Charalambos D. Charalambous

Charalambos D. Charalambous, Umarbek Guvercin, Seddik Djouadi

Witsenhausen, in his seminal 1971 paper [1], introduced decentralized partially observable Markov decision problems (POMDPs), with multiple agents or controls operating under T-step delayed sharing information patterns. A fundamental problem in [1] is the identification of structural properties of optimal strategies that compress the information patterns into multiple information states. In this paper, we develop such structural properties of optimal strategies and associated dynamic programming (DP) equations, using the concept of decentralized sequential team equilibrium (a generalization of person-by-person optimality from static team theory). Within this framework, each strategy is assigned an individual value function conditioned on its delayed sharing information pattern, while the strategies of all other agents are held fixed. The resulting DP framework yields several new DP equations and characterizations of decentralized team equilibrium. Moreover, these DP equations exhibit fundamental properties analogous to those of centralized DP of POMDPs: the optimization in each agent's DP equations is performed over the agent's action space rather than over strategy spaces; each agent's multiple information states satisfy Markov recursions; and a separation principle holds. The DP equations reveal a structural compression property of optimal strategies: each agent compresses its delayed sharing information pattern into three components: 1) a private posterior distribution conditioned on the agent's delayed sharing information pattern, 2) a centralized posterior distribution conditioned on the common information shared by all agents, and 3) the agent's private information component. This structural result substantially extends Witsenhausen's Assertion 8 in [1].

Charalambos D. Charalambous, Nasir U. Ahmed

In this second part of our two-part paper, we invoke the stochastic maximum principle, conditional Hamiltonian and the coupled backward-forward stochastic differential equations of the first part [1] to derive team optimal decentralized strategies for distributed stochastic differential systems with noiseless information structures. We present examples of such team games of nonlinear as well as linear quadratic forms. In some cases we obtain closed form expressions of the optimal decentralized strategies. Through the examples, we illustrate the effect of information signaling among the decision makers in reducing the computational complexity of optimal decentralized decision strategies.

Photios A. Stavrou, Charalambos D. Charalambous

In this paper the relation between nonanticipative rate distortion function (RDF) and Bayesian filtering theory is further investigated on general Polish spaces. The relation is established via an optimization on the space of conditional distributions of the so-called directed information subject to fidelity constraints. Existence of the optimal reproduction distribution of the nonanticipative RDF is shown using the topology of weak convergence of probability measures. Subsequently, we use the solution of the nonanticipative RDF to present the realization of a multidimensional partially observable source over a scalar Gaussian channel. We show that linear encoders are optimal, establishing joint source-channel coding in real-time.

Charalambos D. Charalambous, Photios A. Stavrou

In this paper the relation between nonanticipative rate distortion function (RDF) and Bayesian filtering theory is investigated using the topology of weak convergence of probability measures on Polish spaces. The relation is established via an optimization on the space of conditional distributions of the so-called directed information subject to fidelity constraints. Existence of the optimal reproduction distribution of the nonanticipative RDF is shown, while the optimal nonanticipative reproduction conditional distribution for stationary processes is derived in closed form. The realization procedure of nonanticipative RDF which is equivalent to joint-source channel matching for symbol-by-symbol transmission is described, while an example is introduced to illustrate the concepts.

Photios A. Stavrou, Charalambos D. Charalambous, Christos K. Kourtellaris

In this paper we introduce a definition for nonanticipative Rate Distortion Function (RDF) on abstract alphabets, and we invoke weak convergence of probability measures to show various of its properties, such as, existence of the optimal reproduction conditional distribution, compactness of the fidelity set, lower semicontinuity of the RDF functional, etc. Further, we derive the closed form expression of the optimal nonstationary reproduction distribution. This expression is computed recursively backward in time. Throughout the paper we point out an operational meaning of the nonanticipative RDF by recalling the coding theorem derive in \cite{tatikonda2000}, and we state relations to Gorbunov-Pinsker's nonanticipatory $ε-$entropy \cite{gorbunov-pinsker}.

Charalambos D. Charalambous, Umarbek Guvercin, Seddik Djouadi

This paper develops a dynamic programming (DP) approach for decentralized stochastic optimal control problems with delayed sharing information patterns, which exhibits the fundamental Properties of classical DP of centralized partially observable Markov decision problems (POMDPs): the value functions and information states depend on the actions of the minimizing controls and not their strategies. This is achieved by invoking the concept of Person-by-Person (PbP) optimality, in which each control strategy is associated with a value function conditioned on its assigned delayed sharing information pattern, when all other strategies are fixed to their optimal responses. The value functions satisfy generalized and simplified DP equations. These are used to derive necessary and sufficient conditions for PbP optimality. The simplified DP equations are obtained by invoking the structural property that optimal strategies are separated and functionals of two information states: 1) a private a posteriori probability distribution based on the information pattern of the strategy, and 2) a centralized a posteriori probability distribution based on the shared or common information to all strategies, each satisfying a Markov recursion. The DP approach of this paper, settles a long standing open problem since the appearance of T-step delayed sharing patterns in [1, Section IV.G], in terms of generalizing the fundamental properties of classical DP approach.

Photios A. Stavrou, Jan Ostergaard, Charalambos D. Charalambous

We deal with zero-delay source coding of a vector-valued Gauss-Markov source subject to a mean-squared error (MSE) fidelity criterion characterized by the operational zero-delay vector-valued Gaussian rate distortion function (RDF). We address this problem by considering the nonanticipative RDF (NRDF) which is a lower bound to the causal optimal performance theoretically attainable (OPTA) function and operational zero-delay RDF. We recall the realization that corresponds to the optimal "test-channel" of the Gaussian NRDF, when considering a vector Gauss-Markov source subject to a MSE distortion in the finite time horizon. Then, we introduce sufficient conditions to show existence of solution for this problem in the infinite time horizon. For the asymptotic regime, we use the asymptotic characterization of the Gaussian NRDF to provide a new equivalent realization scheme with feedback which is characterized by a resource allocation (reverse-waterfilling) problem across the dimension of the vector source. We leverage the new realization to derive a predictive coding scheme via lattice quantization with subtractive dither and joint memoryless entropy coding. This coding scheme offers an upper bound to the operational zero-delay vector-valued Gaussian RDF. When we use scalar quantization, then for "r" active dimensions of the vector Gauss-Markov source the gap between the obtained lower and theoretical upper bounds is less than or equal to 0.254r + 1 bits/vector. We further show that it is possible when we use vector quantization, and assume infinite dimensional Gauss-Markov sources to make the previous gap to be negligible, i.e., Gaussian NRDF approximates the operational zero-delay Gaussian RDF. We also extend our results to vector-valued Gaussian sources of any finite memory under mild conditions. Our theoretical framework is demonstrated with illustrative numerical experiments.