Novel Adaptive and Iterative Multi-density DBSCAN Method for Space Signal Sorting in Complex Electromagnetic Environment

Due to the exponential growth of wireless devices, recent years have witnessed a dramatic increase in the density and complexity of the space electromagnetic environment. As a crucial task in the field of space cognitive spectrum sensing, the emitter signal sorting (ESS) task becomes increasingly challenging in such complex electromagnetic environments. Especially when the pulse description word (PDW) dataset of the multiple received emitters exhibits extremely different density level, the existing clustering based ESS methods fail to acquire the satisfactory performance due to its single pair of adjustable parameters. To overcome this difficulty, this work formulates the ESS task into a multi-density clustering problem for unevenly-distributed dataset. By exploiting the idea of density-based spatial clustering of applications with noise (DBSCAN), we propose a novel adaptive and iterative multidensity (AIMD)-DBSCAN based ESS method, which is able to automatically determine a series of key parameter pairs corresponding to different density level and perform a specially-designed iterative DBSCAN algorithm to find new clusters or merge to existing ones in descending density order. Simulation results have demonstrated that the proposed method can achieve better performance than its counterparts in terms of clustering accuracy and normalized mutual information (NMI) in different scenarios.