IIR model identification via evolutionary algorithms

Infinite impulse response (IIR) filters are often preferred to FIR filters for modeling because of their better performance and reduced number of coefficients. However, IIR model identification is a challenging and complex optimization problem due to multimodal error surface entanglement. Therefore, a practical, efficient and robust global optimization algorithm is necessary to minimize the multimodal error function. In this paper, recursive least square algorithm, as a popular method in classical category, is compared with two well-known optimization techniques based on evolutionary algorithms (genetic algorithm and particle swarm optimization) in IIR model identification. This comparative study illustrates how these algorithms can perform better than the classical one in many complex situations.