EM based Per-Subcarrier ML Channel Estimation for Filter Bank Multicarrier Systems

Abstract:
Filter bank based multicarrier (FBMC) systems present an alternative solution to cyclic prefix based orthogonal frequency division multiplexing (CP-OFDM) in wireless environments with multipath propagation. In this contribution, we propose a novel method of per-subcarrier maximum likelihood (ML) narrowband channel estimation as an extension of the scheme recently developed by the same authors. The main difference is that our new estimation method assumes that only the training sequence transmitted in the observed subcarrier is known and unknown data symbols are transmitted in the two immediately adjacent subcarriers. The method is based on the expectation maximization (EM) algorithm and allows iteratively to converge to the ML solution. The main advantage of the method is the increase in the spectral efficiency, since less subcarriers need to be filled with training symbols. Our simulation results show that if enough training and number of iterations are employed, a similar performance to the original ML algorithm, where the 3 subcarriers are filled with training, can be achieved.