Radar Detection Architecture Based on Interference Covariance Structure Classification

This paper proposes a new radar detection architecture, which is composed of an interference covariance structure classifier and a bank of adaptive radar detectors. The classifier is based on model order selection theory and the Bayesian information criterion to determine the covariance structure that is deemed to be suitable for a specific set of radar data [1]. This decision stage drives the choice of the radar detector within a specific class of adaptive detectors to establish the possible target presence. The critical issue concerning the constant false alarm rate behavior of the architecture is discussed and two techniques for the threshold setting process are proposed. Finally, the detection performance analysis, conducted on both simulated and measured data, shows that the proposed architecture can guarantee better performance than classic radar decision schemes in scenarios where the interference covariance exhibits structural properties.