A comparative study of some simplified RLS-type algorithms

The recursive least squares (RLS) algorithm has established itself as the "ultimate" adaptive filtering algorithm in the sense that it is the adaptive filter exhibiting the best convergence behavior. Unfortunately, practical implementations of the algorithm are often associated with high computational complexity and/or poor numerical properties. Rather than focusing on full RLS algorithm implementations aiming directly at remedying these problems, we argue that the use of simplified or partial RLS algorithms may be a viable alternative to full RLS. In particular, we point out that two recently introduced algorithms, fast Euclidian direction search (FEDS) and recursive adaptive matching pursuit (RAMP) can indeed be interpreted as such partial RLS algorithms exhibiting a nice tradeoff between complexity and performance. We support our presentation by a comprehensive set of simulation results.