Aujourd'hui, l'intelligence artificielle (IA) est une composante essentielle d'un jeu vidéo et de plus en plus d'efforts sont portés sur cet aspect afin de rendre les jeux plus ludiques et plus immersifs. Cette amélioration va cependant de pair avec une augmentation sans cesse croissante des ressources informatiques nécessaires au fonctionnement de l'IA. De fait, il arrive que ces besoins soient si importants qu'ils dégradent le taux de rafraîchissement (TR) du jeu et ainsi la qualité d'expérience (QoE) du joueur. Dans ce contexte, notre objectif est de de maintenir le TR au dessus d'un certain seuil en modulant la quantité de ressources requises par l'IA. Pour ce faire, nous proposons de donner la possibilité au programmeur de définir plusieurs niveaux de détails pour l'IA (Level Of Details LOD), à l'instar de ce qui se fait pour afficher une scène graphique.Les travaux utilisant ce type d'approches proposent généralement d'utiliser des critères de distance à la caméra et de visibilité. Cependant, élaborés dans le contexte du rendu graphique, ces critères sont finalement assez peu adaptés au contexte IA car ils ne permettent pas toujours de rendre compte de l'importance réelle d'un personnage pour le joueur. Dans cette thèse, nous proposons d'utiliser des concepts organisationnels tels que le groupe et le rôle pour définir l'importance d'un personnage pour l'IA. De cette façon, un jeu vidéo est considéré comme une société d'agents (les personnages du jeu) dont l'importance individuelle ou collective est déterminée en fonction de leurs positions dans l'organisation, ce qui permet de déterminer une distribution des ressources de calcul disponibles adaptée : les entités les plus importantes dans l'histoire du jeu sont privilégiées.Notre approche a été implémentée et intégrée au moteur de jeu AGDE (Moteur Agent de développement du jeu). L'évaluation expérimentale a été réalisée à l'aide d'un système de mesures répétées pour évaluer la différence entre les QoE d'un jeu avec et sans notre approche. / In recent years there have been many efforts to develop original video games by improving both their aesthetic and mechanics. The more the mechanics is rich and realistic, the more advanced models of programming are required. However, using advanced models of programming such as agent-oriented programming often comes with an overhead in terms of computational resources. Furthermore, this overhead on computational resources may degrade the frame rate and subsequently quality of experience (QoE) for the players.In this context, our aim is to propose the QoE support means for ensuring that, in any case, the frame rate does not fall below a given lower bound. We suggest adapting the amount of time allocated for agents depending upon the importance of their organization roles. In this regard, we use a level of detail (LoD) approach to compute the dynamics of the game.LoD in game AI is based on the idea to use the most of the computational effort on the game characters that are the most important to the player(s). One critical issue in LoD for game AI is to determine the criterion for defining the importance of game characters. Existing work propose to use the criteria of camera distance and visibility. However such criteria have been developed from the perspective of graphics. In this thesis, we have used the roles played by the game characters (in the context of a video game) as the criterion for determining their importance. In this way, a video game has been considered as an agent society, where the game characters get priority and relatively higher share in distribution of the computational resources based on their relative importance in the game story.Our approach has been implemented and integrated to the AGDE (Agent Game Development Engine) game engine. The experimental evaluation has been carried out using a repeated measure scheme to assess the difference in QoE metrics between a game implemented our approach and a control game. The null hypothesizes have been rejected using t-paired test: the players have found significant positive difference in the QoE.

Le Cloud computing émerge comme le nouveau paradigme informatique dans lequel la virtualisation des ressources fournit des services fiables correspondant aux demandes des utilisateurs. De nos jours, la plupart des applications interactives et utilisant beaucoup de données sont développées sur le cloud: Le jeu vidéo en est un exemple. Avec l'arrivée du cloud computing, l'accessibilité et l'ubiquité du jeu ont un brillant avenir; Les jeux peuvent être hébergés dans un serveur centralisé et accessibles via l'Internet par un client léger sur une grande variété de dispositifs avec des capacités modestes : c'est le cloud gaming. Le Cloud computing, dans le contexte de jeu vidéo a beaucoup attiré l'attention en raison de ses facilités d'évolution, de disponibilité et capacité de calcul. Cependant, les systèmes de cloud gaming actuels ont des exigences très fortes en termes de ressources réseau, réduisant ainsi l'accessibilité et l'ubiquité des jeux dans le cloud, car les dispositifs clients avec peu de bande passante et les personnes situées dans la zone avec des conditions de réseau limitées et/ou instables, ne peuvent pas bénéficier de ces services de cloud computing. Dans cette thèse, nous présentons une technique d'adaptation inspirée par l'approche du niveau de détail (Level of detail) dans les graphiques 3D. Elle est basée sur un paradigme du cloud gaming dans l'objectif de fournir une accessibilité multi-plateforme, tout en améliorant la qualité d'expérience (QoE) du joueur en réduisant l'impact des mauvaises conditions réseau (delai, perte, gigue) sur l'interactivité et réactivité du jeu. Notre première contribution se compose de modèles de jeu reliant les objets du jeu à leurs besoins en termes de communication représentés par leurs importances dans le jeu. Nous avons ensuite fourni une approche de niveau de détail pour gérer la distribution des ressources réseau basée sur l'importance des objets dans la scène et les conditions réseau. Nous validons notre approche en utilisant des jeu prototypes et evaluons la QoE du joueur, par des expériences pilotes. Les résultats montrent que le framework proposé fournit une importante amélioration de la QoE. / With the arrival of cloud computing technology, game accessibility and ubiquity havea bright future. Games can be hosted in a centralize server and accessed through theInternet by a thin client on a wide variety of devices with modest capabilities: cloudgaming. Some of the advantages of using cloud computing in game context includes:device ubiquity, computing exibility, affordable cost and lowered set up overheads andcompatibility issues. However, current cloud gaming systems have very strong requirementsin terms of network resources, thus reducing their widespread adoption. In factdevices with little bandwidth and people located in area with limited network capacity,cannot take advantage of these cloud services. In this thesis we present an adaptationtechnique inspired by the level of detail (LoD) approach in 3D graphics. It is based ona cloud gaming paradigm in other to maintain user's quality of experience (QoE) byreducing the impact of poor network parameters (delay, loss, bandwidth) on game interactivity.Our first contribution consist of game models expressing game objects and theircommunications needs represented by their importance in the game. We provided twodifferent ways to manage objects' importance using agents organizations and gameplaycomponents. We then provided a level of detail approach for managing network resourcedistribution based on objects importance in the game scene and network conditions. Weexploited the dynamic objects importance adjustment models presented above to proposeLoD systems adapting to changes during game sessions. The experimental validation ofboth adaptation models showed that the suggested adaptation minimizes the effects oflow and/or unstable network conditions in maintaining game responsiveness and player'sQoE.

Dans un scénario avec plusieurs réseaux sans fil de différentes technologies, ce travail a comme objectif la conception d'un moteur cognitif capable de reconnaitre l'environnement radio et de sélectionner un réseau avec le but final de maximiser la "qualité d'expérience" (QoE) de l'utilisateur. Un accent particulier est mis sur la simplicité de tous les éléments impliqués, du hardware aux algorithmes, afin de garder la faisabilité pratique de ce dispositif.Deux aspects ont été étudiés. Pour la reconnaissance de l'environnement radio une identification de réseau et une classification automatique sur la base de caractéristiques de la couche MAC a été proposée et testée. En ce qui concerne la sélection du réseau, des "Key Performance Indicators" (KPIs), qui sont des paramètres de la couche application, ont étés pris en compte afin d'obtenir la QoE désirée. Un modèle général pour la sélection du réseau a été proposé et testé avec de différents types de trafic par des simulations et par la réalisation d'un démonstrateur (application pour Android). De plus, comme il y a le problème de quand mesurer pour estimer la performance d'un réseau et quand l'utiliser effectivement pour transmettre et recevoir, le problème du bandit manchot ("Multi-armed bandit", MAB) a été appliqué à ce contexte et un nouveau modèle de MAB a été proposé afin de mieux répondre aux cas réels considérés. L'impact du nouveau modèle, qui introduit la distinction de deux actions différentes, mesurer et utiliser, a été testé par des simulations en utilisant des algorithmes déjà disponibles dans la littérature et deux algorithmes conçus spécifiquement. / In a scenario where multiple wireless networks of different technologies are available, this work addresses the problem of the design of a cognitive engine, core of a cognitive radio device, able to perform the surrounding radio environment recognition and the network selection with the final goal of maximization of final user Quality of Experience (QoE). Particular focus is put on the requirement of simplicity of all the elements involved, from hardware to algorithms, in order to keep in mind the importance of its practical realizability.Two aspects were investigated. For the surrounding radio environment recognition step, a network identification and automatic classification method based on MAC layer features was proposed and tested. As regards the network selection, Key Performance Indicators (KPIs), i.e. application layer parameters, were considered in order to obtain the desired goal of QoE. A general model for network selection was proposed and tested for different traffic types, both with simulations and a practical realization of a demonstrator (implemented as an application for Android OS). Moreover, as a consequence of the originated problem of when measuring to estimate a network performance and when effectively using the network for data transmission and reception purposes, the multi-armed bandit problem (MAB) was applied to this context and a new MAB model was proposed, in order to better fit the considered real cases scenarios. The impact of the new model, that introduces the distinction of two different actions, to measure and to use, was tested through simulations using algorithms already available in literature and two specifically designed algorithms.

FundaÃÃo Cearense de Apoio ao Desenvolvimento Cientifico e TecnolÃgico / Ericsson Brasil / Uma nova geraÃÃo de sistemas de comunicaÃÃes sem fio, 5a GeraÃÃo (5G), à prevista para 2020. Para a 5G, à esperado o surgimento de diversos serviÃos baseados em comunicaÃÃes mÃquina à mÃquina em diferentes Ãreas, como assistÃncia mÃdica, seguranÃa e redes de mediÃÃo inteligente. Cada um com diferentes requerimentos de taxa de transmissÃo, latÃncia, capacidade de processamento, eficiÃncia energÃtica, etc. Independente do serviÃo, os clientes precisam ficar satisfeitos. Isto està impondo uma mudanÃa de paradigmas em direÃÃo à priorizaÃÃo do usuÃrio como fator mais importante no gerenciamento de redes sem fio. Com esta mudanÃa, criou-se o conceito de qualidade de experiÃncia (do inglÃs, Quality of Experience (QoE)), que descreve de forma subjetiva como o serviÃo à percebido pelo usuÃrio. A QoE normalmente à avaliada por uma nota entre 1 e 5, chamada nota mÃdia de opiniÃo (do inglÃs, Mean Opinion Score (MOS)). Neste contexto, conceitos de QoE podem ser considerados com diferentes objetivos, como: aumentar a vida Ãtil de baterias, melhorar a seleÃÃo para acesso à rede e aprimorar a alocaÃÃo dos recursos de rÃdio (do inglÃs, Radio Resource Allocation (RRA)). Com relaÃÃo à RRA, nesta dissertaÃÃo consideram-se requerimentos de QoE na gestÃo dos recursos disponÃveis em um sistema de comunicaÃÃes sem fio, como espectro de frequÃncia e potÃncia de transmissÃo. Mais especificamente, estuda-se um problema de assinalamento de recursos de rÃdio e de alocaÃÃo de potÃncia que objetiva maximizar a mÃnima MOS do sistema sujeito a satisfazer um nÃmero mÃnimo de usuÃrios prÃ-estabelecido. Inicialmente, formula-se um novo problema de otimizaÃÃo considerando restriÃÃes quanto à potÃncia de transmissÃo e quanto à fraÃÃo de usuÃrios que deve ser satisfeita, o que à um importante tÃpico do ponto de vista das operadoras. Este à um problema nÃo linear e de difÃcil soluÃÃo. Ele à entÃo reformulado como um problema linear inteiro e misto, que pode ser resolvido de forma Ãtima usando algoritmos conhecidos de otimizaÃÃo. Devido à complexidade da soluÃÃo Ãtima obtida, propÃe-se uma heurÃstica chamada em inglÃs de Power and Resource Allocation Based on Quality of Experience (PRABE). O mÃtodo proposto à avaliado por meio de simulaÃÃes e os resultados obtidos mostram que sua performance à superior à de outros existentes, sendo prÃxima à da Ãtima. / A new generation of wireless networks, the 5th Generation (5G), is predicted for beyond 2020. For the 5G, it is foreseen an emerging huge number of services based on Machine-Type Communications (MTCs) in different fields, such as, health care, smart metering and security. Each one of them requiring different throughput rates, latency, processing capacity, energy efficiency, etc. Independently of the service type, the customers still need to get satisfied, which is imposing a shift of paradigm towards incorporating the user as the most important factor in wireless network management. This shift of paradigm drove the creation of the Quality of Experience (QoE) concept, which describes the service quality subjectively perceived by the users. QoE is generally evaluated by a Mean Opinion Score (MOS) ranging from 1 to 5. In this context, QoE concepts can be considered with different objectives, such as, increasing battery life, optimizing handover decision, enhancing access network selection and improving Radio Resource Allocation (RRA). Regarding the RRA, in this masterâs thesis we consider QoE requirements when managing the limited available resources of a communication system, such as frequency spectrum and transmit power. More specifically, we study a radio resource assignment and power allocation problem that aims at maximizing the minimum MOS of the users in a system subject to attaining a minimum number of satisfied users. Initially, we formulate a new optimization problem taking into account constraints on the total transmit power and on the fraction of users that must be satisfied, which is an important topic from an operatorâs point of view. The referred problem is non-linear and hard to solve. However, we get to transform it into a simpler form, a Mixed Integer Linear Problem (MILP), that can be optimally solved using standard numerical optimization methods. Due to the complexity of obtaining the optimal solution, we propose a heuristic solution to this problem, called Power and Resource Allocation Based on Quality of Experience (PRABE). We evaluate the proposed method by means of simulations and the obtained results show that it outperforms some existing algorithms, as well as it performs close to the optimal solution.

<p>In this thesis we study algorithms and architectures that can provide a better Quality of Experience (QoE) for streaming video systems and services. With cases and examples taken from the application scenarios of football on mobile phones, we address the fundamental problems behind streaming video services. Thus, our research results can be applied and extended to other networks, to other sports and to other cultural activities.</p><p>In algorithm development, we propose five different schemes. We suggest a blind motion estimation and a trellis based motion estimation with dynamic programming algorithms for Wyner-Ziv coding. We develop a trans-media technology, vibrotactile coding of visual signals for mobile phones. We propose a new bandwidth prediction scheme for real-time video conference. We also provide an effective method based on dynamic programming to select optimal services and maximize QoE.</p><p>In architecture design, we offer three architectures for real-time interactive video and two for streaming live football information. The former three are: a structure of motion estimation in Wyner-Ziv coding for real-time video; a variable bit rate Wyner-Ziv video coding structure based on multi-view camera array; and a dynamic resource allocation structure based on 3-D object motion. The latter two are: a vibrotactile signal rendering system for live information; and a Universal Multimedia Access architecture for streaming live football video.</p><p>In QoE exploration, we give a detailed and deep discussion of QoE and the enabling techniques. We also develop a conceptual model for QoE. Moreover we place streaming video services in a framework of QoE. The new general framework of streaming video services allows for the interaction between the user, content and technology.</p><p>We demonstrate that it is possible to develop algorithms and architectures that take into account the user's perspective. Quality of Experience in video mobile services is within our reach.</p>

In this thesis we study algorithms and architectures that can provide a better Quality of Experience (QoE) for streaming video systems and services. With cases and examples taken from the application scenarios of football on mobile phones, we address the fundamental problems behind streaming video services. Thus, our research results can be applied and extended to other networks, to other sports and to other cultural activities. In algorithm development, we propose five different schemes. We suggest a blind motion estimation and a trellis based motion estimation with dynamic programming algorithms for Wyner-Ziv coding. We develop a trans-media technology, vibrotactile coding of visual signals for mobile phones. We propose a new bandwidth prediction scheme for real-time video conference. We also provide an effective method based on dynamic programming to select optimal services and maximize QoE. In architecture design, we offer three architectures for real-time interactive video and two for streaming live football information. The former three are: a structure of motion estimation in Wyner-Ziv coding for real-time video; a variable bit rate Wyner-Ziv video coding structure based on multi-view camera array; and a dynamic resource allocation structure based on 3-D object motion. The latter two are: a vibrotactile signal rendering system for live information; and a Universal Multimedia Access architecture for streaming live football video. In QoE exploration, we give a detailed and deep discussion of QoE and the enabling techniques. We also develop a conceptual model for QoE. Moreover we place streaming video services in a framework of QoE. The new general framework of streaming video services allows for the interaction between the user, content and technology. We demonstrate that it is possible to develop algorithms and architectures that take into account the user's perspective. Quality of Experience in video mobile services is within our reach.

Le trafic vidéo a subi une augmentation fulgurante sur Internet ces dernières années. Pour pallier à cette importante demande de contenu vidéo, la technologie du streaming adaptatif sur HTTP est utilisée. Elle est devenue par ailleurs très populaire car elle a été adoptée par les différents acteurs du domaine de la vidéo streaming. C’est une technologie moins couteuse qui permet aux fournisseurs de contenu, la réutilisation des serveurs web et des caches déjà déployés. En plus, elle est exempt de tout blocage car elle traverse facilement les pare-feux et les translations d’adresses sur Internet. Dans cette thèse, nous proposons une nouvelle méthode de vidéo streaming adaptatif appelé “Backward-Shifted Coding (BSC)”. Il se veut être une solution complémentaire au standard DASH, le streaming adaptatif et dynamique utilisant le protocole HTTP. Nous allons d’abord décrire ce qu’est la technologie BSC qui se base sur le codec (encodeur décodeur) à multi couches SVC, un algorithme de compression extensible ou évolutif. Nous détaillons aussi l’implémentation de BSC dans un environnement DASH. Ensuite,nous réalisons une évaluation analytique de BSC en utilisant des résultats standards de la théorie des files d’attente. Les résultats de cette analyse mathématique montrent que le protocole BSC permet de réduire considérablement le risque d’interruption de la vidéo pendant la lecture, ce dernier étant très pénalisant pour les utilisateurs. Ces résultats vont nous permettre de concevoir des algorithmes d’adaptation de qualité à la bande passante en vue d’améliorer l’expérience utilisateur. Ces algorithmes permettent d’améliorer la qualité de la vidéo même étant dans un environnement où le débit utilisateur est très instable.La dernière étape de la thèse consiste à la conception de stratégies de caching pour optimiser la transmission de contenu vidéo utilisant le codec SVC. En effet, dans le réseau, des serveurs de cache sont déployés dans le but de rapprocher le contenu vidéo auprès des utilisateurs pour réduire les délais de transmission et améliorer la qualité de la vidéo. Nous utilisons la programmation linéaire pour obtenir la solution optimale de caching afin de le comparer avec nos algorithmes proposés. Nous montrons que ces algorithmes augmentent la performance du système tout en permettant de décharger les liens de transmission du réseau cœur. / Due to the growth of video traffic over the Internet in recent years, HTTP AdaptiveStreaming (HAS) solution becomes the most popular streaming technology because ithas been succesfully adopted by the different actors in Internet video ecosystem. Itallows the service providers to use traditional stateless web servers and mobile edgecaches for streaming videos. Further, it allows users to access media content frombehind Firewalls and NATs.In this thesis we focus on the design of a novel video streaming delivery solutioncalled Backward-Shifted Coding (BSC), a complementary solution to Dynamic AdaptiveStreaming over HTTP (DASH), the standard version of HAS. We first describe theBackward-Shifted Coding scheme architecture based on the multi-layer Scalable VideoCoding (SVC). We also discuss the implementation of BSC protocol in DASH environment.Then, we perform the analytical evaluation of the Backward-Sihifted Codingusing results from queueing theory. The analytical results show that BSC considerablydecreases the video playback interruption which is the worst event that users can experienceduring the video session. Therefore, we design bitrate adaptation algorithms inorder to enhance the Quality of Experience (QoE) of the users in DASH/BSC system.The results of the proposed adaptation algorithms show that the flexibility of BSC allowsus to improve both the video quality and the variations of the quality during thestreaming session.Finally, we propose new caching policies to be used with video contents encodedusing SVC. Indeed, in DASH/BSC system, cache servers are deployed to make contentsclosed to the users in order to reduce network latency and improve user-perceived experience.We use Linear Programming to obtain optimal static cache composition tocompare with the results of our proposed algorithms. We show that these algorithmsincrease the system overall hit ratio and offload the backhaul links by decreasing thefetched content from the origin web servers.

The popularity of multimedia services over Internet has increased in the recent years. These services include Video on Demand (VoD) and mobile TV which are predominantly growing, and the user expectations towards the quality of videos are gradually increasing. Different video codec’s are used for encoding and decoding. Recently Google has introduced the VP8 codec which is an open source compression format. It is introduced to compete with existing popular codec namely H.264/AVC developed by ITU-T Video Coding Expert Group (VCEG), as by 2016 there will be a license fee for H.264. In this work we compare the performance of H.264/AVC and WebM/VP8 in an emulated environment. NetEm is used as an emulator to introduce delay/delay variation and packet loss. We have evaluated the user perception of impaired videos using Mean Opinion Score (MOS) by following the International Telecommunication Union (ITU) Recommendations Absolute Category Rating (ACR) and analyzed the results using statistical methods. It was found that both video codec’s exhibit similar performance in packet loss, But in case of delay variation H.264 codec shows better results when compared to WebM/VP8. Moreover along with the MOS ratings we also studied the effect of user feelings and online video watching experience impacts on their perception. / Yeshwanth Alahari Phone : +91-9986739097 Buddhiraja Prashant Phone : +46-734897359

To be successful in today&apos;s competitive market, service providers should look at user&apos;s satisfaction as a critical key. In order to gain a better understanding of customers&apos; expectations, a proper evaluations which considers intrinsic characteristics of perceived quality of service is needed. Due to the subjective nature of quality, the vagueness of human judgment and the uncertainty about the degree of users&apos; linguistic satisfaction, fuzziness is associated with quality of experience. Considering the capability of Fuzzy logic in dealing with imprecision and qualitative knowledge, it would be wise to apply it as a powerful mathematical tool for analyzing the quality of experience (QoE). This thesis proposes a fuzzy procedure to evaluate the quality of experience. In our proposed methodology, we provide a fuzzy relationship between QoE and Quality of Service (QoS) parameters. To identify this fuzzy relationship a new term called Fuzzi ed Opinion Score (FOS) representing a fuzzy quality scale is introduced. A fuzzy data mining method is applied to construct the required number of fuzzy sets. Then, the appropriate membership functions describing fuzzy sets are modeled and compared with each other. The proposed methodology will assist service providers for better decision-making and resource management.

Nowadays, Third Generation (3G) mobile phones equipped with powerful hardware are becoming popular and dominating the market of cellular com-munication systems. Features such as music and video players, in-built Global Positioning System (GPS) receivers, navigation maps, Internet connectivity and high resolution cameras has converted mobile phones into so called Smartphones. With the increase in applications and services, limitations on energy consumption are also increasing. Therefore, it is more im-portant for manufacturers to nd e ective means of increasing battery life of mobile phones, as the mobile device itself is energy consuming and longer operational times are demanded by customers. Good energy mangement in mobile phones requires a good understanding of the energy usage in mobile phones. To this end, this thesis report presents the results of power and energy consumption measurements conducted on available Smartphones. The services under investigation includes basic Smartphone functionalities and few Internet services are studied and conclusions were drawn. This paper presents a cost e ective methodology for reliable measurements of energy on Smartphones. It also examines the e ect of operating systems on en-ergy consumption in mobile phones and reports less energy consumption for Android supported phones. The energy consumption of these Smartphones with respect to various applications has been studied and related to Quality of Experience of users&apos;. Based on the results conclusions were drawn. / This Thesis will serve as a guide to manufacturers and Users of smartphones while choosing smartphones and designing energy-efficient protocols. / +46 (0)723260656