隨著電腦網路的規模與複雜性快速的成長，傳統的基礎設施、網路系統與通訊協定已經無法為當代的網路需求提供充分的解決方案。雲端計算、大數據、物聯網等以及其他技術的快速發展，網路管理比以往更加困難，演算法的設計也比以往更複雜，為了改善網路品質以及網路管理的難易度，軟體定義網路是近年興起的新型網路技術，將控制層與資料層分離，透過控制器負責網路管理，大幅提升網路管理的彈性，且易於操作。目前許多網路的優化演算法皆是基於分析模型的，也就是說只能根據網路所能蒐集到的資訊進行評估，但在複雜的電腦網路中可能存在一些隱藏特徵是我們無法蒐集到的，需要透過神經網路來找出這些隱藏特徵，所以我們考慮基於人工智慧的演算法，在這篇論文中使用RouteNet，是一個基於圖神經網路(GNN)的網路模型，能夠理解網路複雜的特徵像是拓樸、路由規則、網路流之間的關係，並根據輸入的特徵推論出績效指標(e.g. path delay、jitter等)。我們提出的方法為將RouteNet Model與SDN控制器做結合，SDN控制器可以透過RouteNet預測出來的績效指標來實現網路優化，進而改善網路品質像是提高網路利用度、降低傳輸時間等。本論文探討了該架構的可行性，測試了模型的泛化能力，在利用率和吞吐量方面的實驗結果表明，我們提出的方法比傳統方法提高了5%以上。

With the rapid growth in the size and complexity of computer networks, traditional infrastructure, network systems, and protocols no longer provide adequate solutions for contemporary networking needs. To improve network quality and ease of network management, software-defined networks are regarded as a new network technology that has emerged in recent years, separating the control layer from the data layer and making the controller responsible for network management, greatly increasing the flexibility of network management and making it easier to operate. Currently, many network optimization algorithms are based on analytical models, i.e., they can only be evaluated based on the information that can be collected by the network. In this paper, we use RouteNet, a graphical neural network GNN-based network model that can understand the relationship between complex features of the network such as topology, routing rules, and network flows, and infer performance metrics (e.g., path delay, jitter,etc) based on the input features. Our proposed approach is to integrate RouteNet Model with SDN controller, which can implement network optimization according to RouteNet's predicted performance metrics to improve network quality such as improving network utilization, reducing transmission time, etc. This paper explores the feasibility of this architecture, tests the generalization ability of the model, and the experimental results in utilization and throughput show that our proposed method is improved by more than 5% compared with the traditional method.

[1] N Wang et al. “An overview of routing optimization for internet traffic engineering”. IEEE Communications Surveys Tutorials 10.1 (2008), pp. 36-56, 2008.
[2] Fortz, B., & Thorup, M. (2000, March). “Internet traffic engineering by optimizing OSPF weights”. Proceedings IEEE INFOCOM 2000. conference on computer communications. Nineteenth annual joint conference of the IEEE computer and communications societies (Cat. No. 00CH37064) (Vol. 2), pp. 519-528, 2000.
[3] Mestres, Albert, et al. "Knowledge-defined networking." ACM SIGCOMM Computer Communication Review 47.3 (2017), pp. 2-10, 2017.
[4] Shortest Job First(SJF) Algorithm,
Avaible on July 19, 2021:http://os.etf.bg.ac.rs/OS2/Stud/Rad01/sjf.htm
[5] (2019). Knowledge-Defined Networking Repository. [Online]. Available: https://github.com/knowledgedefinednetworking/Papers/wiki/RouteNet:- Leveraging-GNN-for-network-modeling-and-optimization-in-SDN
[6] Nunes, Bruno Astuto A., et al. "A survey of software-defined networking: Past, present, and future of programmable networks." IEEE Communications surveys & tutorials 16.3 (2014), pp. 1617-1634, 2014.
[7] McKeown, Nick, et al. "OpenFlow: enabling innovation in campus networks." ACM SIGCOMM computer communication review 38.2 (2008), pp. 69-74, 2008.
[8] Wang, Xiaohui Yang Lei. "SDN load balancing method based on K-Dijkstra." International Journal of Performability Engineering 14.4 (2018), pp. 709-716, 2018.
[9] Nkosi, M. C., Lysko, A. A., & Dlamini, S. Multi-path load balancing for SDN data plane. 2018 International Conference on Intelligent and Innovative Computing Applications (ICONIC), pp. 1-6, 2018.
[10] Xue, Hai, Kyung Tae Kim, and Hee Yong Youn. "Dynamic load balancing of software-defined networking based on genetic-ant colony optimization." Sensors 19.2, pp. 311-328, 2019.
[11] Akin, Erdal, and Turgay Korkmaz. "Rate-based dynamic shortest path algorithm for efficiently routing multiple flows in SDN." ICC 2019-2019 IEEE International Conference on Communications (ICC), pp. 1-7, 2019.
[12] Quang, Pham Tran Anh, et al. "Multi-objective multi-constrained QoS Routing in large-scale networks: A genetic algorithm approach." 2018 International conference on smart communications in network technologies (SaCoNeT), pp. 55-60, 2018.
[13] Guo, Aipeng, et al. "Research on SDN/NFV network traffic management and optimization based on big data and artificial intelligence." 2018 18th International Symposium on Communications and Information Technologies (ISCIT), pp. 377-382, 2018.
[14] Battaglia, Peter W., et al. "Relational inductive biases, deep learning, and graph networks." arXiv preprint arXiv:1806.01261 (2018), pp. 1-40, 2018.
[15] Shu, Zhaogang, et al. "Traffic engineering in software-defined networking: Measurement and management." IEEE access 4 (2016), pp. 3246-3256, 2016.
[16] Olariu, Cristian; Zuber, Martin; Thorpe, Christina. Delay-based priority queueing for VoIP over Software Defined Networks. 2017 IFIP/IEEE Symposium on Integrated Network and Service Management (IM), pp. 652-655, 2017.
[17] Abugabah, Ahed, et al. "Intelligent traffic engineering in software-defined vehicular networking based on multi-path routing." IEEE Access 8 (2020): pp. 62334-62342, 2020.
[18] Jiang, Jehn-Ruey, et al. "Extending Dijkstra's shortest path algorithm for software defined networking." 16th Asia-Pacific Network Operations and Management Symposium, pp. 1-4, 2014.
[19] Dorigo, Marco, Mauro Birattari, and Thomas Stutzle. "Ant colony optimization." IEEE computational intelligence magazine 1.4 (2006): 28-39.
[20] Whitley, Darrell. "A genetic algorithm tutorial." Statistics and computing 4.2 (1994), pp. 65-85, 1994.
[21] Sgambelluri, Andrea, et al. "Evaluating Link Latency in Distributed SDN-Based Control Plane Architectures." ICC 2019-2019 IEEE International Conference on Communications (ICC), pp. 1-6, 2019.
[22] Liao, Lingxia, and Victor CM Leung. "LLDP based link latency monitoring in software defined networks." 2016 12th International Conference on Network and Service Management (CNSM), pp. 330-335, 2016.
[23] Guo, Wenjing, and Wei Zhang. "A survey on intelligent routing protocols in wireless sensor networks." Journal of Network and Computer Applications 38 (2014), pp. 185-201, 2014.
[24] Huang, Yue-Cai, Jie Zhang, and Siyuan Yu. "Self-Learning Routing for Optical Networks." International IFIP Conference on Optical Network Design and Modeling, pp. 467-478 2019.
[25] Chen, Xiaoliang, et al. "DeepRMSA: A deep reinforcement learning framework for routing, modulation and spectrum assignment in elastic optical networks." Journal of Lightwave Technology 37.16 (2019), pp. 4155-4163, 2019.
[26] Rusek, Krzysztof, et al. "Unveiling the potential of Graph Neural Networks for network modeling and optimization in SDN." Proceedings of the 2019 ACM Symposium on SDN Research, pp. 140-151, 2019.
[27] Rusek, Krzysztof, et al. "RouteNet: Leveraging Graph Neural Networks for network modeling and optimization in SDN." IEEE Journal on Selected Areas in Communications 38.10 (2020), pp. 2260-2270, 2020.
[28] Ryu Controller, Available on July 19:https://github.com/faucetsdn/ryu
[29] Mininet emulator, Available on July 19, 2021:https://github.com/mininet/mininet
[30] OMNet++, Available on July 19, 2021:https://omnetpp.org/
[31] Ryubook,
Available on July 19, 2021: https://osrg.github.io/ryu-book/zh_tw/Ryubook.pdf
[32] Gilmer, Justin, et al. "Neural message passing for quantum chemistry." International Conference on Machine Learning. PMLR, pp. 1263-1272, 2017.
[33] Ping, Available on July 19, 2021:https://linux.die.net/man/8/ping
[34] Mills, David L., and Hans-Werner Braun. "The NSFNET backbone network." Proceedings of the ACM workshop on Frontiers in computer communications technology, pp. 191-196, 1987.
[35] NetworkX library, Available on July 19, 2021:https://networkx.org/
[36] Dijkstra’s algorithm,
Available on July 19,2021:https://en.wikipedia.org/wiki/Dijkstra%27s_algorithm