NeTS: Small: Lightweight, Accurate Network Estimation at the Wireless Edge

点击翻译按钮获取中文摘要

Wireless technology has emerged as a critical piece of our modern, national economy. As more users, devices, and services lean heavily on a blend of WiFi and cellular service, it becomes important to understand whether our national wireless infrastructure will meet the demands of tomorrow. Unfortunately, the instruments for understanding such environments have lagged woefully behind and have not adapted to modern wireless technologies. The focus of this work is to bridge that gap to deliver significant improvements in handling WiFi and cellular. This research seeks to deliver instrumentation ten times faster than current technology while minimally draining the battery or impacting other users. The proposed work further explores whether one can achieve such instrumentation entirely for free, taking advantage of work the mobile device is already doing at present but simply ignoring. The research will create new software development kits to provide a ready-made wireless toolbox for app developers to cleverly improve the vast suite of apps available today.The proposed work seeks to explore the extent to which both modern WiFi and cellular links can be rapidly characterized, conducting both extensive lab experiments for validation as well as realizations of the software as additions to popular open source web serving and video streaming software. The proposed work centers on taking advantage of the intrinsic aggregation properties of most modern wireless systems. Through the clever use of sliced, structured, and reordered packet trains, the proposed work operates within existing TCP constructs to deliver accurate network characterization from 0-10 Mb/s in under 250 ms while requiring less than 100 KB of bandwidth. The net result is an characterization that presents that Available Bandwidth on the end-to-end path with minimal network and energy cost. In addition to the impacts of the software, multiple PhD graduate students will be trained, multiple REU (Research Experience for Undergraduate) students in under-represented groups will be mentored, and results will be disseminated through top tier publications. Final broader impacts include releasing extensive tutorials through an YouTube for the broader public on wireless issues.