Approximate message passing (AMP) algorithms are devised under the Gaussianity assumption of the measurement noise vector. In this work, we relax this assumption within the vector AMP (VAMP) framework to arbitrary independent and identically distributed (i.i.d.) noise priors. We do so by rederiving the linear minimum mean square error (LMMSE) to accommodate both the noise and signal estimations within the message passing steps of VAMP. Numerical results demonstrate how our proposed algorithm handles non-Gaussian noise models as compared to VAMP. This extension to general noise priors enables the use of AMP algorithms in a wider range of engineering applications where non-Gaussian noise models are more appropriate.

This paper investigates distributed estimation problems with factorized structures over factor graphs. By building upon the recent progress in the approximate message passing (AMP) paradigm, this paper extends the vector AMP (VAMP) algorithm to the distributed scenario where multiple agents collaboratively estimate the same signal using different measurement channels. We do so by deriving the new collaborative linear minimum mean square error (LMMSE) messages within the estimation steps through message passing. The new algorithm — coined D-VAMP — allows distributed agents to be heterogeneous thereby handling a broader class of practical applications. Our numerical results demonstrate the trade-off between the reconstructed accuracy and the level of heterogeneity measured in terms of the number of correlated agents and signal-to-noise ratio.

Machine learning (ML)-aided wireless communication studies are initiating the investigation of the domain generalization capabilities of deep neural networks (DNNs) when applied to communication problems. They do so by adopting the out-of-distribution (OOD) performance evaluation by comparing it to the in-distribution (ID) performance as usually done within the ML community. In this paper, we demonstrate that such blind adoption can yield a misleading OOD performance analysis of DNNs unless wireless communication metrics are involved in the OOD evaluation. By analyzing the OOD performance of an end-to-end (E2E) ML communication system over additive white Gaussian noise (AWGN) channels in terms of bit error rate (BER), we show that smaller (resp. larger) BER gaps between ID and OOD performance do not necessarily translate into a high (resp. low) reconstruction accuracy. Our results suggest that the comparison between ID and OOD performances is not enough to judge whether the OOD performance is acceptable or not. The ID and OOD performances of E2E communication systems should instead be carried out based on wireless metrics.

With the proliferation of deep learning techniques for wireless communication, several works have adopted learning-based approaches to solve the channel estimation problem. While these methods are usually promoted for their computational efficiency at inference time, their use is restricted to specific stationary training settings in terms of communication system parameters, e.g., signal-to-noise ratio (SNR) and coherence time. Therefore, the performance of these learning-based solutions will degrade when the models are tested on different settings than the ones used for training. This motivates our work in which we investigate continual supervised learning (CL) to mitigate the shortcomings of the current approaches. In particular, we design a set of channel estimation tasks wherein we vary different parameters of the channel model. We focus on Gauss-Markov Rayleigh fading channel estimation to assess the impact of non-stationarity on performance in terms of the mean square error (MSE) criterion. We study a selection of state-of-the-art CL methods and we showcase empirically the importance of catastrophic forgetting in continuously evolving channel settings. Our results demonstrate that the CL algorithms can improve the interference performance in two channel estimation tasks governed by changes in the SNR level and coherence time.

We present a unified model for connected antenna arrays with a large number of tightly integrated (i.e., coupled) antennas in a compact space within the context of massive multiple-input multiple-output (MIMO) communication. We refer to this system as tightly-coupled massive MIMO. From an information-theoretic perspective, scaling the design of tightly-coupled massive MIMO systems in terms of the number of antennas, the operational bandwidth, and form factor was not addressed in prior art. We investigate this open research problem using a physically consistent modeling approach for far-field (FF) MIMO communication based on multi-port circuit theory. In doing so, we turn mutual coupling (MC) from a foe to a friend of MIMO systems design, thereby challenging a basic percept in antenna systems engineering that promotes MC mitigation/compensation. We show that tight MC widens the operational bandwidth of antenna arrays thereby unleashing a missing MIMO gain that we coin “bandwidth gain”. Furthermore, we derive analytically the asymptotically optimum spacing-to-antenna-size ratio by establishing a condition for tight coupling in the limit of large-size antenna arrays with quasi-continuous apertures. We also optimize the antenna array size while maximizing the achievable rate under fixed transmit power and inter-element spacing. Then, we study the impact of MC on the achievable rate of MIMO systems under line-of-sight (LoS) and Rayleigh fading channels. These results reveal new insights into the design of tightly-coupled massive antenna arrays as opposed to the widely-adopted “disconnected” designs that disregard MC by putting faith in the half-wavelength spacing rule.

We present a unified model for connected antenna arrays with a large number of tightly integrated (i.e., coupled) antennas in a compact space within the context of massive multiple-input multiple-output (MIMO) communication. We refer to this system as tightly-coupled massive MIMO. From an information-theoretic perspective, scaling the design of tightly-coupled massive MIMO systems in terms of the number of antennas, the operational bandwidth, and form factor was not addressed in prior art. We investigate this open research problem using a physically consistent modeling approach for far-field (FF) MIMO communication based on multi-port circuit theory. In doing so, we turn mutual coupling (MC) from a foe to a friend of MIMO systems design, thereby challenging a basic percept in antenna systems engineering that promotes MC mitigation/compensation. We show that tight MC widens the operational bandwidth of antenna arrays thereby unleashing a missing MIMO gain that we coin “bandwidth gain”. Furthermore, we derive analytically the asymptotically optimum spacing-to-antenna-size ratio by establishing a condition for tight coupling in the limit of large-size antenna arrays with quasi-continuous apertures. We also optimize the antenna array size while maximizing the achievable rate under fixed transmit power and inter-element spacing. Then, we study the impact of MC on the achievable rate of MIMO systems under line-of-sight (LoS) and Rayleigh fading channels. These results reveal new insights into the design of tightly-coupled massive antenna arrays as opposed to the widely-adopted “disconnected” designs that disregard MC by putting faith in the half-wavelength spacing rule.

This paper introduces a novel information-theoretic approach for studying the effects of mutual coupling (MC), between the transmit and receive antennas, on the overall performance of single-input-single-output (SISO) near-field communications (NFC). By incorporating the finite antenna size constraint using Chu’s theory and under the assumption of canonical-minimum scattering (CMS), we derive the MC between two radiating volumes of fixed sizes. Expressions for the self and mutual impedances are obtained by the use of the reciprocity theorem. Based on a circuit-theoretic two-port model for SISO radio communication systems, we first establish its input-output relationship where the noise depends on the self/mutual impedances of the antennas, unlike the conventional assumption of independent additive white Gaussian noise. We then characterise the achievable data rate for a given pair of transmit and receive antenna sizes, thereby providing an upper bound on the system performance under physical size constraints. Through the lens of these findings, we shed new light on the influence of MC on the information-theoretic limits of near-field communications using compact antennas.

This paper introduces a new algorithmic solution to the massive unsourced random access (mURA) problem. The proposed uncoupled compressed sensing (UCS)-based scheme relies on slotted transmissions and takes advantage of the inherent coupling provided by the users’ spatial signatures in the form of channel correlations across slots to completely eliminate the need for concatenated coding. As opposed to all existing methods, the proposed solution combines the steps of activity detection, channel estimation, and data decoding into a unified mURA framework. It capitalizes on the bilinear vector approximate message passing (Bi-VAMP) algorithm, tailored to fit the inherent constraints of mURA. Exhaustive computer simulations demonstrate that the proposed scheme outperforms recent coupled and uncoupled mURA schemes in massive connectivity/MIMO setup.

This paper introduces a novel information-theoretic approach for studying the effects of mutual coupling (MC), between the transmit and receive antennas, on the overall performance of single-input-single-output (SISO) near-field communications. By incorporating the finite antenna size constraint using Chu’s theory and under the assumption of canonical-minimum scattering, we derive the MC between two radiating volumes of fixed sizes. Expressions for the self and mutual impedances are obtained by the use of the reciprocity theorem. Based on a circuit-theoretic two-port model for SISO radio communication systems, we establish the achievable rate for a given pair of transmit and receive antenna sizes, thereby providing an upper bound on the system performance under physical size constraints. Through the lens of these findings, we shed new light on the influence of MC on the information-theoretic limits of near-field communications using compact antennas.

Using ideas from Chu and Bode/Fano theories, we characterize the maximum achievable rate over the single-input single-output wireless communication channels under a restriction on the antenna size at the receiver. By employing circuit-theoretic multiport models for radio communication systems, we derive the information-theoretic limits of compact antennas. We first describe an equivalent Chu’s antenna circuit under the physical realizability conditions of its reflection coefficient. Such a design allows us to subsequently compute the achievable rate for a given receive antenna size thereby providing a physical bound on the system performance that we compare to the standard size-unconstrained Shannon capacity. We also determine the effective signal-to-noise ratio (SNR) which strongly depends on the antenna size and experiences an apparent finite-size performance degradation where only a fraction of Shannon capacity can be achieved. We further determine the optimal signaling bandwidth which shows that impedance matching is essential in both narrowband and broadband scenarios. We also examine the achievable rate in presence of interference showing that the size constraint is immaterial in interference-limited scenarios. Finally, our numerical results of the derived achievable rate as function of the antenna size and the SNR reveal new insights for the physically consistent design of radio systems.

During the last few decades, wireless communication technologies and services have radically changed the way we live and interact at the personal, social, local and global levels. Such changes were mainly driven by the continuous emergence of innovative wireless communication services and products. These services and products represents a direct upshot of enduring research outcomes within the area. Nevertheless, the blessing of such innovation was accompanied by extremely high demands in forms of data traffic, per-user transmission rate, minimum transmission delay and in the number of wireless devices per unit area. Tackling these issues through cellular network densification was faced by many technical issues related to high interference levels, tedious user scheduling processes, and complicated network resource allocation algorithms. Trying to address these imperative technical issues in future wireless networks, this thesis develops several innovative enabling techniques for massive wireless multiple access. Specifically, we commence this work by introducing a new concept of partial spectrum overlapping among active users equipment (UEs). The proposed scheme represents a trade-off between fully orthogonal multiple access schemes (e.g. time division multiple access [TDMA], frequency division multiple access (FDMA) and orthogonal frequency division multiple access (OFDMA)) and that of non-orthogonal multiple access (NOMA). Second, we develop several innovative dynamic cell-free network architectures that support massive wireless connectivity through adaptive access points (APs)/base stations (BSs) coordination and/or cooperation. The proposed network models are then evaluated under different state-of-the-art enabling wireless techniques such as millimeter wave (mmWave) channel links and massive multiple-input multiple-output (mMIMO) systems. Furthermore, the performance of the proposed architectures is investigated through the derivation of several closed-form expressions of exact and/or asymptotic performance metrics (example, probability of outage, asymptotic outage, instantaneous rate and outage-capacity). Finally, for practical control and monitoring of the proposed access techniques and network models, we develop several low-complexity deep reinforcement learning (DRL)-based modeling frameworks that can efficiently learn the solution of several combinatorial optimization problems related to network partitioning (clustering) and uplink/downlink beamforming. This is achieved through innovative nested DRL designs that utilizes continuous and discrete deep neural networks (DNN) agents based on the nature of the problem. Several operating scenarios of the proposed techniques are evaluated through extensive Monte-Carlo simulations (Matlab and Python) with practical parameters and assumptions.

We propose a novel antenna clustering-based method for simultaneous wireless information and power transfer (SWIPT) in a multiple-input multiple-output (MIMO) full-duplex (FD) system. For a point-to-point communication set up, the proposed method enables a wireless device with multiple antennas to simultaneously transmit information and harvest energy using the same time-frequency resources. And the energy transmitting device with multiple antennas simultaneously receives information from the energy harvesting (EH) device. This is achieved by clustering the antennas into two MIMO subsystems: one for information transmission (IT) and another for EH. Furthermore, the self-interference (SI) signal at the EH device caused by the FD mode of operation is harvested by the device. For implementation-friendly antenna clustering and MIMO precoding, we propose two methods: (i) a sub-optimal method based on relaxation of objective function in a combinatorial optimization problem, and (ii) a hybrid deep reinforcement learning (DRL)-based method. For the proposed DRL solution, we design a hybrid discrete/continuous action agent that jointly clusters the MIMO antennas between EH and IT, and at the same time, find the best values for MIMO precoding matrices at both devices. This is achieved by using two interacting agent learning subsystems, namely, deep double Q-learning (DDQN), for antenna clustering and deep deterministic policy gradient (DDPG), for MIMO precoding. The effect of imperfect CSI is also studied and investigated. Finally, we study the performances of the two implementation methods and compare them with the conventional time switching-based simultaneous wireless information and power transfer (SWIPT) technique. Our findings show that the proposed MIMO clustering-based SWIPT method gives a significant improvement in spectral efficiency compared to the time switching-based SWIPT method. In particular, the DRL-based method provides the highest spectral efficiency. Besides, the numerical results show that, for the considered system set up, the number of antennas in each device should exceed three to mitigate self-interference to an acceptable level.

Current algorithms for deep learning probably cannot run in the brain because they rely on weight transport, where forward-path neurons transmit their synaptic weights to a feedback path, in a way that is likely impossible biologically. An algorithm called feedback alignment achieves deep learning without weight transport by using random feedback weights, but it performs poorly on hard visual-recognition tasks. Here we describe two mechanisms — a neural circuit called a weight mirror and a modification of an algorithm proposed by Kolen and Pollack in 1994 — both of which let the feedback path learn appropriate synaptic weights quickly and accurately even in large networks, without weight transport or complex wiring. Tested on the ImageNet visual-recognition task, these mechanisms outperform both feedback alignment and the newer sign-symmetry method, and nearly match backprop, the standard algorithm of deep learning, which uses weight transport.