EM-based H-inf channel estimation in MIMO-OFDM systems

A robust and reduced-complexity H-infinity (H-inf) channel estimator for multiple-input multiple-output orthogonal frequency division multiplexing (MIMO-OFDM) systems is addressed. The proposed estimator is realized by taking the following procedures: first, a simplified objective function is considered to guarantee the design simplicity of the H-inf channel estimator. Second, an expectation maximization (EM) algorithm is adopted to make the problem less complex. Third, an equivalent signal model (ESM) is utilized to relieve the non-Gaussian noise (NGN) features of practical channels, which are due to various natural or man-made impulsive sources. According to the simulation results, it is shown that the H-inf estimator has almost the same mean square error (MSE) performance as an optimal maximum a posteriori (MAP) estimator. Moreover, during implementing H-inf estimator via EM process, only few iterations are needed. Compared to traditional signal models (TSM), ESM can improve the robustness of the estimator substantially, especially in case of NGN channels.