无线网络突发流量数据分区调度方法  

A Partition Scheduling Method for Bursty Traffic Data in Wireless Networks

作　　者：余华东 张步忠 濮光宁[1] YU Huadong;ZHANG Buzhong;PU Guangning(Information Engineering College,Anhui Finance&Trade Vocational College,Hefei 230601,China;School of Computer Science and Information,Anqing Normal University,Anqing Anhui 246001,Anhui)

机构地区：[1]安徽财贸职业学院信息工程学院,合肥230601 [2]安庆师范大学计算机与信息学院,安徽安庆246001

出　　处：《海南热带海洋学院学报》2024年第5期80-87,共8页Journal of Hainan Tropical Ocean University

基　　金：安徽省质量工程项目(2023jyxm1548);安徽省高校自然科学研究项目(2023AH052070);安徽省高校自然科学研究重点项目(2022AH052536)。

摘　　要：针对无线网络中突发流量数据导致网络存在传输时延和拥塞问题,提出基于多核神经网络的无线网络突发流量数据分区调度方法。首先,利用空闲时槽侦听的无线网络分区方法,通过计算节点分布密度和节点流量容量,判断无线网络中节点处于稀疏区域还是密集区域。其次,利用多核极限学习机,对无线网络稀疏和密集两个区域的突发流量数据进行预测。对于稀疏区域的突发流量数据,直接通过CSMA/CA协议进行调度;对于密集区域的突发流量数据,通过统计方法计算出突发流量数据利用率以及可利用率,再对各信道数据进行归一化处理,获取突发流量数据的转发概率。再次,依据突发流量数据的转发概率进行节点发送,并通过设置信道的队列长度阈值来发送或丢弃突发流量数据,实现无线网络突发流量数据分区调度。实验结果表明,该方法能够灵活地调度无线网络突发流量数据,提升其吞吐量,使数据丢弃率始终低于0.03%、平均传输时延始终低于0.1 s,可准确预测无线网络突发流量。Aiming at the problems of transmission delay and congestion caused by bursty traffic data in wireless networks,a parti‐tion scheduling method based on multi‐core neural networks for bursty traffic data in wireless networks was proposed.Firstly,the wire‐less network partitioning method of idle time slot listening was utilized to determine whether the nodes in the wireless network are in sparse or dense areas by calculating the node distribution density and node traffic capacity.Secondly,multi kernel extreme learning machine was utilized to predict the bursty traffic data in sparse and dense regions of the wireless network.For bursty traffic data in sparse areas,scheduling was directly carried out through CSMA/CA protocol;for sudden traffic data in dense areas,the statistical method was used to calculate the utilization and availability of sudden traffic data,and then the data of each channel was normalized to obtain the forwarding probability of sudden traffic data.Again,the nodes were sent based on the forwarding probability of sudden traffic data,and sudden traffic data was sent or discarded by setting a queue length threshold for the channel,to realize the partition schedul‐ing of sudden traffic data of the wireless network.The experimental results demonstrated that the current method can flexibly schedule the wireless network bursty traffic data,improve its throughput,and ensure that the data dropout rate is always below 0.03%,the aver‐age transmission delay is always below 0.1 second and the sudden traffic in the wireless networks can be accurately predicted.

关 键 词：无线网络 RBF核函数 多核神经网络 突发流量数据 分区调度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]