Convergence implementation of adaptive lattice algorithm with fuzzy based adaptive gain

In the LMS based adaptive filter, the choice of the adaptive gain is important to the adaptive algorithm performances and the realization structure is also important dependent on application areas. This paper 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 the 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-squared forward prediction error and one step previous adaptive gain. The proposed algorithm is applied to echo cancelling problem of long distance communication channel. The simulation results showed that the convergence speed is improved and the ERLE of echo canceller is also enhanced by comparison with the conventional LMS lattice algorithm.