Using a fast-recursive-least-squared algorithm in a feedback-controller

In active noise control applications (ANC) recursive least-squared algorithms have become more and more common. The reason is that their convergence behavior is quite independent of the statistics of the input sound signal. Using such algorithms with digital finite impulse response filters, FIR, of an order N they show a typical convergence within N samples. There exist several algorithms for FIR-filters which are often called fast recursive least-squared (FRLS) or as fast transversal filters (FTF) in the related literature. The controller is usually built as a feedforward controller when these fast algorithms are used. The crucial integration of the secondary or error path is not done in the same way as for the well known LMS-algorithm. For the LMS usually the filtered-X-LMS is used. But this integration scheme does not work with those fast algorithms. Integration methods like the FASPIS-scheme are used instead. We describe a way to use these efficient algorithms in a adaptive feedback control structure.