Joint source and relay optimization for a MIMO communication system

In this paper, we consider the joint source and relay optimization problem for a Multi-Input- Multi-Output (MIMO) communication system employing a non-regenerative MIMO relay. Given a fixed total transmission power budget for the source and the relay, we formulate the MIMO transmitter and relay design problem using the Minimum Mean Square Error (MMSE) criterion. Since the original formulation is nonconvex (thus difficult to solve), we present equivalent reformulations which are amenable to solutions by modern convex optimization techniques. In particular, we show that when the channel matrices are diagonal, the optimal MMSE joint source and relay design problem can be solved iteratively as a sequence of Second Order Cone Programs (SOCP). The latter can be solved using highly efficient interior point methods. Computer simulation through SeDuMi (Self-Dual-Minimization) software shows that this new approach (optimal joint source-relay power control strategy) is not only efficient, but also effective, leading to substantially improved mean square error performance than the non-optimized uniform power control strategy.