Full text
| |
| Author(s): |
Daniela Maria Casas Velasco
Total Authors: 1
|
| Document type: | Master's Dissertation |
| Press: | Campinas, SP. |
| Institution: | Universidade Estadual de Campinas (UNICAMP). Instituto de Computação |
| Defense date: | 2020-10-15 |
| Examining board members: |
Nelson Luis Saldanha da Fonseca;
Jéferson Campos Nobre;
Ricardo da Silva Torres
|
| Advisor: | Oscar Mauricio Caicedo Rendón; Nelson Luis Saldanha da Fonseca |
| Abstract | |
In communication networks, routing determines the path followed by packets from a source node to a destination node. In traditional Internet routing protocols, routing decisions are based on limited information and derived from the calculation of shorted paths, which can lead to slow adaptation to traffic variability, and the restricted support of the Quality of Service (QoS) requirements of applications. Software-Defined Networking (SDN) was conceived to favor the adoption of innovation in network protocols. Some solutions have shown the improvement of traditional routing protocols, taking advantage of SDN resources, such as programmability, global view, logically centralized control, and decoupling of network control, and packet forwarding. However, these solutions do not fully exploit knowledge about network operation to perform routing intelligently. Other works have explored Machine Learning (ML) techniques, such as Neural Networks, Logistic Regression, and K-means in conjunction with SDN. However, the acquisition of training data sets, in these works, is dependent on the information available in traditional routing protocols. Also, the distributed form of routing is assumed, which tends to generate signaling overhead. This thesis introduces two approaches for routing in SDN called RSIR and DRSIR. RSIR stands for Reinforcement Learning and Software-Defined Networking Intelligent Routing, which adds a Knowledge Plane and defines a routing algorithm based on Reinforcement Learning (RL). The RSIR algorithm considers network state metrics to produce intelligent routing that adapts to dynamic traffic changes. RSIR is based on the interaction with the environment and the global view and control of the network, to proactively compute and install optimal routes on the packet flow routing devices. RSIR is presented in two versions, RSIR-links and RSIR-paths, which use network status information with metrics at the level of links and paths. DRSIR (Deep RSIR) is an extended version of RSIR based on Deep Reinforcement Learning, which improves performance in terms of link throughput, delay and loss rate, concerning RL-based approaches. RSIR and DRSIR were evaluated extensively by emulation using traffic matrices (real and synthetic). The results show that our solutions surpass the routing algorithms based on Dijkstra, concerning the metrics stretching (stretch), packet loss, and delay. Furthermore, the results show the effectiveness of the algorithms regarding the throughput. The results obtained demonstrate that RSIR and DRSIR are a practical and viable solution for routing in SDN (AU) | |
| FAPESP's process: | 19/03268-0 - Software defined networking routing with machine learning |
| Grantee: | Daniela Maria Casas Velasco |
| Support Opportunities: | Scholarships in Brazil - Master |