Evaluation of Bluetooth for Traffic Stream Travel Time Estimation Under Indian Conditions

Travel time estimation has been a high interest topic in traffic operation and management for years. Travel time information is more useful to the user than other parameters such as speed, flow, density, headway etc. However, travel time being spatial in nature, requires a spatial sensor for its measurement. The spatial sensors available for such a data collection include GPS, AVI, mobile phones, Bluetooth, etc. These can measure the travel time directly; however techniques like GPS and AVI require participation of the user for data collection, which is difficult especially in private vehicles. Use of mobile phones is another possibility, which has lesser location accuracy since it uses triangulation to find the location of vehicles. Bluetooth, Zigbee, and wifi sensors are other possibilities and are reported as possible source of travel time data from western countries. They have the advantage of no participation requirement and reasonably good location accuracy. However, the disadvantage is that they can capture only a small percentage of the entire traffic stream. The present study explores Bluetooth sensor as a traffic data source under Indian conditions. The first part of the study checks the penetration rate of Bluetooth to check the possibility of using it as a traffic data source under Indian conditions. The second part of the study analysed and identified the type of vehicles being represented by Bluetooth. This is important under Indian conditions, where several different types of vehicles are sharing the same roadway. And the last part of the study presents the stream travel time estimation from these selected classes of vehicles being represented by Bluetooth data. This involved the use of a weighted linear regression model. The results of modeling are promising and show potential for real time ITS applications.