A Novel Optimization Algorithm for Resilient T-SDN Control-Plane Design in Optical Transport Networks

Conference: Photonische Netze - 24. ITG-Fachtagung
05/09/2023 - 05/10/2023 at Berlin

Proceedings: ITG-Fb. 310: Photonische Netze

Pages: 7Language: englishTyp: PDF

Authors:
Sultana, Shabnam (Chair of Communication Networks, Technische Universität Chemnitz, Germany & highstreet-technologies GmbH, Berlin, Germany)
Reyes, Ronald Romero; Nguyen, Khai Tuan; Bauschert, Thomas (Chair of Communication Networks, Technische Universität Chemnitz, Germany)

Abstract:
With the evolution of SDN and NFV, the design of the control plane with performance and resilience guarantees has gained momentum. While numerous studies exist on the controller placement problem, most of them focus on objectives such as the minimization of the controller-to-switch and controller-tocontroller latency and bandwidth. However, there are only a few papers addressing the design of the control plane network itself, with focus on IP networks only. Concerning software-defined optical transport networks, to the best of our knowledge, no such study exists yet. To close this gap, we propose an algorithm for the design of the control plane network in optical transport systems. The algorithm optimizes the number and placement of the control plane interfaces along with the routing of the control traffic. The approach determines the set of network nodes which are directly connected to the controller. The control traffic of the remaining nodes is carried towards the controller through these directly connected nodes. For that, the algorithm optimizes the routing of control traffic between directly and non-directly connected nodes. Considering reliability constraints, such as protection against single link and/or node failures, the algorithm minimizes the cost of the control plane network w.r.t the number of control plane interfaces and the capacity required to carry control traffic . We evaluate our method on selected optical transport networks and analyze the influence of the cost parameters on the control plane design