Popularity and Energy-Aware Dynamic Programming Cache Strategy Based on User Preference Prediction
Lingjie Zou · 2024
In today’s rapid development of the Internet of Things (IoT), edge servers face significant challenges in handling massive data and meeting users’ personalized demands. Existing caching strategies, primarily based on static data popularity prediction, often fail to adapt to the dynamic changes in user requests and real-time requirements, resulting in low cache hit rates and high energy consumption. To address this issue, this paper proposes a novel Popularity and Energy-aware Dynamic Programming (PEDP) caching strategy based on user preference prediction.This strategy first utilizes Bidirectional Long Short-Term Memory networks (Bi-LSTM) to capture the time-series features of user requests, and then extracts user request similarity representations from macro and entity categories, enhancing the performance of user preference prediction. Based on a comprehensive understanding of user preferences and group behaviors, this paper adopts a dynamic programming approach, modeling the caching decision problem as a 0-1 knapsack problem, to minimize total caching costs and maximize cache hit rates. Experimental results demonstrate that the PEDP strategy can adapt to the dynamic changes in content popularity, exhibiting good dynamic popularity adaptability and effectiveness, with better optimization in hit rates and data transmission energy consumption.