The lecture provides a deeper look into the world of adaptive filter algorithms. Starting with classical definitions like Least Mean Square (LMS) and Wiener solutions, low complexity gradient type algorithms for such solutions are developed and analyzed. Numerous application examples will be presented. The contradiction between the stochastic approach to describe LMS and the deterministic approach to describe RLS algorithms is shown. In order to adapt to a permanently changing environment, the tracking nature of adaptive algorithms is studied and the Kalman algorithm is introduced as optimal solution. Finally, robust descriptions as they are common in modern control theory (H_{00} control) allow to solving the classical contradiction and broaden the theory in order to understand more complex algorithms like adaptive IIR filtering and blind equalization. Applications in the field of hands free telephony, active noise control, car engine control, echo control in communication systems, linear prediction, adaptive equalization and neural networks will be presented and taught by Matlab exercises.

This course will be given the next time in WS 2014!!!

Part I: written exercises during the semester (math calculus and matlab)

Part II: oral exam at the semester end

Nicht erforderlich

Working knowledge of random variables and deterministic signal processing