Research

Dr Nittel’s main research interest is in data management technologies for large sensor networks. Dr. Nittel is the lead of the Geosensor Networks Laboratory at the School of Computing and Information Science at the University of Maine. She currently advises 3 Ph.D. students.

Research at  the University of Maine:

Real-time Analytics of Massive Sensor Data Streams

The current main research focus in on extending data stream engines for real-time analytics of sensor data streams. First work in this area comprised an approach to use a data stream paradigm to perform statistical compression of massive amounts of remote sensing data in order to reduce the data sets but to preserve interesting characteristics within the data sets using a data stream based k-means algorithm (ICDE’04, SSDBM’04 and Journal of Computational and Graphical Statistics 2004).

Currently, we focus on real-time processing of massive amounts of distributed sensor streams as found in crowd sensing applications over large metropolitan areas.

  1. Our interest is on supporting continuous environmental phenomena through a high-level abstraction of a dynamic field. In our work, the field concept is the foundation of the novel dynamic spatial field data model and query language for data stream engines.
  2. Much of our research has focused on exploring and implementing efficient data stream operator frameworks for data stream engines to achieve near real-time processing of sampled fields. In our tests, we experimented with sampled field (radiation, Japan) based on up to 250 sensor update/sec from stations distributed over a geographic region. We have designed and developed stream operators that implement spatial interpolation of up to 250K sample points in under 2 seconds. In  “Geostreaming 2014”, we published novel stream operators for spatio-temporal interpolation of up to 250K asynchronuous sensor updates within a query window.
  3. We are currently working on a field data model and operator algebra for sensor data streams.

For more information please see the following publications:

From Streams to Fields: The Field Data Model for Spatio-Temporal Data Streams

Massive sensor data streams are created from the automatic collection of sensor data in high frequency and in near real-time today. This project aims to advance the analytical potential of live-streamed data, historical data streams, and model simulations by creating an overarching representation in the form of the field data model with a set of operators that establish the field algebra.

This project develops the field algebra, which is an intuitive, yet mathematically defined formalism to represent real-world phenomena as fields and to express analytical needs as canonical operations over fields. The field model represents phenomena as continuous entities again, and the implementation hides the fact that their Spatio-temporal continuity is calculated on-the-fly based on real-time measurement streams. Extending sensor data streams to fields is transformative, as rarely a domain scientist is interested in the readings of individual sensors. Allowing scientists to work with high-level abstractions will significantly enhance their analytical tasks such as finding insights about changes, trends, or unexpected events happening in the real world.

This work has been funded under an NSF Grant (2015-2020). See here for more information.

Data Management for Ad-hoc Geosensor Networks

From 2003-2010, Dr Nittel’s research focused mainly on data management for ad-hoc geosensor networks. Geosensor networks comprise sensor network technology (such as Intel Motes and TinyOS/TinyDB) deployed for environmental applications.  The current main research interest is in the detection, monitoring and tracking of continuous phenomena such as toxic plumes or regions of toxic algae blooms in the ocean. She developed several approaches using discrete measurements of sensor network nodes to estimate the continuous nature of such phenomena. Another research approach is constituted the fundamental idea of switching from processing quantitative information about such phenomena to a qualitative approach. In this case, information, communication and message size between neighboring nodes can be significantly reduced while still being able to track the boundary and/or behavior of continuous phenomena (ACMGIS 2005 publication).

For more information see the following publications:

A third larger research focus is on mobile ad-hoc geosensor networks. In this domain, most of the nodes in the sensor networks are mobile in geographic space. Dr. Nittel developed several approaches of efficient information dissemination in mobile geosensor networks detecting different types of spatio-temporal events. A simple type of such an event is e.g. an icy patch on a street network, and automobiles detecting the event and strategies to efficiently and effectively disseminate the information to other (mobile) nodes in the network (GiScience 2004 publication). This work was eventually extended to the problem ad-hoc shared ride systems in an urban transportation network consisting of pedestrians, cars, taxi cabs and public transportation. In this case, the information of spatio-temporal ‘event’ (the pedestrian in need of a ride to a certain destination) has to be disseminated to transportations hosts (cars, taxis, etc). At the same time, ride offers have to be routed back to clients in time, and clients need to perform routing planning based on the current, dynamically changing situation (IJGIS 2006 paper, “Societies in the Age of Instant Access” 2006 book chapter). 

For more information see the following publications:

Earlier work before UMaine:

Dr. Nittel received her Ph.D. in 1994 from the Computer Science Department of the University of Zurich where she specialized on storage systems for extensible and object-oriented DBMS. She joined the UCLA Computer Science Departmentas postdoctoral researcher in 1995, and worked on high-performance data integration and mining platforms for large heterogeneous remote sensing datasets. In detail, her research focused on data stream-based tools for scientific data mining and scientific collaboration (i.e. the Conquest system). She also was the Co-Director of the UCLA Data Mining Lab (with Dr. R. Muntz), managing a 5 institution NASA grant. She was a co-author of the OGC Simple Feature Specification, and a “Tiger Team” member of interoperability committee within the NASA Earth Science Data Federation.