Secure Internet-of-Things Communications: Joint Precoding and Power Control

In this paper, we consider a downlink internet-of-things (IoT) multiple-input multiple-output (MIMO) network wherein an access point (AP), multiple IoT users, and a single eavesdropper coexist. The eavesdropper attempts to wiretap confidential messages of the IoT users. In the considered system, we solve a sum secrecy rate maximization problem in the finite blocklength (FBL) regime. Due to the FBL, the secrecy rate has a back-off factor with respect to blocklength, decoding error probability, and information leakage, which makes the problem more challenging. The main challenges are: i) the problem is not tractable because of the back-off factor, ii) an objective function is inherently non-convex, and iii) information leakage by the eavesdropper needs to be considered. To address these difficulties, we first obtain a lower bound of the secrecy rate and transform the problem into a product of Rayleigh quotients form. Then, we derive a first-order Karush-Kuhn-Tucker (KKT) condition to find a local optimal solution and interpret the condition as a generalized eigenvalue problem. Consequently, we develop a low-complexity algorithm by adopting a generalized power iteration-based (GPI) method. Via simulations, we validate the secrecy rate performance of the proposed method for the short-packet IoT communication systems.