Adding context on sensor selection problem
Wooyeong Cho Bayesian Network sensor fusion sensor selectionFor my current work, I'm trying to put context into sensor selection problem. To do this, we need to ask
What does context mean in your scenario? (How do you define this - definition)
I don't think I can use linear Kalman filter to integrate the context in this problem. Let's say we have measurement equation like \(y_t = C_tx_t+v_t\), \(x_t\) is the volume of water and \(y_t\) is flow rate. If we set the Context_1 = cloudy, Context_2 = Sunny, \(C_t\) can cover those different condition.
Should we stick to utilizing the Kalman filter or should we use another approach that is able to include more complexity on this problem such as Unscented Kalman filter.
Why context is important?
What impact can context make in this problem?
Previous research using context
In data fusion, information fusion research field, (2016, De Paola) built dynamic bayesian system that is able to deal with context-awareness. By using context, the system was able to not only achieve higher inference accuracy but also substantial energy savings. Here is how they used the context to acheive their goals.
(2016, De Paola) "An adaptive bayesian system for context-aware data fusion in smart environments." IEEE Transactions on Mobile Computing 16.6 (2016): 1502-1515.