Intelligent Transportation Systems

Friday Seminar - Marta C. Gonzalez on Characterizing Urban Road Usage Patterns

Traffic I Missed

This week's Friday Seminar is MIT's Marta C. Gonzalez presenting "Charactericing Urban Road Usage Patterns with a New Metric." The seminar will take place from 4-5 PM in 506 Davis Hall on 23 September. 

Mobility data from half million anonymous mobile phone users are used for this presentation to study the road usage patterns in the Bay Area. Using this mobility data based on our modeling framework each trip’s route is predicted. Surprisingly, it is found that on average 60% of the vehicles passing through a road segment come from 1% of its drivers’ home locations, hinting to high predictability of the vehicle sources. To quantify the heterogeneous traffic contributions of the vehicle sources we use the Gini coefficient and find that a road segment’s Gini coefficient is poorly correlated with its betweenness, traffic volume, and volume over capacity, suggesting that Gini coefficient is a new metric on top of the traditional measures, quantifying road usage patterns in the perspective of drivers’ demographic distribution. Finally, based on the road usage patterns, we find an efficient strategy to mitigate traffic congestion through a tiny decrease of car usage rates in a few targeted neighborhoods.

Don't forget about Cookie Hour in the library at 3:30! See you then.

Special Wednesday Seminar! Lin Zhang discusses Crowd-sourced Mobile Urban Sensing


This afternoon, Spetember 21,  there will be a special Wednesday Seminar. From 2-3 PM in 406 Davis Hall, Lin Zhang of Tsinghua University will present "Crowd-sourced Mobile Urban Sensing".

Wide area urban sensing is a topic of interest both within industry and academia, as well as a technique urgently needed by both city governments and urban residents.  In today’s rapidly urbanizing world, the urban sensing system provides up-to-date, complete and detailed observations of the climate, environment, traffic, and population of a city, all information which can aid government officials in the decision-making process.  The urban sensing system is also a frontier of Internet development, enabling cyber space to sense the ambient environments in which it is embedded.  However, there are two major challenges to urban sensing: communication capacity and sensing capability. 

This presentation will introduce a taxi-cab based mobile sensor system that was designed for wide-area urban sensing purposes.  The presented system addresses both of the aforementioned challenges as well as  considerations of economic and technical feasibility.  The system crowd-sources the sensing tasks to a group of taxi cabs roaming the city, and uses the store-carry-and-forward mechanism to collect and send sensory data to the data center for processing.  Compared to a static, dedicated sensor network, the system enjoys extremely low deployment costs with fairly strong coverage and performance.  The presentation will also describe the details of the system design, including the wireless channel measurement, an energy efficient neighbor discovery method, a utility-based routing protocol for data delivery, and a compressive sensing field recovery algorithm that exploits the sparsity of the physical field in order to reduce the volume of the required sensing data.  The presentation closes with a a discussion of future deployment plans and research directions for the system.


Syndicate content