Deep neural networks are recognized as a promising approach for solving non-convex problems related to resource allocation in wireless communication systems. This letter introduces a novel neural-network based solution that transforms a probability simplex to implement the feasible region of an optimization problem with polytope constraints described by non-negative and monotone matrices. We utilize the proposed solution for optimizing the power allocation of multiple base stations serving multiple users simultaneously in the presence of inter-cell interference while guaranteeing the individual users’ data-rate constraints. Simulation results demonstrate that the proposed scheme significantly outperforms the existing state-of-the-art solutions in terms of the network average sum rate, while guaranteeing meeting the users’ data-rate constraints.

中文翻译：

---

基于单纯形变换的深度无监督学习优化：多用户干扰信道中具有 QoS 约束的功率控制


深度神经网络被认为是解决无线通信系统中与资源分配相关的非凸问题的有前途的方法。这封信介绍了一种新颖的基于神经网络的解决方案，该解决方案转换概率单纯形以实现具有由非负单调矩阵描述的多面体约束的优化问题的可行区域。我们利用所提出的解决方案来优化在存在小区间干扰的情况下同时服务多个用户的多个基站的功率分配，同时保证各个用户的数据速率限制。仿真结果表明，所提出的方案在网络平均总速率方面显着优于现有的最先进的解决方案，同时保证满足用户的数据速率约束。