Improved performance of blind equalization using prior knowledge of transmitter filter

Abstract

Recent progress by several groups around the world shows that moving average linear time-invariant channels that distort digital communication signals can be identified using only second-order statistics, provided that certain conditions on the channel and data are met. In this paper, a subchannel response matching channel identification method is modified to exploit prior knowledge of the pulse-shaping filter. Computer simulations demonstrate the method's performance as a channel estimator and as an equalizer compared to appropriate lower bounds on the normalized channel estimation error and the bit error rate.<>