The Research Of Radar Signal Sorting Algorithm Based On Grid-density Clustering

The radar signal sorting is the must have function of the modern radar reconnaissanceequipment, which is playing an important role in the ECM and ECCM. With thedevelopment of the radar counterreconnaissance technology and the constant increase of theradar radiation source signals, the number of radar signals interception of electronicreconnaissance equipment is huge and complex style. At the same time, the radar signal ofthe special system is increasing and the signal environment of different density of mixedoverlapping, which makes the task of radar signal sorting become very difficult. Theconventional radar signal sorting algorithm is difficult to meet the requirements of signalsorting today, so it is necessary to study the new pattern algorithm for radar signal sorting.Firstly, the paper introduces some classical clustering algorithms, and analyzes theiradvantages and disadvantages. By comparing the performances can be known thatgrid-density clustering algorithm is very suitable for radar signal sorting. Therefore, in orderto solve the problems that the fixed grid-density clustering has the low speed and the clusterresults are imprecision, the paper proposes a radar signal sorting algorithm based ondynamic grid density clustering. In addition, in view of the conventional density-gridclustering algorithm can’t process the multi-density radar signal sorting, the paper proposesa radar signal sorting algorithm based on multi-density of dynamic grid.The algorithm based on dynamic grid-density clustering which uses dynamic-creatinggrids technique to reduce greatly the number of grid partitioned in data space, adoptsshifting grids to eliminate the data input reliance and uses improved double-densitythreshold to improve the clustering precision. The algorithm can discover cluster of variousshapes without the input parameters. The simulation results show that the algorithm caneliminate noises effectively and sort the unknown radar signal rapidly and successfully.The algorithm based on multi-density clustering of dynamic grid which proposed theconcept of grid density-reachable on the basis of the density-reachable idea in the densityclustering, and used dynamic-creating grids technique and boundary point extraction forsolving above problems; In addition, a clustering that used breadth-first strategy searched itsgrid density-reachable cells to form a cluster without the density threshold. The simulationresults show that the algorithm can wipe off the noise effectively and sort the multi-densityradar signals successfully and rapidly.