Design and modeling of survivable network planning for software‐defined data center networks in smart city

Abstract

SummarySmart city can integrate cloud computing, big data, Internet of Things, edge computing, and other modern information technologies, which focuses on integration, sharing, utilization, and service of information resources, and emphasizes the cooperation and coordination of urban management, to achieve deep integration of industrialization and informatization. Data center network planning based on big data is involved in rational deployment and interconnection of network infrastructures, which can improve the network survivability, energy‐efficiency, flexibility, etc. Therefore, a survivable network planning model based on software‐defined networking is presented in this paper. The elastic network planning problem is formulated as a mixed‐integer optimization whose objective is to minimize the number of unprotected nodes. Then, 3 effective schemes including K‐means clustering algorithm based on simulated annealing, greedy routing algorithm, and Lagrangian relaxation algorithm are proposed for feasible solutions. Numerical results on practical network topology reveals that (1) K‐means clustering algorithm based on simulated annealing can make effective classification of different scale topologies; (2) compared with the traditional random deployment method, the greedy routing algorithm and Lagrangian relaxation algorithm can generate the controller deployment strategy more reasonably, and thus improve network survivability and deployment cost‐benefit.