This letter presents the first work introducing a deep learning (DL) framework for channel estimation in large intelligent surface (LIS) assisted massive MIMO (multiple-input multiple-output) systems. A twin convolutional neural network (CNN) architecture is designed and it is fed with the received pilot signals to estimate both direct and cascaded channels. In a multi-user scenario, each user has access to the CNN to estimate its own channel. The performance of the proposed DL approach is evaluated and compared with state-of-the-art DL-based techniques and its superior performance is demonstrated.

Intelligent reflecting surface (IRS), consisting of low-cost passive elements, is a promising technology for improving the spectral and energy efficiency of the fifth-generation (5G) and beyond networks. It is also noteworthy that an IRS can shape the reflected signal propagation. Most works in IRS-assisted systems have ignored the impact of the inevitable residual hardware impairments (HWIs) at both the transceiver hardware and the IRS while any relevant works have addressed only simple scenarios, e.g., with single-antenna network nodes and/or without taking the randomness of phase noise at the IRS into account. In this work, we aim at filling up this gap by considering a general IRS-assisted multi-user (MU) multiple-input single-output (MISO) system with imperfect channel state information (CSI) and correlated Rayleigh fading. In parallel, we present a general computationally efficient methodology for IRS reflecting beamforming (RB) optimization. Specifically, we introduce an advantageous channel estimation (CE) method for such systems accounting for the HWIs. Moreover, we derive the uplink achievable spectral efficiency (SE) with maximal-ratio combining (MRC) receiver, displaying three significant advantages being: 1) its closed-form expression, 2) its dependence only on large-scale statistics, and 3) its low training overhead. Notably, by exploiting the first two benefits, we achieve to perform optimization with respect to the RB that can take place only per several coherence intervals, and thus, reduces significantly the computational cost compared to other methods based on instantaneous CSI which require frequent phase optimization. Among the insightful observations, we highlight that the unrealistic assumption of uncorrelated Rayleigh fading does not allow optimization of the SE, which makes the application of an IRS ineffective. Also, in the case that the phase drifts, describing the distortion of the phases in the RBM, are uniformly distributed, the presence of an IRS provides no advantage. The analytical results outperform previous works and are verified by Monte-Carlo (MC) simulations.

In multi-user millimeter wave (mmWave) multiple-input-multiple-output (MIMO) systems, hybrid precoding is a crucial task to lower the complexity and cost while achieving a sufficient sum-rate. Previous works on hybrid precoding were usually based on optimization or greedy approaches. These methods either provide higher complexity or have sub-optimum performance. Moreover, the performance of these methods mostly relies on the quality of the channel data. In this work, we propose a deep learning (DL) framework to improve the performance and provide less computation time as compared to the conventional techniques. In fact, we design a convolutional neural network for MIMO (CNN-MIMO) that accepts as input an imperfect channel matrix and gives the analog precoder and combiners at the output. The procedure includes two main stages. First, we develop an exhaustive search algorithm to select the analog precoder and combiners from a predefined codebook maximizing the achievable sum-rate. Then, the selected precoder and combiners are used as output labels in the training stage of CNN-MIMO where the input-output pairs are obtained. We evaluate the performance of the proposed method through numerous and extensive simulations and show that the proposed DL framework outperforms the conventional techniques. Overall, CNN-MIMO provides a robust hybrid precoding scheme in the presence of imperfections regarding the channel matrix. On top of this, the proposed approach exhibits less computation time with comparison to the optimization and codebook based approaches.

Recent works have identified massive multiple-input multiple-output (MIMO) as a key technology for achieving substantial gains in spectral and energy efficiency. Additionally, the turn to low-cost transceivers, being prone to hardware impairments, is the most effective and attractive way for cost-efficient applications concerning massive MIMO systems. In this context, the impact of channel aging, which severely affects performance, is investigated herein by considering a generalized model. Specifically, we show that both Doppler shift due to the users' relative movement, as well as phase noise due to noisy local oscillators, contribute to channel aging. To this end, we first propose a joint model, encompassing both effects, to investigate the performance of a massive MIMO system based on the inevitable time-varying nature of realistic mobile communications. Then, we derive the deterministic equivalents for the signal-to-noise-and-interference ratios (SINRs) with maximum ratio transmission (MRT) and regularized zero-forcing (RZF) precoding. Our analysis not only demonstrates a performance comparison between MRT and RZF under these conditions but, most importantly, also reveals interesting properties regarding the effects of user mobility and phase noise. In particular, the large antenna limit behavior profoundly depends on both effects, but the burden due to user mobility is much more detrimental than phase noise even for moderate user velocities (≈ 30 km/h), whereas the negative impact of phase noise is noteworthy at lower mobility conditions. Moreover, massive MIMO systems are favorable even in general channel aging conditions. Nevertheless, we demonstrate that the transmit power of each user to maintain a certain quality of service can be scaled down, at most, by 1√M (M is the number of base station antennas), which indicates that the joint effects of phase noise and user mobility do not degrade the power scaling law but only the achievable sum-rate.

Cell-free (CF) massive multiple-input-multiple-output (MIMO) has emerged as an alternative deployment for conventional cellular massive MIMO networks. As revealed by its name, this topology considers no cells, while a large number of multi-antenna access points (APs) simultaneously serves a smaller number of users over the same time/frequency resources through time-division duplex (TDD) operation. Prior works relied on the strong assumption (quite idealized) that the APs are uniformly distributed, and actually, this randomness was considered during the simulation and not in the analysis. However, in practice, ongoing and future networks become denser and increasingly irregular. Having this in mind, we consider that the APs locations are modeled by means of a Poisson Point Process (PPP) which is a more reaslistic model for the spatial randomness. In particular, by virtue of stochastic geometry tools, we derive both the downlink coverage probability and achievable rate. Notably, this is the only work providing the coverage probability and shedding light into this aspect of CF massive MIMO systems. Focusing on the extraction of interesting insights, we consider small-cells (SCs) as a benchmark for comparison. Among the findings, CF massive MIMO systems achieve both higher coverage and rate with comparison to SCs due to properties of favorable propagation and channel hardening. Especially, we showed that increasing the APs density results in lower coverage and increasing the number of users decreases the achievable rate in both architectures but CF massive MIMO systems take advantage of the aforementioned properties and outperform SCs. In general, the performance gap between CF massive MIMO systems and SCs is enhanced with increasing the APs density. Another interesting observation concerns that a higher path-loss exponent decreases the rate while the users closer to the APs affect more the performance in terms of the rate.

Rate-splitting (RS) has recently been shown to provide significant performance benefits in various multiuser transmission scenarios. In parallel, the huge degrees-of-freedom provided by the appealing massive multiple-input multiple-output (MIMO) necessitate the employment of inexpensive hardware, being more prone to hardware imperfections, in order to be a cost-efficient technology. Hence, in this paper, we focus on a realistic massive multiple-input single-output broadcast channel hampered by the inevitable hardware impairments. We consider a general experimentally validated model of hardware impairments, accounting for the presence of multiplicative distortion due to phase noise, additive distortion noise and thermal noise amplification. Under both scenarios with perfect and imperfect channel state information at the transmitter (CSIT), we analyze the potential robustness of RS to each separate hardware imperfection. We analytically assess the sum-rate degradation due to hardware imperfections. Interestingly, in the case of imperfect CSIT, we demonstrate that RS is a robust strategy for multiuser MIMO in the presence of phase and amplified thermal noise, since its sum-rate does not saturate at high signal-to-noise ratio (SNR), contrary to conventional techniques. On the other hand, the additive impairments always lead to a sum-rate saturation at high SNR, even after the application of RS. However, RS still enhances the performance. Furthermore, as the number of users increases, the gains provided by RS decrease not only in ideal conditions, but in practical conditions with residual transceiver hardware impairments as well. Notably, although a deterministic equivalent analysis is employed, the analytical and simulation results coincide even for finite system dimensions. As a consequence, the applicability of these results also holds for current “small scale” multiantenna systems.