The UCD CASL SenseTile System
Funded by Science Foundation Ireland
Co-PI: Dr. Scott Rickard
Sensor networks are widely seen as an essential technology for supporting next-generation scientific and engineering challenges, including environmental monitoring, climate change, assisted living, national security and intensive agriculture. Addressing these problems requires two complementary strands of research:
- to understand the specific physical and/or social phenomena of interest against which to fit data collection and analysis
- to understand the general scientific and technological principles governing scalable sensor networking in order to maximize the leverage gained from on-going developments.
Amongst the core general challenges are the collection of large volumes of multimedia data, its storage, cataloging, retrieval and processing in such a way as to minimize physical and intellectual costs of accessing the available data. This is especially important in pursuit of cross-disciplinary projects involving different analysis methodologies and constraints. At the same time, it is impossible to study sensor networking completely separate from practical applications which provide real-world validation and verification of the techniques being developed.
This proposal requests funding for a facility, the SenseTile system, to support large-scale experiments with complex multimedia sensing and processing at terabyte scales. The proposed equipment consists of a rich and reusable sensor platform easily deployable into the built environment for easy experimentation and representative of platforms suitable for wider uses, together with data storage and processing capacity and associated high-speed interconnect.
The delivery of this facility will allow the research teams to attack problems on a scale and in a manner not previously possible. The key research challenges this facility will address include:
- dealing with uncertainty in sensor systems
- robust management of large (multi-terabyte) sensing data on heterogeneous platforms
- interaction and activity recognition
- sparse signal analysis
- coding and securing sensor interactions
Underpinning each of these key challenges is a fundamental need for an integrated, multi-modal sensor test bed that has two unique characteristics. firstly, it must be reconfigurable and be easily targeted to collect defined data sets. Secondly, and perhaps most importantly, it must be of sufficient size and complexity to ensure that the challenges listed above, and the resulting theories and techniques, can be evaluated in a scientifically rigorous manner. It is our contention that this facility will provide a capability unique in the world in terms of its scale and flexibility for performing in-depth investigation of both the specific and general issues in large-scale sensor networking. Furthermore, it will provide synergistic benefits in supporting the development of reference data sets and services for use by other researchers internationally.