NeTS: Small: Support for Interactive AR/VR Video: Learning and Optimizing at the Network Edge

点击翻译按钮获取中文摘要

Augmented and virtual reality (AR/VR) are increasingly used in new application areas including entertainment, medicine, public safety, architecture, personnel training, etc. Mobile AR/VR systems allow users to be untethered, thus making it convenient for use in new settings. Edge networking and computing is a promising way to support demanding, low latency AR/VR applications, however there is a need for a deeper understanding of what functionality an edge system should provide to support those applications. This project improves user experience with AR/VR applications by using compute capability in edge cloud nodes located close to the user. The goal is to implement a system for real-time streaming and analysis of AR/VR videos that jointly uses the network efficiently while maintaining a high user Quality of Experience (QoE). The first activity is to develop a common set of functional building blocks that implement a set of generic AR/VR applications.The second goal is to develop the techniques to chain these building blocks together in a way that effectively uses an edge server composed of a mix of both set Central Processing Units (CPUs) and Graphics Processing Units (GPUs). An evaluation will be performed on a testbed of AR/VR devices and smartphones through trace-driven simulations, using publicly-available user datasets.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.