Subject: Cutting-Edge Advances in Wireless Communications: RIS Optimization, Semantic Communications, and Beyond

Hi Elman,

This newsletter covers a collection of recent papers exploring cutting-edge topics in wireless communications, from channel estimation and hardware impairments to integrated sensing and communication (ISAC) and semantic communications. A key focus area is the optimization of reconfigurable intelligent surfaces (RIS), with several works addressing channel estimation, mutual coupling effects, and the potential of beyond diagonal RIS (BD-RIS). The mitigation of hardware impairments also features prominently, with research investigating the impact of non-ideal hardware on large intelligent surfaces (LIS), over-the-air (OTA) pre-distortion techniques, and RIS-assisted space shift keying (SSK). Beyond these core themes, the collection delves into emerging applications such as multi-connectivity in satellite-terrestrial integrated networks (STINs), RIS-assisted high-altitude platform (HAP) networks, and UAV applications in communication systems. Finally, several contributions explore novel signal processing techniques, including compressive spectrum sensing, semantic communication frameworks based on foundation models, and advanced graph neural networks for relationship modeling in large-scale networks.

General Overview

This collection of papers explores several cutting-edge topics in wireless communications, including channel estimation, hardware impairments, integrated sensing and communication (ISAC), and semantic communications. Several works focus on enhancing the efficiency and performance of reconfigurable intelligent surfaces (RIS). Demir et al. (2024) Demir et al. (2024) introduce a novel reduced-subspace least squares (RS-LS) channel estimator for RIS-aided systems, leveraging spatial correlation and array geometry to reduce pilot overhead. Their proposed estimator, optimized for mean square error (MSE) minimization under electromagnetic interference, outperforms conventional LS estimation. Concurrently, Nerini et al. (2024) Nerini et al. (2024) delve into the impact of mutual coupling on RIS, deriving closed-form solutions for channel gain maximization in fully- and tree-connected architectures. They demonstrate that, counterintuitively, mutual coupling can enhance channel gain under certain conditions when properly accounted for in the optimization process. Li and Clerckx (2024) Li & Clerckx (2024) further explore beyond diagonal RIS (BD-RIS), specifically examining the benefits of non-reciprocal BD-RIS in full-duplex systems. Their multiport network model reveals the potential for significant performance gains due to the asymmetric behavior of non-reciprocal surfaces.

Another prominent theme is the mitigation of hardware impairments in advanced communication systems. Sheikhi et al. (2024a) Sheikhi et al. (2024a) analyze the performance of large intelligent surfaces (LIS) with non-ideal hardware, considering the impact of distortion on scalability and proposing antenna/panel selection schemes to optimize the complexity-performance trade-off. Furthermore, Sheikhi et al. (2024b) Sheikhi et al. (2024b) investigate over-the-air (OTA) digital pre-distortion (DPD) and reciprocity calibration in massive MIMO, offering a closed-form calibration method that leverages mutual coupling measurements to compensate for both non-linearity and non-reciprocity. Basu et al. (2024) Basu et al. (2024) study the impact of non-ideal transceivers on RIS-assisted space shift keying (SSK), deriving closed-form expressions for error probabilities under Nakagami-m fading and greedy detection.

Several papers address emerging applications and system architectures. Li and Shang (2024) Li & Shang (2024) provide a comprehensive overview of multi-connectivity in satellite-terrestrial integrated networks (STINs), discussing architectures, challenges, and applications. Tanash et al. (2024) Tanash et al. (2024) analyze the performance of RIS-assisted high-altitude platform (HAP) networks using stochastic geometry, demonstrating the potential for improved connectivity. Two papers explore the application of UAVs in communication systems. Sun et al. (2024) Sun et al. (2024) propose a joint optimization approach for service delay minimization in aerial MEC-assisted industrial cyber-physical systems. Li et al. (2024a) Li et al. (2024a) investigate secure resource allocation and UAV trajectory optimization using model predictive control, while Li et al. (2024b) Li et al. (2024b) focus on joint antenna positioning and beamforming optimization in movable antenna-enabled full-duplex ISAC networks.

Finally, several contributions explore novel signal processing techniques and applications. Yang et al. (2024) Yang et al. (2024) introduce a non-cooperative compressive spectrum sensing method using 1-bit ADCs and multicoset sampling. Xu et al. (2024) Xu et al. (2024) propose a generative semantic communication framework based on foundation models, analyzing the relationship between transmission reliability and perceptual quality. Giraldo et al. (2024) Giraldo et al. (2024) introduce Sparse Sobolev Graph Neural Networks (S2-GNNs) for efficient higher-order relationship modeling in large-scale networks. Shen et al. (2024) Shen et al. (2024) develop a dynamic-attention-based EEG state transition model for emotion recognition. Qin et al. (2024) Qin et al. (2024) investigate optimal Pauli measurement allocation for low-rank quantum state tomography. These diverse contributions collectively advance the state-of-the-art in various aspects of wireless communications and signal processing.

Paper Highlights

Beyond Diagonal RIS with Mutual Coupling: Global Optimality Achieved, Performance Enhanced

Global Optimal Closed-Form Solutions for Intelligent Surfaces With Mutual Coupling: Is Mutual Coupling Detrimental or Beneficial? by Matteo Nerini, Hongyu Li, Bruno Clerckx https://arxiv.org/abs/2411.04949

Caption: This figure shows the channel gain achieved by Beyond Diagonal RIS (BD-RIS) and Diagonal RIS (D-RIS) with and without mutual coupling awareness, for varying inter-element distances. The results demonstrate that mutual coupling, when considered in the optimization, enhances the channel gain, particularly for BD-RIS, and that smaller inter-element distances lead to higher gains.

Reconfigurable Intelligent Surfaces (RISs) hold immense promise for dynamically shaping wireless propagation, but conventional Diagonal RIS (D-RIS) architectures have limitations. Beyond Diagonal RIS (BD-RIS), with interconnections between elements, offers greater flexibility. This paper addresses a critical gap in the field by considering the often-neglected impact of mutual coupling between RIS elements. It derives global optimal closed-form solutions for BD-RIS with mutual coupling, specifically for fully- and tree-connected architectures, maximizing channel gain. Furthermore, it establishes the maximum achievable channel gain in the presence of mutual coupling and its scaling law, providing crucial insights into fundamental performance limits.

The study utilizes multiport network theory to model the RIS-aided system, incorporating mutual coupling effects. For fully-connected RIS, the Z-parameter representation is used to formulate the channel gain maximization problem, subject to constraints of reciprocity and losslessness. A key innovation is the "diagonalization" of the mutual coupling matrix Z_II through a transformation based on its real part, enabling the application of existing solutions for RIS optimization without mutual coupling. A similar approach is applied to tree-connected RIS using the Y-parameter representation and diagonalization of the admittance matrix Y_I. Remarkably, both fully- and tree-connected RISs achieve the same channel gain upper bound when optimized using the proposed solutions. This upper bound is given by: |h|² = (|S_RT| + ||Ŝ_RI||₂ ||Ŝ_IT||₂)², where S_RT, Ŝ_RI, and Ŝ_IT are S-parameters derived from the corresponding Z-parameters.

The paper derives scaling laws for average channel gain under Rayleigh fading, both with and without mutual coupling. With mutual coupling, the scaling law depends on the coupling matrix; without it, it depends on self-impedance and element count. Crucially, the study analytically proves that mutual coupling, when properly considered in optimization, increases the average channel gain. This counterintuitive result challenges the traditional view of mutual coupling as detrimental. Numerical simulations validate the theoretical findings, confirming the global optimality of the proposed methods. Smaller inter-element distances (stronger coupling) lead to higher gains. The study underscores the importance of mutual coupling-aware optimization, showing that neglecting it can cause significant performance degradation (up to 5 dB). BD-RIS outperforms D-RIS by approximately 2 dB, consistent with previous findings (without coupling), and this advantage slightly increases with stronger coupling, highlighting BD-RIS's flexibility in exploiting these effects.

Generative Semantic Communication Gets a Foundation Model Upgrade

Generative Semantic Communications with Foundation Models: Perception-Error Analysis and Semantic-Aware Power Allocation by Chunmei Xu, Mahdi Boloursaz Mashhadi, Yi Ma, Rahim Tafazolli, Jiangzhou Wang https://arxiv.org/abs/2411.04575

Caption: This diagram illustrates a generative semantic communication (SemCom) framework using pre-trained foundation models. Semantic features are extracted, converted to bit sequences, and transmitted through a channel. The receiver uses a generative foundation model to reconstruct the original signal from the received semantic data streams, discarding erroneous data in coded transmissions.

This paper proposes a groundbreaking generative semantic communication (SemCom) framework based on pre-trained foundation models. Instead of training dedicated end-to-end models, this framework leverages the power of pre-trained models like BERT and GPT for semantic encoding and generative models like DALL·E and Sora for decoding. This approach allows for knowledge sharing, eliminates resource-intensive training, and enables generalization across diverse data and channel models. The framework extracts semantic features (e.g., textual prompts, edge maps for images), converts them into bit sequences for transmission, and employs two decoding schemes: uncoded forward-with-error (handling errors in uncoded transmissions) and coded discard-with-error (discarding erroneous data in coded transmissions to maintain synthesis quality).

A key contribution is the theoretical analysis of how transmission reliability affects perceptual quality. Using rate-distortion-perception theory, the authors establish a non-decreasing relationship between perception value and transmission errors, formalized as P(R) = min_PXX** δ(X, X) subject to I(X; X) < R, where δ(·) represents perceptual distance. This leads to the definition of "semantic values" (L**i) quantifying each semantic data stream's importance in signal reconstruction, where L**i = 1 - P**i, and P**i is the perception value when using only the ith stream. This provides a nuanced understanding of how different semantic features contribute to overall meaning, enabling efficient resource allocation.

To minimize power while maintaining semantic performance, the paper introduces semantic-aware power allocation methods. Two methods are proposed: a semantic-aware proportional method (with a closed-form solution achieved by decoupling the perception constraint) and a semantic-aware bisection method (for scenarios with two semantic features). These methods prioritize power allocation to data streams with higher semantic value. Simulations using the Kodak dataset validate the theoretical analysis and demonstrate the methods' effectiveness. Perception-error functions confirm the non-decreasing relationship between errors and perceptual degradation. The semantic-aware bisection method significantly outperforms conventional approaches, achieving up to 10% and 90% power savings in uncoded and coded cases, respectively. Furthermore, the results suggest that some semantic data streams can be omitted entirely under certain conditions, further optimizing resource allocation and potentially motivating research into adaptive semantic coding rates. This work paves the way for more efficient and robust SemCom systems, focusing on conveying meaning rather than simply transmitting bits.

Conclusion

This newsletter highlights significant advancements in wireless communications. The work by Nerini et al. on RIS optimization with mutual coupling challenges conventional wisdom, demonstrating that mutual coupling, often seen as a detriment, can be exploited to enhance performance. Their closed-form solutions offer practical tools for optimizing BD-RIS, showcasing its potential to surpass D-RIS. The research by Xu et al. on semantic communications introduces a paradigm shift, leveraging foundation models to build more efficient and robust SemCom systems. Their focus on the relationship between transmission reliability and perceptual quality, coupled with semantic-aware power allocation methods, opens new avenues for optimizing resource utilization in meaning-centric communication. Together, these works represent a significant step towards the future of wireless communication, where intelligent surfaces and semantic understanding play crucial roles.