A Bernoulli-Gaussian model for multipath channel estimation in downlink WCDMA

This paper presents a new algorithm for channel estimation in downlink WCDMA-FDD mode where a Bernoulli-Gaussian (BG) model is introduced for multipath channel. We have investigated a deconvolution algorithm called single most likely replacement (SMLR) especially derived for complex baseband signals and we have experienced the failure of such tentative for multipath channel estimation. This leads us to develop a new algorithm noted E-SMLR which works as an interference cancelling technique to enhance the SMLR performance. E-SMLR performs very well, it gives good performance while retains a low complexity. Simulations are given that support these claims.