Recursive Inverse Adaptive Filtering for Fading Communication Channels

Adaptive filtering is an important and rapidly developing discipline in the signal processing and communications field. Many different algorithms were developed throughout the years. LMS and RLS being the most well-known and studied of the algorithms. Like in many disciplines in engineering there exists many disadvantages in these algorithms and so variations were proposed. The main parameters sought after in a new algorithm were convergence speed, stability, and computational complexity. LMS suffers in the former as it has generally low convergence speed. Meanwhile, RLS main disadvantage is the computational complexity it introduces, in addition to being unstable in varying channel settings. RI adaptive filtering algorithm proposed in 2009 provides some solutions to the aforementioned issues. In this thesis, the use of the RI algorithm in common communication channel setting is simulated. The performance of the algorithm is studied and compared to some other adaptive filtering algorithms in terms of BER. Both channel estimation and channel equalization performances are investigated and commented upon. The conclusions reached will be that RI outperforms RLS in terms of convergence speed and computational complexity in all cases studied while giving similar results to NLMS in frequency selective channels but outperforming it in flat fading channels. Keywords: Adaptive filter, recursive inverse, RLS, communication channel