Two-stage Update Strategy for Edge Caching Deployment Based on Access Information in Ultra Dense Network

摘要

The explosion of data traffic has an unfavorable effect on communication latency and the overhead of content access in Ultra Dense Network (UDN). Content-based edge caching is deemed to be an efficient solution. In this paper, we present a caching structure and propose a two-stage update strategy in order to reduce the file-access latency and cost in an application scenario covered by the dense base-stations and Wi-Fi signals where users are stationary or moving at a low speed. In the UDN architecture, the proposed strategy can obtain access information based on the combination of user preferences and file attributes to cache high-popularity files on the Radio Access Network (RAN) side. We also apply a transfer actor-critic learning framework to save energy consumption of the base station groups. In addition, filtering of junk files is taken into account to increase the utilization of cache storage space. Finally, we assess the proposed scheme by plenty of simulations. The results show significant reduction on file-access latency and cost as a result of the two-stage update strategy for edge caching deployment method.