Fig. 1

Multi-layer structure model of an underwater communication system block diagram. When the signal detection is required, the control center will issue an indication signal, and all members of the grading center will perform the signal detection operation synchronously. Each SU collects the compressed data, uses the Bayesian model to obtain the shared super-parameters, and transmits the results to the cognitive radio user base station to obtain the preliminary signal detection results. When members of the hierarchical center collect perceptual data, distributed computing is performed and the results are broadcast to the hierarchical control center. After a lot of information exchange, the frequency spectrum of the spectrum signal detection algorithm is proposed to be close to a fixed value, the next step is through the data fusion center on the individual level data fusion, the center of the implement multiple layers joint spectrum signal detection, this paper puts forward a kind of based on Dirichlet process (DP) hierarchical Bayesian model to achieve compression perception. The data are grouped and the shared super-parameters are mined in each group to obtain the optimal signal detection results