Poster: Adaptive Video Offloading in Mobile Edge Computing

By 2022, videos will account for 82% of global Internet traffic. Many camera-based mobile devices, though with limited resources, require ultra low-latency video analytics such as object detection and action recognition. To facilitate these devices with video offloading solutions, Mobile Edge Computing (MEC) is facing several key challenges due to uncertainties associated with the problem. This paper addresses these challenges by proposing a Two-stage Stochastic Program and a novel clustering-based Sample Average Approximation to effectively solve the video offloading problem in uncertain dynamic environments, while satisfying the required latency. The video offloading quality and other offloading decisions are adaptively made to jointly optimize the video quality and migration cost under the consideration of uncertain device mobility.