Download Help (Windows Only) |

When you use the least mean squares (LMS) or filtered-x LMS algorithms to create an adaptive filter, you must specify a value for the step size. The step size value affects the convergence speed, steady state error, and stability of the adaptive filter.

Take the following guidelines into consideration when you specify the step size for an adaptive filter.

- Use a small step size to ensure a small steady state error. However, a small step size also decreases the convergence speed of the resulting adaptive filter.
- Increase the step size to improve the convergence speed of the resulting adaptive filter. However, a large step size might cause the adaptive filter to become unstable.

The following figure shows the performance of an adaptive filter with different step size values.

In the previous figure, notice that when the step size is 0.0035, the learning curve of the adaptive filter has a steady state error of 0.0001. The learning curve becomes steady after approximately 700 iterations. If you increase the step size to 0.035, the convergence speed increases. However, the steady state error also increases. As you increase the step size to 0.055, the learning curve becomes unstable.

Use the AFT Estimate Maximum Step Size for FIR LMS VI to estimate the maximum step size value that you can specify for an adaptive filter. This VI supports the standard LMS algorithm only.