Cette thèse étudie des problèmes d’optimisation distribuée avec différentes structures d’observationset leurs applications aux réseaux sans fil et aux problèmes de Smart Grids. Spécifiquement,une structure d’observation asymétrique entre deux agents est considérée, où un premieragent a connaissance complète à propos de la réalisation d’un état aléatoire, et l’autre agent neconnaît rien à propos de cet état. Dans ce contexte, la question est de savoir comment transmettrede l’information depuis le premier agent vers le second agent dans le but d’utiliser de manièreoptimale les ressources de communication. Plusieurs modèles sont étudiés dans cette thèse. Pourtous, un élément commun est le fait que la source d’information doit être encodée de manièreappropriée pour optimiser l’utilisation de la configuration du système. Un premier modèle estétudié où aucun canal de communication n’est disponible entre les agents et ils ont une fonctiond’utilité commune. Cependant, le seul moyen de communiquer est via les actions choisiespar les agents. Comme les actions ont une influence sur le paiement, l’agent informé encode saconnaissance à propos de l’état dans ses actions, qui seront observées de manière imparfaite parle second agent. Ce dernier décodera l’information et choisira ses actions dans le but de maximiserla fonction objectif commune. Nous utilisons des outils de théorie de l’information pourcaractériser ce compromis optimal par une contrainte d’information, et appliquons ce scénario àun problème de contrôle de puissance pour un canal à interférence. Notre nouvelle stratégie (lecontrôle de puissance codé) donne des gains très prometteurs comparés aux approches classiques.Dans une seconde partie, nous considérons qu’il existe un canal dédié de communication, c’està-dire que les actions de l’agent informé n’ont pas d’influence sur le paiement et sont seulementutiles pour la transmission d’information. De plus, les agents sont supposés avoir des intérêtsdivergents, si bien que l’agent informé n’a pas nécessairement d’incitation à envoyer tout sonsavoir à l’agent non informé. La théorie des jeux et les jeux de « Cheap talk » en particulier sontle bon cadre pour analyser ce genre de problème. Nous caractérisons le schéma de signal sur lequelles agents se seront mis d’accord. Ce schéma amènera à un équilibre de Nash, est donc optimiserala façon dont la communication est faite. Ce modèle est d’un intérêt particulier pour les réseauxde véhicules électriques où un véhicule électrique doit envoyer son besoin en terme de puissancede charge à un aggrégateur qui choisira un niveau de charge effectif pour le véhicule électrique.Ce dernier ne se souciera que de son besoin, alors que l’aggrégateur se soucie également de l’étatdu réseau. Ce modèle aide à optimiser la façon dont le réseau est utilisé.Enfin, nous considérons un modèle avec plus de deux agents, où le but principal est pourtous les agents de retrouver l’observation parfaite des actions passées de tous les agents. Ceci estd’un intérêt très particulier d’un point de vue de la théorie des jeux pour caractériser les utilitésespérées de long terme des agents. Dans ce modèle, nous ajoutons un encodeur qui observeparfaitement toutes les actions passées et aidera les agents à obtenir l’observation parfaite. Enfait, ceci sera possible si la bonne contrainte d’information est satisfaite. Nous caractérisonsdonc cette dernière, en utilisant un schéma de codage hybride combinant des outils classiques dethéorie de l’information ainsi que des outils de la théorie des graphes / This thesis studies distributed optimization problems with different observation structuresand application to wireless network and Smart Grids problems. Specifically, an asymmetricobservation structure between two agents is considered, where a first agent has full knowledgeabout the realization of a random state, and the other agent does not know anything about thisstate. In this context, the question is how to transmit information from the first agent to thesecond agent in order to use in an optimal way the communication resources. Several modelsare studied in this thesis. For all of them, a common element is that the information source hasto be encoded in an appropriate manner to optimize the use of the system’s configuration. Afirst model is studied where no dedicated channel for communication is available between agentsand they have the same objective function. Therefore, the only way communication is possible isthrough the actions chosen by agents. As actions are payoff relevant, the first agent has to findthe optimal tradeoff between transmission of information and payoff maximization. The informedagent encodes his knowledge about the state into his actions, which will be imperfectly observedby the second agent. The latter will decode the information and choose his actions in order tomaximize the common objective function. We use tools from information theory to characterizethis optimal tradeoff by an information constraint, and apply this scenario to a power controlproblem in an interference channel setting. Our new strategy (the coded power control ) givessome promising gains compare to classical approaches.In a second part, we consider that there exists a dedicated channel for communication, that isto say the actions of the informed agent are not payoff relevant and are only useful for transmissionof information. Furthermore, agents are supposed to have diverging interests, so that the informedagent does not necessarily have an incentive to send all his knowledge to the uninformed agent.Game theory and Cheap talk game in particular appears to be the right framework to analyzethis problem. We characterize the signal scheme that agents will agree on. This scheme willlead to a Nash Equilibrium, thus will optimize the way communication is done. This model is ofparticular interest for electrical vehicles networks where an electrical vehicle has to send his needin term of power to an aggregator which will choose an effective charging level for the electricalvehicle. The latter only cares about his need in term of power whereas the aggregator also takesinto account the network status. The considered model help to optimize the way the network isused.We finally consider a model with more than two agents, where the main goal is for all agentsto retrieve perfect observations of all past actions of all agents. This is of particular interest ina game theory point of view to characterize the long term expected utilities of the agents. Inthis model, we add an encoder who perfectly oberves all past actions and will help agents tohave perfect monitoring. In fact, this is possible if the right information constraint is satisfied.We thus characterized the latter, using a hybrid coding scheme combining classical informationtheoretic scheme and tools from graph theory.

Due to the sheer volume of data involved, video coding is an important application of lossy source coding, and has received wide industrial interest and support as evidenced by the development and success of a series of video coding standards. All MPEG-series and H-series video coding standards proposed so far are based upon a video coding paradigm called predictive video coding, where video source frames Xᵢ,i=1,2,...,N, are encoded in a frame by frame manner, the encoder and decoder for each frame Xᵢ, i =1, 2, ..., N, enlist help only from all previous encoded frames Sj, j=1, 2, ..., i-1. In this thesis, we will look further beyond all existing and proposed video coding standards, and introduce a new coding paradigm called causal video coding, in which the encoder for each frame Xᵢ can use all previous original frames Xj, j=1, 2, ..., i-1, and all previous encoded frames Sj, while the corresponding decoder can use only all previous encoded frames. We consider all studies, comparisons, and designs on causal video coding from an information theoretic point of view. Let R*c(D₁,...,D_N) (R*p(D₁,...,D_N), respectively) denote the minimum total rate required to achieve a given distortion level D₁,...,D_N > 0 in causal video coding (predictive video coding, respectively). A novel computation approach is proposed to analytically characterize, numerically compute, and compare the minimum total rate of causal video coding R*c(D₁,...,D_N) required to achieve a given distortion (quality) level D₁,...,D_N > 0. Specifically, we first show that for jointly stationary and ergodic sources X₁, ..., X_N, R*c(D₁,...,D_N) is equal to the infimum of the n-th order total rate distortion function R_{c,n}(D₁,...,D_N) over all n, where R_{c,n}(D₁,...,D_N) itself is given by the minimum of an information quantity over a set of auxiliary random variables. We then present an iterative algorithm for computing R_{c,n}(D₁,...,D_N) and demonstrate the convergence of the algorithm to the global minimum. The global convergence of the algorithm further enables us to not only establish a single-letter characterization of R*c(D₁,...,D_N) in a novel way when the N sources are an independent and identically distributed (IID) vector source, but also demonstrate a somewhat surprising result (dubbed the more and less coding theorem)---under some conditions on source frames and distortion, the more frames need to be encoded and transmitted, the less amount of data after encoding has to be actually sent. With the help of the algorithm, it is also shown by example that R*c(D₁,...,D_N) is in general much smaller than the total rate offered by the traditional greedy coding method by which each frame is encoded in a local optimum manner based on all information available to the encoder of the frame. As a by-product, an extended Markov lemma is established for correlated ergodic sources. From an information theoretic point of view, it is interesting to compare causal video coding and predictive video coding, which all existing video coding standards proposed so far are based upon. In this thesis, by fixing N=3, we first derive a single-letter characterization of R*p(D₁,D₂,D₃) for an IID vector source (X₁,X₂,X₃) where X₁ and X₂ are independent, and then demonstrate the existence of such X₁,X₂,X₃ for which R*p(D₁,D₂,D₃)>R*c(D₁,D₂,D₃) under some conditions on source frames and distortion. This result makes causal video coding an attractive framework for future video coding systems and standards. The design of causal video coding is also considered in the thesis from an information theoretic perspective by modeling each frame as a stationary information source. We first put forth a concept called causal scalar quantization, and then propose an algorithm for designing optimum fixed-rate causal scalar quantizers for causal video coding to minimize the total distortion among all sources. Simulation results show that in comparison with fixed-rate predictive scalar quantization, fixed-rate causal scalar quantization offers as large as 16% quality improvement (distortion reduction).

Due to the sheer volume of data involved, video coding is an important application of lossy source coding, and has received wide industrial interest and support as evidenced by the development and success of a series of video coding standards. All MPEG-series and H-series video coding standards proposed so far are based upon a video coding paradigm called predictive video coding, where video source frames Xᵢ,i=1,2,...,N, are encoded in a frame by frame manner, the encoder and decoder for each frame Xᵢ, i =1, 2, ..., N, enlist help only from all previous encoded frames Sj, j=1, 2, ..., i-1. In this thesis, we will look further beyond all existing and proposed video coding standards, and introduce a new coding paradigm called causal video coding, in which the encoder for each frame Xᵢ can use all previous original frames Xj, j=1, 2, ..., i-1, and all previous encoded frames Sj, while the corresponding decoder can use only all previous encoded frames. We consider all studies, comparisons, and designs on causal video coding from an information theoretic point of view. Let R*c(D₁,...,D_N) (R*p(D₁,...,D_N), respectively) denote the minimum total rate required to achieve a given distortion level D₁,...,D_N > 0 in causal video coding (predictive video coding, respectively). A novel computation approach is proposed to analytically characterize, numerically compute, and compare the minimum total rate of causal video coding R*c(D₁,...,D_N) required to achieve a given distortion (quality) level D₁,...,D_N > 0. Specifically, we first show that for jointly stationary and ergodic sources X₁, ..., X_N, R*c(D₁,...,D_N) is equal to the infimum of the n-th order total rate distortion function R_{c,n}(D₁,...,D_N) over all n, where R_{c,n}(D₁,...,D_N) itself is given by the minimum of an information quantity over a set of auxiliary random variables. We then present an iterative algorithm for computing R_{c,n}(D₁,...,D_N) and demonstrate the convergence of the algorithm to the global minimum. The global convergence of the algorithm further enables us to not only establish a single-letter characterization of R*c(D₁,...,D_N) in a novel way when the N sources are an independent and identically distributed (IID) vector source, but also demonstrate a somewhat surprising result (dubbed the more and less coding theorem)---under some conditions on source frames and distortion, the more frames need to be encoded and transmitted, the less amount of data after encoding has to be actually sent. With the help of the algorithm, it is also shown by example that R*c(D₁,...,D_N) is in general much smaller than the total rate offered by the traditional greedy coding method by which each frame is encoded in a local optimum manner based on all information available to the encoder of the frame. As a by-product, an extended Markov lemma is established for correlated ergodic sources. From an information theoretic point of view, it is interesting to compare causal video coding and predictive video coding, which all existing video coding standards proposed so far are based upon. In this thesis, by fixing N=3, we first derive a single-letter characterization of R*p(D₁,D₂,D₃) for an IID vector source (X₁,X₂,X₃) where X₁ and X₂ are independent, and then demonstrate the existence of such X₁,X₂,X₃ for which R*p(D₁,D₂,D₃)>R*c(D₁,D₂,D₃) under some conditions on source frames and distortion. This result makes causal video coding an attractive framework for future video coding systems and standards. The design of causal video coding is also considered in the thesis from an information theoretic perspective by modeling each frame as a stationary information source. We first put forth a concept called causal scalar quantization, and then propose an algorithm for designing optimum fixed-rate causal scalar quantizers for causal video coding to minimize the total distortion among all sources. Simulation results show that in comparison with fixed-rate predictive scalar quantization, fixed-rate causal scalar quantization offers as large as 16% quality improvement (distortion reduction).

In the first part of this thesis, we generalize a coding theorem and a converse of Kadota and Wyner (1972) to abstract channels with time structure. As a main contribution we prove the coding theorem for a significantly weaker condition on the channel output memory, called total ergodicity for block-i.i.d. inputs. We achieve this result mainly by introducing an alternative characterization of information rate capacity. We show that the ψ-mixing condition (asymptotic output-memorylessness), used by Kadota and Wyner, is quite restrictive, in particular for the important class of Gaussian channels. In fact, we prove that for Gaussian channels the ψ-mixing condition is equivalent to finite output memory. Moreover, we derive a weak converse for all stationary channels with time structure. Intersymbol interference as well as input constraints are taken into account in a flexible way. Due to the direct use of outer measures and a derivation of an adequate version of Feinstein’s lemma we are able to avoid the standard extension of the channel input σ-algebra and obtain a more transparent derivation. We aim at a presentation from an operational perspective and consider an abstract framework, which enables us to treat discrete- and continuous-time channels in a unified way. In the second part, we systematically analyze infinite output memory conditions for abstract channels with time structure. We exploit the connections to the rich field of strongly mixing random processes to derive a hierarchy for the nonequivalent infinite channel output memory conditions in terms of a sequence of implications. The ergodic-theoretic memory condition used in the proof of the coding theorem and the ψ-mixing condition employed by Kadota and Wyner (1972) are shown to be part of this taxonomy. In addition, we specify conditions for the channel under which memory properties of a random process are invariant when the process is passed through the channel. In the last part, we investigate cascade and integration channels with regard to mixing conditions as well as properties required in the context of the coding theorem. The results are useful to study many physically relevant channel models and allow a component-based analysis of the overall channel. We consider a number of examples including composed models and deterministic as well as random filter channels. Finally, an application of strong mixing conditions from statistical signal processing involving the Fourier transform of stationary random sequences is discussed and a list of further applications is given. / Im ersten Teil der Arbeit wird ein Kodierungstheorem und ein dazugehöriges Umkehrtheorem von Kadota und Wyner (1972) für abstrakte Kanäle mit Zeitstruktur verallgemeinert. Als wesentlichster Beitrag wird das Kodierungstheorem für eine signifikant schwächere Bedingung an das Kanalausgangsgedächtnis bewiesen, die sogenannte totale Ergodizität für block-i.i.d. Eingaben. Dieses Ergebnis wird hauptsächlich durch eine alternative Charakterisierung der Informationsratenkapazität erreicht. Es wird gezeigt, dass die von Kadota und Wyner verwendete ψ-Mischungsbedingung (asymptotische Gedächtnislosigkeit am Kanalausgang) recht einschränkend ist, insbesondere für die wichtige Klasse der Gaußkanäle. In der Tat, für Gaußkanäle wird bewiesen, dass die ψ-Mischungsbedingung äquivalent zu endlichem Gedächtnis am Kanalausgang ist. Darüber hinaus wird eine schwache Umkehrung für alle stationären Kanäle mit Zeitstruktur bewiesen. Sowohl Intersymbolinterferenz als auch Eingabebeschränkungen werden in allgemeiner und flexibler Form berücksichtigt. Aufgrund der direkten Verwendung von äußeren Maßen und der Herleitung einer angepassten Version von Feinsteins Lemma ist es möglich, auf die Standarderweiterung der σ-Algebra am Kanaleingang zu verzichten, wodurch die Darstellungen transparenter und einfacher werden. Angestrebt wird eine operationelle Perspektive. Die Verwendung eines abstrakten Modells erlaubt dabei die einheitliche Betrachtung von zeitdiskreten und zeitstetigen Kanälen. Für abstrakte Kanäle mit Zeitstruktur werden im zweiten Teil der Arbeit Bedingungen für ein unendliches Gedächtnis am Kanalausgang systematisch analysiert. Unter Ausnutzung der Zusammenhänge zu dem umfassenden Gebiet der stark mischenden zufälligen Prozesse wird eine Hierarchie in Form einer Folge von Implikationen zwischen den verschiedenen Gedächtnisvarianten hergeleitet. Die im Beweis des Kodierungstheorems verwendete ergodentheoretische Gedächtniseigenschaft und die ψ-Mischungsbedingung von Kadota und Wyner (1972) sind dabei Bestandteil der hergeleiteten Systematik. Weiterhin werden Bedingungen für den Kanal spezifiziert, unter denen Eigenschaften von zufälligen Prozessen am Kanaleingang bei einer Transformation durch den Kanal erhalten bleiben. Im letzten Teil der Arbeit werden sowohl Integrationskanäle als auch Hintereinanderschaltungen von Kanälen in Bezug auf Mischungsbedingungen sowie weitere für das Kodierungstheorem relevante Kanaleigenschaften analysiert. Die erzielten Ergebnisse sind nützlich bei der Untersuchung vieler physikalisch relevanter Kanalmodelle und erlauben eine komponentenbasierte Betrachtung zusammengesetzter Kanäle. Es wird eine Reihe von Beispielen untersucht, einschließlich deterministischer Kanäle, zufälliger Filter und daraus zusammengesetzter Modelle. Abschließend werden Anwendungen aus weiteren Gebieten, beispielsweise der statistischen Signalverarbeitung, diskutiert. Insbesondere die Fourier-Transformation stationärer zufälliger Prozesse wird im Zusammenhang mit starken Mischungsbedingungen betrachtet.

In the first part of this thesis, we generalize a coding theorem and a converse of Kadota and Wyner (1972) to abstract channels with time structure. As a main contribution we prove the coding theorem for a significantly weaker condition on the channel output memory, called total ergodicity for block-i.i.d. inputs. We achieve this result mainly by introducing an alternative characterization of information rate capacity. We show that the ψ-mixing condition (asymptotic output-memorylessness), used by Kadota and Wyner, is quite restrictive, in particular for the important class of Gaussian channels. In fact, we prove that for Gaussian channels the ψ-mixing condition is equivalent to finite output memory. Moreover, we derive a weak converse for all stationary channels with time structure. Intersymbol interference as well as input constraints are taken into account in a flexible way. Due to the direct use of outer measures and a derivation of an adequate version of Feinstein’s lemma we are able to avoid the standard extension of the channel input σ-algebra and obtain a more transparent derivation. We aim at a presentation from an operational perspective and consider an abstract framework, which enables us to treat discrete- and continuous-time channels in a unified way. In the second part, we systematically analyze infinite output memory conditions for abstract channels with time structure. We exploit the connections to the rich field of strongly mixing random processes to derive a hierarchy for the nonequivalent infinite channel output memory conditions in terms of a sequence of implications. The ergodic-theoretic memory condition used in the proof of the coding theorem and the ψ-mixing condition employed by Kadota and Wyner (1972) are shown to be part of this taxonomy. In addition, we specify conditions for the channel under which memory properties of a random process are invariant when the process is passed through the channel. In the last part, we investigate cascade and integration channels with regard to mixing conditions as well as properties required in the context of the coding theorem. The results are useful to study many physically relevant channel models and allow a component-based analysis of the overall channel. We consider a number of examples including composed models and deterministic as well as random filter channels. Finally, an application of strong mixing conditions from statistical signal processing involving the Fourier transform of stationary random sequences is discussed and a list of further applications is given. / Im ersten Teil der Arbeit wird ein Kodierungstheorem und ein dazugehöriges Umkehrtheorem von Kadota und Wyner (1972) für abstrakte Kanäle mit Zeitstruktur verallgemeinert. Als wesentlichster Beitrag wird das Kodierungstheorem für eine signifikant schwächere Bedingung an das Kanalausgangsgedächtnis bewiesen, die sogenannte totale Ergodizität für block-i.i.d. Eingaben. Dieses Ergebnis wird hauptsächlich durch eine alternative Charakterisierung der Informationsratenkapazität erreicht. Es wird gezeigt, dass die von Kadota und Wyner verwendete ψ-Mischungsbedingung (asymptotische Gedächtnislosigkeit am Kanalausgang) recht einschränkend ist, insbesondere für die wichtige Klasse der Gaußkanäle. In der Tat, für Gaußkanäle wird bewiesen, dass die ψ-Mischungsbedingung äquivalent zu endlichem Gedächtnis am Kanalausgang ist. Darüber hinaus wird eine schwache Umkehrung für alle stationären Kanäle mit Zeitstruktur bewiesen. Sowohl Intersymbolinterferenz als auch Eingabebeschränkungen werden in allgemeiner und flexibler Form berücksichtigt. Aufgrund der direkten Verwendung von äußeren Maßen und der Herleitung einer angepassten Version von Feinsteins Lemma ist es möglich, auf die Standarderweiterung der σ-Algebra am Kanaleingang zu verzichten, wodurch die Darstellungen transparenter und einfacher werden. Angestrebt wird eine operationelle Perspektive. Die Verwendung eines abstrakten Modells erlaubt dabei die einheitliche Betrachtung von zeitdiskreten und zeitstetigen Kanälen. Für abstrakte Kanäle mit Zeitstruktur werden im zweiten Teil der Arbeit Bedingungen für ein unendliches Gedächtnis am Kanalausgang systematisch analysiert. Unter Ausnutzung der Zusammenhänge zu dem umfassenden Gebiet der stark mischenden zufälligen Prozesse wird eine Hierarchie in Form einer Folge von Implikationen zwischen den verschiedenen Gedächtnisvarianten hergeleitet. Die im Beweis des Kodierungstheorems verwendete ergodentheoretische Gedächtniseigenschaft und die ψ-Mischungsbedingung von Kadota und Wyner (1972) sind dabei Bestandteil der hergeleiteten Systematik. Weiterhin werden Bedingungen für den Kanal spezifiziert, unter denen Eigenschaften von zufälligen Prozessen am Kanaleingang bei einer Transformation durch den Kanal erhalten bleiben. Im letzten Teil der Arbeit werden sowohl Integrationskanäle als auch Hintereinanderschaltungen von Kanälen in Bezug auf Mischungsbedingungen sowie weitere für das Kodierungstheorem relevante Kanaleigenschaften analysiert. Die erzielten Ergebnisse sind nützlich bei der Untersuchung vieler physikalisch relevanter Kanalmodelle und erlauben eine komponentenbasierte Betrachtung zusammengesetzter Kanäle. Es wird eine Reihe von Beispielen untersucht, einschließlich deterministischer Kanäle, zufälliger Filter und daraus zusammengesetzter Modelle. Abschließend werden Anwendungen aus weiteren Gebieten, beispielsweise der statistischen Signalverarbeitung, diskutiert. Insbesondere die Fourier-Transformation stationärer zufälliger Prozesse wird im Zusammenhang mit starken Mischungsbedingungen betrachtet.