In this paper, a novel mechanism is studied to improve the performance of the channel state information (CSI) feedback in massive multiple-input multiple-output (MIMO) systems. The proposed mechanism encompasses convolutional neural network (CNN)-based CSI compression and reconstruction structure. In this structure, the long-short term memory (LSTM) is adopted to learn temporal correlation of channels, and then, an attention mechanism is developed to perceive local information and automatically weight feature information. In addition, the CNN framework is further adjusted to reduce the number of training parameters and accelerate CSI recovery. The CNN structure with optimal training parameters can be achieved via offline iterative training and learning based on various training datasets. Comparative experimental studies demonstrate the effectiveness of the proposed approach that the trained CNN can obtain the higher feedback accuracy and better system performance in massive MIMO CSI online feedback reconstruction. Moreover, the proposed scheme in the less parameters-based neural network owns a higher performance with lower computational complexity compared to the conventional algorithms.

In this work, an effective adaptive block-sparsity log-sum least mean square (BSLS-LMS) algorithm is proposed to improve the convergence performance of cluster sparse system identification. The main idea of the proposed scheme is to add a new block-sparsity induced term into the cost function of the LMS algorithm. We utilize the ${l}_{{1}}$ norm of adaptive tap weights and log-sum as a mixed constraint, via optimizing the cost function through the gradient descent method, the proposed adaptive filtering method can iteratively move the identified signals towards the optimal solutions, and finally identify the unknown system accurately. The cluster-sparse system response, with block length and arbitrary average sparsity, is generated by a Markov-Gaussian (M-G) model. For the white Gaussian input data, the theoretical formulas of the steady-state mis-adjustment and convergence behaviors of the BSLS-LMS are derived in a general sparse system and a block sparse system, respectively. Numerical experiments demonstrate that the proposed adaptive BSLS-LMS algorithm achieves much better convergence behavior than conventional sparse signal recovery solutions. The experimental study also verifies the consistency between the simulation results and the theoretical analysis.

Channel estimation is essential for millimeter-wave (mmWave) multiple-input multiple-output (MIMO) systems with hybrid precoding. However, accurate channel estimation is a challenging task as the number of antennas is huge, while the number of RF chains is limited. Traditional methods of compressed sensing for channel estimation lead to serious loss of accuracy due to channel angle quantization. In this brief, we propose a new iterative reweight-based log-sum constraint channel estimation scheme. Specifically, we exploit the structure sparsity of the mmWave channels by formulating the channel estimation problem as an objective optimization problem. We utilize the log-sum as a constraint, via optimizing an objective function through the gradient descent method, the proposed algorithm can iteratively move the channel estimated angle-of-arrivals (AOAs) and angle-of-departures (AODs) towards the optimal solutions, and finally improve the angle estimation performance significantly. In addition, to guarantee the accuracy of channel estimation, we introduce a dynamic regularization factor to leverage between the channel sparsity and the data fitting error. Numerical experiments demonstrate that the proposed algorithm achieves better convergence behavior than conventional sparse signal recovery approaches.

Resource allocation with sensing-based interference price is considered for multi-users cognitive radio (CR) network, in which the primary base station (PBS) controls the secondary users’ (SUs) transmission by pricing the SUs’ interference power. SUs firstly initiates data transmission based on the sensing decision and then PBS sets the interference price according to each SU’s interference power. Stackelberg game is formulated to jointly obtain the maximum revenue for PBS and optimize the resource allocation to maximize the transmission gain for SUs. Two practical CR network models are investigated: the sensing based spectrum sharing(SBSS) and the opportunistic spectrum access(OSA). For each scenario, the resource allocation strategy is investigated under the two pricing schemes, namely uniform interference pricing and non-uniform interference pricing. Especially, the stackelberg equilibriums for the proposed games is characterized, and the distributed sensing based interference price bargaining algorithm is proposed according to different channel state information (CSI) for the non-uniform interference pricing case. Numerical examples are carried out to demonstrate the effectiveness of the proposed game algorithm under different pricing scheme.

Reconfigurable intelligent surface (RIS) holds great promise as communication aid which is capable of controlling the electromagnetic propagation environment by adjusting the phase shift of reflective elements. However, in RIS-assisted multi-user wireless communication systems, channel estimation is challenging due to the inclusion of a large number of passive reflective elements within the RIS. To address this problem, we take advantage of the sparsity inherent in multi-user cascade channels and propose a novel cascade channel estimation strategy with low pilot overhead. Specifically, inspired by matrix vectorization and total variational regularization, we estimate the cascaded sparse channels by following Sparse Bayesian Learning (SBL) inference after vectorizing the channel matrix. By introducing regularization to virtual angle-domain sparse channel hyperparameter priors during hyperparameter updates, which penalize non-zero and zero regions of sparse channels differently, thereby enhance the sparseness objective exploited by SBL, this innovation results in a significant reduction in pilot overhead. We then perform SBL inference by updating hyperparameters in parallel using an Expectation-Maximization-based segment alternating optimization method. Simulation results demonstrate the effectiveness of our proposed algorithm that it can substantially reduce the pilot overhead.