Adaptive filtering is an important component of modern signal processing. In many applications, a signal must be processed in a manner which depends on unknown or time-varying system parameters. Adaptive filters provide a means of automatically ``learning'' these unknown parameters and ``tracking'' them as they change in time so that the signal of interest is correctly processed. Applications in which adaptive filtering has become a key component include communications, control, radar, sonar, seismology, and biomedical engineering. This course provides the fundamental concepts in modern adaptive filtering.

- Here are some matlab-based adaptive filter simulation toys that might be fun to play with: DALE and ALI.

- R.M. Gray, Toeplitz and circulant matrices: A review (A recently updated version of some classical work.)
- N.R. Yousef and A.H. Sayed, ``A unified approach to the steady-state and tracking analyses of adaptive filters'' (This is the method of steady-state analysis we adopt in this class.)
- B.D. Rigling and P. Schniter, ``Subspace leaky LMS''(The outcome of an 801.01 project!)
- W.A. Sethares, ``The Least Mean Square Family'', in Efficient System Identification and Signal Processing Algorithms, ed. N. Kalouptsidis and S. Theodoridis, Springer-Verlag, 1993. (A good summary of LMS modifications and an overview of different approaches to LMS analysis: stochastic, deterministic, etc.)
- P. Schniter and C.R. Johnson Jr., Dithered Signed-Error CMA (If you are curious how dither can be used to restore LMS-like mean behavior in signed-error algorithms. Here, CMA is the focus, though it would be easy to extend this idea to any cost function.)
- P. Schniter, A Derivation of the Steady-State MSE of RLS: Stationary and Nonstationary Cases.
- A. Sayed and T. Kailath, Recursive Least-Squares Adaptive Filters, from the DSP Handbook. (An excellent summary of the RLS algorithm and various numerically-stable implementations.)

schniter@ece.osu.edu

(