A New Technique using GA style and LMS for Structure Adaptation

In this paper an adaptation technique using both genetic algorithm (GA) and least mean square algorithm (LMS) is proposed. The key control parameters tap-length and tap-weight are updated using the GA and the LMS algorithm respectively according to the value of mean square error (MSE). The proposed algorithm results optimal solutions very fast by reducing initialization and derivative problems. It can give good convergence than that of only LMS style algorithms. In this paper, LMS style adaptive algorithms where both tap-length and tap-weights are updated using the same algorithm, and its difficulties in application field are discussed. And effectiveness of the proposed adaptation technique using both the GA and the LMS is verified by its extensive simulations.