LSQR-based ICI equalization for multicarrier communications in strongly dispersive and highly mobile environments

For OFDM transmission over rapidly varying channels, intercarrier interference (ICI) constitutes a major source of performance degradation. We propose a low-complexity ICI equalization technique that uses the iterative LSQR algorithm for regularized inversion of a triple-band approximation to the frequency-domain channel matrix. The LSQR algorithm achieves regularization of the (typically ill-conditioned) channel inversion via early termination of the iteration process, and thus yields good results at low complexity. We also consider pulse shaping as a means of reducing the bandwidth of the matrix approximation. Simulation results demonstrate the excellent performance of the proposed ICI equalizer even for strongly dispersive and rapidly varying channels, as well as the potential of pulse shaping for reducing the "ICI bandwidth."