Research Statement
As sensor hardware becomes a commodity, the ability to deploy multiple types of sensors is becoming extremely easy.
Therefore, there is a necessity of developing smart systems able to organize and understand the different inputs from the
heterogeneous collection of deployed sensors. These inputs represent the view of the collection of sensors to different
activities in the environment.
We believe that two main areas of research are necessary:
- The study of the causal relations between the different sensors as they react to different activities.
- The study of how to do sensor fusion to increase activity detection, and to decrease use of resources.
In our research, we propose the study of fuzzy and non-fuzzy Bayesian networks because they are able to represent
the possible causal relations between different sensors, and to learn the structure of these relations. These properties can
be used to study sensor fusion, and the response of the sensors to activities in the environment.
|
