In this paper, we study optimal decentralized power allocation strategies for multiple access channels with additive white generalized Gaussian noise (AWGGN). The generalized Gaussian (GG) noise can model a wide variety of noise distributions. The Gaussian and Laplacian distributions are special cases of the GG distribution. First, we provide tight lower and upper bounds on the sum-rate of the generalized Gaussian MAC (GG-MAC) with per user peak transmit power and average transmit power constraints. The problem of optimizing transmit power allocation which maximizes this lower bound on the sum rate is formulated as a convex optimization problem. Next, we prove that the greedy power allocation policy is asymptotically optimal as the number of transmitters increases. Here, the greedy power allocation policy is to transmit with the peak power when the channel gain is a above a fixed threshold; this threshold is chosen such that the average transmit power constraint is met. The computation of the greedy power allocation policy does not require the channel state information (CSI) of other users and, hence, it allows for a completely decentralized power allocation strategy. More importantly, we rigorously show that the greedy power allocation policy also becomes optimal forfinite, but sufficiently large, number of users. An upper bound on this minimum number of users beyond which the greedy power allocation policy becomes optimal is also provided. From simulations, we see that this finite threshold number of users is within practical limits.

This paper presents an algorithm to jointly optimize the precoder matrix and intelligent reflective surfaces’ (IRS) phases for MIMO-IRS system with discrete alphabet input. First, we show that, as the number of IRS elements increases, the capacity of the unprecoded MIMO-IRS channel converges to that of interference-free, parallel AWGN channels. This channel hardening effect occurs even for small number of transmit and receive antennas. Using the channel hardening effect, we identify the suitable concave objective function to maximize for obtaining the optimal IRS phases. A joint optimization algorithm that alternatively optimizes the IRS phases and precoder matrix for finite alphabet MIMO-IRS is developed. The proposed optimization is concave and can attain the global maxima. Through simulations, we show that the proposed algorithm outperforms the state-of-the-art methods. Furthermore, we show that the channel hardening of the MIMO-IRS channel occurs with relatively small number of IRS elements.

In visible light communication (VLC) systems, the uplink and downlink channel gains depend on environmental geometry. This geometric nature induces a relationship between the uplink and downlink VLC channel gains. In this paper, the reciprocity relation between the uplink and downlink VLC channel gains are analyzed. It is shown that the reciprocity relationship can be accurately modeled as affine or polynomial. This reciprocity analysis can be a useful tool that enables several VLC system analysis and design. As an interesting application, using this channel reciprocity and model of the solar cell, the relationship between the downlink rate, harvested power and uplink rate are obtained for a solar cell based VLC-SLIPT system. This enables the formulation of the uplink-downlink rate region for a solar cell-based SLIPT system. The rate region gives the set of all simultaneously achievable rates in the uplink and downlink. The rate analysis, enabled by the reciprocity result, gives the trade-off between downlink and uplink rates that is essential in the system design of solar cell based SLIPT receivers.

We consider the problem of link adaptation with quality of service (QoS) constraints and present a novel solution using the theory of multi-arm bandits from reinforcement learning literature. In link adaptation, the transmit modulation and coding scheme (MCS) is adapted in accordance to the time varying communication channel. The goal is to learn an adaptation technique that keeps the packet errors below a certain threshold, as specified by the QoS constraint, while maximizing the transmission data rate. We formulate this as a multi-armed bandits problem with linear constraints, where the MCS are the arms of the bandit, the expected packet error is a linear constraint and the data rate is the reward. To solve this problem, we propose a novel Thompson sampling-based algorithm with Dirichlet prior. Furthermore, the packet error rate is a function of the channel quality, the structure of this function is exploited in the development of this link adaptation algorithm. We derive bounds on the optimality of the proposed algorithm and experimentally show that the performance of it is better than that of state-of-the-art methods. We also propose a neural network-based faster variant of this algorithm.

We study the problem of decentralized power allocation in a multi-access channel (MAC) with non-cooperative users, additive noise of arbitrary distribution and a generalized power constraint, i.e., the transmit power constraint is modeled by an upper bound on$\mathrm{E}[\phi(\vert S\vert)$, where$S$is the transmit signal and$\phi(.)$is some non-negative, increasing and bounded function. The generalized power constraint captures the notion of power for different wireless signals such as RF, optical, acoustic, etc. We show that the optimal power allocation policy takes a simple closed form when there a large number of users. Further, once the number of users in the MAC crosses a finite threshold, this optimal power allocation policy remains invariant irrespective of the actual number of users. We derive the above results under the condition that the entropy power of the MAC,$e^{2h(S)+c}$, is strictly convex, where$h(S)$is the maximum achievable entropy of the transmit signal and$c$is a finite constant corresponding to the entropy of the additive noise.

We study the problem of maximizing the secrecy capacity in multiple input multiple output (MIMO) communication systems in the presence of an eavesdropper using optimal power allocation. One of the key bottlenecks arise in identifying the appropriate spatial streams to securely transmit data and the corresponding transmit powers to maximize capacity. We employ the generalized singular value decomposition (GSVD) technique to identify these spatial streams and propose a power allocation scheme over these spatial streams. We also present an upper bound for the maximum achievable secrecy rate under certain conditions. We show that the proposed power allocation technique not only achieves better secrecy, but also converges to the upper bound faster than those in the literature.

Indoor optical wireless communication (IOWC) is a key technology complementing radio communication in 6G. Typically, the downlink of IOWC is implemented using visible light and uplink is implemented using infrared. Though the downlink of IOWC has been studied well, the uplink of multiple-access IOWC requires further research. In this paper, power allocation strategies in the uplink of IOWC with multiple non-cooperative users is studied. Optimizing power allocation in IOWC that maximize sum-rate with power constraints becomes a non-convex problem. First, this paper proposes an alternating maximization algorithm to compute near optimal power allocation policies for small number of users. Next, a phenomenon of invariance is observed when the number of users in the system is large. Under invariance, the optimal power allocation policy of a user becomes independent of other users, thereby facilitating fast and decentralized computation of power allocation strategies. From simulations, it was observed that the invariance can occur even with five to ten users. For the first time, a rigorous analysis of the invariance phenomenon is provided. An algorithm to compute the optimal power allocation at invariance is proposed; this algorithm is proven to achieve the global maxima in sum-rate without the channel knowledge of other users.

In this paper, we consider the problem of on-demand source coding for collaborative data dissemination in vehicular ad-hoc networks (VANETs). Specifically, we address the problem of index code construction for a big-data network. In this setup, users request a certain subset of data from a central server, while possessing some side information. The central server broadcasts a compressed message to meet all the users' demands. We consider a practical and generalized scenario where the side information and requested data packets of a user may be coded. Constructing index codes for such a setup is known to be an NP-hard problem. For this scenario, we propose a deep learning based approach to construct index codes through low-rank matrix completion. The proposed deep neural network performs matrix factorization to obtain a low-rank index code. We show that the proposed deep learning based approach to construct index codes is not only computationally feasible for networks with large data packets, but also provides better performance compared to the alternating minimization based approach for index code construction.

Time-indexed media-based modulation (TI-MBM) is an index modulation scheme where time slots in a transmission frame are indexed to convey additional information bits in media-based modulation (MBM). It was shown in the literature that, in frequency selective fading channels with inter-symbol interference, TI-MBM with cyclic-prefixed single-carrier (CPSC) scheme can achieve better transmission rates, bit error performance, and multipath diversity gain compared to conventional MBM. In this paper, we analyze the capacity of TI-MBM with CPSC, which has not been reported before. We show that, for a given channel matrix, selecting the index of the activated time slots and RF mirrors according to uniform distribution is suboptimal. We derive the probabilities with which time slots and RF mirrors in TI-MBM can be activated such that the achievable transmission rate is maximized. Further, we prove that the transmit symbols from a Gaussian mixture distribution, whose mixture weights are chosen to be the product of these probabilities, can achieve capacity.

Automatic modulation classification (AMC) is the problem of identifying the modulation type of a given radio frequency (RF) signal. This operation is one of the key steps in a cognitive radio based spectrum sharing communication network. It is known that the optimal classification algorithms for AMC are computationally intensive which renders real-time implementation almost impossible. In this paper, we propose a practical AMC algorithm that employs multiple stages of clustering to identify the modulation type of the received RF signal. Here, we consider the communication signals to be modulated using the most common digital modulation types: phase shift keying (PSK) or quadrature amplitude modulation (QAM). First, we present a novel algorithm that performs multiple stages of clustering to identify the clusters present in the received data and classifies it to one of the several possible modulation types. Second, we validate our proposed algorithm through practical implementation using software defined radios (SDR). Our results show that the proposed multistage clustering based AMC algorithm works well in practical conditions.

In this paper, we present an efficient methodology to design precoders for distributed detection of unknown high dimensional signals. We consider a wireless sensor network, where several distributed sensors collaborate to perform binary hypothesis testing based on observations of an unknown high dimensional signal corrupted by noise. The sensors collect data over both temporal and spatial domains. Due to network resource constraints, each sensor performs a linear compression (through precoding) of the observed high dimensional signal at each time instant and forwards the compressed signal to the fusion center (FC). The FC then employs the generalized likelihood ratio test (GLRT) to make a decision on the presence or absence of the signal. We propose online linear precoding/compression strategies for such sensors that collect data over spatio-temporal domain, so that the detection performance at the FC is maximized under certain network resource constraints. Through the measure of non-centrality parameter and receiver operating characteristics (ROC), we show that our proposed precoder design achieves very good detection performance.

In this paper, we design and evaluate a practical AMC system that can be readily deployed to provide robust performance in various real-time commercial scenarios. Thus, our main goal is to develop a robust AMC algorithm with low computational complexity for easy implementation and practical deployment. To this end, we utilize recently revitalized machine learning based approaches used for various classification purposes. In our proposed AMC architecture, we first propose various statistics that serve as features of the AMC signals; next, we design an artificial neural network (ANN) based classifier that performs AMC over a wide range of SNRs. We employ Nesterov accelerated adaptive moment (NADAM) estimation technique to improve the classification performance of our ANN. Further, to establish the practical feasibility of our proposed architecture, we implement it on a SDR testbed. The proposed ANN-based classifier is shown to outperforms the hybrid hierarchical AMC (HH-AMC) system and is flexible enough to easily expand the dictionary of modulation formats for other applications.

In this paper, we present an iterative best-response algorithm to compute the Nash equilibrium of a power allocation game in a multiple access channel (MAC), where each user greedily chooses a power allocation policy to maximize their average transmission rate. We consider a MAC where the fading channel gains are assumed to be stationary and ergodic processes, taking values from a finite set. The receiver decodes the message of a user by assuming the messages of the rest of the users as noise. The Shannon capacity of a user is considered to be the achievable rate of that user. A user transmits with a transmit power chosen from a finite set of power values, in a selfish manner, such that their average rate of transmission is maximized. We show that the Nash equilibrium of this game is unique, provided the number of users in the system is sufficiently large, but finite. Under this condition, we also show that the equilibrium policy does not change with more number of users coming into the system. We propose a simple greedy algorithm to compute the Nash equilibrium when the number of users is sufficiently large, but finite. The proposed algorithm does not depend upon the parameters of other users and hence, can be computed without any feedback or side-information from other users. We also present numerical results to illustrate the performance of the proposed algorithm.

Automatic modulation classification (AMC) with multiple sensors is a challenging problem when the channel conditions are unknown at the receiver. In this paper, using the Markov chain Monte Carlo (MCMC) approach, we develop a novel algorithm for AMC when the amplitude and phase of the channel gains are unknown. Using sampling techniques, we marginalize over the unknown channel parameters that follow a certain probability distribution. This improves the estimate of the a posteriori distribution of the modulation formats, thereby improving the overall classification performance. Further, to overcome the problem of local extrema traps encountered in sampling algorithms, we introduce the idea of adding artificial noise beyond a certain threshold of signal-to-noise (SNR). This improves the performance of the sampling based AMC algorithm in the high SNR regime. Simulation results and comparisons are provided to show the efficiency of the proposed algorithm over the most related works in the literature.