Expectation maximization channel estimation for nonlinear OFDM systems

Abstract

This paper introduces the expectation maximization (EM)-based channel estimation for the high power amplifier (HPA)-incurred nonlinear orthogonal frequency division multiplexing (OFDM) systems based on the linearization using extended Bussgang decomposition. Analyses and numerical simulations show that the proposed algorithm only requires reasonable computation complexity with relatively small number of iterations while vastly improves the estimation performance compared to the other conventional estimation methods such as least square error (LSE) or minimum mean square error (MMSE) applied to such system. Moreover, the EM-LSE estimator could give almost the same performance as the EM-MMSE counterpart while does not require channel statistics, forming a robust estimator for both fading and nonlinear channels with reduced computation complexity. This makes the estimator more applicable.