Data fusion and missing data estimation in road networks

• Main Author: Foo, Her Yiow
• Other Authors: Justin Dauwels
• Format: Final Year Project
• Language: English
• Published: 2015
• Subjects: DRNTU::Engineering::Electrical and electronic engineering
• Online Access: http://hdl.handle.net/10356/64363
• Institution: Nanyang Technological University

Often in urban area, road users would like know the traffic condition and how long it would take to reach the destination. The focus of the project will be providing such information for traffic applications. This includes recovering low resolution large scale urban traffic data, predicting future traffic data and recovering missing traffic data using tensor methods which is essential for intelligence transport system (ITS) application. We developed a model for recovering of low resolution traffic data with the availability of higher resolution data in real-time, drivers can plan their journeys high low uncertainty. Partial least square regression method is used in estimating large scale urban traffic with fusion of traffic speed, flow and/or speed band as inputs. CANDECOMP/PARAFAC (CP) Tensor factorization specifically Bayesian CP and CP weight optimization (CPWOPT) will be used in estimating missing traffic data. The estimated results are compared with the actual speed for performance evaluation and demonstrated to be optimistic.