Detection and estimation a scattering signal with unknown location and length in multiple measurement

This paper considers the problem of detecting and estimating a scattering signal with unknown location and length in multiple measurement, which is often arisen in signal processing society, e.g., identifying a seismic wave measured by the observation stations around the earthquake hypocenter and detecting a distributed target in unknown occurring range cells with High Resolution (HR) Multiple-Input Multiple-Output (MIMO) radars. We propose the novel algorithm, which is consisted of an recursive signal energy accumulated process and a novel Generalized Likelihood Ratio Test (GLRT). Unlike the classic method which detects the signal and estimate it´s location and length in each measurement independently, the proposed algorithm makes it equivalent to a random walking process, then estimates the moving velocity during the recursive process and obtains the Maximum Likelihood Estimations (MLEs) of the unknown signal length through a Dynamic Programming (DP) method. Meanwhile, the decision result and the estimations output at the same time. The performances of the proposed algorithm are evaluated by numerical simulations.