Caching at the network edge can significantly reduce users' perceived latency and relieve backhaul pressure, hence invigorating a new set of innovations toward latency-sensitive applications. Nevertheless, the efficacy of caching policies relies on the users' content preference to be 1) known a priori and 2) highly homogeneous, which is not always the case in the real world. In this article, we explore how artificial intelligence (AI) techniques and recommendation can be leveraged to address those core issues and reap the potentials of cache-enabled wireless networks. Specifically, we present the hierarchical, cache-enabled wireless network architecture, in which AI techniques and recommendation are utilized, respectively, to estimate users' content requests in real time using historical data and to reshape users' content preference. Through case studies, we further demonstrate the effectiveness of an AI-based predictor in estimating users' content requests as well as the superiority of joint recommendation and caching policies over conventional caching policies without recommendation.
All Science Journal Classification (ASJC) codes
- Automotive Engineering