Design an optimum Hierarchical clustering algorithm for radar pulse clustering in ESM processor

The first step of an ESM system is extracting features of the received signal in a noisy environment. The next step is clustering and deinterleaving of radar pulses on the basis of data achieved from the previous procedure, and the final step is identifying the emitters and threat factors. Proper selection of initial centers for clusters is the fundamental problem in clustering methods. In the proposed algorithm, first a similarity matrix for the input data based on defined neighborhood radius is formed. Then, with analysis of the similarity matrix and selection of lines with maximum similar codes, denser clusters are separated sequentially. The structure of this algorithm is an optimum hierarchical clustering method. In this algorithm similarity criterion has a very important role. This technique reduces the iterations and computations considerably and separates data into a not predetermined number of clusters using neighborhood radius. The selection priority is with dense sets. One of the advantages of this algorithm compared to the algorithms based on k-mean is careful selection of the initial center of clusters. Results of the proposed method for a data sample consisting of 200 radar pulses is compared with the results of clustering around the leader which is one of the main clustering algorithms in the field of radar pulses and the desired advantage is achieved. In this way, regarding the high stream radar pulse and without iteration requirement, we have optimum pulse strings separation.