Estimation of Very Large MIMO Channels Using Compressed Sensing

Daniel C. Araújo, André L. F. de Almeida⋆, João C. M. Mota, Dennis Hui†

DOI: 10.14209/sbrt.2013.95

Evento: XXXI Simpósio Brasileiro de Telecomunicações (SBrT2013)

Keywords: Very-large MIMO channels compressed sensing sparsity channel estimation matching pursuit.

Abstract

In this paper, we propose an efficient pilot-assisted technique for the estimation of very-large MIMO (multiple-input multiple-output) channels exploiting the inherent sparsity of the channel. We first obtain an appropriate sparse decomposition model from a virtual channel representation of the very-large MIMO channel. Based on this model, we capitalize on a fundamental result of the compressed sensing (CS) to show that the channel matrix can be accurately estimated from very short training sequences compared to the number of used transmit antennas. We compare the normalized mean square error (NMSE) obtained using the proposed CS-based channel estimator, the least-square (LS) estimator and the Cramer-Rao lower bound (CRLB). The simulation results show that the proposed estimator obtains good performance, being 5 dB from the CRLB.

Download