Research On Adaptive Echo Cancellation Algorithms In Sparse Paths

Echo cancellation is a traditional problem in the field of communication and signal processing. The echo will seriously affect the voice quality and the communication system stability. It is of great significance for echo cancellation to improve communication quality and increase system stability. Echo cancellation becomes an important research theme with the rapid development of wireless communication and teleconference system.This dissertation analyzes the harm of echo in brief at first. Then adaptive echo cancellation algorithms are discussed based on the model of system identification. Echo cancellation in sparse environment is researched as an important problem. Impulse responses are often sparse, that is, most of the taps are small, only some of the taps are large. This includes the traditional notion of sparsity as a few large taps separated by many negligible taps. This dissertation provides a measurement method of sparsity related to other measurement methods.LMS algorithm is a common and effective adaptive algorithm, which has simple structures and low complexity. However, the performance of LMS algorithm will degrade when the input signals are correlated. And LMS algorithm does not make full use of the prior information that the echo path is usually sparse. Using the sparsity in adaptive algorithms is feasible. Based on the LMS algorithm and its LMS-based forms, we propose some new algorithms using the prior information of sparsity.Sparse LMS algorithm is effective to adjust the parameters in the iterative process. If the value of a tap is large, we use the large step size. If the value of a tap is small, we use the small step size accordingly. So the sparse algorithm would convergence faster than the normal LMS algorithm. For the first sparsity that some...