公网数字集群系统的动态并发请求调度处理队列算法  

Dynamic concurrent request scheduling and processing queue algorithm for digital trunking system based on public network

作　　者：李耀升 孙昕[1] LI Yaosheng;SUN Xin(College of Electronic and Information Engineering,Beijing Jiaotong University,Beijing 100044,China)

机构地区：[1]北京交通大学电子信息工程学院,北京100044

出　　处：《计算机应用》2022年第S01期236-241,共6页journal of Computer Applications

基　　金：国家重点研发计划项目(2018YFB1601504)。

摘　　要：针对公网数字集群系统高并发时吞吐量低、响应时间长、失败率高等问题,提出一种公网数字集群系统的动态并发请求调度处理队列(DCRSPQ)算法。首先建立请求失败率和平均响应时间的优化目标,利用请求的各类优先级系数得到每个请求的加权平均优先级;然后,采用K-means聚类算法,根据每个请求的加权平均优先级确定所属的优先级队列,将请求分类到不同优先级的队列中;最后,利用自适应资源反馈调整机制,将系统资源动态分配给各优先级队列处理器,同时动态改变各优先级队列的长度,实现各类请求的高效快速处理。仿真结果表明,与分层级赤字加权轮询队列调度(HDWRR)和基于队列长度的动态加权公平队列调度(DQLWFQ)等算法相比,DCRSPQ算法的平均响应时间能缩短23.8%以上,吞吐量可提高3.5%以上,请求成功率可提升0.4个百分点。DCRSPQ算法具有更低的平均响应时间、更高的吞吐量以及更好的请求成功率,在公网数字集群系统并发情形下能有效提升系统的处理效率。Aiming at the problems of low throughput,long response time and high failure rate for high concurrency in digital trunking system based on public network,a Dynamic Concurrent Request Scheduling and Processing Queue(DCRSPQ)algorithm for digital trunking system based on public network was proposed.First,the optimization goals of request failure rate and average response time were established,the weighted average priority of each request was obtained by using various priority coefficients of requests;then,the K-means clustering algorithm was used to determine the priority queue to which each request belonged according to its weighted average priority,and the requests were classified into queues of different priorities;finally,adaptive resource feedback adjustment mechanism was used to achieve efficient and fast processing of various requests,which dynamically allocated system resources to each priority queue processor and changed the length of queues at all levels.The simulation results show that compared with algorithms such as Hierarchical Deficit Weighted Round Robin(HDWRR)queuing scheduling and Dynamic Queue Length Weighted Fair Queuing(DQLWFQ)scheduling,DCRSPQ algorithm shortens the average response time by more than 23.8%,increases the throughput by more than 3.5%and increases the request success rate by 0.4%.The DCRSPQ algorithm has lower average response time,larger throughput,and higher request success rate,which can effectively improve the processing efficiency of the system under the concurrent situation of the digital trunking system based on public network.

关 键 词：数字集群 队列 高并发 优先级 资源调度 

分 类 号：TP393.09[自动化与计算机技术—计算机应用技术]