A Contextual Bandit Approach for Network Service Selection

Zhenyu Li,Rui Zhang,Shutian Zhou,Haiyang Yu,Feng Hong

DOI: https://doi.org/10.1109/ithings-greencom-cpscom-smartdata-cybermatics60724.2023.00118

2024-01-01

Abstract:As large companies expand their operations, the importance of the business-to-business (B2B) model is also on the rise. The network service selection problem serves as a typical example of B2B services, but existing solutions often rely heavily on static statistics, which do not perform well in complex and dynamic environments. Therefore, there is a need for a method that can address the volatile demands of network service selection and rapidly provide the optimal service provider choice. We primarily face three challenges: (1) dealing with the immense scale of business traffic, making real-time analysis challenging and demanding advanced and high-efficiency algorithms; (2) static statistical methods fall short when adapted to the complexity and volatility of real-world environments; (3) the trade-off between duration of processing, success rate, and profit, making it difficult to simultaneously optimize all three criteria. To address these issues, we propose a network service selection method based on contextual bandits and restructure the reward function. We evaluate the accuracy and performance of our method through experimental testing, which validates the improvement of our network service selection method.