The experimental WSAN testbed in this paper only has one route generator and one
schedule generator. We can further investigate whether it is better to have multiple
route generators and schedule generators. This paper doesn’t give reason why 2 ms
guard time is suitable. Numerical analysis or experiments should be done on the
impact of longer or shorter guard time.
For both the fixed priority scheduling and dynamic priority scheduling, we may
apply machine learning approach to assign suitable priority. Instead of man-made
rules, the machine learning approach can learn from large amounts of data.
There are only simulation results for the 4 optimization methods used in selecting
sampling rate. It is much better to include the performance for real applications. It is
also desirable to give numerical analysis of the optimization performance and
running time of these 4 algorithms. We can consider using special hardware to
speed up the execution.
In the emergency communication section, this paper assumes the control networks
consist of regular flows and emergency flows and they have predetermined
deadlines. These assumptions are generally true. But if we have signals of various
kind of emergency, how we set the threshold to determine whether it is emergent
signal. The deadlines may not be predetermined and related to other factors beside
the signal itself.
There are only one experiment (water tank systems) for the proposed emergency
communication protocol. It needs more experiments for many different applications
to prove the protocol is effective.