Channel Estimation of MIMO Relay Systems With Multiple Relay Nodes

In this paper, we consider the channel estimation problem for multiple-input multiple-output wireless relay communication systems with multiple relay nodes. In particular, all individual channel matrices of the first-hop and second-hop links are estimated at the destination node by applying the superimposed channel training algorithm, where training sequences are superimposed at the relay nodes to assist the estimation of the relays-destination channel matrices. To improve the performance of channel estimation, we consider the estimation error inherited from the second-hop channel estimation and develop a new minimal mean-squared error-based algorithm to estimate the first-hop channel matrices. Furthermore, we derive the optimal power allocation and training sequences at the source and relay nodes. Numerical examples demonstrate a better performance of the proposed superimposed channel training algorithm.