CONVERGENCE IMPROVEMENT OF ADAPTIVE LATTICE ALGORITHM WITH FUZZY BASED ADAPTIVE GAIN

Abstract

The choice of the adaptive gain is important to the performance of LMS-based adaptive filters. Depending on application areas, the realization structure of the filters is also important. This letter presents an adaptive lattice algorithm which adjusts the adaptive gain of LMS using fuzzy if-then rules determined by matching input and output variables during adaptation procedure. In each lattice filter stage, this filter adjusts the adaptive gain as the output of the fuzzy logic which has two input variables, normalized squared forward prediction error and one step previous adaptive gain. The proposed algorithm is applied to echo canceling problem of long distance communication channel. The simulation results are compared with NLMS on TDL and lattice structures.