|
Activity Number:
|
474
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 9, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Physical and Engineering Sciences
|
| Abstract - #307122 |
|
Title:
|
Dynamic Origin-Destination Matrix Estimation from Partially Observed Trajectories in a Large Network
|
|
Author(s):
|
Jaimyoung Kwon*+ and Pravin Varaiya
|
|
Companies:
|
California State University, East Bay and University of California, Berkeley
|
|
Address:
|
Department of Statistics, Hayward, CA, 94542,
|
|
Keywords:
|
origin-destination matrix ; network ; graph theory ; partially observed data ; space-time data
|
|
Abstract:
|
We recently developed Method of Moment and bootstrap methods for utilizing a novel data type, partially observed packet or vehicle trajectories, for making inferences about time-varying origin-destination (OD) matrices in computer or traffic networks. There are various challenges when the algorithm is applied to a large network with complicated topology. Put in graph theoretic terms, it stops working if the graph representing the network has cycles or is not uniquely traversable. In the current paper, we address those issues and present a systematic method for resolving the problem by incorporating trip pattern information in the model. Improving the estimator by smoothing over time and space will also be discussed. The new approach is applied to partially observed vehicle trajectories data from FasTrak infrastructure in a very large freeway network in San Francisco Bay Area.
|
