Abstract
In this thesis, we combine the least-mean square (LMS) and Kalman filter (KF) during channel (time varying frequency selective fading, modeled by AR(2)) estimation base on per-survivor processing-maximum likelihood sequence estimation (PSP-MLSE) algorithm in MIMO OFDM system. In transmitter, data stream is encoded by space-frequency trellis codes, then map into constellation points and modulated by OFDM. At the receiver, receiving data is degraded by fast fading channels (cause by Doppler effect). Using adaptive filter estimate the channel matrix H which is an important information to PSP-MLSE detection. Different from conventional recursive processing, it estimates and update the channel of for one step prediction. According to the velocity of MS, we can adapt the ratio of using KF or LMS algorithm. Finally, we can decode the information bits by utilizing the PSP-MLSE via Viterbi-algorithm with lower complexity than conventional pilot-aided channel estimation.