Assessment of GPS-enabled smartphone data and its use in traffic state estimation for highways

GPS-enabled cell phones provide new opportunities for location-based services and traffic estimation. When traveling on-board vehicles, these phones can accurately provide position and velocity of the vehicle, and therefore can be used as probe traffic sensors. The focus of this thesis is to assess the feasibility of using GPS-enabled cell phones for traffic monitoring purposes.;This thesis first presents a field experiment nicknamed Mobile Century, which was conceived as a proof of concept of a GPS-enabled cell phone based traffic monitoring system. Mobile Century included 100 vehicles carrying a GPS-enabled Nokia N95 phone driving loops on a 10-mile stretch of I-880 near Union City, California, for 8 hours. The data obtained in the experiment was processed in real-time and successfully broadcast on the internet, demonstrating the feasibility of the proposed system for real-time traffic monitoring. Results presented in this thesis suggest that a 2-3% penetration of cell phones in the driver population is sufficient to provide accurate velocity of the traffic flow.;The second part of the thesis proposes and assesses methods to perform traffic state estimation in the presence of data provided by GPS-enabled cell phones. Traffic state estimation uses the measurement collected to produce estimates for locations for which no measurement is available. Three methods to incorporate mobile probe measurements into highway flow models are proposed. The first one is based on Kalman filtering, the second is a heuristic method derived from the first method, and the third technique is an extension of a technique used in oceanography called Newtonian relaxation. The three methods assume the knowledge of the fundamental diagram and the conditions at both boundaries of the section of interest. Data from intermediate ramps is not required, since mobile sensors are expected to provide data to infer the state of the system at intermediate locations. The performance of the methods is assessed using two different datasets. The first dataset consist of NGSIM data, in which all vehicle trajectories are known. The second dataset consists of the Mobile Century data. It is shown how the accuracy in the estimation depends on the number of measurements available and their accuracy. For the cases investigated, one observation per mile-lane per minute provides sufficient data to identify most of the congestion.