【摘要】 随着高比例新能源大规模并网，配电网分布式能源调控、用户信息采集等业务对通信时延、能效等提出更高要求。5G边缘计算高效赋能配电网电力数据传输与处理。然而，任务卸载优化仍面临多设备决策耦合、多服务质量(quality of service，QoS)性能指标耦合等挑战。面向配电网多业务差异化QoS需求，针对能效敏感型与时延敏感型业务分别建立能效与时延优化问题，提出基于差异化QoS驱动与配额升价匹配的任务卸载优化算法，利用Lyapunov优化将长期随机优化问题解耦为单时隙确定性问题，基于QoS性能偏差判定业务类型，并采用配额升价匹配实现不同业务优先级设备与服务器之间的多对一稳定匹配。仿真结果验证了所提算法在能效与时延差、业务优先级感知等方面的性能优势。更多还原

【Abstract】 With the large-scale integration of high-proportion new energy,the power distribution network services such as distributed energy regulation and user information collection pose higher requirements on the communication delay and energy efficiency.5G edge computing effectively empowers the data transmission and processing of the power distribution network.However,task offloading optimization is still facing several challenges such as multi-device decision coupling and multiple quality of service (QoS) metric coupling.Aiming at differentiated QoS demands of multiple services in the power distribution network,the energy efficiency optimization problem and delay optimization problem are respectively established for energy efficiency sensitive services and delay sensitive services and a task offloading algorithm based on differentiated QoS driven and quota price matching is proposed.Specifically,the long-term stochastic optimization problems are decoupled into single-slot deterministic problems,and the service type is judged based on QoS performance deviation.Quota price matching is leveraged to achieve the many-to-one stable matching between devices and edge computing servers.The simulation results prove the performance advantages of the proposed algorithm in the difference in energy efficiency and delay as well as service priority awareness.更多还原