Spatiotemporal framework for smart city event processing

Abstract

Events are activities happening in the real world that cannot entirely be foreseen. They can be measured by sensors and represented in a computer system by data. Data produced by sensors has become more voluminous in size, is high in variety and is quickly made available through real time data streams. Event processing is the set of operations required to handle events, including detection, tracking and analyzing of events and it provides the possibility to take immediate action. Event processing has proven useful in business environments, for example for the detection of patterns on stock markets, but in a smart city context it can be used for the detection of urban phenomena, such as a pollution cloud forming. The aim of this research is to contribute to the development of mechanisms that facilitate processing and analytic workflows to exploit sensor data in a smart city context. To this end, the research questions is as follows: To what extend can real time event processing be used to process multiple data streams in a smart city context, using open standards? The components of an event based geoprocessing system are identified. The methods for the specification, detection and notification of spatiotemporal events are explored. A second research question relates to presenting results of event processing: To what extend can users be provided with a web application that allows for specifying and notifying events? In this context, users are city staff monitoring a city for air pollution events. A web application is used for notifying city staff in case of events, but it also allows users to specify interests in particular events without having to know the technical details behind it. These questions are answered by applying event processing in an air pollution case study. Using real world sensor data from the SmartSantander project, it is demonstrated that event processing can be used for detecting urban phenomena.