Decentralized/distributed control
editDecentralized control is about control of a system by multiple agents that help obtain advantages of decentralization in control systems. The agents often coordinate their actions to perform improved control. Geographically, the agents can be distributed in space, thus monitoring and controlling a larger area. Such decentralization is often used in controlling room temperature, detecting forest fires. Small sensors and devices, such as smartdust offer technologies that allow for introducing such control cheaply and in different environments.
Decentralized control theory
editTheoretically, decentralized control problems have proven to be much harder than centralized control problems. For instance, while Linear-quadratic-Gaussian_control (LQG control) is well understood for centralized systems, its understanding is very limited in decentralized control systems. This is best exemplified by Witsenhausen's counterexample, a decentralized LQG control problem that has remained unsolved for the last four decades.
There has been considerable success in understanding toy problems where there are explicit channels connecting controllers.
Decentralized control in practice
editReferences
editcategory:control theory