Event detection of any process like weather monitoring, animal habitat monitoring, vehicle traffic monitoring etc employ WSNs to detect any abnormal changes in the environment.
Most kind of outliers detection.
So what is the best algorithms to detect above common events.
Like if I want to check whether train is coming, then I can put a accelerometer sensor and put some threshold which if crossed the train has passed.
But what could be a better and more mathematical way of accurately knowing an even.
Whether neural networks or statistical based methods work well.
Any suggestions are welcome.
Possibly a combination of the two methods proposed would help in event detection.
Following open to correction.
1. Obtain PMF of suspected event from predetermined variables.e.g at quiescent
states and when event occurs .
2 Use above training sets in ANN
3. Finally Bayesian decision techniques for categorization etc depending on
severity of event.
4. It would appear that sensor fusion techniques for multiple sensor
location/type is an additional possibility.
Good luck
Cheers
Dear sir
I read some paper when they use GoF tests for event detection combined with evidence theory
http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=6140536&url=http%3A%2F%2Fieeexplore.ieee.org%2Fiel5%2F7361%2F6189422%2F06140536.pdf%3Farnumber%3D6140536
Refer to this.
Good luck
- Badges
- Science topic