Max–Min Rate Optimization for STAR‐RIS‐Assisted SWIPT MIMO Cognitive Radio Systems

ABSTRACT This paper studies a downlink multiple‐input multiple‐output (MIMO) simultaneous wireless information and power transfer (SWIPT) cognitive radio (CR) system with the assistance of a simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR‐RIS). Different from conventional RISs which can only serve users in their reflection region, STAR‐RISs using energy splitting (ES) techniques simultaneously support both reflection users (RUs) and transmission users (TUs). In our proposed system, the RUs utilize power splitting (PS) techniques to simultaneously harvest energy and detect information while TUs exclusively receive information. Our aim is to jointly optimize the precoding matrices (PCMs) at the secondary transmitter (S‐Tx), the transmission and reflection coefficient matrices (TRCMs) at the STAR‐RIS, and the PS ratios at the RUs to maximize the minimum achievable data rate among TUs while ensuring the minimum harvested energy and data rate requirements for RUs, and maintaining interference power (IP) at primary users (PUs) below a predefined threshold. To address the nonconvex nature of the design problem, we divide it into two sub‐problems. In each sub‐problem, we employ the minorization–maximization (MM) method to identify suitable surrogate functions and establish convex inner sets and, thus, transform the original sub‐problems into convex ones. Subsequently, by capitalizing on an alternating optimization (AO) framework, we develop a computationally efficient iterative algorithm to find the optimal solutions. Through extensive simulations, we demonstrate that the MIMO SWIPT CR systems with the optimal TRCM STAR‐RIS provide the superior performance as compared with the counterparts with the fixed TRCM STAR‐RIS, those without the STAR‐RIS, and those with the optimal phase shift conventional RIS.