Spatio-Temporal Content Caching: Leveraging Deep Learning and Stochastic Optimization

The proliferation of smart devices and the rising demand for video data traffic present substantial problems for Internet and content service providers. A potential solution to this problem is content caching on mobile edge computing (MEC) servers, which lowers the latency of content downloads. Nevertheless, current caching solutions sometimes presume the popularity of stationary information or necessitate real-time popularity knowledge, which is inconsistent with practical situations. To overcome these restrictions, we propose the ProCache algorithm. ProCache aims to decrease active users' long-term expected content download latency by accounting for spatio-temporal financial budget sharing for caching. YouTube dataset is used to run trace-driven simulations, and we show that ProCache performs better than current prediction models and content caching methods.