QoE-Driven Proactive Caching With DRL in Sustainable Cloud-to-Edge Continuum

Cloud-enabled edge computing scenarios can intelligently cache and update the content on a periodic basis, thereby enhancing users' overall perception of quality, which is called quality of experience (QoE). To enhance the QoE, we aim to the multi-objective optimization, which maximizes the cache hit ratio while simultaneously minimizing traffic load and time latency. To address this issue, we focus on employing an innovative algorithm named HT-PAD, which provides a complete solution for prediction and decision-making for proactive caching. First, to improve the prediction accuracy of the cached content, we use the encoding layer in hyperdimensional computing to extract the information features. Second, HD-Transformer, as the prediction part of HT-PAD, is proposed to make predictions based on user preferences, historical information, and popular information. HD-Transformer uses DNN to predict user preferences and process time series data by combining hyperdimensional computation with Transformer. Third, to avoid error in the prediction content, we employ PER-MADDPG as the decision-making part of HT-PAD, which consists of Multi-Agent Deep Deterministic Policy Gradient (MADDPG) and Prioritized Experience Replay (PER). We use MADDPG to enhance the content decision-making and utilized PER to select appropriate training samples for PER-MADDPG. Finally, our experiments have shown that our proposed approach achieves the strong performance in terms of the edge hit ratio, the latency, and the traffic load, thus improving the QoE