软件定义网络(Software Defined Networking , SDN)作为一个新时代的网络系统架构，由于其本身具有全局视图和可编程的优势，为有效解决由于网络流量快速增长而导致的路径拥塞问题提供了可行的方法。但是传统的流量调度算法与SDN架构不能完美兼容，表现出一定的局限性。因此，本文基于SDN架构来研究网络流量的调度算法与策略，以达到实现网络负载均衡，提升网络性能的目的。

       本文在分析流量调度算法的基础上，首先提出了一种最小花销流量调度算法，该算法综合考虑了时延和带宽对于网络性能的影响，通过综合评估计算出网络链路的最小花销值，最后选取其中花销最小的路径作为最优路径。其次，针对工业数据中心流量特点，在最小花销算法的基础上设计了一种融合了蚁群优化算法的流量调度策略。在流量调度阶段，该算法按照一定周期对网络负载进行检测，计算出负载均衡参数和网络链路利用率，根据阈值将流量调度方式分为两类，负载均衡参数及链路利用率小于阈值时选择最小花销算法对流量进行调度处理，若二者均大于阈值，将流量划分成大象流和老鼠流两类，对老鼠流使用默认的最小花销算法进行调度，对大象流则使用本文提出的改进蚁群算法进行调度。本文针对基础蚁群算法的全局搜索能力弱、收敛速度慢等缺点，对蚁群算法进行了三方面的改进：蚁群寻径策略，信息素的更新方式以及信息素重置。

       最后，对本文提出的最小花销算法和基于蚁群的流量调度策略进行有效性验证。通过与常用的流量调度算法进行对比实验测试，分析了平均流吞吐量，带宽利用率标准差，平均时延和平均丢包率四个性能指标的算法实验结果，说明了本文提出的方法对于网络性能有所提升，有效实现了网络链路负载均衡，达到了预期的目标。

    Software Defined Networking (SDN) as a network system architecture of new era, due to its own advantages of global view and programmability, provides a feasible way to effectively solve the path congestion problem caused by the rapid growth of network traffic .However, the traditional traffic scheduling algorithm is not perfectly compatible with the SDN architecture, showing certain limitations. Therefore, this paper is based on SDN technology to study network traffic scheduling algorithms and strategies to achieve the goal of achieving network load balancing and improving network performance.

   Based on the analysis of the traffic scheduling algorithm, this paper first proposes a minimum cost traffic scheduling algorithm. This algorithm comprehensively considers the impact of delay and bandwidth on network performance, and calculates the minimum cost of the network link through comprehensive evaluation, and selects the path with the least cost as the optimal path.Secondly, in view of the traffic characteristics of industrial data centers, a traffic scheduling strategy incorporating the ant colony optimization algorithm is designed on the basis of the minimum cost algorithm. In the traffic scheduling stage, the algorithm detects the network load according to a certain period, calculates the load balancing parameters and network link utilization, and divides the traffic scheduling methods into two types according to the threshold. When load balancing parameters and link utilization are less than the thresholds, the least cost algorithm is selected to schedule traffic. If both are more than the thresholds, divide the flow into two types: elephant flow and mouse flow, and use the default minimum cost algorithm to schedule the mouse flow.For the elephant flow, the improved ant colony algorithm proposed in this paper is used for scheduling.Aiming at the shortcomings of the basic ant colony algorithm such as weak global search ability and slow yield, this paper proposes three improvements to the ant colony algorithm: ant colony path finding strategy, pheromone update method and pheromone reset.

    Finally, the effectiveness of the minimum cost algorithm and ant colony-based traffic scheduling strategy proposed in this paper is verified.Through experimental testing and comparison of results with commonly used traffic scheduling algorithms, the algorithm experiment results of the four performance indicators of average flow throughput, bandwidth utilization standard deviation, average delay, and average packet loss rate are analyzed, which shows that the method proposed in this paper is effective for the network. The network performance has been improved, the network link load balance has been effectively realized, and the expected goal has been achieved.

[1]Tahmasebi Shirin, Rasouli Nayereh, Kashefi Amir Hosein. SYNCOP: An evolutionary multi-objective placement of SDN controllers for optimizing cost and network performance in WSNs[J]. Computer Networks,2021,185.

[2]周宇. 面向IP及光网络融合的控制技术研究[D].北京邮电大学,2018.

[3]赵伟,高华,张建辉等.SDN中的网络拓扑发现[J].信息工程大学学报,2018,19(03):270-274+281.

[4]周一青, 潘振岗, 翟国伟. 第五代移动通信系统5G标准化展望与关键技术研究[J].数据采集与处理, 2015, 6(4): 714-724.

[5]冯誉. OpenFlow网络技术综述[J].科技创新与应用,2018(19):13-14+17.

[6]程东亮, 曹小刚. 谷歌、阿里巴巴和腾讯SDN应用案例分析及金融业应用研究[J].互联网天地,2018(05):50-54.

[7]Masayoshi Kobayashi, Srini Seetharaman, Nick McKeown. Maturing of OpenFlow and Software-defined Networking through deployments[J]. Computer Networks,2014,61.

[8]乔平安, 任泽乾.基于SDN的负载均衡动态流量调整策略[J].信息技术,2019,43(08):24-28.

[9]李道全, 黄泰铭, 于波. 基于流量分配倾向度的SDN多路径负载均衡[J].计算机工程与设计,2020,41(10):2718-2723.

[10]Wang R, Butnariu D, Rexford J. OpenFlow-based server load balancing gone wild[C]// Usenix Conference on Hot Topics in Manageme of Internet, Cloud, and Enterprise Networks and Services. USENIX Association, 2011:12-12.

[11]Hopps C E. Analysis of an equal-cost multi-path algorithm[J]. Journal of Allergy & Clinical Immunology, 2010:281-196.

[12]Li Y. Pan D. OpenFlow based load balancing for fat-tree networks with multipath support[C]//Proc. 12th IEEE International Conference on Communications (ICC’13), Budapest, Hungary. 2013: 1-5.

[13]Chou L D ,Yang Y T ,Hong Y M. A Genetic-Based Load Balancing Algorithm in OpenFlow Network[M]//Advanced Technologies, Embedded and Multimedia forHuman-centric Computing. 2014:411-417.

[14]Yahya W, Basuki A, Jiang J R. The extended dijkstra's-based load balancing for OpenFlow network[J]. International Journal of Electrical & Computer Engineering ,2015,5(2):289-296.

[15]Guo Aipeng,Yuan Chunhui. Network Intelligent Control and Traffic Optimization Based on SDN and Artificial Intelligence[J]. Electronics,2021,10(6).

[16]Haibo Wang, Hongli Xu, Chen Qian. PrePass: Load balancing with data plane resource constraints using commodity SDN switches[J]. Computer Networks,2020,178.

[17]Nallusamy P, Saravanen S, Krishnan M. Decision Tree‐Based Entries Reduction scheme using multi‐match attributes to prevent flow table overflow in SDN environment[J]. International Journal of Network Management, 2020: e2141.

[18]王文东,刘继梅.基于蚁群算法的云计算资源调度研究综述[J].电脑知识与技术,2017,13(23):161-163.

[19]向永靖.蚁群算法中参数设置的研究——以TSP为例[J].现代信息科技,2020,4(22):95-98+102.

[20]Chen X, Beiijebbour A, Li A, et al. Consideration on successive interference canceller (SIC) receiver at cell-edge users for non-orthogonal multiple access (NOMA) with SU-MIMO[C]// 2015 IEEE 26th Annual International Symposium on Personal, Indoor, and Mobile Radio Communications (PIMRC). Hong Kong: IEEE Press, 2015: 522-526.

[21]Dong L , Li J , Xia W , et al. Double ant colony algorithm based on dynamic feedback for energy-saving route planning for ships[J]. Soft Computing, 2021, 25(1997).

[22]Liu X, He D. Ant colony optimization with greedy migration mechanism for node deployment in wireless sensor networks[J]. Journal of Network & Computer Applications, 2014, 39:310-318.

[23]A M Koushika, S T Selvi. Load balancing Using Software Defined Networking in cloud environment[C]// Fourth International Conference on Recent Trends in Information Technology. 2014:1-8.

[24]张焕青, 张学平, 王海涛等. 基于负载均衡蚁群优化算法的云计算任务调度[J].微电子学与计算机, 2015, 32(5):31-35.

[25]李春艳,王茜.云计算平台联合资源调度优化算法研究[J].数字技术与应用,2017(02):146-147+149.

[26]Wenxuan Gao. Research on Load Balancing Control of Internet of Things (IoT) Link Based on Sparse Random Clustering[J]. Journal of Physics: Conference Series,2020,1693(1).

[27]Khaled Alghamdi,Robin Braun. Software Defined Network (SDN) and OpenFlow Protocol in 5G Network[J]. Communications and Network,2020,12(01).

[28]Min Jia, Yuejie Shu, Suofei Xie. DDoS Attack Detection Method for Space-Based Network Based on SDN Architecture[J].ZTE Communications,2020,18(04):18-25.

[29]张雯雯,许天予,章玥等.SDN数据平面软件一致性测试用例生成方法[J].软件学报,2020,31(09):2709-2722.

[30]金勇,刘亦星,王欣欣. SDN-based Multipath Traffic Scheduling Algorithm for Data Center Network[J]. 计算机科学, 2019, 046(006):90-94.

[31]Shuker F M . Improvement of Routing Protocol for IoT Network Using SDN/OpenFlow[C]// 2019 1st AL-Noor International Conference for Science and Technology (NICST). IEEE, 2019.

[32]Yuki Goto, Bryan Ng, Winston K.G. Seah. Queueing analysis of software defined network with realistic OpenFlow–based switch model[J]. Computer Networks,2019,164.

[33]李成博, 加云岗.基于OpenFlow的SDN技术浅析[J].电脑知识与技术,2019,15(27):15-16.

[34]陈雷. OpenFlow 1.3协议一致性测试平台研究与开发[J].现代电信科技,2016,46(03):34-40.

[35]Xu X , Dai J , Zhi Y . An Intrusion Prevention Scheme for Malicious Network Traffic Based on SDN[J]. IOP Conference Series Materials Science and Engineering, 2020, 790:012030.

[36]Isyaku Babangida, Bte Kamat Maznah, Ghaleb Fuad A. Software Defined Networking Flow Table Management of OpenFlow Switches Performance and Security Challenges: A Survey[J]. Future Internet,2020,12(9).

[37]陈珂,刘亚志,王思晗.基于流量特征的流调度策略研究综述[J].计算机应用研究,2020,37(10):2889-2894.

[38]Tran A K, Piran M, Pham C. SDN controller placement in IoT networks: An optimized submodularity-based approach[J]. Sensors, 2019, 19(24): 5474.

[39]Morzhov S, Alekseev I, Nikitinskiy M. Firewall application for Floodlight SDN controller[C]//2016 International Siberian Conference on Control and Communications (SIBCON). IEEE, 2016: 1-5.

[40]Yang Li, Zhi-Ping Cai, Hong Xu. LLMP: Exploiting LLDP for Latency Measurement in Software-Defined Data Center Networks[J]. Journal of Computer Science and Technology,2018,33(2).

[41]Selvakumar K, Revathy G. Escalating quality of services with channel assignment and traffic scheduling in wireless mesh networks[J]. Cluster Computing,2019,22(5).

[42]张罡. 基于蚁群优化算法的SDN流量调度研究[D].湖北工业大学,2017.

[43]叶颖诗, 魏福义, 蔡贤资.基于并行计算的快速Dijkstra算法研究[J].计算机工程与应用,2020,56(06):58-65.

[44]居建涛, 罗琴, 郑弘迪.SDN中基于链路特征的路由控制机制[J].计算机应用,2019,39(S1):138-142.

[45]李道全, 魏艳婷, 张玉霞等. 基于改进蚁群算法的WSN能量均衡路由算法[J].计算机工程与应用,2019,55(17):117-124.

[46]覃远年,梁仲华.蚁群算法研究与应用的新进展[J].计算机工程与科学,2019,41(01):173-184.

[47]刘耀,毛剑琳.动态环境下基于改进蚁群算法的路径规划研究[J].电子测量技术,2020,43(07):82-87.

[48]李士勇, 陈永强, 李研. 蚁群算法及其应用[M]. 哈尔滨工业大学出版社, 2004.

[49]黄宇达, 王迤冉, 牛四杰. 基于混沌和自适应搜索策略的GSO算法分析与优化[J].计算机工程与应用,2019,55(03):147-153.

[50]Shahryari S, Hosseini-Seno S A, Tashtarian F. An SDN based framework for maximizing throughput and balanced load distribution in a Cloudlet network[J]. Future Generation Computer Systems, 2020, 110: 18-32.

[51]Rezaee M , Yaghmaee Moghaddam M H. SDN-based Quality of Service Networking for Wide Area Measurement System[J]. IEEE Transactions on Industrial Informatics, 2019:1-1.

[52]段元新, 倪晓军, 章韵. 多指标综合评价的负载均衡路由策略研究[J].小型微型计算机系统,2017,38(02):209-212.