In the Internet-of-Things (IoT), massive sensitive and confidential information is transmitted wirelessly, making security a serious concern. This is particularly true when technologies, such as non-orthogonal multiple access (NOMA), are used, making it possible for users to access each other's data. This paper studies secure communications in multiuser NOMA downlink systems, where each user is potentially an eavesdropper. Resource allocation is formulated to achieve the maximum sum secrecy rate, meanwhile satisfying the users' data requirements and power constraint. We solve this non-trivial, mixed-integer non-linear programming problem by decomposing it into power allocation with a closed-form solution, and user pairing obtained effectively using linear programming relaxation and barrier algorithm. These subproblems are solved iteratively until convergence, with the convergence rate rigorously analyzed. Simulations demonstrate that our approach outperforms its existing alternatives significantly in the sum secrecy rate and computational complexity.

翻译：在物联网（IoT）中，大量敏感和机密信息通过无线传输，使得安全性成为一个严峻的问题。当采用非正交多址接入（NOMA）等技术时尤其如此，这些技术使用户能够访问彼此的数据。本文研究了多用户NOMA下行链路系统中的安全通信，其中每个用户都可能是窃听者。我们通过资源分配来实现最大总和保密速率，同时满足用户的数据需求与功率约束。我们通过将该问题分解为具有闭式解的功率分配子问题以及利用线性规划松弛和障碍算法高效求解的用户配对子问题，解决了这个非平凡且混合整数非线性的规划问题。这些子问题迭代求解直至收敛，并严格分析了收敛速率。仿真结果表明，我们的方法在总和保密速率和计算复杂度方面显著优于现有替代方案。