Communication Relationship Discovery Based on Spectrum Data

With the continuous advancement of communication technology, accurately and efficiently discovering the communication relationships between communication emitters in large-scale, unevenly distributed data sets has become an important direction of current research. To address this issue, this paper proposes an adaptive clustering method called Balanced Iterative Re-ducing and Clustering Using Hierarchies- Density-Based Spatial Clustering of Applications with Noise (BIRCH-DBSCAN) communication relationship discovery algorithm. First, the BIRCH algorithm is used to compress the large-scale spectrum data set with uneven density distribution. Next, the compressed data set is constructed into an M-nearest neighbor directed graph, and the clustering parameters for the DBSCAN algorithm are adaptively determined using the K-Nearest Neighbor (KNN) algorithm. Subsequently, by combining breadth-first search with matrix algebra operations, the strongly connected components in the directed graph are searched to identify the core set of classes. Finally, by matching the clustering sets, the communication relationships between transmitters are identified. Simulation results show that, compared to existing algorithms, the proposed method demonstrates superior performance in both discovery accuracy and time efficiency.