基于Q-Learning的动态BLE Mesh网络高能效路由算法  

Energy-efficient Routing Algorithm for Dynamic BLE Mesh Network Based on Q-Learning

作　　者：蒋龙荣 刘静 刘文超 王兰 JIANG Longrong;LIU Jing;LIU Wenchao;WANG Lan(College of Electronics and Information Engineering,Shenzhen University,Shenzhen 518060,China;Research Institute of Tsinghua University in Shenzhen,Shenzhen 518057,China;E-surfing Digital Life Technology Co.,Ltd.,Shanghai 200072,China)

机构地区：[1]深圳大学电子与信息工程学院,广东深圳518060 [2]深圳清华大学研究院,广东深圳518057 [3]天翼数字生活科技有限公司,上海200072

出　　处：《无线电通信技术》2024年第4期664-673,共10页Radio Communications Technology

摘　　要：针对动态低功耗蓝牙(Bluetooth Low Energy, BLE)Mesh网络规范采用的管理式泛洪路由机制所导致的数据包冗余和高能耗的问题,提出了动态BLE Mesh网络高能效路由算法。通过建立基于Q-Learning的BLE Mesh网络路由模型,将BLE Mesh网络中节点的剩余能量、转发成本、移动因子以及接收信号强度指示(Received Signal Strength Indicator, RSSI)纳入到奖励函数的设计中。同时对BLE Mesh网络中的心跳机制进行改进,使网络中的节点更加高效地获取环境信息。采用基于探索的路由发现机制来更新网络中节点的Q值,使其能更加准确反映节点移动所导致的网络拓扑变化。仿真结果表明,该算法在节点能耗和网络开销上均优于传统的管理式泛洪路由机制。To address issues of packet redundancy and high energy consumption associated with the managed flooding routing mechanism used in the Bluetooth Low Energy(BLE)Mesh network specifications,an energy-efficient routing algorithm for dynamic BLE Mesh network is proposed.By establishing a Q-Learning based routing model for BLE Mesh network,this algorithm incorporates factors such as residual energy of nodes,forwarding costs,mobility factors,and Received Signal Strength Indicator(RSSI)into the design of the reward function.Improvements to the heartbeat mechanism in the BLE Mesh network enable nodes to more effectively gather environmental information.An exploration-based routing discovery mechanism is utilized to update the Q-values of the nodes within the network,thereby more accurately reflecting changes in network topology caused by node movements.Simulation results demonstrate that this algorithm outperforms traditional managed flooding routing mechanism in terms of node energy consumption and overall network overhead.

关 键 词：低功耗蓝牙 MESH网络 物联网应用 Q-LEARNING 

分 类 号：TN92[电子电信—通信与信息系统]