Smart Service Function Chain System for Dynamic Traffic Steering using Reinforcement Learning (CHRL)

Authors

Ahmed Nadhum, Department of Computer Engineering, University of Technology- IraqFollow
Ahmed Al-Saadi, Department of Computer Engineering, University of Technology- IraqFollow

Abstract

The rapid development of the Internet and network services coupled with the growth of communication infrastructure necessitates the employment of intelligent systems. The complexity of the network is heightened by these systems, as they offer diverse services contingent on traffic type, user needs, and security considerations. In this context, a service function chain offers a toolkit to facilitate the management of intricate network systems. However, various traffic types require dynamic adaptation in the sets of function chains. The problem of optimizing the order of service functions in the chain must be solved using the proposed approach, along with balancing the network load and enhancement of net-work security. In addition, the delay issue must be resolved by selecting an optimal path to establish a connection. The proposed system provides a set of intelligent function chains that can adaptively optimize the network performance while considering dynamic traffic demands using SDNs and Q-learning. The proposed system can significantly improve the overall efficiency, scalability, and adaptability of the network while also providing a better quality of service to end-users. Compared with traditional software-defined networks, the simulation results of the proposed system showed an improvement in throughput of up to 76%, accompanied by a reduction in the level of link congestion. The results also exhibit an improvement of up to 54% compared with state-of-the-art load balancing. In particular, in terms of the FTP performance, our proposed system outperforms existing approaches by up to 20%.

Recommended Citation

Nadhum, Ahmed and Al-Saadi, Ahmed (2023) "Smart Service Function Chain System for Dynamic Traffic Steering using Reinforcement Learning (CHRL)," Karbala International Journal of Modern Science: Vol. 9 : Iss. 4 , Article 8.
Available at: https://doi.org/10.33640/2405-609X.3326

Creative Commons License

This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.