arrow
Return

Modularity-based controller deployment scheme for scalable software defined networks

delete2026-04-01
delete0
PRE
AI
B
Bipal Khanal
K
Kumar, Chandan
M
Md. Sarfaraj Alam Ansari *
DOI:10.1016/j.comnet.2026.112269delete
deleteOriginal
deleteOriginal request for help
deleteShare
deleteSave
Abstract

Abstract

En 中文
The inception of software-defined networking simplifies network management by uncoupling control and infrastructure planes. The controller in the control plane provides ease of management of various network functions in the infrastructure plane. However, as the network grew in size and complexity, a single controller became a bottleneck. Introducing multiple distributed controllers, each managing a part of a network, solves this issue by enhancing redundancy, scalability, reliability and flexibility. Despite having numerous benefits, multiple controllers create new challenges such as controller placement, synchronization between controllers, increased network cost, and complexity. This article focuses on the controller placement problem, aiming to determine the number and optimal placement of controllers. The Clauset-Newman-Moore Greedy Modularity Maximization algorithm is modified to incorporate load balancing while dividing the large-scale network into balanced partitions. Moreover, four placement strategies are proposed based on average and worst-case latencies. The proposed Greedy Modularity Controller Deployment (GMCD) algorithm is evaluated extensively on seven real topologies of varying network scales obtained from SND-lib, Internet Topology Zoo, and Internet2 OS3E, demonstrating effective controller load balancing and minimizing latency while preserving the network's internal structure.
Keywords:
Software defined networking
Controller placement problem
Community detection
Modularity
Load balancing

Journal

Computer Networks cover
Computer Networks
IF:
4.6
Papers:
1.6K
Citations:
1.6W

Organization

N
national institute of technology (nit system)
Scholars:
4.0W
Papers: 3.7W
Citations: 31
Cited Papers

Cited Papers

No cited papers available