MU-MIMO Research Articles

Defending Adversarial Attacks on Deep Learning-Based Power Allocation in Massive MIMO Using Denoising Autoencoders

Recent work has advocated for the use of deep learning to perform power allocation in the downlink of massive MIMO (maMIMO) networks. Yet, such deep learning models are vulnerable to adversarial attacks. In the context of maMIMO power allocation, adversarial attacks refer to the injection of subtle perturbations into the deep learning model’s input, during inference (i.e., the adversarial perturbation is injected into inputs during deployment after the model has been trained) that are specifically crafted to force the trained regression model to output an infeasible power allocation solution. In this work, we develop an autoencoder-based mitigation technique, which allows deep learning-based power allocation models to operate in the presence of adversaries without requiring retraining. Specifically, we develop a denoising autoencoder (DAE), which learns a mapping between potentially perturbed data and its corresponding unperturbed input. We test our defense across multiple attacks and in multiple threat models and demonstrate its ability to (i) mitigate the effects of adversarial attacks on power allocation networks using two common precoding schemes, (ii) outperform previously proposed benchmarks for mitigating regression-based adversarial attacks on maMIMO networks, (iii) retain accurate performance in the absence of an attack, and (iv) operate with low computational overhead.

Zero-Crossing Precoding Techniques for Channels With 1-Bit Temporal Oversampling ADCs

A promising approach to reducing the energy consumption is to consider coarse quantization at the receiver. In this study, we investigate novel precoding techniques in space and time for bandlimited multiuser MIMO downlink channels with 1-bit quantization and oversampling at the receiver, considering zero-crossing modulation. The proposed time-instance zero-crossing modulation conveys the information into the time-instances of zero-crossings. Two design criteria for time-instance zero-crossing modulation are investigated, namely, the minimum distance to the decision threshold and the mean-square error between the received and the desired signal. The maximization of the minimum distance to the decision threshold can be formulated as a quadratically constraint quadratic program. As an alternative, an equivalent problem can be formulated based on power minimization, which reduces computational complexity. Departing from the conventional mean-square error based technique, a more sophisticated algorithm is developed, which implies active constellation extension in order to improve the performance at high SNR. The extended problem is solved with two approaches, namely by formulating the problem as a second-order cone program and by considering an alternating optimization algorithm. Numerical results show that the proposed time-instance zero-crossing precoding methods significantly improve the bit error rate compared to the state-of-the-art methods.

Beamforming Design for In-Band Full-Duplex Multi-Cell Multi-User MIMO Networks With Global and Local CSI

This paper investigates beamforming techniques for in-band full-duplex multi-cell multi-user (IBFD-MCMU) wireless networks considering hardware impairments (HWIs), channel uncertainty, and limited channel state information (CSI). For the global CSI scenario, we first enhance zero-forcing (ZF), maximum ratio transmission and combining (MRTC), and minimum mean-squared error (MMSE) beamforming to be compatible with multi-antenna users and IBFD base stations. Then, we investigate beamforming schemes for the local CSI assumption where only intra-cell channel knowledge is fully available at the base stations, and inter-cell channels are known statistically due to the limited training resources. With these limitations, two MMSE-based methods are proposed, which regard unknown CSI as random instances (i.e., eMMSE-RI) and noise (i.e., eMMSE-N). We explore the self-interference cancellation (SIC) capability of beamformers and evaluate the performance of these methods in 3GPP scenarios. Numerical results reveal that the enhanced MMSE beamforming can achieve the desired IBFD with low analog SIC depth, inspiring a low-cost IBFD transceiver design. The practical imperfections (e.g., transceiver HWIs, channel uncertainty, limited CSI) decrease the achievable sum rate but do not reduce the IBFD gain. With CSI limitations, eMMSE-RI outperforms eMMSE-N for microcells, where the probability of the presence of LOS is high, whereas eMMSE-N performs better when cell size is enlarged.

Deep Reinforcement Learning for Energy Efficiency Maximization in Cache-Enabled Cell-Free Massive MIMO Networks: Single- and Multi-Agent Approaches

Cell-free massive multiple-input multiple-output (CF-mMIMO) is an emerging beyond fifth-generation (5 G) technology that improves energy efficiency (EE) and removes cell structure limitation by using multiple access points (APs). This study investigates the EE maximization problem. Forming proper cooperation clusters is crucial when optimizing EE, and it is often done by selecting AP–user pairs with good channel quality or aligning AP cache contents with user requests. However, the result can be suboptimal if we determine the clusters based solely on either aspect. This motivates our joint design of user association and content caching. Without knowing the user content preferences in advance, two deep reinforcement learning (DRL) approaches, i.e., single-agent reinforcement learning (SARL) and multi-agent reinforcement learning (MARL), are proposed for different scenarios. The SARL approach operates in a centralized manner which has lower computational requirements on edge devices. The MARL approach requires more computation resources at the edge devices but enables parallel computing to reduce the computation time and therefore scales better than the SARL approach. The numerical analysis shows that the proposed approaches outperformed benchmark algorithms in terms of network EE in a small network. In a large network, the MARL yielded the best EE performance and its computation time was reduced significantly by parallel computing.

Secure Linear Precoding in Overloaded MU-MIMO Wireless Networks

We investigate signaling and linear precoding designs for secure downlink communications in a multiuser MIMO (MU-MIMO) network. The network is tapped by an external eavesdropper (Eve) who is equipped with a large number of antennas. In addition to the transmission of several (unicast) information signals to downlink users (Bobs), the transmitter (Alice) generates friendly jamming (FJ), aiming to prevent Eve from decoding Alice’s signals. To mitigate both multiuser interference (MUI) and FJ on Bobs, MU-MIMO systems often rely on zero-forcing (ZF) precoders. A condition for the application of such precoders is that the network must be <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">underloaded</i> , i.e., Alice has more antennas than all Bobs combined. For <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">overloaded</i> scenarios, ZF-based precoding cannot guarantee nullification of both MUI and FJ at legitimate receivers. Accordingly, we propose a combined signaling-and-precoding scheme that can also be applied to an overloaded MU-MIMO network. In an underloaded scenario, our technique is shown to impose a more stringent antenna requirement on Eve than a classical ZF-based precoding, i.e., Eve must utilize more antennas if she is to successfully eavesdrop on Alice’s transmissions. Although our scheme is comparable to antenna selection (AS) approaches in terms of Eve’s antenna requirement, it incurs much less delay and complexity by avoiding frequent on/off switching of the RF chains at Bobs. For underloaded scenarios, we provide security analysis that establishes the conditions under which our scheme is superior to the ZF scheme. Using computer simulations, we validate the advantages of our design and contrast it with ZF and AS schemes in terms of the symbol error rate, the EVM, and the achievable rate in both underloaded and overloaded scenarios.

Low-Complexity Pareto-Optimal 3D Beamforming for the Full-Dimensional Multi-User Massive MIMO Downlink

Full-dimensional (FD) multi-user massive multiple-input multiple output (m-MIMO) systems employ large two-dimensional (2D) rectangular antenna arrays to control both the azimuth and elevation angles of signal transmission. We introduce the sum of two outer products of the azimuth and elevation beamforming vectors having moderate dimensions as a new class of FD beamforming. We show that this low-complexity class is capable of outperforming 2D beamforming relying on the single outer product of the azimuth and elevation beamforming vectors. It is also capable of performing close to its FD counterpart of massive dimensions in terms of either the users' minimum rate or their geometric mean rate (GM-rate), or sum rate (SR). Furthermore, we also show that even FD beamforming may be outperformed by our outer product-based improper Gaussian signaling solution. Explicitly, our design is based on low-complexity algorithms relying on convex problems of moderate dimensions for max-min rate optimization or on closed-form expressions for GM-rate and SR maximization.

Spherical Wavefronts Improve MU-MIMO Spectral Efficiency When Using Electrically Large Arrays

Modern MIMO communication systems are almost exclusively designed assuming locally plane wavefronts over antenna arrays. This is known as the far-field approximation and is soundly justified at sub-6-GHz frequencies at most relevant transmission ranges. However, when higher frequencies and shorter transmission ranges are used, the wave curvature over the array is no longer negligible, and arrays operate in the so-called radiative near-field region. This letter aims to show that the classical far-field approximation may significantly underestimate the achievable spectral efficiency of multi-user MIMO communications operating in the 30-GHz bands and above, even at ranges beyond the Fraunhofer distance. For planar arrays with typical sizes, we show that computing combining schemes based on the far-field model significantly reduces channel gain and spatial multiplexing capability. When the radiative near-field model is used, interference rejection schemes, such as the optimal minimum mean-square-error combiner, appear to be very promising, when combined with electrically large arrays, to meet the requirements of next-generation networks.

Unsupervised Learning-Based Joint Power Control and Fronthaul Capacity Allocation in Cell-Free Massive MIMO With Hardware Impairments

A deep learning-based resource allocation algorithm that maximizes the sum rate of a limited fronthaul cell-free massive MIMO network with transceiver hardware impairments is proposed in this paper. The sum rate maximization problem with user power constraints and total fronthaul capacity constraints for channel state information (CSI) and data transmission is considered. The deep neural network (DNN) PowerNet is proposed to learn solutions to the joint power control and capacity allocation problem in a low-complex, flexible, and scalable way. An unsupervised learning approach is used which eliminates the need of knowing the optimal resource allocation vectors during model training, hence having a simpler and more flexible model training stage. Numerical simulations show that PowerNet achieves close sum rate performance compared to the existing optimization-based approach, with a significantly lower time complexity which does not exponentially scale with the number of users and access points (APs) in the network. Furthermore, the addition of the online learning stage resulted in a better sum rate than the optimization-based method.

Learning to Beamform in Heterogeneous Massive MIMO Networks

Learning to Beamform in Heterogeneous Massive MIMO Networks

Low complexity nonlinear detection for multiuser hybrid MIMO systems and performance analysis

This paper proposes an analytical framework and a system of distortion-aware receiver structures for single-carrier uplink hybrid beamforming-based multiple-input-multiple-output (MIMO) channels. Proposed single-user processing units cooperate at the base station (BS) in an iterative manner for multi-user symbol detection under hardware impairments. Firstly, in the proposed system, an iterative nonlinear interference cancellation (nIC) based on feedforward neural network (NN) modeling is developed to eliminate the nonlinear distortion effects of power amplifiers (PAs). Furthermore, depending on the fading pattern of the instantaneous channel, it is analytically shown that a large gain may be applied to the distortion signal during the equalization stage, which yields distortion amplification causing a significant decrease in the signal-to-distortion-plus-noise ratio (SDNR). A reduced complexity distortion aware detector, based on a bank of fractionally delayed iterative block decision feedback equalization (IB-DFE) receivers, is introduced to adjust the combiner weights for fractionally delayed IB-DFE outputs so that distortion power is minimized before detection. Consequently, multi-user detection capability is enhanced considerably. Secondly, an asymptotic SDNR analysis is performed. The theoretical analysis, verified via numerical evaluations, clearly shows that distortion amplification limits the achievable SDNR at the BS, and the proposed reduced complexity multi-user receiver architecture improves the BER performance by effectively mitigating this effect.

Selective interference alignment and neutralization in coordinated multipoint using multiuser MIMO

SummaryIn wireless communication, the concept of coordinated multipoint (CoMP) transmission is more attractive, and it transpired the notion based on interference management techniques. Interference alignment (IA) and interference neutralization (IN) methods can substantially align and neutralize the interference signals. The existing work in the CoMP transmission using multiuser multi‐input and multi‐output (MU‐MIMO) had long‐held problems such as specific limit of intercell (ICI) and cochannel interference (CCI) cancelation that contain low performance in cell‐edge users. The proposed framework of the transmission signal in selective interference alignment and neutralization (SIAN) CoMP MU‐MIMO system transmits multiple data streams in multipath by using downlink coordination between base stations and receiver side. This work contributes the individual perspectives to implement the IA and IN to align and cancel the interfered signals at the receiver side. Once the perspectives mentioned above are executed, zero‐forcing (ZF) and rechanneling filter are applied on the receiver side to eliminate residual ICI. In addition, spectral efficiency significantly improves the achievable data rate and enhances the performance of cell‐edge users and reduces the CCI at the receiver side. Furthermore, the antenna configuration signals are decoded to get the exact version of interference‐free signals with null path loss of signal transmission. The effectiveness of the proposed framework analyzes and verifies the numerical evaluation of the achievable degree of freedom. Finally, the simulation demonstrates the comparison of the proposed CoMP scheme with MIMO ZF and non‐CoMP schemes, which significantly improve the performance of spectral efficiency for cell‐edge users.

Underlay Spectrum Sharing in Massive MIMO Systems

We consider an underlay spectrum sharing scheme, where a multi-user massive MIMO network (referred to as the primary network (PN)) allows another multi-user massive MIMO network (called the secondary network (SN)) to use the spectrum, owned by the PN, to communicate with the SN’s users. For such a scheme, we devise joint power allocation and beamforming schemes at the SN for both conventional time-division duplexing (C-TDD) and reverse time-division duplexing (R-TDD) protocols. This is accomplished by maximizing the weighted uplink and downlink sum-rates of the SN, while protecting the PN from the SN’s interference. Computationally efficient solutions are presented for both protocols. Our simulation results show that our solution for the R-TDD protocol mostly outperforms the solution to the C-TDD mode in terms of the SN’s total sum-rate.

Opportunistic Schedule Selection for Multiuser MIMO FSO Communications: A Security- Reliability Trade-Off Perspective

The security and reliability of free space optical (FSO) communications and the trade-off between the two are the most critical characteristics to emphasize in FSO systems, especially as optical wireless communications continue to evolve. In this work, we investigate the impact of opportunistic transmit aperture selection (TAS) on the security-reliability trade-off (SRT) of the multiuser multiple-input multiple-output (MU-MIMO) free space optical system. We consider a general scenario in which the FSO system consists of a transmitter side with <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">K</i> users and a single receiver with <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">N</i> apertures communicating over generalized Malaga-M turbulence channels. We derive the TAS combined signal-to-noise ratio (SNR) distributions for the FSO MIMO channel between the transmitting users and the receiver. A statistical analysis of the TAS scheme is performed, where the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">k</i> th best user with the highest SNR is selected. Based on this analysis, the closed-form expressions for the outage probability, average bit error rate, intercept probability, and SRT are derived. Then, the diversity advantages provided by the channel variations considering the atmospheric turbulence and generalized nonzero boresight pointing errors are further discussed. The results show that the considered TAS significantly improves the security, reliability, and SRT performance and has the potential to solve existing FSO communication problems. Monte Carlo simulations are used to verify the correctness of the numerical results.

Bit-Metric Decoding Rate in Multi-User MIMO Systems: Applications

This is the second part of a two-part paper that focuses on link-adaptation (LA) and physical layer (PHY) abstraction for multi-user MIMO (MU-MIMO) systems with non-linear receivers. The first part proposes a new metric, called bit-metric decoding rate (BMDR) for a detector, as being the equivalent of post-equalization signal-to-interference-noise ratio (SINR) for non-linear receivers. Since this BMDR does not have a closed form expression, a machine-learning based approach to estimate it effectively is presented. In this part, the concepts developed in the first part are utilized to develop novel algorithms for LA, dynamic detector selection from a list of available detectors, and PHY abstraction in MU-MIMO systems with arbitrary receivers. Extensive simulation results that substantiate the efficacy of the proposed algorithms are presented.

Hybrid Beamforming With Sub-Connected Structure for MmWave Massive Multi-User MIMO Relay Systems

In this paper, we investigate the hybrid beamforming design with sub-connected structure for millimeter-wave (mmWave) massive multi-user multiple-input multiple-output (MU-MIMO) amplify-and-forward (AF) relay systems. Considering the constant-modulus and block-diagonal constraints, an iterative algorithm is proposed to sequentially design the analog beamforming of the base station, relay, and users. Next, the relay baseband combiner is designed by transforming the highly complicated non-convex mutual information maximization problem into an easily tractable weighted minimum mean squared error one. To mitigate the inter-user interference, a successive serial interference cancellation based and a piecewise successive approximation based methods are developed for the uniform sparse distribution and dense distribution scenarios, respectively. Simulation results demonstrate that the two proposed hybrid beamforming schemes can achieve good performance in terms of sum-rate and energy efficiency, and outperform benchmarks significantly. In addition, the two proposed schemes are shown to be robust to imperfect channel state information (CSI), even though they are studied based on perfect CSI.

Reducing Inter-CPU Coordination in User-Centric Distributed Massive MIMO Networks

This paper proposes a method for reducing inter-central processing unit (CPU) coordination in user-centric (UC) distributed massive multiple-input multiple-output (D-mMIMO) networks, also called cell-free networks. A two-stage heuristic approach is proposed to reduce the number of inter-coordinated user equipment (UE) (UEs connected to multiple CPUs) in each CPU. Firstly, each UE connects to a master access point (AP) and a primary CPU; then, AP selection is executed. Secondly, each CPU performs a fine-tuning algorithm to drop the worst inter-coordinated UEs. Simulation results reveal that one can reduce inter-CPU coordination while yielding slight degradation in spectral efficiency and improving energy efficiency.

Numerical Evaluation of the MU-MIMO Beamforming Performance with User Selection Algorithm

This paper presents the numerical evaluation of the ZF beamforming algorithm using the user selection in the multiuser multiantenna (MU-MIMO) downlink system. Two user selection algorithm – semiorthogonal user selection and greedy user selection are numerically evaluated based on the open source MIMO channel model. The sum rate performance depending on number of users are presented. The arising inter user correlation degrades the sum rate (spectral efficiency) performance of multiuser MIMO system especially in scenarios where the number of users is larger than the number of antennas at the BS. The selection of users is based on the orthogonality of the channels among selected users. For MIMO channel simulation the QUADRIGA channel model reflecting the real propagation conditions is used. The obtained performance of MU-MIMO ZF precoding in spatially correlated channel are compared based on the empirical cumulative density function of the sum rate of multiple users. Numerical results show that the ZF precoder using user selection (G ZF) outperforms the ZF precoder with random user selection in spectral efficiency. The greedy user selection in spatially correlated channel has advantage to semi-orthogonal user selection. It isobserved that as the increasing the number of served users used for selection the greedy user selection gives better performance than semi-orthogonal algorithm.

Full-duplex multi-user MIMO communication systems performance optimization using leakage-based precoding

The spectral efficiency (SE) can approximately double when using full-duplex (FD) multiuser MIMO communications. However, there are difficulties because of multiuser interferences, self-interference (SI), and co-channel interference (CCI). To improve the SE of the downlink (DL), this paper proposes CCI-aware enhancement to SLNR (signal-to-leakage-and-noise-ratio) signal-to-leakage-and-noise-ratio (SLNR). It considers a suppressing filter at the receiver to cancel the interferences again designing a beamformer based on CCI-plus-noise covariance matrices for every user at the transmitting side. Additionally, we propose an improvement in the SLNR method by using SI-plus-noise covariance matrices to design uplink (UL) beamformers. Unlike zero-forcing and block-diagonalization, the SLNR approach serves numerous antennas at users and BS (base station). The total SE of the communication yielded using the optimized precoder, i.e., obtained from the SLNR-based precoding. To achieve maximum energy efficiency (EE), we use a power consumption model. Simulation results confirm that full-duplex performs well compared to half-duplex (HD) when the number of antennas at every user in uplink as well downlink channels grow, for all Rician factors, for slight powers of the CCI and SI, and a limited number of antennas at the BS. With the proposed scheme for given transmit power and circuit power, we demonstrate that FD has a higher EE than HD.

Performance Evaluation of Hybrid Precoder Design for Multi-User Massive MIMO Systems with Low-Resolution ADCs/DACs

This paper presents a comprehensive analysis and design of a hybrid precoding system tailored for mmWave multi-user massive MIMO systems in both downlink and uplink scenarios. The proposed system employs a two-stage precoding approach, incorporating UQ and NUQ techniques, along with low-resolution DACs in downlink and ADCs in uplink to address hardware limitations. The system considers Zero Forcing and Minimum Mean Square Error algorithms as digital precoding methods for the uplink scenario, while exploring the impact of different DAC resolutions on system performance. Extensive simulations reveal that the proposed system surpasses conventional analog beamforming methods, particularly in multi-user scenarios involving inter-user interference. In downlink, the system demonstrates a trade-off between SE and EE, achieving higher Energy Efficiency with NUQ. In uplink, NUQ and UQ converters exhibit similar performance trends regardless of the chosen combiner algorithm. The proposed system attains enhanced Spectral and Energy Efficiency while maintaining reduced complexity and overhead. The study significantly contributes to the advancement of efficient and effective mmWave multi-user massive MIMO systems by providing a thorough analysis of various quantization schemes and precoding techniques. The findings of this research are expected to aid in the optimization of 5G and beyond technologies, particularly in high-density deployment scenarios.

Secrecy energy efficiency optimization for multiuser massive MIMO wiretap systems with transmit antenna selection

Massive multiple-input-multiple-output (Massive MIMO) significantly improves the capacity of wireless communication systems. However, large-scale antennas bring high hardware costs, and security is a vital issue in Massive MIMO networks. To deal with the above problems, antenna selection (AS) and artificial noise (AN) are introduced to reduce energy consumption and improve system security performance, respectively. In this paper, we optimize secrecy energy efficiency (SEE) in a downlink multi-user multi-antenna scenario, where a multi-antenna eavesdropper attempts to eavesdrop the information from the base station (BS) to the multi-antenna legitimate receivers. An optimization problem is formulated to maximize the SEE by jointly optimizing the transmit beamforming vectors, the artificial noise vector and the antenna selection matrix at the BS. The formulated problem is a nonconvex mixed integer fractional programming problem. To solve the problem, a successive convex approximation (SCA)-based joint antenna selection and artificial noise (JASAN) algorithm is proposed. After a series of relaxation and equivalent transformations, the nonconvex problem is approximated to a convex problem, and the solution is obtained after several iterations. Simulation results show that the proposed algorithm has good convergence behavior, and the joint optimization of antenna selection and artificial noise can effectively improve the SEE while ensuring the achievable secrecy rate.