Conference Contribution

LQG Control Over a Markov Communication Network

Johan Nilsson, Bo Bernhardsson


We formulate and solve a control problem where data are sent over a communication network that introduces random time delays. Past time delays are assumed known by the use of timestamps and the probability distribution of future delays are modeled with a Markov chain with continuous observation densities. We generalize previous results on LQG control of jump linear systems to cover this situation.

In Proceedings of the 36th IEEE Conference on Decision and Control, San Diego, California, January 1997.

Download full document