Application of genetic algorithms to infinite impulse response adaptive filters

Adaptive infinite impulse response filters require the exploration of a multimodal error surface. We describe the use of genetic algorithms for this particular problem and demonstrate that our approach is capable of discovering the global minimum of the performance surface of a multimodal adaptive filter example in a reasonable length of time compared with other methods. We also compare the performance of the genetic algorithm with the recently-proposed method of very fast simulated reannealing.