6G:What is it and where do we stand?

Do you wonder what will come after 5G?

In the Summer 2024, I published an article*** about 6G in the French language “Annales des Mines” journal, going through a few underlying concepts that seem to reach consensus to date. Below I attach an English-language, slightly edited, version of this 6G primer. Note that this brief article is focusing on technical concepts. It will be accompanied soon by another piece on more (and no less important) social aspects of these technologies.

*** https://annales.org/enjeux-numeriques/2024/resumes/septembre/07-en-resum-FR-AN-septembre-2024.html

INTRODUCTION

The arrival of 6G, planned for 2030 or soon after, will be marked by a new wireless standard under the framework of 3GPP, with meanwhile several evolution stages expected to appear (“standard releases” in the 3GPP terminology).

So what can we already say about 6G in 2024? In reality, the debate is still too lively to be able to announce for sure at this stage what the key use cases of 6G will be or the design bricks that will be chosen to support them. After all, the technology is still in the R&D phase. The objective of this article is therefore twofold. First of all, take a quick look at certain expectations that seem to have reached a consensus in the tech community. Secondly, to shed light on some scientific concepts which are likely to play a role in the future wireless generation. Finally, we briefly raise a few societal challenges some of which will however be the main thread of another short article.

PROMISES AND EXPECTATIONS OF 6G

New modes of use: virtual realities and connected robotics

By adding industrial use cases and the notion of Internet of Things (IoT), alongside the more typical use as internet access “pipe”, 5G was the first wireless communication standard having to take into consideration the balance of spectral resources serving both  the needs of human and non-human users, thanks to the use of so-called “network slicing” techniques. 6G will certainly face similar challenges, but with increased requirements on the expected quality of service as the transition to machine-centric services is expected to accelerate.

On the side of human users, use cases are evolving towards so-called immersive communication by which context data (real or synthetic 3D video environment) are transported in the radio channel alongside usual data (voice), or with holographic communications. Furthermore, the increased use of digital virtual worlds (such as metaverse) for educational or recreational purposes through virtual or augmented reality wireless headsets will place new demands for very high-speed communication [1].

On the machine side, the share of generated data traffic not directly under the control of a human is expected to rise fast. The future will see an increasing use of connected sensors of all kinds which will add up to the low-speed connected sensors devices served by 4G and 5G (road traffic sensors, pollution, weather, agriculture, everyday household appliances, etc.). To low-end devices will be added machine customers who are much more demanding in terms of quality of the radio link, particularly in the context of robotics. By 2030, factory/personal robots and autonomous vehicles, that rely on connectivity to the network to make good and fast response decisions, are expected to become prevalent. For example, unmanned drones using the wireless network for guidance algorithms, and autonomous car fusing of local data from on-board sensors (radar, lidar) with more global information from the cellular network to determine their routes. In industrial settings, factory production will depend on “cobots” (collaborative robots) communicating with each other at very high speeds to resolve complex mechanical tasks. All these machine use cases pose considerable unsolved challenges to future wireless communication networks [2].

High-performance networks

To allow for the above use cases, future wireless communication standard will have to raise the level of quality in order to exceed the performance of current 5G networks [3]. The need to transport large video streams generated by virtual or augmented reality or vision sensors on robots (drones or cars) leads to significantly improve the quality of the radio link. Quality is here measured by the maximum achievable throughput for which the objective of 1 TB/sec is given (under conditions where the entire spectral resource is assigned to a single machine) but also by the average throughput measured under more realistic conditions. Capacity is measured by the number of links with a target performance that it is possible to maintain simultaneously per m2 or km2. Note that it is difficult to give a safe target figure other than for peak rates at this stage since it will depend on the density of use and network deployment. However, by integrating modes of use from the Internet of Things alongside traditional human terminals (phones, tablets), the average figure of 1 radio user per m2 in coverage areas can be reached. Of course, peak performance is not the only issue here. In fact, global connectivity over a maximum surface area of ​​the planet is also a key criterion, which will also require evolving the standard. Concepts such as integrating so-called non-terrestrial technologies (NTN) like satellite communications and high altitude platforms together with terrestrial cellular network are being experimented already in 5G. A full integration of these technologies will likely occur in 6G.

Coverage and speed are key but other metrics are gaining importance in the 5G-6G transition. For use cases linked to cobotics such as human-robot and robot-robot collaboration), metrics measuring latency and accuracy of positioning, already present for 5G design, will be considerably reinforced in the context of time/position-sensitve networking.

Global latency measures the response time of the network from the moment a command is transmitted at the control point to the time the action can be executed at the machine on the receiving end of the network. Very low latency is particularly necessary for the proper functioning of cooperating robots, or to a lesser extent for some of the tactile-sensitive internet use cases, where for instance haptic sensors are connected to radio nodes.

Dedicated communication protocols capable of delivering a latency of the order of 1 msec or even lower (excluding machine control internal latency) have been investigated in the context of 5G, but are still slow to materialize in market-ready products. Such protocols will have to be implemented by 6G.

In addition to latency, connected robotics pose geolocation challenges that cannot be addressed by satellite systems alone (GNSS technologies including the famed GPS) because of their lack of precision in non-military scenarios and their inability to deal with situations where propagation satellite is obstructed (for example in dense urban or indoor environment). In the context of the factory of the future, it is essential for industrial collaborative robots carrying out precision mechanical tasks to be able to self-position themselves in order to ensure maximum operation safety. We then aim for a precision of the order of 1 cm or less. Similar requirements are found for cases where humans and robots interact directly in a factory or business area. For these reasons, very high precision geolocation is one of the performance criteria receiving particular attention and which will require new methodologies.

THE 6G TECHNOLOGICAL BUILDINGS BLOCKS

Several technical approaches are being studied to meet the above performance expectations. Here is a quick, though non-exhaustive, overview of the main concepts.

Frequencies

6G will firstly rely on the foundations of 5G, and this includes the combined use of several frequency ranges. The zone below 6 GHz will remain key for its ability to facilitate propagation in urban environments and penetrate structures for indoor coverage. The deployment of millimeter waves (30 to 300 GHz) has proven difficult in 5G because it is very sensitive to obstacles. Hence it will continue for limited use cases, alongside the introduction of THz (bands from 300 GHz to 3 THz). ) which can give excellent performance over very short distances (around a meter) [4]. The Upper Mid Band zone (7 to 30 GHz) is another option which could provide a compromise between throughput and coverage. The notion of dynamic spectrum sharing between operators, under study in 5G, also remains a promising area for improving spectral efficiency.

The spatialization of information

This generic term designates the processes at the physical layer by which wireless communication no longer takes place in broadcast mode. Classical antennas tend to flood air space with radio waves in an indiscriminate manner direction-wise. Instead, spatial communication directs energy along beam directions. The basic technology known as multiple-input multiple-output (MIMO) dates back to the 1990s (and even earlier in the military) and has established itself as one of the fundamental building blocks of 5G in the form of Massive MIMO, by which antennas can be made up of a large number (several dozen) of compact radio frequency (RF) elements interacting together. 6G will extend these concepts to make communication ever more spatialized and, thus, more energy efficient. Being intelligently combined with each other, the antenna elements will allow the formation of highly discriminating beams in the spatial domain, that is to say that the radio energy generated at the transmitter will end up being concentrated only at the points in space where the targeted receivers are located. Another variant of this is the notion of reflective intelligent surfaces (RIS) [5], conceptualized around ten years ago. RIS generalizes the antenna into a synthetic and integrated surface whose properties in terms of reflection of electromagnetic waves are controlled by algorithms. RIS can then play the role of intelligent relay antennas allowing radio waves to be conducted towards a target and circumventing propagation obstacles.

Artificial intelligence

Faced with the growing complexity of wireless communication networks, AI is called upon to play a crucial role to enable self-optimization, for instance for the management of spectral resources [6] or more recently as a tool for generating protocols and code via LLMs. By directly feeding on observed data, AI competes with, or preferably supports, algorithms which have have been derived on the basis of models based on decades of system-level and physics expertise.  There are two fundamental levels of AI integration in 6G: integration at the core network level and integration at the edge level (edge-AI), that is to say the devices at the edge of the network to which we might offload part of the functions and calculations in order to make decisions with a minimum of latency.

Machine learning in the core network (or the cloud) already plays an important role for network optimization, resource allocation based on service classes, maintenance, and detection of intrusion or malicious behavior. More specific to future 5G or 6G deployments, the use of edge AI on individual devices (smartphones, sensors, connected robots, cellular antennas, etc.) will depend on the evolution of their capacity to integrate computing platforms from tiny-ML platforms to full scale GPUs. Edge-AI is key as it will enable distributed, local or collaborative decision-making, in a manner close to the signal or sensors. Thus, the devices will take an active role, yielding some powerful collective intelligence to the network [7].  

Sensing

This notion redefines the function of radio waves, whose role until now has been confined to the transport of information, to enable effective probing of the environment. For instance, by analyzing (with AI tools) the way in which radio waves are reflected by RF obstacles, it is possible to recover precise mapping information and information on the presence of objects around such as walls, humans, and vehicles. This information is key for the safe operation of autonomous vehicles for instance. Because the transmitted waveform is typically optimized either for data transport or for detection (radar), 6G introduces the novel concept of hybrid waveforms allowing an optimal compromise between the two goals, under the name ISAC (integrated sensing and communication) [8]. Note that in some cases, waves can be analyzed at selected points in space using autonomous robots, allowing even more precise mapping and geolocation [15]. Needless to say, sensing and positioning information constitute highly sensitive digital data and must be handled while adhering to the strictest form of government policies for data privacy.

Semantic Communication

This idea revisits the fundamental precepts of Shannon's information theory, according to which encoding functions are optimized for the reliable transport of binary data without regard for the nature of the information which is carried [9]. In contrast, the semantic framework highlights the purpose of the data, its degree of relevance over time, and the type of action carried out with it at the receiver, in order to encode the information more effectively.  For example, the communication of a video sequence for detection purposes will be based on the (AI-aided) extraction of the key elements of the scene. Such key elements may be the only ones having to be transmitted. With this approach, semantic communication can discriminate between more and less relevant bits of information, thus lightening the load on the network and gain capacity or energy efficiency [10].

 SECURITY AND ECOLOGICAL IMPACT

The success of 6G will not be measured solely by its ability to deliver better data pipes. As we have already observed during the deployment of 5G, some issues not related to performance have become essential for the acceptance of the technology by the general public. We will take two examples here, firstly the question of the security level associated with these systems and secondly the question of their ecological footprint.

Security aspects cover quite diverse issues. It covers for instance the resilience and redundancy of the equipment, to be reinforced compared to 5G. Another essential aspect is the integrity of the transported data, involving the implementation of robust encryption and authentication protocols, including new methods from quantum and from AI. Such methods must protect against data breaches and unauthorized takeovers which could endanger the operation of cyber-physical systems essential to national infrastructures (factories, nuclear power plants, airports, etc.). Note that these threats will be broader in scope than during the 5G era, due to the multiplication of devices accessing the network [11] as well as the growing number of entry points and vendors in the context of open RAN (ORAN) architectures.

The third objective is that of enforcing individual security, which concerns data confidentiality and the protection of private life. In a heavily digitalized world where everyone is required to share ever-increasing quantities of personal data on the network, 6G, just like wired networks, must put in place advanced mechanisms for the secure management of personal data.

The other major societal issue of 6G concerns its environmental impact, which again covers diverse factors ranging from energy efficiency in terms of joules per bit transported, to the overall footprint. The issue also includes questions of durability and recyclability of the various equipments. With a strong emphasis placed in 5G on energy efficiency from the design phase, wireless systems have already shown an improvement in efficiency of up to 10x. Such gains are obtained by a combination of traffic-based deactivation protocols, more efficient modulations, and network densification which reduces the average distance from transmit to receive. Despite all this, it is known that the overall energy footprint of 5G remains greater than that of previous generations. Why is this?

The paradox is explained by the multiplication of devices, increasing density of use, and the generalization of very high bandwidth applications such as VHD video on smartphones. For this reason, improving energy efficiency remains a fundamental objective of 6G design, which may rely on the spatialization methods mentioned above, the use of AI-native design to optimize resources, and the implementation of semantic mechanisms which will significantly reduce the quantity of raw data to be transported in the pipes.

Drawing lessons from 5G however, it is reasonable to think that an overall reduction in the energy footprint might not be achievable solely through technical improvement. Hence, 6G may have to be accompanied by new incentive mechanisms aimed at rationalizing the use of these future communication tools. Additionally, our communication to the general public should highlight much more the positive impact that these systems can have in helping other industries consume less resources [12] [13] [14].

CONCLUSIONS

To a great extent, 6G is likely to appear as an evolution of 5G capable of materializing some of the most ambitious use cases initially envisioned under 5G, but that could not yet be achieved. Still, 6G will also target novel and challenging objectives in terms of performance and new application scenarios such as immersive communication and high-precision connected robotics. Novel technological bricks are being studied, such as the spatialization of transmission, integration of AI into core network functions or embedded in edge equipments, semantic communication, among other advances. More broadly, human success around 6G will be helped by taking a broader view of the challenges moving beyond the traditionnal performance-centric approach, including taking security, the protection of individual digital rights, and the environmental impact into (very) serious account.

REFERENCES

[1] TANG F., CHEN X., ZHAO M. & KATO N. (2023), “The roadmap of communication and networking in 6G for the Metaverse”, IEEE Wireless Communications, vol. 30, n°4, pp. 72-81.

[2] NGUYEN D. C. et al. (2022), “6G Internet of Things: a comprehensive survey”, IEEE Internet of Things Journal, vol. 9, n°1, pp. 359-383.

[3] WANG C.-X. et al. (2023), “On the road to 6G: visions, requirements, key technologies, and testbeds”, IEEE Communications Surveys & Tutorials, vol. 25, n°2, pp. 905-974.

[4] TRIPATHI S., SABU N.V., GUPTA A.K. & DHILLON H.S. (2021), "Millimeter-wave and Terahertz spectrum for 6G wireless", In WU Y. et al., 6G Mobile Wireless Networks, Computer Communications and Networks, Springer.

[5] LIU Y. et al. (2021), “Reconfigurable intelligent surfaces: principles and opportunities”, IEEE Communications Surveys & Tutorials, vol. 23, n°, pp. 1546-1577.

[6] GÜNDÜZ D., DE KERRET P., SIDIROPOULOS N. D., GESBERT D., MURTHY C. R. & VAN DER SCHAAR M. (2019), “Machine Learning in the Air”, IEEE Journal on Selected Areas in Communications, vol. 37, n°10, pp. 2184-2199.

[7] XU W., YANG Z., NG D. W. K., LEVORATO M., ELDAR Y. C. & DEBBAH M. (2023), “Edge learning for B5G networks with distributed signal processing: semantic communication, edge computing, and wireless sensing”, IEEE Journal of Selected Topics in Signal Processing, vol. 17, n°1, pp. 9-39.

[8] PIN TAN D. K. et al. (2021), “Integrated sensing and communication in 6G: motivations, use cases, requirements, challenges and future directions”, 2021 1st IEEE international online symposium on Joint Communications & Sensing (JC&S), Dresden, Germany.

[9] LUO X., CHEN H.-H. & GUO Q. (2022), “Semantic communications: overview, open issues, and future research directions”, IEEE Wireless Communications, vol. 29, n°1, pp. 210-219.

[10] STRINATI E. C., BARBAROSSA S. (2021), “6G networks: beyond Shannon towards semantic and goal-oriented communications”, Computer Networks, vol. 190.

[11] PORAMBAGE P., GÜR G., M. OSORIO D. P., LIYANAGE M., GURTOV A. & YLIANTTILA M. (2021), “The roadmap to 6G security and privacy”, IEEE Open Journal of the Communications Society, vol. 2, pp. 1094-1122.

[12] IMOIZE A. L., ADEDEJI O., TANDIYA N. & SHETTY S. (2021), “6G enabled smart infrastructure for sustainable society: opportunities, challenges, and research roadmap”, Sensors, 21(5), article 1709.

[13] Numéro « Transitions énergétique et numérique », Annales des mines, série Responsabilité & Environnement, avril 2023,https://annales.org/re/2023/re_110_avril_2023.html

[14] FAURE A. & ROUSSILHE G. (2024), Note d’analyse de France Stratégie, « Quelle contribution du numérique à la décarbonation ? », juillet, https://www.strategie.gouv.fr/publications/contribution-numerique-decarb...

[15] D. Gesbert, O. Esrafilian, J. Chen, R. Gangula, U. Mitra, "UAV-aided RF Mapping for Sensing and Connectivity in Wireless Networks", in IEEE  Communications Magazine, May 2022.