|
|
Mark Kranz
kranzm01@csse.uwa.edu.au
Entry year: 2004
Enrolment status: checked and confirmed
Degree: BCompSci
Degree status: complete semester 1, 2005
Project: Rule-Based Programming of Wireless Sensor Networks
Supervisor(s): Rachel Cardell-Oliver
Project status: complete semester 1, 2005
|
|
In the field of environmental monitoring, computer applications are often employed to detect occurrences of complex events patterns with spatial and temporal characteristics.
The gathering and delivery of the associated data is typically carried out by deeply embedded systems such as wireless sensor networks.
Traditional approaches to event pattern matching involve the communication of sensor data to a central database, where powerful computers can perform offline data mining.
However, this approach is only sufficient for passive monitoring, as it does not allow the network to react to events.
It also wastes valuable energy by transmitting all data regardless of its relevance.
A novel approach to these problems utilises in-network processing to perform on-line detection of event patterns.
Development as a middleware application allows applications to subscribe to situations of interest, whilst ignoring irrelevant data.
A prototype system has been designed and built to illustrate this, using Berkeley Mica2 motes and the TinyOS operating system.
This system has shown that in-network situation detection is a feasible approach.
It has been shown to be capable of expressing a variety of situations, from explosion detection to frost monitoring.
It is also shown acceptable performance in typical environmental monitoring applications, despite hardware limitations of speed and memory.
|
 Proposal
Dissertation
|
Last update: Wed Jul 27 10:37:17 2005
|
|
