A new composite-adaptive algorithm with reduced computational complexity

In the IA and HA algorithms, the convergence factors are updated every iteration, producing an increase in the number of operations. We propose a new composite-adaptive algorithm in which the IA and HA algorithms are complementarily used for a reduction in the number of operations. In particular, we apply the thinned-out method to the IA algorithm in which the convergence factors are updated only once a block. The results of the computer simulations for system identification show that the proposed algorithm works successfully.