Abstract: Probabilistic data association is an important issue in multi-source information fusion algorithms. The data association problem based on radar and electronic support measurement (ESM) sensors is mainly discussed in this paper. As the azimuthal data of radar and ESM sensors have approximately the same distribution, the discriminant function can be obtained through the analysis of ESM data, and the corresponding decision rules can be used to associate the dots and tracks. The association issue can be essentially regarded as a pattern recognition problem. In this paper, considering the good performance of support vector machine(SVM) in pattern classification, we establish a dotting and tracking association model for radar ESM systems based on SVM algorithm. We train the SVM model with ESM data, and classify the radar data to acquire association result. Finally, the simulation results show that the association accuracy can be effectively improved compared with the classical multiple hypothesis tracking algorithm.

Key words: radar; electronic support measurement (ESM); tracking association; support vector machine (SVM)