An Effective Controller Placement Algorithm Based on Clustering in SDN

Rong Yujie,Wu Muqing,Chen Yiming

DOI: https://doi.org/10.1109/ICCC51575.2020.9345045

2020-01-01

Abstract:Different from the traditional TCP/IP network, Software Defined Network (SDN) uses a layered architecture to decouple the control plane and data plane. The controller is the core device of the control plane, and the control plane is mainly responsible for the decision-making control of the routing strategy. Because the processing power of a single controller is limited, the multi-controller network architecture has become an inevitable choice in the WAN. The layout of the controllers in SDN has an impact on the performance of the entire network. The controller placement problem (CPP) is an NP-hard problem. The challenges of CPP solution are as follows. 1) the number of controllers; 2) the position of the controllers and the position of the switches under their management in the network topology. The clustering by fast search and find of density peaks (CFSFDP) algorithm is a density-based clustering algorithm. It has few initial parameters and does not require iterative solution. However, the important parameter $d_{c}$ of the CFSFDP algorithm is usually selected based on experience, and the number of cluster centers cannot be determined automatically. This paper proposes a new controller placement algorithm by improving the CFSFDP algorithm. We use the idea of information entropy to automatically determine the $d_{c}$ value, and automatically determine the number and location of the controllers based on the jumping point in the decision graph. It can be seen from the experimental results that compared with random placement, k-means and CFSFDP algorithm, our proposed algorithm can effectively reduce network propagation delay and improve network performance.