Secrecy-Fair Resource Allocation and Decoding Design for IRS-Assisted Untrusted NOMA Systems

Abstract

In this paper, we investigate an intelligent reflecting surface (IRS)-assisted secrecy-fair non-orthogonal multiple access (NOMA) framework in the presence of untrusted users, where legitimate users may act as potential eavesdroppers. For a twouser untrusted NOMA system, we first show that with the conventional NOMA decoding order, a positive secrecy rate for the far user cannot be achieved under untrusted settings. To overcome this limitation, we propose a revised secure decoding order and derive corresponding power allocation conditions under which both users can achieve positive secrecy rates, thereby enabling secrecy-fairness. Building on this proposed decoding strategy, we formulate a secrecy-fairness maximization problem that jointly optimizes transmit power allocation and IRS phase shifts to ensure balanced security performance between users. The resulting problem is non-convex due to variable coupling and unit-modulus IRS constraints. To efficiently solve it, we develop a framework combining semidefinite relaxation, closedform power allocation, and alternating optimization. Numerical results validate the analytical findings and highlight the impact of key parameters on system performance. In particular, unlike conventional NOMA, allocating more power to the far user is not always optimal from a secrecy-fairness perspective in untrusted settings. The results further show that the proposed IRS-assisted optimal design achieves significant secrecy-fairness gains over different benchmarks.