Excessive delay spread in mobile channels causes severe Inter-Symbol-Interference (ISI). The problem can be reduced by adaptive equalizers that implicitly estimates the channel impluse response and filter the received signal through the inverse channel. For reliable communications, the adaptive filter should be able to converge quickly, and maintain its convergence in presence of occasional errors and fading hits.

The Least Mean Square (LMS) adaptation algorithm is commonly used to adapt the equalizer coefficients because of its simplicity. The LMS algorithm, in its stochastic gradient form, adapts the coefficients using fixed small step size which makes the algorithm too slow for appropriate coping with the rapid changes in the channel impulse response.

In this thesis, we introduce a new adaptation algorithm, where the coefficients are adapted using a new cost function which adjust the step size automatically to speed up the convergence. The choice of the cost function is derived from a simple geometrical interpretation of the equalization process under multipath propagation conditions.

The new algorithm was evaluated using a Decision Feedback Equalizer (DFE) structure and simulated wireless channel conditions. Mathematical formula were developed to predict the performance of the equalizer under the new adaptation algorithm. The analysis and the simulation studies show good agreement.