Adaptive Filter Performance Analysis: A Unified Approach

Based on the theory of simple preconditioned iterative linear equation solvers, we recently introduced a framework within which all major adaptive filter algorithms can be viewed as special cases through the specification of a few parameters. This resulted in two versions of a generic adaptive filter update equation which we in this paper use in deriving general explicit expressions for the learning curve, the steady state excess mean square error (EMSE) and the steady state mean square coefficient deviation (MSD) that are applicable to many families of adaptive filter algorithms. We present experimental results supporting the usefulness and validity of our approach.