Convex combination of two adaptive filters for PBS-LMS algorithm

Combination approaches can improve the performance of adaptive filters. Recently a convex combination of adaptive filters was proposed to improve the performance of LMS algorithm. In this paper we propose to use the PBS-LMS algorithm instead of LMS algorithm in the structure of this convex combination. Our simulations show that this structure not only has the optimality of first one, but also it has the features of PBS-LMS algorithm such as regularity. By using PBS-LMS algorithm in this structure we can save in total number of samples needed by filter to converge about 22.2%, for example the fast filter converges to the steady state in 254 samples, the slow one in 397 samples and the overall filter in 309 samples.