Energy-Efficient Proactive Content Caching For On-Demand Video Streaming Under Uncertainty

Abstract

Proactive content caching using <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">edge service providers</i> (ESPs) present in wireless heterogeneous networks is an emerging research topic to deal with stringent client requirements. The limited computing resources of an ESP (e.g., small-cell base stations) can be utilized for transcoding the high resolution contents and thereby supporting the bit rate requirements of the clients. Unlike existing works, this paper deals with the case where the storage and computing capacities of the ESPs are random and time varying (e.g., due to internal workloads). This aspect is important because a <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">content provider</i> (CP) cannot be aware of the exact resource availability at an ESP when determining proactive content placement. Different from existing works, we also ensure that the bit rate requirements of the clients are maintained throughout their video streaming sessions. Further, uncertainties in the demand (i.e., number of content requests) and link capacities are also considered. Given these unique challenges, we propose a two-stage stochastic problem which jointly determines the content placement decisions and demand allocation decisions for the CP under various uncertainties. The performance of the proposed model is rigorously analyzed in terms of energy consumption and percentage demand allocation, and compared to some benchmark models through extensive simulations.