Adaptive Filter Convergence

This example shows the convergence path taken by different adaptive filtering algorithms. The plot is a sequence of points of the form (w1,w2) where w1 and w2 are the weights of the adaptive filter. The blue dots in the figure indicate the contour lines of the error surface. Zoom into the graph to see properties of convergence path by selecting Zoom In from the Tools menu.

This example does not depict the convergence speed of the different algorithms. You can experiment with different step-size values for the adaptive filters to see the change in convergence paths.

Each of the adaptive filters can be enabled or disabled separately:

  • LMS - Least Mean Square algorithm

  • NLMS - Normalized LMS algorithm

  • SELMS - Sign-Error LMS algorithm

  • SSLMS - Sign-Sign LMS algorithm