EECI09: Information patterns

From Murray Wiki
Jump to navigationJump to search
Prev: Optimization-based control Course home Next: Graph theory

In this lecture, we look at one of the reasons why networked control systems problems are difficult. Due to the distributed nature of the problem, different controllers may have access to different information sets at every time. We introduce the concept of information patterns, and identify some patterns that permit simple solutions to the optimal control problem.

Lecture Materials

Further Reading

  • "Separation of Estimation and Control for Discrete Time Systems"], H. S. Witsenhausen, Proceedings of the IEEE, vol. 59, no. 11, pp. 1557-1566, Nov. 1971. This paper covers the concept of information patterns and discusses why non-classical information patterns may turn the optimal control problem difficult to solve.

  • "A Counterexample in Stochastic Optimum Control," H. S. Witsenhausen, SIAM Journal of Control, vol. 6, no. 1, pp. 131-147, 1968. This paper introduced the counterexample, and proved that a non-linear controller will perform better than the optimal affine controller.

  • "Team Decision Theory and Information Structures in Optimal Control Problems - Part I," Y. C. Ho and K. C. Chu, IEEE Transactions on Automatic Control, vol. AC-17, no. 1, pp. 15-22, Jan. 1972. This paper considers the partially nested information structure.

Additional Information