Adaptive zero-tracking complemented by LMS weight update for guaranteed convergence

A low-complexity zero-tracking algorithm with fast and guaranteed convergence behavior is proposed and investigated. The algorithm implements a zero-tracking algorithm and a least mean square (LMS) weight update algorithm in parallel, with the former adjusting the zeros of the array in a time-multiplexed manner to achieve fast asymptotic and tracking behavior, while the latter improves the initial transient behavior of the algorithm and guarantees its convergence to the global optimum. By comparing their output powers, the relative performances of the two component algorithms are monitored, and re-initialization of one algorithm by the other may occur periodically. This gives the algorithm both fast and guaranteed convergence behavior, even though the zeros are directly available and the implementation complexity is only two times that for the conventional LMS algorithm