Exponential function echo cancellation method based on one-norm and zero-attractor

Abstract

The invention relates to an exponential function echo cancellation method based on one-norm and zero-attractor, and discloses a zero-norm sub-band echo cancellation method. The method comprises the following steps: A, remote signal sampling; B, echo signal estimation; C, echo signal cancellation; D, filter tap weight coefficient updating; and E, assuming that n is equal to n+1, and repeating steps B, C and D to achieve real-time echo cancellation. According to the method disclosed by the invention, a one-norm of a weight coefficient vector is adopted in the derivation of a weight coefficient vector updating formula, the norm relates to FORMULA, wherein gamma is a proportion parameter of the one-norm of the weight coefficient vector, zero-attractor Rho (n) is obtained from the derivation, Rho (n) is equal to b.sgnW(n), which means that the speed that the weight coefficient vector is updated to zero is faster under sparse systems; a method for taking an exponential function as a cost function is adopted during weight update, and thus the cost function is changed to FORMULA when updating the weight coefficient vector, and a new step length factor is introduced to ensure that output signals of a filter can obtain faster convergence and lower steady state maladjustment under Gaussian signals and the sparse systems.

Description

technical field [0001] The invention relates to an exponential function echo cancellation method based on a norm zero attraction. Background technique [0002] As an important branch of information technology, adaptive signal processing has been widely used in the field of communication. In the field of communication, echo cancellation is a hot spot that attracts much attention and challenges. The sound will form an echo after multiple reflections in a closed space, and the echo will also be formed in the signal transmission due to the impedance mismatch in the transmission medium. The communication echo can be eliminated by the system identification model: the identified system is the echo channel, and the output of the system identification is the estimate of the echo signal, and the echo can be eliminated by subtracting the voice signal containing the echo signal from the estimate of the echo signal, which is The principle of adaptive echo canceller. [0003] Least mea...

AI Technical Summary

Problems solved by technology

For the characteristics of the sparse system, although the LMS algorithm adds a norm to form a zero attractor to speed up the convergence speed of the original algorithm, the linear combination in the LMS algorithm is still used in the update formula of the filter weight coefficient vector In this way, the weight coefficient vector of the filter maintains a fixed speed in the update, so the sensitivity to useful signals at different times decreases. Therefore, the convergence speed of the original algorithm still needs to be improved.

Images