In control theory, the classical problem of linear quadratic (LQ) optimal control has a simple and elegant solution. The optimal controller separates into a Kalman filter followed by a ``Certainty Equivalence'' controller leading to a complete design obtained by solving Riccati equations. Thus, in the classical setting, design of the mean square optimal controller is no more difficult than designing a mean square optimal estimator. Networked control loops are those loops where the properties of communication links in a loop also matter for the control design. Examples are process control loops completed over wireless networks, and mobile robot teams communicating over wireless networks. Here the data network properties such as the available bandwidth, and packet loss rates affect control performance. Like in the classical case, one can setup the LQ optimal control problem. But here, one has to design the controller and the communication strategies jointly. It will be useful to have design methods of low complexity, and good performance. In the networked version, the optimal designs are typically hard. Only in a special case do we obtain that ``separation'' and ``Certainty Equivalence'' are optimal. In this talk, we will explain this using only some examples, some elementary ideas from signals and systems, and some elementary random processes. Special attention will be given to setups with Event-triggered communications.
Maben Rabi obtained his B.Tech. in Electrical Engineering from IIT Madras in 1998, and his Ph.D. in Electrical and Computer Engineering from the University of Maryland College Park in 2006. He has been a postdoc at KTH Stockholm and the University of Cambridge. At present he is an assistant professor at Chalmers University of Technology. His research is on Control theory, with special attention to problems of Networked control.