Robust design for linear non-regenerative MIMO relays

Optimal source precoding matrix and relay amplifying matrices have been developed in recent works for linear non-regenerative multiple-input multiple-output (MIMO) relay communication systems. All these works require the channel state information (CSI) knowledge in order to perform the optimization. However, in practical relay systems, the CSI is unknown and has to be estimated. There is always mismatch between the true and the estimated CSI. In this paper, the true CSI is modelled as a Gaussian random matrix with the estimated CSI as the mean value. First, by using the averaged mean-squared error (MSE) matrix of the signal waveform estimation, we develop the source and relay matrices which optimize most commonly used objective functions subjecting to the averaged transmission power constraints. Second, we consider the quality-of-service (QoS) issue by minimizing the averaged total transmission power subjecting to the averaged MSE constraints at each data stream. Simulation results demonstrate the improved robustness of the proposed algorithms against CSI errors.