Novel adaptive filter for fading channel estimation in coherent CDMA receivers

Wireless communications have to cope with fading multipaths. Coherent receivers require the estimation of path amplitudes and phases. The choice of adequate filter length and coefficients is one of the important tasks in the channel-coefficient estimation process. There is a fundamental tradeoff between reducing the noise level and keeping the fading signal spectrum undistorted. We introduce here a novel channel estimator filter with fixed coefficients and adaptive length, which has a performance near to the optimum Wiener filter performance and a much lower complexity. The proposed filter length depends on both the mobile speed and the noise level. We compare our channel estimator filter with some other existent approaches in the literature, for a CDMA scenario, and we analyze the impact of speed estimation errors and signal quality estimation errors on the new filter design. The proposed theoretical filter design is also validated via system studio simulations.