Comparison of the stochastic iteration algorithm and the dual sign algorithm in nonstationary environments

Abstract

An adaptive algorithm, called the dual sign algorithm (DSA), has been proposed recently to simplify the implementation of the widely-used stochastic iteration algorithm (SIA). Previous study has shown that, in stationary environments, the DSA, although simpler, exhibits faster convergence compared with the SIA, Here, computer simulation is used to compare the performance of the two algorithms in identifying some typical time-varying models. The results show that if both algorithms are designed to attain the same steady-state accuracy, the DSA always has a better tracking capability. >