Performance Analysis of Gradient Adaptive Lattice Joint Processing Algorithm

Abstract

Tracking speed and stability of adaptive gradient filtering algorithms represented by least mean square (LMS) are restricted for non-stationary circumstance. A joint processor which consist of the gradient lattice filter and transversal LMS linear combiner was designed, the performance of processor were investigated when the input signals were interfered by white noise, Volvo noise and pink noise respectively. The noise canceling computer simulation testified that the joint processor could get stabilization only after 20 iterative operations, and provide stronger ability to boost SNR of weak signal compared with transversal LMS filter. All the performance indices including tracking ability and convergence stability are superior to the transversal LMS algorithm in the same circumstance, and it needs less hardware resource.