Research Of MLSE Algorithm And Its Performance

With the development of modern digital communication system, the high transmission speed and reliable performance are required. The effect of intersymbol interference (ISI) need to be solved, which is caused by multi-path fading. Equalization is an effective technology to overcome ISI. The maximum-likelihood sequence estimation algorithm (MLSE) is the optimal equalization technique at bit error rate, which becomes a hotspot of research.MLSE finds the survival path, which has the maximum-likelihood metric, by exhaustive search. It is not practical when the computational power of the receiver is limited. Forney improved the Viterbi algorithm (VA) for MLSE[1], for the implement of MLSE is similar to convolutional encoding. Although the VA for MLSE is more efficient than the exhaustive ML search, the disadvantages of MLSE are high complexity and large delay,and the application range of channel is limited. The VA-MLSE is called conventional MLSE. In addition, the suitable algorithm can’t be directly determined by the simulation results, since the results of simulation will differ under various condition. So the bound of performance need to be studied, which can be used to analysis the algorithm.In this thesis, the improved MLSE algorithms at reduced complexity and shorted delay are introduced, the performance of improved MLSE algorithms are analysed, and upper bounds form the error event are derived. The MLSE over multi-path, which has ISI and additive white Gaussian noise(AWGN), has been presented. But in mobile radio environments, the frequency selective fading channel is common. Then, a combined MLSE equalization algorithm with D-diversity reception is proposed. The principle of new algorithm is introduced and a new upper bound on the bit error rate is derived. A new upper bound is tighter for simulation result than Forney’s bound. On the other hand, a new combined MLSE based on the improved MLSE algorithm is presented. Finally, the BER performance and bound of these algorithms are simulated by Matlab, and the results of performance are analyzed. Through the results show that combined MLSE equalization algorithms with D-diversity reception have a better practical value.