A hybrid-ABC approach to the multi-controller placement problem of software-defined networks

The emergence of Software-Defined Network (SDN) paradigm offers a revolutionary approach to network management, aiming to enhance agility and flexibility by centralizing control of network nodes. However, large-scale SDN deployments face scalability challenges, necessitating the use of multi-controller architectures. This gives rise to the Controller Placement Problem (CPP), a complex NP-hard issue that critically influences network performance. To address the CPP, this paper introduces a hybrid-ABC algorithm designed specifically for multi-CPP scenarios. The proposed hybrid-ABC algorithm leverages the efficiency of Artificial Bee Colony (ABC) metaheuristic, a nature-inspired approach known for quickly reaching optimal solutions. The proposed algorithm integrates K-means clustering with the ABC algorithm to enhance solution quality and address the issue of poor exploitation of solutions inherent in ABC. Thus, the proposed method is a fusion of metaheuristic swarm intelligence and machine learning techniques, combining the efficiency of ABC metaheuristic with K-means clustering. The study evaluates the efficacy of the hybrid-ABC algorithm against the traditional ABC algorithm across diverse datasets. Key performance metrics, including between-controller delays, solution quality, standard deviations, and latency times, are thoroughly assessed to measure effectiveness. Experimental results demonstrate that the proposed hybrid-ABC algorithm outperforms the traditional ABC algorithm in determining optimal controller placements within SDN architectures. This highlights the algorithm's superior performance in solving the multi-CPP problem, highlighting its potential for enhancing SDN network scalability and efficiency. Received: 22 April 2024 Accepted: 29 July 2024 Published: 26 August 2024