The teaching of adaptive signal processing: algorithms, architectures and applications

We discuss the presentation of a one semester course on adaptive signal processing to a class of postgraduates. The course is taught from the four viewpoints of adaptive algorithms, architectures, applications and real world DSP implementation technology. In addition to the course structure and syllabus, this paper also reports on the development of an in-house software package called Adapt-Lab. The aim of this software is to allow students to understand the fundamentals of real world adaptive signal processing by presenting adaptive algorithms, architectures and applications in a visual form. AdaptLab allows the single channel architectures of system identification, inverse system identification, noise cancellation, prediction and filtered-X implementations to be set up. The user can select adaptive filter type and length, algorithm parameters (such as step size, forgetting factor, leakage, update rate, etc.), and an appropriate input signal from a signal generator or from the on-disk signal library of typical input signals