AUCB: Accelerated Upper Confidence Bound Algorithm to Improve Caching Performance in Industrial Network
Yingbiao Ye, Ying Liu, Shanghan Xie, Wenqian Jia · 2021
With the development of industrial network, the coexistence of best effort service and industrial service brings many challenges. An emerging problem is that industrial service will affect the load capacity of the network significantly. Hence, this paper proposes an intra-network caching algorithm based on reinforcement learning to reduce redundant traffic and enhance network scalability. First, we found the problem of insufficient bandwidth and scalability of industrial network. Above pressure can be relieved effectively through intra-network cache. Second, an accelerated upper confidence bound (AUCB) algorithm is designed to cope with the variable industrial request. Third, we evaluate the performance of AUCB through reasonable experiments. Corresponding results verified that AUCB can outperform other candidates for cache hit rate and average latency in both static and dynamic scenarios.