Signal processing on graphs for urban traffic modelling

• Main Author: Vajapeyam Shreyas Nagaraj
• Other Authors: Justin Dauwels
• Format: Theses and Dissertations
• Language: English
• Published: 2015
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Electronic systems
• Online Access: http://hdl.handle.net/10356/64876
• Institution: Nanyang Technological University

The exponential rise in road traffic has led to more congestion on roads, thus resulting in unpredictability and delay in road travel especially in urban centres. This has led to research engineers focusing on congestion avoidance algorithms, in an attempt to maximize road network capacity whilst minimizing travel time delay, thus leading to optimal use of road networks. Consequently it demands for radical approaches to analyse road traffic network for anomalous behaviour In our present model we propose to employ wavelet functions on weighted graphs to detect traffic events in road network. We have made a novel approach of using spatial as well as temporal features to mine the traffic data. The data from sensors employed on traffic networks is very exhaustive and it is very hard to get comprehensive information just by observing the parameters like road occupancy and flow rate per hour. Today's ITS systems are smarter and there needs to be prediction techniques employed and these predictions need to be accurate, so that the commuter is benefited. It is shown that this model can be used to find out traffic events on a particular road on the network. Also the number of links affected by change in the traffic behaviour on a particular network can be inferred. This can in turn be used by the Intelligent Transport System for prediction of events in a future horizon and also to alert drivers well in advance.