Abstract
In this thesis, we consider the problem of detecting an identified signal corrupted by Laplace noise. The noise is zero mean, additive and independent and identically distributed, with imperfectly known standard deviation. We have compared the Maximum Likelihood detector, which uses the Maximum Likelihood parameter estimation algorithm to estimate the standard deviation and detect the signal, with the Neyman-Pearson and sign detectors. The comparison of their performance has been made in terms of the detection probability and the false alarm rate, and the trade-off between these two parameters has been studied. The detectors have been compared for various signal and noise levels as well as the number of samples. We show that the Maximum Likelihood detector is most appropriate for detecting weak signals in applications that require a stable false alarm rate.
Valangiman Raman, Sathya Narayanan (2002). Performance analysis of the Maximum Likelihood detector for nominally Laplace noise. Master's thesis, Texas A&M University. Available electronically from
https : / /hdl .handle .net /1969 .1 /ETD -TAMU -2002 -THESIS -V33.