Bharath Shamasundar

In reconfigurable intelligent surfaces (RIS), the reflective elements increase the dimensionality of the overall channel model and, correspondingly, the training requirements. This paper analyzes how channel training and spectral efficiency guide the choice of the operational dimensionality of RIS. We derive an inner bound on the training-based capacity as a function of pilot power, data transmission power, and RIS array dimensions. We study the outcomes and implications of data/pilot power optimization in RIS-assisted systems according to the achievable rate metric. Further, this work sheds light on the tradeoff between the superior array gains of larger RIS on the one hand, and the respective training requirements on the other hand, when optimizing the end-to-end capacity. Beyond a certain critical array size, further increase of array size is not beneficial for spectral efficiency due to the training requirement. This critical size is calculated, and its dependence on the signal-to-noise ratio (SNR) and coherence interval of the channel is clarified. The benefits and drawbacks of different training schemes are analyzed in this context, and demonstrated by simulations.

Index modulation is a MIMO technology where some transmit antennas are idle during each transmission interval, but the receiver requires knowledge of all link gains at all times. Even though index modulation is particularly sensitive to the cost of estimating the channel state information (CSI), the impact of CSI cost and imperfections on the capacity of index modulation has not been adequately characterized until now. As a result, the marginal cost/benefit of each additional antenna, and the optimal antenna alphabet have remained unclear. This study computes the spectral efficiency of index modulation subject to training and determines optimal antenna alphabets. Our approach involves a comprehensive examination of the influence of pilot power and degrees of freedom on the achievable rate of index modulation. The results include 2.5dB improvement over the best previously known bound for 4 × 2 spatial modulation at 6b/s/Hz. Additionally, we determine the conditions under which single-antenna transmission is superior to spatial modulation, and vice versa. Moreover, this training-based spectral efficiency analysis is extended to the multiuser uplink, identifying the number of users that can be accommodated while maximizing the uplink sum-rate under spatial modulation.

Reconfigurable intelligent surfaces (RIS) are passive controllable arrays of small reflectors that direct electromagnetic energy towards or away from the target nodes, thereby allowing better management of signals and interference in a wireless network. The RIS has the potential for signficantly improving the performance of wireless networks. Unfortunately, RIS also multiplies the number of Channel State Information (CSI) coefficients between the transmitter and receiver, which magnifies the challenges in estimating and communicating the channel state information. Furthermore, the simplicity and cost-effectiveness of the passive RIS also implies that the incoming links are not locally estimated at the RIS, and fresh pilots are not inserted into outgoing RIS links. This introduces new challenges for training and estimation of channel state information. The rapid growth of the literature on CSI acquisition in RIS-aided systems has been accompanied by variations in the underlying assumptions, models, and notation, which can obscure the similarities and differences of various techniques, and their relative merits. This paper presents a comprehensive exposition of principles and approaches in RIS channel estimation. The basic ideas underlying each class of techniques are reduced to their simplest form under a unified model and notation, and various approaches within each class are discussed. Several open problems in this area are identified and highlighted.

Index modulation represents the transmitted information in two parts: by selecting a subset of available transmission dimensions (antennas, sub-carriers, or time-slots) whose selection index carries information, and by modulation symbols transmitted in the selected dimensions. Index modulation is motivated by reducing the transmitter hardware complexity and has attracted significant research attention in the past decade. In practice, knowing the spectral efficiency or capacity is essential for setting the parameters of modulation and coding for index modulation, but approximations and bounds thus far have not been accurate enough for that purpose. We calculate close lower and upper bounds for the spectral efficiency of index modulation. Our lower and upper bounds meet at high-SNR when the number of receive antennas is greater than or equal to the number of transmit antennas, thus the high-SNR capacity of index modulation in these cases has been fully characterized. A catalog of results is provided for spatial modulation, generalized spatial modulation, time and frequency index modulation. Extensive simulations illustrate the usefulness and accuracy of our results. For example, for spatial modulation with 4 × 2 antennas at 8 bits/s/Hz, our results are 2dB tighter than the best available bounds in the literature.

In this letter, we consider the problem of signal detection in generalized spatial modulation (GSM) using deep neural networks (DNN). We propose a novel modularized DNN architecture that uses small sub-DNNs to detect the active antennas and complex modulation symbols, instead of using a single large DNN to jointly detect the active antennas and modulation symbols. The main idea is that using small sub-DNNs instead of a single large DNN reduces the required size of the NN and hence requires learning lesser number of parameters. Under the assumption of i.i.d Gaussian noise, the proposed DNN detector achieves a performance very close to that of the maximum likelihood detector. We also analyze the performance of the proposed detector under two practical conditions: i) correlated noise across receive antennas and ii) noise distribution deviating from the standard Gaussian model. The proposed DNN-based detector learns the deviations from the standard model and achieves superior performance compared to that of the conventional maximum likelihood detector.

Efficient constellation design is important for improving performance in communication systems. The problem of multidimensional constellation design has been studied extensively in the literature in the context of multidimensional coded modulation and space-time coded MIMO systems. Such constellations are formally called as lattice codes, where a finite set of points from a certain high dimensional lattice is chosen based on some criteria. In this paper, we consider the problem of constellation/signal set design for media-based modulation (MBM), a recent MIMO channel modulation scheme with promising theoretical and practical benefits. Constellation design for MBM is fundamentally different from those for multidimensional coded modulation and conventional MIMO systems mainly because of the inherent sparse structure of the MBM signal vectors. Specifically, we need a structured sparse lattice code with good distance properties. In this work, we show that using an (N,K) non-binary block code in conjunction with the lattice based multilevel squaring construction, it is possible to systematically construct a signal set for MBM with certain guaranteed minimum distance. The MBM signal set obtained using the proposed construction is shown to achieve significantly improved bit error performance compared to conventional MBM signal set. In particular, the proposed signal set is found to achieve higher diversity slopes in the low-to-moderate SNR regime.

Media-based modulation (MBM) is an attractive channel modulation scheme with rate, performance, and hardware advantages. In this paper, we are concerned with two important aspects of MBM. The first one is on the capacity of MBM and the other is on low-complexity detection of high-rate MBM signals using structured spare recovery techniques. We derive closed-form expression for the capacity of generalized MBM (GMBM). The main idea in the capacity analysis is to recognize that MBM uses two alphabets to convey information, namely, the source alphabet (e.g., QAM/PSK) and the channel alphabet (fade coefficients). This observation allows us to show that the capacity is achieved when the mutual information is maximized over the source-channel product alphabet. We then propose a greedy structured sparse recovery algorithm for the detection of GMBM signal vectors. The proposed algorithm is a two-stage algorithm in which support recovery is done in greedy manner first, followed by data detection in the nonzero positions. Simulation results show that the proposed algorithm achieves good performance at low complexity even when the system is highly underdetermined.

Mirror activation pattern (MAP) in media-based modulation (MBM) refers to the ON/OFF status of radio-frequency mirrors, which create different near field geometries for different ON/OFF combinations. In this letter, we introduce a novel coding scheme for MBM in which indices are assigned to MAPs, such that these indices are used as labels for the elements in a Galois field. The MAP indices are then coded across time, and the block transmission is carried out. The resulting signal set has good distance distribution, which leads to good bit error performance. Structured sparse matrix sketching-based signal detection algorithms for the proposed scheme are presented.

Media-based modulation (MBM) is an attractive modulation scheme which can achieve high data rates using multiple radio frequency (RF) mirrors (parasitic elements) and fewer transmit antennas/RF chains. In this paper, we consider MBM for the uplink in a massive MIMO system consisting of tens of users and tens to hundreds of base station (BS) receive antennas. Each user employs MBM with one transmit antenna and multiple RF mirrors placed near it. The ON/OFF status of the RF mirrors conveys information bits in addition to the bits conveyed through conventional modulation symbols. We show that the multiuser MBM (MU-MBM) system achieves very good bit error performance compared to other conventional multiuser MIMO systems. We propose two types of low-complexity detection algorithms for massive MU-MBM systems. The first type is based on compressive sensing (CS) that exploits the inherent sparse nature of the MBM transmit vectors. The second type is is based on message passing that performs both detection and channel estimation at low computational complexity. To do this, it exploits the 'channel hardening' phenomenon that occurs in large MIMO channels. It is shown that the proposed algorithms are not only computationally less complex but also provide very good bit error performance in massive MU-MBM systems.