Pilot-Assisted MIMO-V-OFDM Systems: Compressed Sensing and Deep Learning Approaches
Wei Zhang,Xuyang Gao,Zhipeng Li,Yibing Shi
DOI: https://doi.org/10.1109/access.2020.2964046
IF: 3.9
2020-01-01
IEEE Access
Abstract:In this paper, we investigate the channel estimation and decoding methods exploiting the channel sparsity in pilot-assisted Multiple-Input Multiple-Output (MIMO) Vector Orthogonal Frequency Division Multiplexing (V-OFDM) systems. Based on the sparse multipath channels, we utilize orthogonal and non-orthogonal pilot schemes to design the compressed sensing (CS) measurement process. For the optimization of the sensing matrix, we discuss the influence of pilot search algorithms and evaluation criteria and propose a particle swarm optimization (PSO) based pilot search algorithm with the simplified evaluation criterion to improve the pilot design procedure. Meanwhile, the effect of pilot insertion on the Peak-to-Average Power Ratio (PAPR) is reduced by a particular precoding matrix method without affecting the decoding complexity. Simulation data are used to evaluate the classical sparsity adaptive matching (SAMP) algorithms and the proposed Variable Threshold SAMP (VTSAMP) algorithm, and the results show that the improved method has higher channel estimation accuracy with unknown sparsity. On the other hand, to overcome the complexity of CS-based decoding, we design the fully connected Deep Neural Network (FC-DNN) decoders, which combine the results of channel estimation results with the prevalent neural network technology. We observe that when the sparse channels are estimated accurately by CS methods, the proposed FC-DNN can achieve the same performance as the high-precision linear decoder by using the time-domain pilots and channel estimation results.
computer science, information systems,telecommunications,engineering, electrical & electronic