A scalable spatio-temporal query processing engine for linked sensor data

The ever-increasing amount of Internet of Things (IoT) data emanating from sensors and mobile devices is creating new capabilities and unprecedented economic opportunity for individuals, organizations, and states. To fully realize the potential benefits of these sensor datasets, two fundamental requirements need to be addressed, namely interoperability and an effective data management system. Fortunately, a suite of technologies developed in the Semantic Web effort, such as the RDF model, Linked Data, and SPARQL, can be used as some of the principal solutions to help sensor data from the challenge of poor interoperability. However, in this context, providing an effective data management system for sensor data that can combine the benefits of Semantic Web principles, and also be able to deal with the ''big spatio-temporal data'' nature of sensor data, is still an open challenge. Central to this problem is not only knowing how to store a massive volume of sensor data, but also being able to answer a complex spatio-temporal related query on large-scale sensor datasets in a timely manner. Researchers in the Semantic Web community have proposed a substantial number of works that use Semantic Web technologies for effectively managing and querying heterogeneous sensor data. However, our research survey revealed that these solutions primarily focused on semantic relationships and paid less attention to the temporal-spatial correlation of sensor data. Moreover, most semantic approaches do not have spatio-temporal support. Some of them have addressed limitations as regards providing full spatio-temporal support but have poor performance for complex spatio-temporal aggregate queries. In addition, while the volume of sensor data is rapidly growing, the challenge of querying and managing the massive volumes of data generated by sensing devices still remains unsolved. In this work, we propose a scalable spatio-temporal query engine for sensor data based on Linked Data model, called EAGLE. The ultimate goal of our approach is to provide an elastic and scalable system which allows fast searching and analysis on the relationships of space, time and semantic in sensor data. In order to support spatio-temporal computing, we introduce a set of new query operators which is compatible with SPARQL 1.1. For dealing with ''big data'' and a high update throughput of sensor data, EAGLE adopts a loosely hybrid architecture that consists of different clustered databases. This flexible architecture not only helps the engine with the overhead of ''big data'' processing but also allows us to make use of the existing spatio-temporal query functions provided by the underlying databases. The engine also provides a learning optimization approach that can predict query performance based on historical query execution plans. To demonstrate the advantages of the query processing engine in terms of performance, the thesis provides extensive experimental evaluations. The evaluations cover a comprehensive set of parameters that indicate the performance of spatio-temporal queries over Linked Sensor Data