Information Structures, the Witsenhausen Counterexample, and Communicating Using Actions

The concept of “information-structures” in decentralized control is a formalization of the notion of “who knows what and when do they know it.” Even seemingly simple problems with simply stated information structures can be extremely hard to solve. Perhaps the simplest of such unsolved problem is the celebrated Witsenhausen counterexample, formulated by Hans Witsenhausen in 1968. This article discusses how the information structure of the Witsenhausen counterexample makes it hard, and how an information-theoretic approach, that explores the knowledge-gradient due to a non-classical information pattern, has helped obtain insights into the problem.

[1]  Tamer Basar,et al.  Variations on the theme of the Witsenhausen counterexample , 2008, 2008 47th IEEE Conference on Decision and Control.

[2]  M. Rotkowitz,et al.  Linear Controllers are Uniformly Optimal for the Witsenhausen Counterexample , 2006, Proceedings of the 45th IEEE Conference on Decision and Control.

[3]  Christoph Kawan Data Rate of Nonlinear Control Systems and Feedback Entropy , 2015, Encyclopedia of Systems and Control.

[4]  Anant Sahai,et al.  Approximately Optimal Solutions to the Finite-Dimensional Witsenhausen Counterexample , 2013, IEEE Transactions on Automatic Control.

[5]  Pulkit Grover,et al.  Actions can speak more clearly than words , 2010 .

[6]  Thomas Parisini,et al.  Nonlinear approximations for the solution of team optimal control problems , 1997, Proceedings of the 36th IEEE Conference on Decision and Control.

[7]  Sergio Verdú,et al.  Witsenhausen's counterexample: A view from optimal transport theory , 2011, IEEE Conference on Decision and Control and European Control Conference.

[8]  T. Başar,et al.  Stochastic Teams with Nonclassical Information Revisited: When is an Affine Law Optimal? , 1986, 1986 American Control Conference.

[9]  Anant Sahai,et al.  It may be “easier to approximate” decentralized infinite-horizon LQG problems , 2012, 2012 IEEE 51st IEEE Conference on Decision and Control (CDC).

[10]  H. Witsenhausen Separation of estimation and control for discrete time systems , 1971 .

[11]  Max H. M. Costa,et al.  Writing on dirty paper , 1983, IEEE Trans. Inf. Theory.

[12]  A. Sahai,et al.  Witsenhausen's counterexample as Assisted Interference Suppression , 2010 .

[13]  Y. Ho,et al.  Teams, signaling, and information theory , 1977, 1977 IEEE Conference on Decision and Control including the 16th Symposium on Adaptive Processes and A Special Symposium on Fuzzy Set Theory and Applications.

[14]  S. Mitter,et al.  Information and control: Witsenhausen revisited , 1999 .

[15]  Linda Bushnell,et al.  Networked Control Systems: Architecture and Stability Issues , 2015, Encyclopedia of Systems and Control.

[16]  R. Bellman Dynamic programming. , 1957, Science.

[17]  Serdar Yüksel,et al.  Information and Communication Complexity of Networked Control Systems , 2015, Encyclopedia of Systems and Control.

[18]  H. Witsenhausen A Counterexample in Stochastic Optimum Control , 1968 .

[19]  J. Tsitsiklis,et al.  Intractable problems in control theory , 1986 .

[20]  Urbashi Mitra,et al.  On Witsenhausen's counterexample: The asymptotic vector case , 2012, 2012 IEEE Information Theory Workshop.

[21]  Girish N. Nair Quantized Control and Data Rate Constraints † , 2015, Encyclopedia of Systems and Control.

[22]  João Pedro Hespanha,et al.  Networked Control Systems: Estimation and Control Over Lossy Networks , 2015, Encyclopedia of Systems and Control.