Effect of time interval on speed estimation using single loop detector measurements

Abstract

Considering the importance of estimating speed from single loop detector measurements for many Intelligent Transportation System (ITS) related applications, a variety of algorithms have been developed in the literature for specific time intervals. However, the performance of these algorithms over a spectrum of time intervals has not yet received appropriate attention in the transportation research community. In this study, a Kalman filter based algorithm is selected as a typical representative and investigated over a spectrum of 30 time intervals starting from 1-minute to 30-minute with one minute increment. Empirical results using real world data show that the selected algorithm has workable performances for most of the time intervals under investigation. Specifically, the performances of the selected approach improve for intervals from 1-minute to 5-minute, stay stable for intervals between 5-minute and 15-minute, and decrease slightly for longer time intervals greater than 15-minute. The results demonstrate the ability of the selected algorithm to be readily implemented in a variety of transportation applications with specific time interval needs. Future work is recommended to develop a framework of coupling speed estimating algorithms and real world transportation applications through investigating other single loop speed estimation approaches and ITS application related time interval needs.