A Unified Propagation Graph Theory with Bayesian Optimization for ISAC Channel Modeling

Channel modeling based on sensing data is one of the key research fields in Integrated Sensing and Communication (ISAC). Propagation graph theory, as a channel modeling approach based on directed graph topologies, distinguishes itself with high computational efficiency and precise characterization of multipath channel. In this paper, a unified propagation graph modeling method based on measured data calibration is proposed, providing a novel approach for channel reconstruction in ISAC after sensing. Furthermore, We develop a unified graph modeling method integrated with the Directive Model and propose a hyperparameter calibration framework based on Bayesian optimization. This approach enables accurate channel reconstruction modeling in multipath-rich scenarios with a limited amount of measurement data. Through simulation analysis and measurement comparison, the feasibility and accuracy of the proposed method have been verified.