In this paper, we explore radio resource management for a time-varying 5G heterogeneous wireless access network that includes multi-RATs such as 5G new radio (NR) and long-term evolution (LTE). To cope with the practical challenges of a centralized approach such as signalling overhead and computational complexity, we decomposed the process of radio resource management into three parts, 1) RAT selection, 2) optimal radio resource allocation, and 3) congestion control. RAT selection is performed by each user device with network assistance, whereas the problem of radio resource allocation and congestion control is formulated as a stochastic optimization problem. Maintaining network stability, the average throughput utility is maximized subject to admission control and resource allocation. By using Lyapunov optimization, this utility maximization problem is decomposed into two subproblems. Radio resource allocation policy implemented at the central controller node allocates resources at each time slot using the Lagrange dual method, whereas the process of congestion control is carried out at user end based on throughput adaptation according to its current channel conditions. The theoretical and simulation results evaluate the performance of our proposed approach under the assumption of network stability. Simulation results related to individual users throughput and queue length, and performance comparison of equal power and adaptive power allocation techniques, are presented to depict the effectiveness of our proposed scheme. Furthermore, our proposed RAT selection scheme performs better than the traditional centralized and distributive mechanisms.

中文翻译：

---

时变 5G 异构无线接入网络的混合无线电资源管理

在本文中，我们探索了时变 5G 异构无线接入网络的无线电资源管理，该网络包括多 RAT，例如 5G 新无线电 (NR) 和长期演进 (LTE)。为了应对集中式方法的实际挑战，例如信令开销和计算复杂性，我们将无线电资源管理过程分解为三个部分，1) RAT 选择，2) 最佳无线电资源分配，3) 拥塞控制。RAT 选择由每个用户设备在网络协助下执行，而无线电资源分配和拥塞控制问题被表述为随机优化问题。保持网络稳定性，平均吞吐量效用在准入控制和资源分配的条件下最大化。通过李雅普诺夫优化，这个效用最大化问题被分解为两个子问题。中央控制器节点实施的无线资源分配策略采用拉格朗日对偶法在每个时隙分配资源，而拥塞控制过程则在用户端根据其当前信道条件基于吞吐量自适应进行。理论和仿真结果在网络稳定性假设下评估了我们提出的方法的性能。提供了与单个用户吞吐量和队列长度相关的仿真结果，以及等功率和自适应功率分配技术的性能比较，以描述我们提出的方案的有效性。此外，我们提出的 RAT 选择方案比传统的集中式和分布式机制表现更好。