An H∞ filtering approach for robust tracking of OFDM doubly-selective channels

In OFDM systems, fast time-varying channels can be effectively estimated and tracked by coupling a channel basis expansion model with a Kalman filtering approach. Although the Kalman filter is asymptotically MSE-optimal when the channel statistics are known, it may suffer some performance degradation in the presence of statistical mismatches. We propose to estimate OFDM doubly-selective channels by an H_∞ filtering approach that, although suboptimal in the MSE sense, is more robust than classical Kalman filtering to statistical mismatches. Simulation results and performance comparisons assess the effectiveness of the proposed solution.