| Summary: | 碩士 === 國立清華大學 === 資訊工程學系 === 94 === A number of WSN (wireless sensor networks) applications have been deployed to periodically monitor the environment and catch any anomaly event.Anomaly events are defined to occur rarely.However, when an anomaly event occurs, it often carries temporal and spacial locality.
The temporal locality of an event indicates that it is likely to continue on for some time once detected.
The spacial locality implies that this event is likely to be detected by neighboring nodes shortly.To provide better observation on an anomaly event, the period of a task that detects this event should be reduced to monitor the environment more frequently.
We call a task a locality-aware task if it supports temporal and spacial locality to automatically adjust its period.
In this paper, we propose and implement Locality-aware TinyOS or LA-TinyOS to be the first locality-aware WSN
operating system.
By embedding locality-aware features in a kernel component, LA-TinyOS provides a more reliable and efficient framework to develop locality-aware tasks.Compared to TinyOS, our LA-TinyOS reduces the lines of code in such a task by more than 80%.Both TinyOS and LA-TinyOS consume similar amount of RAM memory while LA-TinyOS uses less flash memory.
Finally, we use LA-TinyOS to implement a locality-aware monitoring application using two dozens of micaZ sensors.
This application is currently deployed in the public library of our university to report any occurrence of noisy students.
|
|---|
