Cooperative Communications and Control
This page contains some thoughts on how to do cooperative communications and control.
The idea of using local broadcast capability of wireless communications with ideas on distributed estimation and control could be a good one for some new ideas. In particular, in the context of sensor networks, it would be interesting to think about how to do distributed Kalman filtering when local measurements were available to a group of sensors at the same time. One would want this to be robust with respect to lost information at some nodes (eg, a fading channel that caused packet loss) and at the same time show that the overall system converged quickly to the estimated state. Convergence would probably be a function of the connectivity, which would depend on the transmit power, which is something one could update adaptively.
There has been some interested over the past few years within the control community in the problem of "dynamic consensus". The original problem that we studied came from formation control, where we were trying to get a group of vehicles to move to the right location relative to each other. Since then, several people in the field have extended these results to consider the problem of a group of agents that are trying to compute a common function using only sparse information flow (eg, talking to nearest neighbors). There are a couple of good papers on the topic:
Dynamic Consensus for Mobile Networks, Demetri P. Spanos, Reza Olfati-Saber, Richard M. Murray, 2005 IFAC World Congress. This paper describes how to use the basic consensus idea to track a time varying signal measured by multiple agents.
Distributed Sensor Fusion Using Dynamic Consensus, Demetri P. Spanos and Richard M. Murray, 2005 IFAC World Congress. Extends the results of the first paper to look at how you can do Kalman filtering using consensus.
A natural thing to think about is how any of this would change if we were to use a cooperative communication model instead of point to point communications.