Multiple Correlated Attributes Based Physical Layer Authentication in Wireless Networks

Physical (PHY) layer authentication has been a significant trend towards ensuring the identity security of terminals in wireless networks due to the high security and low complexity. However, the independence assumption of existing literature ignores the inherent correlation of the PHY-layer attributes, which limits its generality. In this paper, we propose a multi-attribute-based PHY-layer authentication scheme by taking the correlation into account. To cope with the exponential growth of computational complexity in correlated analysis, this paper studies the reconstruction and heuristic algorithm to find a suboptimal authentication solution with low complexity. Specific to the inherent volatility nature of the PHY-layer attributes, we propose an unsupervised machine learning (ML) based non-parametric clustering algorithm to enhance the reliability of PHY-layer authentication. Unlike existing PHY-layer authentication schemes based on ML, the proposed PHY-layer authentication scheme does not require any prior information or the training set, which has a more potent universality. Extensive simulations are performed under both synthetic and real data sets, and the figures verify that the proposed authentication scheme can achieve a reliable and robust performance with low complexity.