Structured compressive sensing-based non-orthogonal time-domain training channel state information acquisition for multiple input multiple output systems

In practical multiple input multiple output (MIMO) systems, accurate knowledge of the channel state information (CSI) is a prerequisite to guarantee the system performance. The conventional CSI acquisition methods for MIMO system usually rely on the orthogonal (either time- or frequency-domain) training sequences (TSs) to estimate the channel associated with each transmit-receive antenna pair, which is not spectrally efficient. This study proposes a non-orthogonal time-domain training-based CSI acquisition approach for MIMO systems under the framework of structured compressive sensing. By exploiting the spatial-temporal correlations of the sparse MIMO channels, a spatially-temporally spARsity-adaptivE-simultaneous orthogonal matching pursuit algorithm is proposed, which could use the inter-block interference free region of very small dimension within the received TS to recover the multiple channels. Furthermore, the proposed algorithm could utilise the priori channel partial common support to improve the recovery probability and reduce the complexity. Simulation results show that the proposed scheme has better performance and higher spectral efficiency than the conventional MIMO schemes, which might be an appealing solution for the future wireless communications.