Regularized Preconditioning for Krylov Subspace Equalization of OFDM Systems over Doubly Selective Channels

Orthogonal frequency division multiplexing (OFDM) systems over doubly selective channels may simultaneously suffer from inter-carrier interference (ICI) and imperfect channel estimation. The performance of a linear equalizer can degrade significantly if the presumed system model is mismatched with the actual system. Krylov subspace equalization, which is often combined with preconditioning, can improve the robustness against model mismatch. In this letter, we show that conventional preconditioners degrade performance when the model mismatch is serious. To address this problem, we design regularized preconditioners that cluster only the largest eigenvalues of the relevant system matrix. We demonstrate that the regularized preconditioners can effectively reduce complexity and at the same time preserve the robustness of Krylov subspace equalizers.