Adaptive detection and estimation for an unknown occurring interval signal in correlated Gaussian noise

This paper considers the problem of detecting and estimating an unknown occurring interval signal in correlated Gaussian noise, which is often arisen in signal processing society, e.g., identifying the onset times of a seismic wave and detecting a distributed target in unknown occurring range cells. We propose the novel Generalized Likelihood Ratio Test (GLRT) algorithm, where the Maximum Likelihood Estimations (MLEs) of the unknown occurring interval are obtained through a Dynamic Programming (DP) method adaptively without the secondary data. Unlike the classic Sequential Probability Ratio Test (SPRT) methods which consist of an on-line detector before an off-line estimator, the proposed GLRT outputs the decision result and the estimations at the same time. The performances of the proposed algorithm are evaluated by numerical simulations as well as the applications of detecting and suppressing the transient interference in a radar operated on High-Frequency (HF) band with real data.