Adaptive power optimization in IRS-assisted hybrid OFDMA-NOMA cognitive radio networks with dynamic TDMA slot allocation

Abstract

The large-scale advancement of beyond-fifth-generation (B5G) wireless networks cannot be achieved without addressing the unprecedented requirements of IoT networks, such as massive connectivity, spectrum efficiency, and energy efficiency. Accordingly, integrating non-orthogonal multiple access (NOMA) with cognitive radio (CR) has been identified as a potential solution for B5G due to its ability to support massive number of IoT devices while improving the spectrum utilization. In particular, CR networks (CRNs) permit spectrum sharing by allowing a set of secondary users to under-utilize the available spectrum without interfering with primary users (i.e., licensed users), which improves spectral efficiency. Furthermore, unlike orthogonal multiple access (OMA), NOMA can serve more than one user at each orthogonal resource block (i.e., time or frequency) through power-domain multiplexing, which supports the massive connectivity requirements of B5G networks. Incorporating intelligent-reflecting surfaces (IRS) into NOMA-enabled CRNs can improve coverage, data rates, and power efficiency, especially when CR users lack direct line-of-sight to base stations. However, this IRS-assisted NOMA CRN system cannot be fully exploited without an efficient power-allocation framework that reduces power consumption while adhering to IRS, CR, NOMA, and quality of service (QoS) constraints. This paper introduces an IRS-assisted OMA-NOMA power allocation framework for CRNs that utilizes time and frequency domains with NOMA and IRS to serve more CR users with minimal power by optimizing power allocation and IRS reflection coefficients. The proposed framework dynamically divides every idle channel into time slots, creating adaptive frequency–time resource blocks (RBs) to accommodate more users using power-domain NOMA. The power-minimization problem over these adaptive RBs, considering IRS, CR, NOMA, and QoS constraints, is formulated as a non-convex optimization problem. An iterative approach is applied to convert the problem into a solvable convex optimization. Simulation results demonstrate that the proposed framework significantly outperforms traditional IRS-based approaches across multiple metrics.