Blind equalization for fast frequency selective fading channels

This paper addresses the issue of blind adaptive maximum likelihood sequence estimation (BMLSE) using per survivor processing (PSP) over a fast frequency selective fading (FFSF) channel. Special care is dedicated to the choice of the adaptive filtering (AF) algorithm used and its influence on the receiver symbol error rate (SER) performance. An in-depth investigation of least mean square (LMS) and Kalman filtering (KF) performance is attempted. Regarding the KF algorithm two channel models were considered: first-order autoregressive (AR1) and second-order autoregressive (AR2) models. Three reception BMLSE schemes are discussed: BPSP-LMS, which uses LMS, and BPSP-KF1 and BPSP-KF2, which employ KF, according to the channel model used. For all the simulations accomplished in this work, the M-algorithm was used to obtain the survivors path. Results of several simulations, under varying signal-to-noise ratio (SNR) conditions and maximum Doppler shift (f_D), are reported