This paper proposes a novel downlink precoding method for a cell-free massive multiple-input multiple-output (CF-mMIMO) network, requiring no channel state information sharing between the access points via fronthaul links. By drawing analogies between a CF-mMIMO network and an artificial neural network, the proposed algorithm borrows the idea of backpropagation to train the precoders and the combiners through over-the-air ping-pong signaling between the access points and user equipments. It utilizes manifolds optimization to meet the per-AP power constraint and is named as distributed quasi-neural network precoding on manifold (DQNPM). The DQNPM algorithm can accommodate a large category of objective functions for fully distributed implementation. Numerical simulations show that our method outperforms the state-of-the-art approaches, and is robust against pilot contamination.

Cell-free massive multiple-input multiple-output (MIMO) can resolve the inter-cell interference issue in cellular networks through cooperative beamforming of the distributed access points (APs). This paper focuses on an uplink cell-free massive MIMO network and investigates novel methods to train the central processing unit (CPU), the APs, and the users in the network. To reduce the communication burden posed on the fronthaul, each AP applies receive beamforming to compress the vector signals into scalar ones before passing them to the CPU for centralized processing. By drawing analogies between an uplink cell-free network and a quasi-neural network and borrowing the idea of backpropagation algorithm, we propose a novel scheme named the distributed learning for uplink cell-free massive MIMO beamforming (DLCB), which can achieve the multi-AP cooperation without explicit estimation of their channel state information (CSI). The DLCB has low computational complexity and is applicable to various objective functions, such as the minimum mean squared error criterion and the maximum sum rate criterion. Extensive simulations verify that the proposed scheme achieves superior performance over the state-of-the-art methods.

This paper studies distributed optimization of an uplink cell-free Massive MIMO (CF-mMIMO) network. By observing some interesting analogies between the CF network and an artificial neural network (ANN), we propose to relate the uplink CF network to a so-called quasi-neural network. Borrowing the idea of the back-propagation (BP) algorithm, we propose a novel scheme to optimize the central processing unit (CPU) and the access points (APs) of the network. The proposed scheme can achieve multi-AP cooperation using only the pilot sequences, but without the channel state information (CSI). To reduce the required throughput of the fronthaul, we let each AP beamform the received vector signals into scalar ones before passing them to the CPU. The effectiveness of the proposed scheme is verified by the simulations.

This paper studies a distributed scheme for a multi-input multi-output (MIMO) relay network, where the transmit nodes are subject to the nonlinear instantaneous power constraints. We introduce a novel perspective of regarding a relay network as a so-termed quasi-neural network by drawing its striking analogies with a (four-layer) artificial neural network (ANN). We propose a nonlinear amplify-and-forward (NAF) scheme inspired by the back-propagation (BP) algorithm, namely the NAF-BP, to optimize the transceivers to maximize the output signal-to-interference-plus-noise ratio (SINR) of the data streams. The NAF-BP algorithm can be implemented in a distributed manner with no channel state information (CSI) and no data exchange between the relay nodes. The NAF-BP can also coordinate the distributed relay nodes to form a virtual array to suppress interferences from unknown directions. Extensive simulations verify the effectiveness of the proposed scheme.

The estimation of velocity, angle-of-arrival (AOA), and range of a target has been researched for decades, as it finds wide applications in radar and wireless communications. In recent years, this classic problem has gained renewed interest with the advent of 5G internet of things (IoT) technologies, owing to the numerous emerging localization-related applications. This paper studies the joint estimation of velocity, AOA, and range (JEVAR) of a target in a multipath environment. To solve the JEVAR problem, we propose a novel scheme, which has the target transmit a pair of conjugate Zadoff-Chu (ZC) sequences and has the multi-antenna receiver conduct maximum likelihood (ML) estimation. The simulations verify the effectiveness of the proposed scheme by showing that its performance can approach the Cramer-Rao bound (CRB).