An optimal energy switching-based eavesdropping attacks for anti-user detection in networked multi-sensor fusion systems

This article discusses the optimal design of eavesdropping schemes in networked multi-sensor fusion systems (NMFSs) with energy constraints. Multiple sensors observe the state of the process and transmit the processed data to the remote user fusion center equipped with a detector via wireless channels, under an intelligent eavesdropper with dual attack capabilities of the passive monitoring and active jamming. To tackle the energy supply issue related to eavesdropping attacks in certain scenarios, the eavesdropper adopts energy switching scheduling. Therefore, this article aims at designing an attack strategy to improve the eavesdropping performance while reducing the user estimation performance under energy constraints. The eavesdropper firstly selects the jamming signal power based on the given threshold. Then, the initial problem is transformed into an unconstrained Markov decision process (MDP) by introducing Lagrange multipliers. Finally, sufficient conditions are provided to evade the user detection. The results indicate that the optimal eavesdropping strategy exhibits threshold-type structures. The above conclusion is supported by numerical examples.