To see the applet you need a JDK 1.1.5 or higher compatible browser...

- On this page you can see an online demonstration of adaptive filtering algorithms at work.

- When the demonstration is started a window appears where the impulse response of the unknown system is displayed in the top right corner of the window. This impulse response can be changed by clicking on the window. At the bottom you can see two adaptive algorithms operating in a parallel system identification set-up. The algorithms can be changed by selecting the in the menu bar.

- The following algorithms implemented are:
- FIR-Least Mean Squares (FIR-LMS)
- FIR-Normalised Least Mean Squares (FIR-NLMS)
- FIR-Sign Data Least Mean Squares (FIR-SD-LMS)
- FIR-Sign Error Least Mean Squares (FIR-SE-LMS)
- FIR-Sign Data Sign Error Least Mean Squares (FIR-SD-SE-LMS)
- FIR-Least Squares with QR Decomposition (FIR-QR)
- FIR-Recursive Least Squares (FIR-RLS)
- FIR-Random Walk Algorithm (FIR-RW) assuming that the disturbance of the adaptive parameters is white with the given power
- IIR-Least Mean Squares Algorithm, Feintuch (IIR-LMS)
- IIR-Simplified Gradient Algorithm (IIR-Simp)
- IIR-Full Gradient Algorithm (IIR-Full)
- IIR-QR Pseudo Linear Regression Algorithm (IIR-QR)
- The input sources availble for the input signal and the observation noise are:
- White Noise Source
- White Noise with Slowly Changing Power
- Low-Pass Noise Source
- High-Pass Noise Source
- Band-Pass Noise Source
- Sinusoidal with a Low Frequency
- Sinusoidal with a Medium Frequency
- Sinusoidal with a High Frequency

If under the UNKNOWN SYSTEM Option File is selected the applet is changed such that user supplied files are read. Note that this can cause problems if the demonstration is run as an applet due to security restrictions. The user has to supply the input files which contain the data in ASCII column format and the the algorithms are started together by using the "Start Algorithm" button.

The SNR ratio is the signal-to-noise ratio in the desired signal, i.e. the power of the output of the unknown system divided by the power of the interference noise.

The internal structure of the applet is shown below:

If you have any comments, please email moritz@spd.eee.strath.ac.uk.