Committee: 

Dr. Sivakumar, Advisor 

Dr. Fekri, Chair

Dr. Blough

Abstract: The objective of the proposed research is to provide insights into video content consumption, and develop a set of data-driven prediction and prefetching algorithms, based on machine-learning and deep-learning techniques, which accurately anticipates the video content the user will consume, and caches it on edge nodes during off-peak periods to reduce peak-time usage. Video streaming accounts for over 60% of global fixed downstream Internet traffic and 65% of worldwide mobile downstream traffic; and is expected to grow to 82% by 2022. As a result of the increasing growth and popularity of video content, the network is heavily burdened. Typically, upgrades are triggered when there is a reasonably sustained peak usage that exceeds 80% of capacity. In this context, with network traffic load being significantly higher during peak periods (up to 5x as much), we explore the problem of prefetching video content during off-peak periods of the network even when such periods are substantially separated from the actual usage-time.  To this end, we collect and perform an in-depth analysis on real-world datasets of YouTube and Netflix usage collected from over 1,200 users. Equipped with the datasets and our derived insights, we develop a set of data-driven prediction and prefetching algorithms, based on machine-learning and deep-learning techniques, which anticipates the video content the user will consume, and prefetches it during off-peak periods to reduce peak-time usage.