Robust transceiver optimization for frequency selective MIMO channels

We study the problem of joint transmit and receive filters optimization in a frequency selective, multiple-input multiple-output setup. The information about the channel at the transmitter is imperfect and belongs to a specified uncertainty set, defined by bounding the norm of the error transfer function. The framework for a robust optimization of the system, with mean-square-error (MSE) as the performance measure, is derived. Robustness is defined in the worst-case sense, and a broad range of MSE-optimization problems is supported. The algorithms are constructed in an iterative manner, where each iteration consists of two efficiently solvable semidefinite programs. The proofs of the convergence are provided, as well. Numerical examples show significant performance gains in comparison to the system which performs the optimization of the precoder only.