110 likes | 259 Views
Calling All Cars: Cell Phone Networks and the Future of Traffic. Jeremiah Dunn. Overview. Introduction Mobile Millenium Goal Complexity of the Problem Gathering Data Data Fusion Modeling the Flow of Traffic Mobile Century Conclusion. Introduction.
E N D
Calling All Cars: Cell Phone Networks and the Future of Traffic Jeremiah Dunn
Overview • Introduction • Mobile Millenium • Goal • Complexity of the Problem • Gathering Data • Data Fusion • Modeling the Flow of Traffic • Mobile Century • Conclusion
Introduction • Fixation of humanity on futuristic cars & autonomous travel has drastically changed the modern car. • While not quite going in that direction, cheap sensors and network availability are essentially boosting the “brainpower” of our driving environment. • Between road-side sensors, dashboard GPS, and Smart phones many companies are provided with traffic data collection.
Mobile Millenium • Started in 2007 • One of the first large-scale projects for traffic monitoring • Run by Nokia, NAVTEQ, and UC Berkeley • Only able to be conceived and work thanks to the rise of the “smart-phone” thanks to embedded GPS
Goal • Merge road-side sensor networks with smartphoneGPS feedback to generate a real-time traffic monitoring situation.
Gathering Data • VTLs (Virtual Trip Lines): to prevent constant packet transfer, the phone will only upload statistics when it crosses a “checkpoint” along the VTLs.
Data Fusion • Incoming data from many sources • GPS • Buses • Taxis • Cars • Static Sensors • Loop Detectors • RFID tag readers • GPS may have faulty Data • Walkers/Parked/etc
Modeling the Flow of Traffic • Obvious way to think about modeling traffic is by individual cars • Designed a new set of algorithms based on fluid mechanics
Mobile Century • All this data collection culminated in 2008 in a test • 100 Cars mixed in a 10-mile stretch in Norther California • 10 Hours and accounted for 2-5% of the cars on the highway • Mobile Millinumvs Google Maps w/ Traffic • Noticed a sudden red blotch appeared on the test stretch, but it took several minutes to appear on Google’s system.
Conclusion • Was able to detect and report a slow-down in under a minute • Proved that only a few cars were needed to get the system to run efficiently (2-5%) • Successful test has led to the concept and technology demonstrated to become widespread into Google’s mobile Maps app. • Proved that the mobile scene was better performed than any static detector system.