UAV-assisted ISAC network physical layer security based on Stackelberg game

This paper deals with the physical layer security for multi-unmanned aerial vehicle (UAV) assisted transmission of integrated sensing and communication (ISAC) signals from a dual-function base station (BS) to the ground user. Here, an active eavesdropper (Eve) on the ground eavesdrops on the information transmitted by the UAV network to the user. We propose a Stackelberg game model for the game problem of information security transmission between UAV and Eve, with perfect or imperfect channel state information (CSI) and cost information. To deal with the Stackelberg game problem, the dual optimization theory and KKT conditions are used to solve it, then we prove the existence and uniqueness of Stackelberg equilibrium (SE). At the same time, considering that the influence of ISAC signal's sensing of Eve's location on the distance between Eve and the user, and thus on the utility function of both sides of the game, this paper uses the Cramer-Rao bound (CRB) to represent the accurate performance of location sensing. Then the relaxation method is used for the non-convex optimization problem, and convex optimization tools can be used subsequently. Finally we notice that both operations have their own optimal resolution strategy, it is difficult to reach agreement, therefore an overall algorithm is proposed to solve the whole problem iteratively. In the simulation results, the effect of power on game and location sensing, and the effect of the distance between Eve and the user on the utility function of both sides of the game are given. In addition, we give the relationship between the CRB threshold and the transmit power, and verify the convergence of the game and overall algorithm.

Automated summary

This paper addresses the issue of physical layer security in a UAV network and proposes a Stackelberg game model to solve the game problem of information security transmission. The dual optimization theory and KKT conditions are employed to solve the problem. The research demonstrates that the accuracy of location sensing can be effectively measured using the CRB bound while considering Eve's location awareness. The relaxation method and convex optimization tools are used to solve the non-convex optimization problem.