Tariq Alsahfi

ORCID Identifier(s)


Graduation Semester and Year




Document Type


Degree Name

Doctor of Philosophy in Computer Science


Computer Science and Engineering

First Advisor

Ramez Elmasri


Advanced technologies in location acquisition allow us to track the movement of moving objects (people, planes, vehicles, animals, ships, ..) in geographical space. These technologies generate a vast amount of trajectory data (TD). Several applica- tions in different fields can utilize such trajectory data, for example, traffic control management, social behavior analysis, wildlife migrations and movements, ship tra- jectories, shoppers behavior in a mall, facial nerve trajectory, location-based services (LBS) and many others. Fortunately, there are now many trajectory data sets avail- able that collected from moving objects such as cars with enabled GPS devices. Two main challenges arise when dealing with TD: 1) storing and analyzing TD data due to a large amount of data that arrives in a streaming and unpredictable rate. 2) inaccurate capturing of the exact location of moving objects due to the errors caused by GPS devices. In order to tackle these two problems and gain useful knowledge from TD, in this dissertation, we provide a framework called Trajectory Data Ware- house (TDW). This framework aims to review existing studies on storing, managing, nd analyzing TD using data warehouse technologies. Furthermore, we provide the requirements for building the TDW with different applications using the TDW. Despite the second challenge, in this dissertation also, we utilize the vast amount of TD by building a digital road map. Because road maps are important in our personal lives and are widely used in many different applications; therefore, an up-to- date road map is essential. We propose a novel method to generate road maps using GPS trajectories that is accurate with good coverage area, has a minimum number of vertices and edges, and several details of the road network. Besides, our algorithm extracts road features such as turn restrictions, average speed, road length, road type, and the number of cars traveling in a specific portion of the road. To demonstrate the accuracy of our proposed algorithm, we conduct experiments using two real data sets and compare our results with two baseline methods. The comparisons indicate that our algorithm is able to achieve higher F-score in terms of accuracy and generates a detailed road map that is not overly complex. Lastly, we present a data fusion framework for heterogeneous data Sources for Intelligent Transportation Systems (ITS). This framework aims to provide data fusion techniques to integrate and extract features from heterogeneous data sources to be ready for deep learning training approaches. We also generate preprocessed real-world traffic datasets that are publicly available to solve ITS-related problems. The traffic datasets have rich features such as traffic flow, average speed, vehicle occupancy, weather conditions, incidents information, congestion reports, point of interest locations, and temporal features. Furthermore, we provide two applications to show the importance of our data fusion techniques. (1) Traffic datasets analysis and visualization, where we build a data cube to provide in-depth analysis of the dataset. Also, a visual-interactive GIS tool that presents the results in different methods. (2) Traffic flow forecasting using deep learning, we performed a comprehensive study onhow different features can improve the traffic flow prediction models. The results show that deep learning approaches achieved better results when extra features are considered.


Trajectory data, Map generation, Data fusion, Transportation systems, Deep learning


Computer Sciences | Physical Sciences and Mathematics


Degree granted by The University of Texas at Arlington