信息年龄和能耗联合优化的无线体域网卸载和调度策略
Joint optimization strategy of age of information and energy consumption for offloading and scheduling in WBAN
张政 1谢鑫 2柏桐 3林金朝 4李章勇5
作者信息
- 1. 重庆邮电大学通信与信息工程学院,重庆 400065;重庆邮电大学光电信息感测与微系统重庆市重点实验室;重庆 400065
- 2. 重庆邮电大学自动化学院,重庆 400065
- 3. 重庆邮电大学光电信息感测与微系统重庆市重点实验室;重庆 400065;重庆邮电大学光电工程学院,重庆 400065
- 4. 重庆邮电大学光电信息感测与微系统重庆市重点实验室;重庆 400065
- 5. 重庆邮电大学光电工程学院,重庆 400065
- 折叠
摘要
在无线体域网(WBAN)中,为保障生理数据的实时传输并降低节点能耗,提出了一种联合优化的卸载与调度策略.该策略通过选择在传感节点或Sink处理数据,最小化信息年龄(AoI)与能耗的加权和.为解决卸载和调度决策的强耦合问题,采用双层马尔可夫决策过程(MDP),并通过深度强化学习(DRL)应对维数灾难.仿真结果表明,DRL策略在性能上接近MDP,且在节点增加时,相较于RRG与EG策略,平均加权和分别降低约3.58%和24.9%,收敛速度约为MDP的2倍.
Abstract
Both real-time physiological data transmission and reduced energy consumption are critical to wireless body area network(WBAN).A joint optimization strategy for offloading and scheduling was proposed to minimize the weighted sum of age of information(AoI)and energy consumption by determining whether data should be processed at sensor nodes or the Sink.To handle the strong coupling between offloading and scheduling decisions,a two-layer Mar-kov decision process(MDP)was used to approximate the optimal solution.A deep reinforcement learning(DRL)ap-proach was introduced to address the dimensionality issue.Simulations show that the DRL strategy performs comparably to the MDP under various weight factors and frame lengths.Furthermore,as the number of sensor nodes increases,the DRL strategy reduces the weighted sum by 3.58%and 24.9%compared to RRG and EG strategies,respectively,and con-verges twice as fast as the MDP.
关键词
无线体域网/卸载调度/信息年龄/能耗Key words
wireless body area network/offloading and scheduling/age of information/energy consumption引用本文复制引用
基金项目
国家自然科学基金资助项目(U21A20447)
国家自然科学基金资助项目(62171073)
国家自然科学基金资助项目(62311530103)
重庆市自然科学基金资助项目(CSTB2022NSCQ-MSX1523)
重庆邮电大学博士研究生创新人才基金资助项目(BYJS202206)
出版年
2024
国家科技期刊平台
NSTL
万方数据