Developing Origin-destination Matrices Using Bluetooth Data for Strategic Transport Models- Worcester Case Study
Hariharan Thogulava, CH2M Hill, Viktor Antonov, CH2M Hill, Simon Bingham, CH2M Hill
This paper explores the use of Bluetooth traffic data to generate Origin-Destination matrices to inform the development of a transport model and techniques used to address concerns regarding the multiple detections and missed detections etc.
Developing Origin-destination matrices using Bluetooth Data for Strategic Transport Models- Worcester case study
Authors: H. Thogulava, V. Antonov, S. Bingham
Transport Planning and Advisory, CH2M Hill, Worcester
Robust data is a primary need for building transport models. Growing demand for transport and global budget squeeze of public spending lead to the growing interest for the cheaper and yet reliable sources of traffic data collection. Whilst traffic flow counts are obtained from loop detectors and video counts, travel patterns are generally obtained from roadside interviews and household travel surveys that provide valuable information for transport modelling but are expensive to implement, intrusive, contain a lot of bias and to a certain extent are stated preferences of the travelers. In this regard, data from Bluetooth devices could provide low cost revealed information about vehicle travel patterns to estimate origin-destination matrices and travel times. Previous research into Bluetooth detections reported number of issues such as high misdetection rate, identical MAC IDs, multiple devices within one vehicle, etc.
This paper aims to find alternative means of addressing issues identified with using Bluetooth data to estimate OD matrices by establishing relationships between multiple data sources. This paper is based on an ongoing study to update an existing transport model for Worcester, UK. Worcester is a small city with a population of nearly 100,000, served by a network of radials roads linking it with surrounding population centres. As part of the study, multiple data were collected that included roadside interviews, loop detectors, video counts, parking counts and Bluetooth traffic data collected over three days, including the day of roadside interviews, across a network of detectors laid across multiple cordons, screenlines and parking lots.
Over 100,000 Bluetooth records were collected on each of the three days surveyed. The data will be compared over each day, with count data at each location, with road side interview responses to Bluetooth devices being enabled as well as the vehicle arrival profiles at parking lots to elicit the Bluetooth penetration rate and validate the travel patterns.
The paper will describe the processes adopted to obtain origin-destination matrices from the Bluetooth datasets and demonstrate its viability as an alternative to roadside interviews and household travel surveys.
Association for European Transport