User-Preference-based Proactive Caching in Edge Networks
The Vi Nguyen, Van Dat Tuong, Anh-Tien Tran, Thanh Phung Truong, Demeke Shumeye Lakew, Chunghyun Lee, Yunseong Lee, Sungrae Cho · 2021
Recently, with the rapid growth of data traffic, caching in the edge networks is considered a promising approach which provides low latency and improves user's quality of service (QoS). However, user preference diversity can be a challenge for developing an effective caching algorithm with limited cache capacity. In this paper, we propose a proactive caching scheme considering user preferences to maximize the cache hit ratio using long short-term memory networks (LSTM). First, each demographic user group is trained on an LSTM model for predicting user demand for movie genre. Then, the results are combined by averaging to obtain the average demand over user groups to generate an efficient caching policy. The experimental results show that our caching algorithm outperforms benchmark schemes in terms of the cache hit ratio. The proposed control provides up to 35% higher cache hit ratio than benchmark algorithms and near-optimal cache hit ratio within around 12% of the optimal scheme with perfect prior knowledge of movie popularity.