Abstract:
Detection of low power signals in the presence of high power interferers is a common
problem in spatial signal processing. Notch depth (ND) is defined as the response of the
beamformer in the interferer direction when the beamformer is steered towards a specified
look direction. This thesis analyzes the ND of the constrained LeastMean Squares algorithm
proposed by Frost [1]. Several variants of the LMS algorithm are considered, and the
algorithm is analyzed for the case of single and multiple interferers. The thesis compares
the ND of the LMS beamformer to the ND of the Dominant Mode Rejection beamformer
proposed by Abraham and Owsley [2]. The performance comparison indicates that DMR
attains a deeper notch faster than LMS. The white noise gain of the two beamformers
is approximately the same. Analysis of the computational complexity of the LMS and
DMR algorithms indicates that DMR requires on the order of N times more floating point
operations than LMS, where N is the size of the receiving array. Thus, DMR is a better
choice for applications requiring fast convergence as long as the processor can handle the
increased computational load.
