Distributed Radar PRI Sequence Classification using K-medoids Algorithm and Feedforward Neural Networks

Pulse repetition interval (PRI) is an important feature of distributed radar signals. PRI feature extraction and PRI sequence classification are meaningful in modern electronic warfare (EW). This article recommends applying the k-medoids algorithm and feedforward neural networks (FNN) to classify PRI sequences. The proposed approach first calculates the first-order difference of PRI sequences, then uses the k-medoids algorithm to cluster the elements in each PRI sequence and its difference vector, and extracts the centroids, and feeds the centroids into the FNN classifier to categorize the sequences. Simulation results show the proposed approach achieves excellent performance and has strong robustness.