ZERO TRACKING FOR SEPARATING SOURCES AND ALLEVIATING INCIDENTAL CANCELLATION IN TWO-DIMENSIONAL ARRAYS

Abstract

This thesis presents two algorithms for separating and tracking desired signal with the purpose of enhancing the system's SNR improvement. Both algorithms are proposed with particular array structures to achieve certain goals. The algorithm in Chapter 3 suggests the use of a preprocessor whose mechanism is the riveted part of the whole array structure in source separation and tracking. Chapter 4 describes the two-dimensional array which successfully relates the change of a weight to the change of system zeros with equations. Results of extensive simulations were obtained by passing the combination of noise and signal through one particular array, using algorithm derived and getting the output of the array. To make it converge, principles of stochastic process are applied to minimize the system output. The simulation results of new algorithms are compared with traditional methods of LMS to show the robust aspect of the new algorithms in changing environment while having almost the same computational complexity. In the last four figures of Chapter 4, the simulation results show that the SNR in the LMS algorithm degrades significantly when the source signal changes its direction but the new algorithm maintains a rather high level of SNR under the same situation.