Multiuser Multihop MIMO Relay System Design Based on Mutual Information Maximization

In this paper, we consider multiuser multihop relay communication systems, where the users, relays, and the destination node may have multiple antennas. We address the issue of source and relay precoding matrices design to maximize the system mutual information (MI). By exploiting the link between the maximal MI and the weighted minimal mean-squared error (WMMSE) objective functions, we show that the intractable maximal MI-based source and relay optimization problem can be solved via the WMMSE-based source and relay design through an iterative approach which is guaranteed to converge to at least a stationary point. For the WMMSE problem, we derive the optimal structure of the relay precoding matrices and show that the WMMSE matrix at the destination node can be decomposed into the sum of WMMSE matrices at all hops. Under a (moderately) high signal-to-noise ratio (SNR) condition, this WMMSE matrix decomposition significantly simplifies the solution to the WMMSE problem. Numerical simulations are performed to demonstrate the effectiveness of the proposed algorithm.