Projecte ofert
Títol: Multi-Objective Optimization of Controller Placement in SDN Networks: A Comprehensive Analysis
Per assignar-vos el projecte us heu de dirigir al director perquè us l'assigni.
Director: CERVELLÓ PASTOR, CRISTINA
Departament: ENTEL
Títol: Multi-Objective Optimization of Controller Placement in SDN Networks: A Comprehensive Analysis
Data inici oferta: 19-04-2023 Data finalització oferta: 19-12-2023
Estudis d'assignació del projecte:
MU MASTEAM 2015
Tipus: Individual | |
Lloc de realització: Fora UPC | |
Supervisor extern: Kurdman Rasol | |
Institució/Empresa: I2CAT Foundation | |
Titulació del Director: PhD in Computer Architecture (UPC) | |
Paraules clau: | |
SDN, Optimization, reliability, placement, ML | |
Descripció del contingut i pla d'activitats: | |
This Master's Thesis offer is associated with an internship.
Please apply at: https://borsapractiques.eetac.upc.edu/ca/ofertes/oferta/777 Software-defined networking (SDN) is an approach that separates the control plane from the data plane, enabling network intelligence to be deployed in a centralized SDN controller for a global view of the network. The initial SDN approach used a single centralized controller, but distributed controllers have emerged to address scalability issues. One of the challenges of distributed controller architectures is managing controllers effectively, including allocating them to appropriate network locations [1]. The Hierarchical Controller Placement Problem (HCPP) methodology uses a hierarchical architecture with at least two controller levels, improving network scalability and efficiency. This work is a continuation of our previous work done in [2]. The HCPP methodology solves the problem of controller placement using K-means and K-center algorithms, with a Super Controller (SC) at the top, some Master Controllers (MCs) in the middle, and Domain Controllers (DCs) at the bottom. The optimization problem is known to be NP-hard, so it is commonly formulated as a Mixed Integer Linear Programming (MILP) problem. The metric that is optimized in this case is the latency between controller-to-switch communication (CS) communications. To summarize, the main task of this project includes but is not limited to: 1. The plan is to expand the investigation of larger topologies, such as the Full European NREN emulation model, which has 1157 routers. In previous work, controllers were only placed vertically and oversaw a set of switches within their domain. The goal is to extend the hierarchical control plane among controllers to enable communication between them and allow coordination and consensus algorithms to keep them synchronized. 2. The study aims to enhance the reliability of SDN networks when facing network failures, including single and multiple link failures. It will analyze the impact of these failures on network performance and propose a controller placement method that can handle these failures effectively, ensuring efficient network operation in a practical scenario. 3. Another factor that should be considered when solving this problem is the controller's and the link's capacity. 4. The main tasks are to design and implement multi-level hierarchical controller placement strategies for SDN networks using a machine learning approach (i.e., genetic algorithms). The purpose is to solve the controller placement problem by considering all relevant constraints, including latency, reliability, and capacity, and comparing it to the exact solution. References ---------- 1. Brandon Heller, Rob Sherwood, and Nick Mckeown. The controller placement problem. Computer Communication Review, 42(4):473-478, 2012. ISSN 01464833. doi:10.1145/2377677.2377767. 2. Kurdman Rasol and J. Domingo-Pascual 'Multi-level Hierarchical Controller Placement in Software Defined Networking. In: Ghita B., Shiaeles S. (eds) Selected Papers from the 12th International Networking Conference. INC 2020. Lecture Notes in Networks and Systems, vol 180. Springer, Cham. https://doi.org/10.1007/978-3- 030-64758-210 Additional comment: ------------------- Contact Dr. Kurdman Rasol at kurdman.rasol@i2cat.net for any questions related to the activities. |
|
Orientació a l'estudiant: | |
The minimum skills required for the internship include a thorough
understanding of SDN architectures, protocols, and technologies, practical experience using network simulation and optimization tools like CPLEX, Gurobi, or Pyomo, and strong proficiency in the Python programming language. The main Expectations outcomes are as follows: 1. The program involves developing software tools and writing a report on the Multi-Objective Optimization of Controller Placement. 2. Additionally, the student is expected to produce written works suitable for publication in academic journals or conferences. Desired requirements: -------------------- Desired Skills for Internship Candidates (It is expected that the student has the ability to share software tools and submit the report). The following are some requirements for prospective candidates: 1. Master's degree in computer science, Operations Research, or related field. 2. At least 4 years of experience in software development and network engineering. 3. Deep understanding of SDN architectures, protocols, and technologies. 4. Experience with network simulation and optimization tools such as CPLEX, Gurobi or Pyomo 5. Strong experience with Python programming languages. 6. Strong problem-solving skills and the ability to work independently or in a team environment. 7. Excellent communication and interpersonal skills. Other Information: ----------------- - 20h/ Week - The amount to be paid for undergraduate students is €8/h and for master's students €10/h. - This Master's Thesis offer is associated with an internship. Please apply at https://borsapractiques.eetac.upc.edu/ca/ofertes/oferta/777 - Contact Dr. Kurdman Rasol at kurdman.rasol@i2cat.net for any questions related to the activities. |
|
Horari d'atenció a estudiants per a l'assignació de projecte: |