Most signal processing systems today need to estimate parameters of the underlyingprobability distribution, however quantifying the robustness of this system hasalways been difficult. This thesis attempts to quantify the performance and robustnessof the Maximum Likelihood Estimator (MLE), and a robust estimator, whichis a Huber-type censored form of the MLE. This is possible using diff erential geometricconcepts of slope. We compare the performance and robustness of the robustestimator, and its behaviour as compared to the MLE. Various nominal values ofthe parameters are assumed, and the performance and robustness plots are plotted.The results showed that the robustness was high for high values of censoring andwas lower as the censoring value decreased. This choice of the censoring value wassimplifi ed since there was an optimum value found for every set of parameters. Thisstudy helps in future studies which require quantifying robustness for di fferent kindsof estimators.
展开▼
