Experience-driven learning-based intelligent hybrid beamforming for massive MIMO mmWave communications

We have studied massive MIMO hybrid beamforming (HBF) for millimeter-wave (mmWave) communications, where the transceivers only have a few radio frequency chain (RFC) numbers compared to the number of antenna elements. We propose a hybrid beamforming design to improve the system’s spectral, hardware, and computational efficiencies, where finding the precoding and combining matrices are formulated as optimization problems with practical constraints. The series of analog phase shifters creates a unit modulus constraint, making this problem non-convex and subsequently incurring unaffordable computational complexity. Advanced deep reinforcement learning techniques effectively handle non-convex problems in many domains; therefore, we have transformed this non-convex hybrid beamforming optimization problem using a reinforcement learning framework. These frameworks are solved using advanced deep reinforcement learning techniques implemented with experience replay schemes to maximize the spectral and learning efficiencies in highly uncertain wireless environments. We developed a twin-delayed deep deterministic (TD3) policy gradient-based hybrid beamforming scheme to overcome Q-learning’s substantial overestimation. We assumed a complete channel state information (CSI) to design our beamformers and then challenged this assumption by proposing a deep reinforcement learning-based channel estimation method. We reduced hybrid beamforming complexity using soft target double deep Q-learning to exploit mmWave channel sparsity. This method allowed us to construct the analog precoder by selecting channel dominant paths. We have demonstrated that the proposed approaches improve the system’s spectral and learning efficiencies compared to prior studies. We have also demonstrated that deep reinforcement learning is a versatile technique that can unleash the power of massive MIMO hybrid beamforming in mmWave systems for next-generation wireless communication.     

Decomposition of uplink multi-user OTFS channel with partial CSI feedback

This work studies system design methods for the uplink multi-user orthogonal time–frequency space (OTFS) channel which forms a virtual multiple-input multiple-output (MIMO) communication system. For such system, the received signal contains interferences in the space, frequency, and time domain at the same time. To reduce the computational complexity, this work proposes to decompose the original large MIMO channel into parallel small sub-channels in the following order: first to decompose in the space domain, then to decompose in the time domain, thereby reducing the computational complexity. To help achieve channel decomposition, the proposed method requires the transmitter to perform precoding that needs channel state information (CSI) feedback. However, the proposed method only needs partial CSI feedback including delay and Doppler shift, so the feedback burden is small. Simulation results of the bit error rate (BER) performance verify that the proposed channel decomposition method is effective.     

Deep learning-driven MIMO: Data encoding and processing mechanism

Large-scale Multiple-Input Multiple Output (MIMO) is the key technology of 5G communication. However, dealing with physical channels is a complex process. Machine learning techniques have not been utilized commercially because of the limited learning capabilities of traditional machine learning algorithms. We design a deep learning hybrid precoding scheme based on the attention mechanism. The method mainly includes channel modeling and deep learning encoding two modules. The channel modeling module mainly describes the problem formally, which is convenient for the subsequent method design and processing. The model design module introduces the design framework, details, and main training process of the model. We utilize the attention layer to extract the eigenvalues of the interference between multiple users through the output attention distribution matrix. Then, according to the characteristics of inter-user interference, the loss minimization function is used to study the optimal precoder to achieve the maximum reachable rate of the system. Under the same condition, we compare our proposed method with the traditional unsupervised learning-based hybrid precoding algorithm, the TTD-based (True-Time-Delay, TTD) phase correction hybrid precoding algorithm, and the deep learning-based method. Additionally, we verify the role of attention mechanism in the model. Extensive simulation results demonstrate the effectiveness of the proposed method. The results of this research prove that deep learning technology can play a driving role in the encoding and processing of MIMO with its unique feature extraction and modeling capabilities. In addition, this research also provides a good reference for the application of deep learning in MIMO data processing problems.     

Robust Beamforming Based on Graph Attention Networks for IRS-Assisted Satellite IoT Communications

Satellite communication is expected to play a vital role in realizing Internet of Remote Things (IoRT) applications. This article considers an intelligent reflecting surface (IRS)-assisted downlink low Earth orbit (LEO) satellite communication network, where IRS provides additional reflective links to enhance the intended signal power. We aim to maximize the sum-rate of all the terrestrial users by jointly optimizing the satellite’s precoding matrix and IRS’s phase shifts. However, it is difficult to directly acquire the instantaneous channel state information (CSI) and optimal phase shifts of IRS due to the high mobility of LEO and the passive nature of reflective elements. Moreover, most conventional solution algorithms suffer from high computational complexity and are not applicable to these dynamic scenarios. A robust beamforming design based on graph attention networks (RBF-GAT) is proposed to establish a direct mapping from the received pilots and dynamic network topology to the satellite and IRS’s beamforming, which is trained offline using the unsupervised learning approach. The simulation results corroborate that the proposed RBF-GAT approach can achieve more than 95% of the performance provided by the upper bound with low complexity.     

The Downlink Performance for Cell-Free Massive MIMO with Instantaneous CSI in Slowly Time-Varying Channels

In centralized massive multiple-input multiple-output (MIMO) systems, the channel hardening phenomenon can occur, in which the channel behaves as almost fully deterministic as the number of antennas increases. Nevertheless, in a cell-free massive MIMO system, the channel is less deterministic. In this paper, we propose using instantaneous channel state information (CSI) instead of statistical CSI to obtain the power control coefficient in cell-free massive MIMO. Access points (APs) and user equipment (UE) have sufficient time to obtain instantaneous CSI in a slowly time-varying channel environment. We derive the achievable downlink rate under instantaneous CSI for frequency division duplex (FDD) cell-free massive MIMO systems and apply the results to the power control coefficients. For FDD systems, quantized channel coefficients are proposed to reduce feedback overhead. The simulation results show that the spectral efficiency performance when using instantaneous CSI is approximately three times higher than that achieved using statistical CSI.     

Deep Semantic-Preserving Reconstruction Hashing for Unsupervised Cross-Modal Retrieval

Deep hashing is the mainstream algorithm for large-scale cross-modal retrieval due to its high retrieval speed and low storage capacity, but the problem of reconstruction of modal semantic information is still very challenging. In order to further solve the problem of unsupervised cross-modal retrieval semantic reconstruction, we propose a novel deep semantic-preserving reconstruction hashing (DSPRH). The algorithm combines spatial and channel semantic information, and mines modal semantic information based on adaptive self-encoding and joint semantic reconstruction loss. The main contributions are as follows: (1) We introduce a new spatial pooling network module based on tensor regular-polymorphic decomposition theory to generate rank-1 tensor to capture high-order context semantics, which can assist the backbone network to capture important contextual modal semantic information. (2) Based on optimization perspective, we use global covariance pooling to capture channel semantic information and accelerate network convergence. In feature reconstruction layer, we use two bottlenecks auto-encoding to achieve visual-text modal interaction. (3) In metric learning, we design a new loss function to optimize model parameters, which can preserve the correlation between image modalities and text modalities. The DSPRH algorithm is tested on MIRFlickr-25K and NUS-WIDE. The experimental results show that DSPRH has achieved better performance on retrieval tasks.     

Design of bitstream-based adaptive-connected array architecture for mmWave Multiuser MIMO systems

We investigate a bitstream-based adaptive-connected massive multiple-input multiple-output (MIMO) architecture that trades off between high-power full-connected and low-performance sub-connected hybrid precoding architectures. The proposed adaptive-connected architecture which enables each data stream to be computed independently and in parallel, consists of fewer phase shifters (PS) and switches than the other adaptive-connected architectures. With smaller array groups, the proposed architecture uses fewer PS and switches, so that its power consumption gradually decreases in millimeter wave (mmWave) Multiuser MIMO (MU-MIMO) system. To fully demonstrate the performance of the proposed architecture in mmWave MU-MIMO system with practical constraints, we combine the connection-state matrix with the hybrid precoders and combiners to maximize energy efficiency (EE) of the system equipped with the proposed architecture. We then propose the hybrid precoding and combining (HPC) scheme suitable for multi-user and multi-data streams which utilizes the SCF algorithm to obtain the constant modulus of the analog precoder at convergence. In the digital precoding and combining stage, the digital precoder and combiner are designed to reduce the amount of computation by utilizing the singular value decomposition (SVD) of corresponding equivalent channel. In the mmWave MU-MIMO-OFDM system equipped with the proposed architecture, with the increase of the total number of data streams, simulation results demonstrate that we can exploit the proposed HPC scheme to achieve better EE than the traditional hybrid full-connected architecture exploiting some existing schemes.     

Auction-based spectrum allocation for CoMP systems utilizing hybrid division duplex

The coordinated multi-point (CoMP) transmission is a well-recognized promising technique for achieving high spectral efficiency. In this paper, we study the performance of the spectrum allocation and the system utility of the CoMP systems. First, we combine the hybrid division duplex (HDD) with a CoMP system to form a CoMP–HDD system, for the purpose of guaranteeing the quality demand of the channel state information (CSI) feedback. Second, in order to improve the system utility and the spectrum allocation efficiency of the CoMP–HDD system, we utilize the auction theory for the spectrum allocation. A system utility maximization problem is formulated as an NP-hard problem. Finally, we propose a multi-band multi-winner (MBMW) greedy algorithm to optimize the system utility and the spectrum allocation efficiency. Our simulation results demonstrate the effectiveness of the proposed algorithm.     

Performance of multi-user transmitter preprocessing assisted MIMO system over correlated frequency-selective channels

In this contribution we present the performance of a multi-user transmitter preprocessing (MUTP) assisted multiple-input multiple-output (MIMO) space division multiple access (SDMA) system, aided by double space time transmit diversity (DSTTD) and space time block code (STBC) processing for downlink (DL) and uplink (UL) transmissions respectively. The MUTP is invoked by singular value decomposition (SVD) which exploits the channel state information (CSI) of all the users at the base station (BS) and only an individual user’s CSI at the mobile station (MS). Specifically, in this contribution, we investigate the performance of multi-user MIMO cellular systems in frequency-selective channels from a transmitter signal processing perspective, where multiple access interference (MAI) is the dominant channel impairment. In particular, the effects of three types of delay spread distributions on MUTP assisted MIMO SDMA systems pertaining to the Long Term Evolution (LTE) channel model are analyzed. The simulation results demonstrate that MUTP can perfectly eliminate MAI in addition to obviating the need for complex multi-user detectors (MUDs) both at the BS and MS. Further, SVD-based MUTP results in better achievable symbol error rate (SER) compared to popularly known precoding schemes such as block diagonalization (BD), dirty paper coding (DPC), Tomlinson–Harashima precoding (THP) and geometric mean decomposition (GMD). Furthermore, when turbo coding is invoked, coded SVD aided MUTP results in better achievable SER than an uncoded system.     

Two-level-enforced security with hybrid precoding for multicast massive MIMO wiretap systems

Multicast hybrid precoding reaches a compromise among hardware complexity, transmission performance and wireless resource efficiency in massive MIMO systems. However, system security is extremely challenging with the appearance of eavesdroppers. Physical layer security (PLS) is a relatively effective approach to improve transmission and security performance for multicast massive MIMO wiretap systems. In this paper, we consider a transmitter with massive antennas transmits the secret signal to many legitimate users with multiple-antenna, while eavesdroppers attempt to eavesdrop the information. A fractional problem aims at maximizing sum secrecy rate is proposed to optimize secure hybrid precoding in multicast massive MIMO wiretap system. Because the proposed optimized model is an intractable non-convex problem, we equivalently transform the original problem into two suboptimal problems to separately optimize the secure analog precoding and secure digital precoding. Secure analog precoding is achieved by applying singular value decomposition (SVD) of secure channel. Then, employing semidefinite program (SDP), secure digital precoding with fixed secure analog precoding is obtained to ensure quality of service (QoS) of legitimate users and limit QoS of eavesdroppers. Complexity of the proposed SVD-SDP algorithm related to the number of transmitting antennas squared is lower compared with that of constant modulus precoding algorithm (CMPA) which is in connection with that number cubed. Simulation results illustrate that SVD-SDP algorithm brings higher sum secrecy rate than those of CMPA and SVD-SVD algorithm.     

Unsupervised LoS/NLoS identification in mmWave communication using two-stage machine learning framework

Identification of line-of-sight (LoS)/ non-LoS (NLoS) condition in millimeter wave (mmWave) communication is important for localization and unobstructed transmission between a base station (BS) and a user. A sudden obstruction in a link between a BS and a user can result in poorly received signal strength or termination of communication. Channel features obtained by the estimation of channel state information (CSI) of a user at the BS can be used for identifying LoS/NLoS condition. With the assumption of labeled CSI, existing machine learning (ML) methods have achieved satisfactory performance for LoS/NLoS identification. However, in a real communication environment, labeled CSI is not available. In this paper, we propose a two-stage unsupervised ML based LoS/NLoS identification framework to address the lack of labeled data. We conduct experiments for the outdoor scenario by generating data from the NYUSIM simulator. We compare the performance of our method with the supervised deep neural network (SDNN) in terms of accuracy and receiver characteristic curves. The proposed framework can achieve an accuracy of 87.4% and it outperforms SDNN. Further, we compare the performance of our method with other state-of-the-art LoS/NLoS identification schemes in terms of accuracy, recall, precision, and F1-score.     

Deep neural network inspired by iterative shrinkage-thresholding algorithm with data consistency (NISTAD) for fast Undersampled MRI reconstruction

With the aim of developing a fast algorithm for high-quality MRI reconstruction from undersampled k-space data, we propose a novel deep neural Network, which is inspired by Iterative Shrinkage Thresholding Algorithm with Data consistency (NISTAD). NISTAD consists of three consecutive blocks: an encoding block, which models the flow graph of ISTA, a classical iteration-based compressed sensing magnetic resonance imaging (CS-MRI) method; a decoding block, which recovers the image from sparse representation; a data consistency block, which adaptively enforces consistency with the acquired raw data according to learned noise level. The ISTA method is thereby mapped to an end-to-end deep neural network, which greatly reduces the reconstruction time and simplifies the tuning of hyper-parameters, compared to conventional model-based CS-MRI methods. On the other hand, compared to general deep learning-based MRI reconstruction methods, the proposed method uses a simpler network architecture with fewer parameters. NISTAD has been validated on retrospectively undersampled diencephalon standard challenge data using different acceleration factors, and compared with DAGAN and Cascade CNN, two state-of-the-art deep neural network-based methods which outperformed many other state-of-the-art model-based and deep learning-based methods. Experimental results demonstrated that NISTAD reconstruction achieved comparable image quality with DAGAN and Cascade CNN reconstruction in terms of both PSNR and SSIM metrics, and subjective assessment, though with a simpler network structure.     

Energy Efficiency of User-Centric,Cell-Free Massive MIMO-OFDM with Instantaneous CSI

In the user-centric, cell-free, massive multi-input, multi-output (MIMO) orthogonal frequency division multiplexing (OFDM) system, a large number of deployed access points (APs) serve user equipment (UEs) simultaneously, using the same time–frequency resources, and the system is able to ensure fairness between each user; moreover, it is robust against fading caused by multi-path propagation. Existing studies assume that cell-free, massive MIMO is channel-hardened, the same as centralized massive MIMO, and these studies address power allocation and energy efficiency optimization based on the statistics information of each channel. In cell-free, massive MIMO systems, especially APs with only one antenna, the channel statistics information is not a complete substitute for the instantaneous channel state information (CSI) obtained via channel estimation. In this paper, we propose that energy efficiency is optimized by power allocation with instantaneous CSI in the user-centric, cell-free, massive MIMO-OFDM system, and we consider the effect of CSI exchanging between APs and the central processing unit. In addition, we design different resource block allocation schemes, so that user-centric, cell-free, massive MIMO-OFDM can support enhanced mobile broadband (eMBB) for high-speed communication and massive machine communication (mMTC) for massive device communication. The numerical results verify that the proposed energy efficiency optimization scheme, based on instantaneous CSI, outperforms the one with statistical information in both scenarios.     

Reinforcement Learning-Based Multihop Relaying: A Decentralized Q-Learning Approach

Conventional optimization-based relay selection for multihop networks cannot resolve the conflict between performance and cost. The optimal selection policy is centralized and requires local channel state information (CSI) of all hops, leading to high computational complexity and signaling overhead. Other optimization-based decentralized policies cause non-negligible performance loss. In this paper, we exploit the benefits of reinforcement learning in relay selection for multihop clustered networks and aim to achieve high performance with limited costs. Multihop relay selection problem is modeled as Markov decision process (MDP) and solved by a decentralized Q-learning scheme with rectified update function. Simulation results show that this scheme achieves near-optimal average end-to-end (E2E) rate. Cost analysis reveals that it also reduces computation complexity and signaling overhead compared with the optimal scheme.     

Modified leakage-based transmit filter designs for multi-user MIMO systems

This paper proposes two modified definitions of signal-to-leakage-plus-noise ratio (mSLNR) as criteria for linear transmit filter design in Multi-User MIMO systems. The proposed criteria incorporate receiver structures and priority weight matrices into the precoder design, which can effectively exploit unused receive signal spaces, in the case that available eigenmodes are not fully transmitted, and can potentially prioritise users’ data streams. Iterative SLNR (iSLNR) precoding algorithms based on the proposed mSLNR definitions are also presented. Further, the variations of the iSLNR algorithms are thoroughly studied. In particular, the impact of the mSLNR definitions, choices of receive filters and iteration types on the convergence property and sum-rate performance is discussed. Moreover, the effect of weight matrices on users’ prioritisation is elaborated. For extreme weight values, the proposed algorithms are shown to converge to either eigenbeamforming or null-space decomposition techniques. Further, a robust design of the proposed schemes for the case of imperfect channel state information (CSI) is presented. It is shown that adequate knowledge of effective receive subspaces can be attained by simply assuming matched filters (MF) in the iterative precoding process, while further improvement can be obtained by an actual implementation of interference-mitigation-capable receivers.     

Full-Duplex Relay with Delayed CSI Elevates the SDoF of the MIMO X Channel

In this article, the sum secure degrees-of-freedom (SDoF) of the multiple-input multiple-output (MIMO) X channel with confidential messages (XCCM) and arbitrary antenna configurations is studied, where there is no channel state information (CSI) at two transmitters and only delayed CSI at a multiple-antenna, full-duplex, and decode-and-forward relay. We aim at establishing the sum-SDoF lower and upper bounds. For the sum-SDoF lower bound, we design three relay-aided transmission schemes, namely, the relay-aided jamming scheme, the relay-aided jamming and one-receiver interference alignment scheme, and the relay-aided jamming and two-receiver interference alignment scheme, each corresponding to one case of antenna configurations. Moreover, the security and decoding of each scheme are analyzed. The sum-SDoF upper bound is proposed by means of the existing SDoF region of two-user MIMO broadcast channel with confidential messages (BCCM) and delayed channel state information at the transmitter (CSIT). As a result, the sum-SDoF lower and upper bounds are derived, and the sum-SDoF is characterized when the relay has sufficiently large antennas. Furthermore, even assuming no CSI at two transmitters, our results show that a multiple-antenna full-duplex relay with delayed CSI can elevate the sum-SDoF of the MIMO XCCM. This is corroborated by the fact that the derived sum-SDoF lower bound can be greater than the sum-SDoF of the MIMO XCCM with output feedback and delayed CSIT.     

Deep learning for image reconstruction in thermoacoustic tomography

Microwave-induced thermoacoustic tomography(TAT)is a rapidly-developing noninvasive imaging technique that integrates the advantages of microwave imaging and ultrasound imaging.While an image reconstruction algorithm is critical for the TAT,current reconstruction methods often creates significant artifacts and are computationally costly.In this work,we propose a deep learning-based end-to-end image reconstruction method to achieve the direct reconstruction from the sinogram data to the initial pressure density image.We design a new network architecture TAT-Net to transfer the sinogram domain to the image domain with high accuracy.For the scenarios where realistic training data are scarce or unavailable,we use the finite element method(FEM)to generate synthetic data where the domain gap between the synthetic and realistic data is resolved through the signal processing method.The TAT-Net trained with synthetic data is evaluated through both simulations and phantom experiments and achieves competitive performance in artifact removal and robustness.Compared with other state-of-the-art reconstruction methods,the TAT-Net method can reduce the root mean square error to 0.0143,and increase the structure similarity and peak signal-to-noise ratio to 0.988 and 38.64,respectively.The results obtained indicate that the TAT-Net has great potential applications in improving image reconstruction quality and fast quantitative reconstruction.     

Bit error rate evaluation of relay-aided cooperative NOMA with energy harvesting under imperfect SIC and CSI

In this paper, we investigate a decode-and-forward relay-assisted cooperative non-orthogonal multiple access (NOMA) scheme, where the relay implements a time-switching (TS) based energy harvesting (EH). The impacts of the imperfect channel state information (CSI), inter-cell interference (ICI) and imperfect successive interference cancellation (SIC) are taken into account. We derive the end-to-end bit error rate (BER) expressions under imperfect CSI and ICI for both users. The effect of the EH parameters under the imperfect CSI on users’ BER performance is also examined. Furthermore, we discuss the impact of ICI on BER performance. Computer simulations are used for numerical analysis validation. The results reveal that the CSI deterioration reduces the SIC performance in each node despite the increase in EH parameters and causes an error floor at the higher signal-to-noise ratio (SNR). Furthermore, the BER performance of the users increases by increasing the EH parameters. Also, the ICI affects the SIC and degrades the BER of users.     

Learning the ECSI from binary signals for cognitive multicast secure beamforming

This paper investigates cognitive single-group multicast secure beamforming (SGMC-SBF) in multicast scenario where an eavesdropper who acts as a regular user seeks to intercept the multicast service without authorization. This study emphasizes that the transmitter iteratively learns the eavesdropper’s spatial correlation matrix from the accumulated binary feedback on received signal-to-noise-ratio. In each iteration, the transmitter learns the eavesdropper’s spatial correlation matrix based on the historical beamformings and the historical binary feedback information, which is then used to design the optimal beamforming that will be used to learn eavesdropper’s spatial correlation matrix in the next iteration. Without loss of generality, it is assumed that the transmitter knows instantaneous channel state information (CSI) of the legitimate users (LCSI), but not the instantaneous or statistical CSI of the eavesdropper (ECSI). For comparison, we also established the corresponding genie-aided SGMC-SBF with perfect ECSI and two traditional robust schemes with erroneous and statistical ECSI, respectively. The numerical results verify that the proposed cognitive SGMC-SBF are feasible solutions that provide excellent performance.     

Contract Theory-Based Incentive Mechanism for Full Duplex Cooperative NOMA with SWIPT Communication Networks

Cooperative Non-Orthogonal Multiple Access (NOMA) with Simultaneous Wireless Information and Power Transfer (SWIPT) communication can not only effectively improve the spectrum efficiency and energy efficiency of wireless networks but also extend their coverage. An important design issue is to incentivize a full duplex (FD) relaying center user to participate in the cooperative process and achieve a win–win situation for both the base station (BS) and the center user. Some private information of the center users are hidden from the BS in the network. A contract theory-based incentive mechanism under this asymmetric information scenario is applied to incentivize the center user to join the cooperative communication to maximize the BS’s benefit utility and to guarantee the center user’s expected payoff. In this work, we propose a matching theory-based Gale–Shapley algorithm to obtain the optimal strategy with low computation complexity in the multi-user pairing scenario. Simulation results indicate that the network performance of the proposed FD cooperative NOMA and SWIPT communication is much better than the conventional NOMA communication, and the benefit utility of the BS with the stable match strategy is nearly close to the multi-user pairing scenario with complete channel state information (CSI), while the center users get the satisfied expected payoffs.