This paper investigates the resource allocation problem in non-orthogonal multiple-access (NOMA) cellular networks underlaid with OMA-based device-to-device (D2D) communication. This network architecture enjoys the intrinsic features of NOMA and D2D communications; namely, spectral efficiency, massive connectivity, and low-latency. Despite these indispensable features, the combination of NOMA and D2D communications exacerbates the resource allocation problem in cellular networks due to the tight coupling among their constraints and conflict over access to shared resources. The aim of our work is to maximize the downlink network sum-rate, while meeting the minimum rate requirements of the cellular tier and underlay D2D communication, and incorporating interference management as well as other practical constraints. To this end, many-to-many matching and difference-of-convex programming are employed to develop a holistic sub-channels and power allocation algorithmic solution. In addition to analyzing the properties of the proposed solution, its performance is benchmarked against an existing solution and the traditional OMA-based algorithm. The proposed solution demonstrates superiority in terms of network sum-rate, users’ connectivity, minimum rate satisfaction, fairness, and interference management, while maintaining acceptable computational complexity.

本文研究了基于 OMA 的设备到设备 (D2D) 通信的非正交多址 (NOMA) 蜂窝网络中的资源分配问题。这种网络架构享有 NOMA 和 D2D 通信的内在特征；即频谱效率、大规模连接和低延迟。尽管有这些不可或缺的特性，但 NOMA 和 D2D 通信的结合加剧了蜂窝网络中的资源分配问题，因为它们之间的约束和访问共享资源的冲突之间存在紧密耦合。我们工作的目标是最大化下行链路网络总速率，同时满足蜂窝层和底层 D2D 通信的最低速率要求，并结合干扰管理以及其他实际约束。为此，采用多对多匹配和凸差编程来开发整体子信道和功率分配算法解决方案。除了分析所提出解决方案的特性外，还针对现有解决方案和传统的基于 OMA 的算法对其性能进行了基准测试。所提出的解决方案在网络和速率、用户连接性、最小速率满意度、公平性和干扰管理方面展示了优越性，同时保持了可接受的计算复杂性。