Simpler Multipath Detection for Vehicular OFDM Channel Tracking

The ever increasing requirements in wireless communications have led to the search for unexploited correlations that could improve channel estimation and tracking. Kalman filtering (KF) has been proposed to exploit several such correlations, e.g., the time correlation in each tap in a multipath channel. When making full use of this correlation, however, a capital disadvantage of KF is its weak performance in the face of a significant, occasional nonlinearity, such as the potential birth of a new tap in a multipath channel or an active tap's death. These nonlinearities are typical for vehicular applications. So far, solutions proposed to this birth-death nonlinearity problem have been shown to be computationally prohibitive. In this paper, a simplified detection framework is introduced and a computationally inexpensive simplified maximum a posteriori estimator is derived. Under low to medium SNR conditions, simulations show the channel tracking error can be approximately halved (versus simple KF) by this novel estimator in orthogonal frequency division multiplexing systems.