The potential of low-frequency AVL data for the monitoring and control of bus performance

In this paper we investigate the potential of "low-frequency" bus localization data for the monitoring and control of bus system performance. We show that data with a sampling rate as low as one minute, when processed appropriately, can provide ample information. In particular, we obtain accurate estimates of stop arrival and departure times which in turn allow the analysis of headways and travel times. A three parameter gamma family of distributions is fitted for headways at the stops along a bus line. The evolution of the parameters demonstrates critical points on the line where bus bunching is significantly increased. Moreover, this analysis allows differentiating problems associated with varying passenger demand from uncertainties associated with traffic conditions. Furthermore we show that both expected travel time and travel time variability can be calculated from low-frequency localization data. Finally, we present how our results can be used to calibrate a simulation model which can test bus control strategies. We apply and validate the methods to data obtained from bus route number 1 in Boston.