresses, exploiting frequencies above mmWave, particularly at the terahertz (THz) frequency ranges, will then become critical [10]. To leverage higher THz and mmWave frequencies, the size of some of the 6G cells would have to shrink from small cells to “tiny cells” whose radius is only few tens of meters. As a result, once we go higher in frequency and reach the THz frontier, there will be a need for new network architecture designs that can accommodate much denser deployments of tiny cells. In addition, at higher frequencies, there is a need for developing new mobility management techniques that are tailored to the highly intermittent nature of high-frequency communication links.

3.2 Transceivers with Integrated Frequency Bands

Delivering seamless connectivity for mobile 6G services will not be possible by only relying on dense, high-frequency tiny cells. Instead, we must conceive of an integrated system that can exploit multiple frequencies across the microwave/mmWave/THz bands (e.g., using multimode base stations) in order to provide seamless connectivity at both wide and local area network levels. Early works in [17, 18] have shown the potential of exploring such multiband communication. In addition, one can anticipate the integration of RF and non-RF bands. For instance, one can leverage the use of RF, optical, and visible light communication (VLC) to enhance not only communication efficiency but also resilience of the system to surges in traffic (e.g., in hotspot areas or even disaster-affected areas).

3.3 Communication with Large Reconfigurable Intelligent Surfaces

In order to deliver higher data rates, wireless research activities have mainly focused on two directions: (a) exploring higher frequencies such as mmWave bands and (b) equipping tower-mounted BSs with a massive number of RF antennas via the so-called massive multiple-input multiple-output (MIMO) communication paradigm. Indeed, both massive MIMO and mmWave bands will be integral to both 5G and 6G because they can help overcome the challenge of spectrum scarcity as well as the wireless channel impediments such as fading and interference, thus delivering better SEE and higher data rates at higher frequencies (Trend 1). However, these two paths toward faster wireless networking have their own limitations. For instance, the number of RF antennas and the type of RF circuits that can be used to create a “truly massive” MIMO system are limited by the hardware capability of BS towers. Moreover, due to the high susceptibility of mmWave frequencies to channel variations (e.g., blockage), reaping their benefits requires maintaining constant line-of-sight (LoS) links to the users, a feat only possible through network densification – deploying a significantly large number of massive MIMO BSs. However, densification is again limited by various geographical and hardware constraints. Hence, ushering in the 6G era will require a major rethinking to the architecture of wireless cellular systems. In particular, for many decades, the focus of wireless research and development efforts has been on designing effective transceivers while assuming the wireless channel and its propagation environment to be uncontrollable. In contrast, owing to the recent advances in metamaterial-based devices, it is now possible to transform man-made structures such as walls, buildings, and roads into electromagnetically active metasurfaces that can be employed as RF transceiver, complementing or even replacing traditional tower-mounted BSs. By doing so, one can build large RISs that can be used to not only design more effective massive MIMO transceivers (with antenna arrays spanning a very large surface and an ability to perform near-field LoS communications) but to also control the propagation environment by employing RISs as reflectors of wireless signals. Indeed, for 6G systems, as per Trend 3, we foresee an initial shift from conventional massive MIMO over tower-mounted BSs toward large RISs and smart environments [19,20,21,22,23] that act as both transceivers and reflectors so as to provide massive surfaces for wireless communications and for heterogeneous devices (Trend 7). RIS will hence allow 6G systems to now control the propagation environment, thus yielding many new research challenges and opportunities. They will also enable novel ways for wireless communication such as by using holographic RF radio and holographic MIMO.

3.4 Edge AI

AI is experiencing a major interest from the wireless communications community [4] motivated by some of the recent breakthroughs in deep learning, the increase in the data availability (Trend 4), and the emergence of smart devices (Trend 7). We envision at least three 6G use case scenarios for AI: (a) growth in big data analytics through AI, particularly for prediction, caching, and environment mapping tasks; (b) emergence of distributed AI for network optimization and for creating SSNs (Trend 5), through multi-agent reinforcement learning techniques; and (c) rise of on-device, edge AI techniques that exploit advances in federated learning [24] and related areas to enable the 6G system to exploit distributed, small data that is dispersed across its devices. Moreover, AI will allow 6G to automatically provide MPS to its devices and to generate and transmit 3D radio environment maps (Trend 6). Ultimately, we expect that 6G systems will witness a new paradigm of “collective network intelligence” in which network intelligence is further fostered at the edge to provide fully distributed autonomy. This new leap toward edge AI will lead to a 6G system that can support the services of Sect. 7.2, deliver 3CLS, and potentially substitute classical frame structures that were proposed by 3GPP. Indeed, a major open problem here is whether future wireless systems, starting with 6G, will gradually become fully operated and managed by AI functionalities.

3.5 Integrated Terrestrial, Airborne, and Satellite Networks

As outlined earlier, providing coverage to rural areas and areas with extreme conditions (e.g., disaster-affected areas) has been a major challenge for wireless systems since the rise of cellular networks a couple of decades ago. One seemingly promising solution for this decades-old problem is through the use of drones that can act as flying wireless BSs. Indeed, drone-BSs can complement terrestrial ground networks by delivering wireless connectivity to hotspots and to rural areas that has little to no infrastructure. Drone-BSs can also be used to provide on-demand wireless access in response to emergency situations in disaster-affected areas. In addition to acting as BSs, drones will also be integral users of 5G infrastructure and beyond. Indeed, drones will require wireless connectivity in order to receive control data, transmit sensing data (e.g., maps or videos), or communicate with ground infrastructure. This dual role of drones in tomorrow’s wireless networks means that 6G systems will inherently be 3D wireless systems that must meet volumetric performance targets (Trend 2). In addition, to support communication for drone-BSs as well as terrestrial BSs, there will be a need for guaranteeing satellite connectivity with low orbit satellites (LEO) and CubeSats to provide backhaul links as well as further wide area coverage. Integrating terrestrial, airborne, and satellite networks [12] and [25] into a single wireless system will therefore be an important objective for 6G.

3.6 Energy Transfer and Harvesting

One common feature among all the aforementioned IoE services is their need for energy efficiency. In fact, many IoE devices, including wearables, sensors, and implants, have a very small form factor and very limited energy and computing resources. As such, 6G systems must be able to deliver more energy-efficient communications. To do so, one possibility is to exploit emerging energy harvesting and energy transfer technologies. On the one hand, 6G infrastructure may leverage advances in energy harvesting to equip network devices (e.g., BSs or even drones) with solar-powered energy sources that can provide a clean and continuous source of power. On the other hand, RF energy harvesting and transfer can be exploited to provide RF energy to IoE devices. Indeed, 6G can possibly be the first cellular generation that can deliver energy, as well as 3CLS (Trend 6). As wireless energy transfer technologies start to mature, we envision 6G BSs to be capable of providing basic energy transfer for IoE devices, especially implants, wearables, and sensors (Trend 7). Other energy-related ideas such as energy harvesting and backscatter communication will also constitute important enabling technologies for 6G.

3.6.1 Beyond 6G

A few technologies will start to mature along the same timeline as 6G, and, hence, they can potentially play a role toward the final steps of the 6G standardization and research process. One major example is quantum computing and communications that can provide security and long-distance networking. While 6G system will likely not leverage much quantum technologies, we foresee that quantum communication will start to become more viable and practical along the same timeline as 6G. As such, although the specific role of quantum communications and computing in 6G remains rather unclear, the next few years will see more synergies across these two areas. For instance, quantum computing can potentially speed up much of the algorithms that run in a cellular system, thus contributing to reducing latency. Moreover, quantum computing can also be an important enabler for faster AI at the edge of 6G systems and beyond. Last, but not least, emerging areas such as neuro-inspired designs [26] and molecular communications may also have a role in shaping 6G systems.

4 6G: Open Research Problems

As is evident from the trends that we have identified in Sect. 7.2 and the enabling technologies that we exposed in Sect. 7.3, 6G will bring forth many interesting open problems and challenges, as summarized in Table 7.3 and discussed next.

Table 7.3 Summary of key 6G research areas
Full size table

4.1 3D Rate-Reliability-Latency Fundamentals

Performance analysis is arguably always the first step toward understanding the limits and capabilities of a wireless system. 6G will not be an exception: There is a clear need for characterizing its performance. In particular, for 6G systems, there is a need for new techniques to understand the fundamental 3D performance of the system, in terms of rate-reliability-latency trade-offs as well as volumetric SEE. This analysis must be able to quantify the spectrum, energy, and communication requirements that are needed by 6G in order to support the previously discussed driving applications. Some recent works in [15, 27,28,29] provide a first step in this direction.

4.2 Leveraging Integrated, Heterogeneous High-Frequency Bands

Leveraging high-frequency bands such as mmWave and THz in 6G opens up the door for a diverse set of new research problems. As previously discussed, for mmWave, a central open problem is to develop new mobility management techniques that enable mobile communications at mmWave bands. Here, there is a need for new mobility management protocols that can minimize handover rates (as done in [16]) as well as for fundamental analysis of mobility performance in mmWave networks. It can also be of interest to investigate whether AI techniques can help in improving performance for highly mobile mmWave systems by exploiting environmental information (e.g., image modalities [3]) or by predicting the mobility patterns and user behavior [30]. Meanwhile, at the THz frequencies, there is a need for novel transceiver architectures and propagation models [10, 31]. In order to overcome the high THz path loss, new transceivers must exhibit high power, high sensitivity, and low noise figure. Once physical layer challenges are overcome, there is a need to introduce new link-layer, multiple access, and network protocols so as to optimize the use of cross-band resources while factoring in uncertain and dynamically varying mmWave and THz environments. Another important research direction here is to investigate the co-existence of THz, mmWave, and microwave cells across all layers, building on early works such as [18]. In addition, there is a need to understand whether high-frequency bands can indeed provide reliable communication. For example, in [32], we have shown that molecular absorption can significantly affect the distance at which THz communications can provide reliable links to XR users, and in [33], we have shown that blockages will also play an important role in determining whether THz can provide reliable but high-rate links as required by XR services. Building on this work, one envisions a plethora of open problems related to investigating whether MBRLLC is possible at THz frequency or whether it is inevitable to exploit integrated frequency bands for maintaining high reliability.

4.3 3D Networking

Because ground and aerial networks are becoming largely integrated, as discussed in Sect. 7.3, 6G will have to support communications in 3D space. This includes providing connectivity to 3D flying users, as well as enabling the deployment of 3D drone-carried BSs (e.g., temporary drone-BSs or tethered balloons). This motivates the need for major research efforts across multiple directions related to 3D networking. First, data-driven and measurement-based modeling of the 3D propagation environment is needed. Such modeling requires both theoretical and experimental efforts that can create realistic propagation models for 3D cellular systems. Second, novel techniques for performing 3D frequency and wireless network planning (e.g., where to place BSs, balloons, or drone-BSs) must be investigated. Our prior results in [12] showed that such 3D planning is substantially different from conventional 2D networks due to the new altitude dimension and the associated degrees of freedom. In particular, we showed that designing a fully fledged 3D cellular system puts forward new challenges from network deployment all the way to optimization and network operation. Last, but not least, new techniques for network optimization, 3D mobility management, routing, and dynamic network resource management are also needed.

4.4 Communications with RISs

As a byproduct of Trend 3, 6G will potentially deliver wireless connectivity via intelligent RIS systems that encompass active frequency-selective surfaces, passive metallic reflectors, passive/active reflection arrays, as well as non-reconfigurable and reconfigurable metasurfaces. This naturally leads to many important research problems that range from the optimized deployment of passive reflectors and metasurfaces to the intelligent operation (potentially using edge AI) of reconfigurable metasurfaces. Fundamental performance analysis to understand the performance limitations and benefits of RISs and smart surfaces in terms of data rate, delay, reliability, and achievable coverage is also needed, building on the early works in [19,20,21,22,23]. Other key open problems here include investigating how practical models for metamaterial-based RIS devices and systems can impact the operation of RIS-based RF transceivers or reflectors. In other words, there is a need for metamaterial-informed communication system models that can reflect the real-world constraints of actual RIS devices. Last, but not least, it is of interest to analyze how RISs can leverage high-frequency bands (e.g., THz or mmWave) to provide high-speed connectivity to services such as XR, as studied in our work in [34].

4.5 AI for Wireless

AI brings forward many major research directions for 6G. We can distinguish two major areas: (a) AI for wireless communication and (b) wireless communication for AI. In the first area, beyond the need for massive, small data analytics, there is a need for deploying innovative machine learning algorithms that can provide SSN functions to 6G systems. This includes AI-enabled network optimization, resource management, and distributed network control. This area will particularly leverage advances in multi-agent reinforcement learning, artificial neural networks (ANNs), and game theory so as to instill smart, self-sustaining properties into 6G systems. One prominent problem here is the development of ANN-driven reinforcement learning algorithms that enable a network to optimize the usage of its resources while learning from its environment. This is particularly suitable for emerging applications such as XR and drone networking, as done in [35,36,37,38,39]. In the second direction, there is a need to understand how wireless factors, such as fading, mobility, or interference, can impact the performance of edge AI algorithms, such as federated learning. For example, it was shown in [24] and [40] that the convergence of federated learning will be strongly impacted by wireless packet errors and wireless latency. This, in turn, motivates a need for joint design of wireless and learning algorithms, particularly when dealing with edge AI techniques such as federated learning or distributed generative adversarial networks (GANs) [41]. Indeed, to perform critical edge AI application tasks, low-latency, high-reliability, and scalable AI is needed, along with a reliable infrastructure [4] and [8]. This joint design of ML and wireless networks is a very important 6G research area. Another important open problem for AI in 6G systems is the design of training-free machine learning frameworks that can execute different tasks with very limited training data. One step toward this direction is through the idea of experienced deep learning that we introduced in [27] in which a deep learning agent is allowed to gain experience in a virtual environment that can be created using GANs.

4.6 QoPE Metrics

The development of QoPE metrics that integrate physical factors from human cognition and physiology (for HCS) or from a control system (e.g., for CRAS) is an important 6G research field, particularly in light of the type of emerging network devices (Trend 7). This requires both realistic psychophysics experiments and new, precise mathematical QoPE expressions that merge QoS, QoE, and human user perceptions. In order to perform theoretical modeling of QoPE metrics, one can explore techniques from other domains that include the field of multi-attribute utility theory in operations research (e.g., see [38]) and the field of machine learning (e.g., see [7]). We anticipate that 6G could be one of the first cellular network generations that can support a whole new range of services (wireless BCI) that can capture the multiple cognitive senses of a human.

4.7 Joint Communication and Control

6G will have to provide pervasive connectivity to CRAS services for various purposes such as navigation and control. The performance of services such as CRAS is highly dependent on their practical control systems whose operation requires data input from the wireless communication links of 6G. Hence, effectively integrating CRAS over 6G systems requires a new communication and control co-design paradigm in which the wireless communication performance of the 6G links is optimized in a way to satisfy the control system stability. Meanwhile, the control system must be designed in a way to be cognizant of the wireless network state. Due to the traditional radio-centric focus (3GPP and IEEE fora), this joint communication and control co-design aspect has not yet been investigated in depth. Here, we note that prior art on related ideas, such as networked control systems, often abstracts the wireless network specifics and, hence, they cannot be directly applied to real-world cellular communications. As a result, communication and control co-design will be an important research problem for 6G. However, in order to deliver connectivity to cyber-physical systems such as autonomous vehicle, there is a need to couple the performance of the control, communication, and computing systems. For example, in [42, 43], we provided guidelines on how one can design a wireless system that can meet the control system stability requirements (in terms of delay and reliability) of autonomous vehicles. Similar performance analysis and joint communication, control, and computing designs are needed for a broad range of CRAS applications ranging from vehicular platoons to autonomous swarms of drones [44, 45].

4.8 3CLS

Beyond joint communications and control, one can also envision a need for joint design across all 3CLS functions. For instance, there is little work that rigorously studies the possible interdependence between computing, communication, control, localization, sensing, energy, and mapping, from an end-to-end perspective. Fundamental problems here range from developing new ways to jointly achieve the target performance of all 3CLS functions to introducing multimodal sensor fusion algorithms for faithfully reconstructing 3D images and allowing autonomous vehicles or robots to navigate in unknown environments. 3CLS will be relevant to several 6G applications that include XR, CRAS, and DLT.

4.9 Design of 6G Protocols

From the discussion in Sect. 7.2.2 and the identified challenges, one can see that 6G may require a major redesign of protocols. For instance, conventional 5G protocols may need to be replaced with novel AI-powered protocols for various network functions that range from signaling to scheduling. In contrast to the largely rigid designs of 5G protocols, new 6G protocols must be able to continuously evolve with the dynamic state of the wireless system. Moreover, as the development of 6G progresses, one must investigate the possibility of introducing novel protocols for dynamic multiple access [46]. These multiple access protocols must be able to intelligently switch the type of adopted multiple access (orthogonal or non-orthogonal, random or scheduled) scheme based on the network state and the application requirements. In addition, there is also a need for developing novel protocols for device handover that are cognizant of the 3D nature of 6G and the diverse types of mobile devices that must be served. Authentication and identification protocols will also have to be revisited in order to support a new generation of wireless devices such as vehicles, drones, and implants. Finally, 6G may require all protocols to be distributed in order to exploit the small datasets distributed over the system’s edge.

4.10 RF and Non-RF Link Integration

In 6G, it is expected that multiple forms of RF and non-RF links will co-exist. In particular, a 6G device may potentially be able to leverage optical, visible light communication (VLC), molecular communication, and neuro-communication, among others. Indeed, the design of systems with joint RF/non-RF capabilities is an important research area for 6G and beyond 5G.

4.11 Holographic Radio

By deploying RISs and similar metastructures, it is conceivable that RF holography (including holographic MIMO) and spatial spectral holography will become possible in 6G systems. In essence, holographic RF provides spatial spectral holography and spatial wave field synthesis capabilities that enable a network to control the entirety of a physical space as well as the full closed-loop of an electromagnetic field. This can significantly enhance spectrum efficiency and system capacity. It also will be an enabler for integrating imaging and wireless communication. Clearly, holographic radio is a widely open research problem for 6G.

Finally, to explore all the aforementioned open areas, there will be a need for a broad range of analytical tools. As a result, in Fig. 7.2, we provide a detailed summary of those analytical tools that will play important roles in 6G systems.

Fig. 7.2
figure 2

Necessary foundations as well as associated analytical tools for 6G

Full size image

5 Conclusions

Although it is too early to assert how 6G systems will look like, in this chapter, we provided a rather comprehensive and holistic view on what the main building blocks of 6G systems will be. In particular, we have identified the main limitations of current 5G systems and outlined some of the driving trends and applications behind the leap toward 6G. We have then provided a holistic view on the trends, technologies, and open problems of 6G wireless systems. Although several topics will emerge as a natural evolution from 5G, new research avenues such as RIS and smart surface communication, 3CLS, wireless networking for BCI, and others will create an exciting research agenda for the next decade that will span multiple disciplines as seen in Fig. 7.2. We have also made a few key observations:

  1. (1)

    The first near-term step toward 6G systems will be to enable mobile broadband services at high-frequency mmWave bands which will be necessary to sustain high-speed communications at high frequencies.

  2. (2)

    The next step toward 6G will be to understand the fundamentals of URLLC, with a focus on the notion of reliability. This will require exploring new tools from economics and statistics to quantify the performance of wireless systems in terms of distributions, rather than averages.

  3. (3)

    Future wireless services will likely require high reliability, low latency, and high data rates which is a significant departure from traditional short-packet, low-rate URLLC services. This, in turn, requires a new understanding of the fundamental trade-offs governing the rate-reliability-latency spaces.

  4. (4)

    6G will experience a transition from the smartphone-BS paradigm into a new era of smart surfaces communicating with implants and human-embedded devices. This transition will lead to many new opportunities, but it will also require new ways to define QoS and to seamlessly integrate human users into the wireless communication loop.

  5. (5)

    Performance analysis and optimization of 6G requires operating in 3D space and investigating a rich system that integrates drones, satellites, flying vehicles, and traditional wireless infrastructure.

  6. (6)

    The move toward 6G systems will not be yet another case of leveraging additional, high-frequency spectrum bands to deliver higher network capacity, as has been the case for decades. Instead, it will be driven by a diverse set of applications, technologies, and techniques (see Figs. 7.1 and 7.2) as well as a convergence of 3CLS factors.

  7. (7)

    AI will play an instrumental role in 6G systems. This role ranges from enabling SSNs to embedding collective network intelligence through new notions of edge AI. A new paradigm of joint learning-communication co-design is particularly necessary to widely deploy emerging edge AI algorithms such as federated learning.

In a nutshell, the next decade presents a rich set of opportunities for wireless research that span multi-disciplinary areas. Indeed, the road toward 6G systems will be an exciting era for wireless technologies in which we will see convergence of technologies ranging from AI to computing, control, and cyber-physical systems within the realm of 6G systems.