Assigning channels to data attributes (such as the value or location) rather than to individual sensors is a new paradigm to
benefit from the large amount of redundancy often characterizing the sensor readings as a whole.
In this talk we discuss two architectures that exemplify this design approach and illustrate how the information gathering problem can scale
with the complexity of the data, not the sensor density.
One method is a query and response strategy, using these channels to probe a sequence of guesses on the sensor states efficiently.
The second method is an average consensus algorithm, computing via near neighbors data driven communications functions
of the sensor data. This last algorithm has an interesting connection in the synchronization of pulse coupled oscillators, modeling
synchronism emerging in biological networks.
The talk will discuss the several challenges that lie ahead in solving sensor networking problems that serve large and critical infrastructure.