Communication Anti-jamming Method Based on POMDP Model under Limited State Information Conditions

In circumstances where hardware constraints restrict the communication system's ability to gather comprehensive channel state information, this paper introduces an anti-jamming method for communication, founded on the Partially Observable Markov Decision Process (POMDP) model. This method follows a sequential process: first, it acquires observation results from selected channels during the current time slot and the belief state pertaining to the target frequency band from the previous time slot. Subsequently, it updates the belief state for the current time slot. Finally, using heuristic algorithms based on value functions to make decisions based on the belief state of the current time slot. Simulation results demonstrate that the proposed method consistently attains a high level of accuracy in communication decision-making when countering tracking jamming, linear sweep jamming, and combined tracking and linear sweep jamming. Remarkably, these outcomes are achieved exclusively with the limited communication channel state information obtained.