Data Science and Engineering Seminar II
We invite all who are interested to the next DSE seminar to be presented by Dr. Silvia Nittel, Spatial Informatics, School of Computing and Information Science.
The seminar will be Friday April 24 at 3:00 in room 336 Boardman Hall.
From Sensor Streams to Fields
Silvia Nittel, Spatial Informatics, School of Computing and Information Science
Over the last decade, the world has seen an ever increasing deployment of small, light-weight sensors, connected via various wireless network topologies, each monitoring and collecting data about the geographic world. Such sensors produce continuous streams of updates allowing us to observe the world almost live. With this information at our fingertips, we are facing novel challenges: how to manage this deluge of sensor data? How to model real-world phenomena and their changes that we can now observe both over space and time? How to analyze these phenomena in real-time?
Over the last decade, data stream engines have matured as platforms to plug in sensor data streams and run simple, continuous queries. However, users are better supported if they would be able to interact with higher-level abstractions of the real-world phenomena, rather than taming the individual measurement streams. Dealing with individual streams requires that users need to write code that not only copes with the real-time nature of streams but also that fact that the streams need to be integrated and analyzed, continuously, which is a non-trivial task.
In this talk, I present our recent work on extending data stream engines with a generic interface to dynamic, continuously-distributed spatial phenomena. This easy-to-use, convenient and flexible interface is based on the concept of a field. Similar to a magnetic field in physics — where we assume that a force is present at any point in time and space over an observed area — we talk about fields in geographic environments: for example, temperature fields, soil moisture fields or pollen fields. While fields are widely used to describe continuous phenomena in spatial information science, information systems with field-based interfaces are still rare. Nevertheless, due to their mathematical foundation, fields are excellent candidates to represent higher-level concepts derived from observation streams. In our research work, we formally extended data stream engines with fields, on the data model and query language level, as well as on the computational side. We have built a data model foundation by formalizing the mapping between fields, streams and spatio-temporal relations, and developing the generic, flexible continuous field stream data model. Secondly, we have investigated efficient stream operators algorithms to achieve near real-time evaluations of query operators over fields. These contributions allow users to delegate the sorting, alignment, integration, interpolation, and simple analysis of massive sensor data streams to the data stream engine. This in turn enables users to immediately focus on higher-level analysis. We expect that this generic, well-defined interface to dynamic, continuously-distributed spatial phenomena will have a dramatic impact on the use and the analytical opportunities of real-time sensor data streams.