State transitions are asserted within the DEDS observer model according
to the probability value of the occurrence of an event. Events are
thus defined as ranges for the different parameters.
The problem then reduces to computing the corresponding areas under the
refined distribution curves.
An obvious way of using those probability values is to establish some
threshold values and assert transitions according to those thresholds.
It might be the case that none of the obtained probability values
exceeds the set threshold value and/or all values are very low.
In that case, there is a good chance that we are at either the wrong automata
state. The remedy to such problems can be implemented through time proximity,
that is, wait for a while (which is to be preset) till a strong probability
value is registered and/or *backtrack* in the automaton model for the
observer till a high enough probability value is asserted, a fail state
is reached or the initial ambiguity is asserted. The backtracking strategy
can be implemented using a stack-like structure associated with each state
that has already been traversed, which includes a sorted list of
the computed event probabilities and a father-state variable.

Tue Nov 22 21:30:54 MST 1994