Estimating Control Delays At Signalized Intersection By Using Under-sampling GPS Data

Currently, urban traffic congestion has become acute and most often occurredin the signalized intersection. Signalized intersections are running good or notdirectly related to the operational efficiency and service quality of the entire roadnetwork. The performance of the signalized intersection is mainly decided by thedelay time. By now, the most commonly used method for calculating the delay isstill confined to the site survey or formula. Whichever method requires a lot ofmanpower and material resources to carry out an investigation on the basis of trafficparameters. With the application of GPS technology, it is providing a new methodfor estimating the delay of signalized intersection. However, by computing needs,the existing methods require1second sampling interval of GPS data to estimate thesignal intersection delay. There is no doubt that the collection, transmission and datastorage costs increased. At the same time, subjecting to sampling frequency limits,GPS data still have to be collected by researchers.At the moment, buses and taxis in most of big cities are equipped with GPSequipment. Since the data is only used to achieve operational vehicle monitoringand scheduling, sampling interval is often set to30seconds, and a lot of GPS datalate in the idle state. In addition, buses and taxis are in a large scale and almostacross the entire road network, the GPS data are transmitted in real time. So it isnecessary to develop a method which using30s interval GPS data, namelyunder-sampling GPS data.The focus of this paper is researching on the method of calculating signalizedintersection control delay based on the features of the GPS data under-sampling ofbuses and taxis in big cities.First, the existing methods using GPS data to calculate intersection delay are introduced and under-sampling GPS data is defined and a conclusion that those GPSsampling interval greater than10s are under-sampling GPS data. And then, theprocess of vehicle travel at signalized intersection is analyzed based onunder-sampling GPS data and it makes clear that the key to calculate control delayby using under-sampling GPS data is getting the critical time of delay process, thatis the time when vehicle start decelerating and end accelerating.Then piecewise constant deceleration model and acceleration model are builtunder two conditions: One is when the vehicle passing through a signalizedintersection, it stops. The other is the vehicle doesn’t stop. The models are used torestore the vehicles’ movements at signalized intersection and calculate the criticaltime of delay process. And then, the acceleration and deceleration of vehicle passingthrough the signalized intersection is descripted qualitatively and quantitatively toverify the reliability of deceleration and acceleration models.Next, the appropriate algorithms which converting nonlinear and non-positivedefinite equations to nonlinear programming problem are given based on theunder-sampling GPS data to solve models.And then the GPS data whose sampling interval is1second is resampled by30seconds sampling interval to calculate the signalized intersection control delay andthe calculation accuracy could reach to85%. At last, an example of usingunder-sampling GPS data of Beijing buses to calculate control delay is given as casestudy. The relative errors are below20%and it verifies the feasibility andpracticality of the method.