ICE-T:RC: Optimizing Internet video through support from the network edge

Purdue University, West Lafayette IN

Investigators

Abstract

Despite many advances in recent years, the performance of Internet video often falls below user expectations. The issues are likely to be exacerbated with the emergence of bandwidth-intensive 4K video, and more recently, interactive video such as 360-degree video which is latency sensitive as well. This project investigates new opportunities opened by recent Internet Service Provider (ISP) interest in deploying value-added services at the edge of the provider's network. Such edge-based services can aid video streaming applications through lower latencies, higher throughput, and visibility into network performance bottlenecks. This project will explore novel services that may be deployed at the ISP edge, which can significantly improve the performance of video streaming solutions, as well as develop new approaches to video streaming that can leverage these services. The research will systematically characterize these benefits, develop new mechanisms, and consider practical deployment concerns. Specific research contributions will include: 1. Techniques for proactively estimating the rate at which video applications can transmit informed by explicit information collected at the network edge in real-time. The approach includes edge-based monitoring and estimation to ensure scalability, and extensive use of network data to refine and validate our algorithms. 2. Redesigning state-of-the-art available bit rate (ABR) approaches to exploit proactive rate estimation. The project team will also explore a new design point where the ABR algorithm itself is run on an edge server rather than the client. 3. Measurement-based characterization studies of interactive video streaming to investigate issues around edge support for interactive video. Specifically, the project will consider the interplay between the encoding scheme used for 360-degree video, user interactivity models, and the edge cache hit rate. This project will extend an ongoing research collaboration between Purdue University and the Politecnico di Torino, Italy. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.