A simulation study of adaptive filtering applied to MLSE-PSP receivers

We address adaptive maximum-likelihood sequence estimation using per survivor processing (MLSE-PSP) over frequency selective fast fading channels. Special care is dedicated to the choice of the adaptive filtering algorithm (AFA) used in these schemes and its influence on the receiver bit error rate (BER) performance. We present an in-depth investigation of LMS and KF performance in the MLSE-PSP context. Regarding the KF algorithm, we focus attention on statistical channel models and propose the use of a second-order autoregressive (AR(2)) modeling as an alternative to the first-order autoregressive (AR(1)) modeling, which has been adopted in other works. The performance of the MLSE-PSP schemes is extensively evaluated by Monte Carlo simulation. Several experiments under conditions of varying signal-to-noise ratio (SNR), maximum Doppler shift (f_D) and data block length (BL) are reported. The results presented indicate that the use of AR(2) produces remarkable performance improvement