Secure IoU in Adversarial Environments: User Authentication and Performance Optimization

The emerging UAV network suffers from severe security risks from attackers. If the attackers sneak into an Internet of UAV (IoU) network, they would occupy the scarce resources which should be allocated among authenticated users, degrading the communication performance. Therefore, a two-stage scheme is proposed to both enhance IoU security and optimize network resource allocation. In the first stage, a distributed radio-frequency fingerprinting-based authentication model with consensus voting is designed to enhance the IoU security. Moreover, in the second stage, a performance optimization problem in consideration of authentication accuracy is formulated to maximize the data rate. It is proved that the problem can be transformed into a potential game, which implies that the optimization of the overall network's utility equals the optimization of each individual's utility. Also, the existence of this game's Nash equilibrium (NE) is proved and an iterative algorithm is used to achieve its NE. Finally, simulation results show that that our proposed consensus-based authentication model improves recognition accuracy by 40
$\%$
compared to the conventional Adaboost method under typical parameter settings, and our proposed algorithm can improve the network utility significantly. The effect of various parameters on the utility is also numerically verified.