Download Help (Windows Only) |

When you create an adaptive filter, you must specify a value for the adaptive filter length. The filter length affects the computational resource requirements, convergence speed, and steady state error of the resulting adaptive filter.

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

- Choosing an appropriate filter length for an application is a trial-and-error process. You must simulate the adaptive filter to determine the most appropriate filter length.
- The filter length must be long enough to satisfy the application requirements.
- The filter length must be greater than the number of significant taps in the impulse response of the unknown system.
- A long filter length can reduce the steady state error. However, if the filter length is too long, the steady state error cannot become minimal. A long filter length also requires more computational resources.
- Use the minimum filter length that satisfies the application requirements. A small filter length can increase the convergence speed. A small filter length also can save computational resources.

The following figure shows the learning curves of an example VI. This VI uses an adaptive filter with three different filter lengths and the same step size to perform system identification.

In the previous figure, when the filter length is 20, the adaptive filter becomes steady after approximately 800 iterations. The steady state error is approximately 0.002. When you increase the filter length to 30, the steady state error becomes smaller, but the adaptive filter takes more time to converge. When you increase the filter length to 300, the steady state error becomes larger instead of becoming smaller, which indicates that the filter length is too large. The convergence of this adaptive filter also is slow. In this situation, reduce the filter length to reduce the steady state error and improve the convergence speed.