摘要
5G移动通信输出端口的有效任务量取决于端口输入流量的速率和流量分配缓冲区大小,通信多端口并行流量过大会导致网络压力过大,流量偏小不利于数据的高效传输。因此,需要对多端口的并行流量实现合理控制。提出一种5G移动通信多端口并行流量控制方法。对5G移动通信多端口性能建模,根据性能优势对多端口各条链路流量完成并行分配。利用建立的模糊逻辑控制器获取每条链路的网络拥堵反馈情况,并通过模糊逻辑控制器判断是否需要丢弃或标记输入的违约流量信元,以此保证每条链路网络流量的畅通性,实现并行流量的控制。实验结果表明,研究方法下并行流量达到2500Mbs时,CPU占用率低于50%。当链路数不断增加时,研究方法的总端口平均带宽可保持在500M,以下,端口带宽最大利用率可达95%以上,平均时延低于20ms。
The effective workload of 5G output port depends on the rate of input traffic and the size of the allocation buffer.Excessive parallel traffic will lead to high network pressure and small traffic.Therefore,it is necessary to control the multi-port parallel traffic reasonably.In this paper,a method of controlling multi-port parallel traffic of 5G mobile communication was proposed.Firstly,the multi-port performance of 5G mobile communication was modeled.Based on performance advantages,the traffic of each link was distributed in parallel.After that,the fuzzy logic controller was used to obtain the network congestion feedback of each link and judge whether it was necessary to discard or mark the input default traffic information elements,thus ensuring the smoothness of network traffic of each link.Finally,the control of parallel traffic was achieved.Experimental results show that after the proposed method was adopted,the CPU utilization is less than 50%while the parallel traffic reaches 2500mbs.When the number of links increases constantly,the average bandwidth of all ports can be less than 500M.Meanwhile,the maximum utilization rate of port bandwidth can reach more than 95%,and the average delay is less than 20ms.
作者
龚万炜
邢军
GONG Wan-wei;XING Jun(School of Optoelectronic Information Minnan Science and Technology University,Quanzhou Fujian 362332,China;Institute of Information Science and Engineering,Dalian Polytechnic University,Dalian Liaoning 116034,China)
出处
《计算机仿真》
北大核心
2023年第4期176-179,185,共5页
Computer Simulation
关键词
移动通信多端口
并行流量控制
并行流量分配
模糊逻辑控制器
控制性能
Mobile communication multi-port
Parallel traffic control
Parallel traffic distribution
Fuzzy logic controller
Control performance