基于多时隙业务联合整形的低能耗资源调度方法

摘要：面向未来6G移动通信系统超高数据密度的业务需求场景，为保障用户服务质量（QoS）并降低系统能耗，首先分析了移动通信系统的能耗构成，发现了系统能耗的非线性特征。然后在此基础上，设计了多时隙业务联合整形的低能耗资源调度方法。该方法通过感知用户业务流量和时延要求等需求侧的数据特征，利用深度强化学习算法在给定的多个时隙内动态调整基站资源分配策略。该资源分配策略降低了用户业务请求的非平稳性，从而减少了基站的非线性传输特性产生的额外能耗。最后通过软件仿真对比不同方法，验证了基于多时隙业务联合整形的理论和算法的正确性和有效性。

关键词：系统能耗的非线性特征；多时隙业务联合整形；低能耗资源调度；深度强化学习

Abstract: Facing the business requirements scenario of ultra-high data density of future 6G mobile communication systems, to ensure user quality of service (QoS) and reduce system energy consumption, the energy consumption composition of mobile communication systems is analyzed and the nonlinear characteristics of system energy consumption is studied. On this basis, a low-energy resource allocation method for multi-slot traffic joint shaping is designed. By sensing the data characteristics of the demand side such as user service traffic and delay requirements, this method uses the deep reinforcement learning algorithm to dynamically adjust the base station resource allocation strategy within a given multiple time slots. The resource allocation strategy reduces the non-stationarity of user service requests, thereby reducing the additional energy consumption caused by the nonlinear transmission characteristics of the base station. Finally, by comparing different methods through software simulation, the correctness and effectiveness of the theory and algorithm based on multi-time slot business joint shaping are verified.

Keywords: nonlinear characteristics of system energy consumption; multi-slot traffic joint shaping; low energy resource allocation; deep reinforcement learning