摘要
在定位移动通信网络空洞节点时,易受路径选择阶段的局部最优问题影响,使定位效果难以达到理想状态.为此,提出基于深度强化学习的移动通信网络空洞节点智能定位方法.在确定移动通信网络中不同节点之间的位置信息阶段时,采用加权处理的方式对每跳距离信息进行量化,再结合加权平均跳距和节点与锚节点的距离,确定移动通信网络空洞节点定位步距.在空洞节点定位阶段引入深度强化学习中的灰狼算法,并从快速收敛的角度出发,利用模拟退火和混沌映射对原始的灰狼算法进行优化,将最终灰狼奔走围攻对象作为移动通信网络空洞节点的定位结果.测试结果表明,本文设计的定位方法能够降低空洞节点定位阶段的路径开销,在定位精度方面有良好表现,定位效率也明显提升.
When locating hollow nodes in mobile communication networks,it is difficult to achieve an ideal location effect due to the local optimal problem in the path selection stage.Therefore,an intelligent location method of hollow nodes in mobile communication networks based on deep reinforcement learning is proposed.In the stage of determining the location information between different nodes in the mobile communication network,the distance information of each hop is quantized by weighted processing,and then the positioning step of the hollow node in the mobile communication network is determined by combining the weighted average hop distance and the distance between the node and the anchor node.In the stage of location of hollow nodes,the grey wolf algorithm in deep reinforcement learning is introduced,and the original grey wolf algorithm is optimized by simulated annealing and chaotic mapping from the point of view of rapid convergence,and the final grey wolf running around is taken as the location result of hollow nodes in mobile communication network.The test results show that the design of location method can reduce the path overhead in the location stage of hollow nodes,and has a good performance in the location accuracy,and the specific location efficiency is also significantly improved.
作者
倪强
NI Qiang(Anhui Institute of Industrial Economics and Vocational Technology,Information Development Office,Hefei 230000,China)
出处
《常熟理工学院学报》
2024年第5期78-82,共5页
Journal of Changshu Institute of Technology
基金
安徽省高校自然科学研究项目“基于智能数据云平台的高校一体化平台设计应用探索”(2022AH052667)。
关键词
深度强化学习
移动通信网络空洞节点
智能定位
模拟退火
混沌映射
deep reinforcement learning
hollow nodes in mobile communication networks
intelligent positioning
simulated annealing
chaotic mapping
