Twelve Scientific Challenges for 6G: Rethinking the Foundations of Communications Theory

In a LinkedIn post earlier in the year, Prof. Merouane Debbah pointed out that he was asked by students about the important topics to work on in the field of 6G. As a result he along with colleagues summarised many of the ideas in the paper 'Twelve Scientific Challenges for 6G: Rethinking the Foundations of Communications Theory', available here.

Quoting from the paper:

In this article, we aim to dive deep into the fundamentals of wireless communications systems and identify the roots of the highlighted challenges anticipated to be experienced in future wireless generations. In particular, we cover the following scientific challenges needed to revisit the foundations of communication theory:

1) Electromagnetic information theory: In Section III, we discuss the limitations imposed by the laws of electromagnetism on the Shannon information capacity and the number of degrees of freedom. We further discuss several cutting-edge technologies that are envisioned to realize the full potential of 6G. 

2) Non-linear signal processing: in Section IV, we explain the source of nonlinearities in communications, the limitations of the linear model, and the need of designing signal processing algorithms beyond the linear assumption. We also discuss possible approaches in modeling the non-linearities and designing non-linear transceivers, either based on non-linear signal processing tools such as non-linear Fourier transform or an end-to-end approach using machine learning algorithms. We discuss challenges and open questions for each approach.

3) Multi-agent learning systems: in Section V, we overview the drawbacks of conventional centralized learning systems, and we point out the motivation behind the need for distributed/federated learning architecture. We explore the distributed AI theory, with emphasis on multiagent systems, its advantages, and shortcomings of such a learning scheme on the individual node as well as the system-wide levels. We also overview the concept of emergent communications in multi-agent systems and discuss the main challenges of this emerging topic.

4) Super-resolution theory: in Section VI, we well-define the task of super-resolution in reconstructing the fine details of an observation from coarse-scale information. We review the classical methods of addressing this problem and show their limitations. We then focus on the frequency response of the multipath channel and provide the limitations of state-of-the-art solutions in estimating the different channel parameters, especially when ignoring the noise and non-linearity effects. We then explain some challenges and research opportunities such as considering realistic assumptions and fusing different measurements at different bands and from different sensors.

5) Thermodynamics of computation and communication: the thermodynamics of communication and computation theory may help the development of novel algorithms for energy-efficient 6G networks. In Section VII, besides investigating the energy required for information processing, we also describe how it is essential to understand the fundamental energy limits of information processing and how they are connected to the fundamental limits of communication performance. For example, what is the optimal trade-off between communication performance and network energy consumption?

6) Signals for time-varying systems theory: in Section VIII, we explain how signals should be designed in time-varying systems, in particular doubly selective wireless channels. We review the proposed waveforms in the literature within the general framework of spreading waveforms that spread the symbol energy in the domain of selectivity of the channel. We then explain the limitations and challenges of spreading waveforms in time varying systems and provide future research directions.

7) Semantic communication theory: Section IX digs deeply into Shannon’s information theory, and rethinks the classical communication theory, in which the meaning and the semantic aspect of transmitted messages are ignored. We further identify the limitation of such classical definitions in the era of native-AI, where machines enjoy a considerable level of intelligence. Also, we thoroughly discuss the fundamentals of semantic communications and its role in future 6G networks, redefining the concept of information structure.

8) Signals for integrated sensing and communication: in Section X, we first explain the benefits of the integration of sensing and communications, and then we focus on waveform design challenges for different joint communications and radar scenarios e.g. monostatic active radar, bistatic active and passive radar. Then we describe other challenges beyond the waveform design, such as performance indicators, channel modeling, and representation and sharing of sensing data.

9) Large-scale communication theory: in Section XI, we shed light on the challenges anticipated to be experienced in current and future wireless networks, which are attributed to the modeling and optimization of largescale networks. We further explore the limitations of current computer simulation and artificial neural network based approaches. Subsequently, we discuss three major tools that have manifested themselves as promising, generalized yet resilient methods for modeling, analyzing, and optimizing large-scale dynamic wireless networks. This includes random matrix theory, decentralized stochastic optimization, tensor algebra, and low rank tensor decomposition.

10) Non-equilibrium information theory: information theory deals with fundamental limits of communication (or information processing in general), and it allows to development of strategies to achieve these limits. These fundamental limits may then be used to assess the performance of actual communication systems, and the strategies to construct frameworks for the system design. Hidden in the mathematical proofs is the assumption of infinite-length sequences, such that asymptotic techniques can be applied. Real-world applications, however, operate on finite lengths. In this finite-length regime, classic information-theoretic results still apply as ultimate limits but they are usually not achievable, and thus bounds may have little value. In recent years, finite-length information has been developed, aiming at fundamental information-theoretic limits (and strategies) subject to finite-length constraints. In Section XII, we review the main concepts related to Non-equilibrium Information Theory.

11) Combining queuing and information theory: in Section XIII, we shed light on the impact of information theory on networking. In particular, we review several topics related to communication networks with queuing and information-theoretic aspects, including multiaccess protocols and effective bandwidth of bursty traffic. We also discuss how layering can be formulated based on different optimization techniques including the optimization decomposition for systematic network design.

12) Non-coherent communication theory: in Section XIV, We revisit the challenge of channel state information acquisition within the context of large-scale networks, and emphasize the shortcomings of coherent and blind based approaches in achieving the needed reliability, latency, and spectral and energy efficiency requirements of such networks. Also, we will reveal the role of Grassmannian modulation and non-coherent tensor modulation as potential candidates for fulfilling the massive scale network vision of 6G.

Embedded below is the video of a recent keynote by Dr. Merouane Debbah on "10 Scientific Challenges for 6G" delivered at IEEE LCN 2022.

Let us know what you believe is the most important from amongst them.

Related Posts