Improved Change Point Detection via Bias Reduction in Parameters Estimation

The recognition of unknown radar signals requires the implementation of detection algorithms in order to extract signals of interest. This paper presents an implementation of the Cumulative Sum (CuSum) algorithm for radar applications and the segmentation of unknown received signals. The study focuses on the parameters estimation of the signal distribution with an evaluation of the bias of several estimators along with the proposition of a correction factor to obtain a more robust estimator. Simulations and real data experiments validate the algorithm's capacity to retrieve both the onset and end of unknown waveforms.