Background: In the recent years, the focus of research has shifted to Quality of Experience(QoE) to maintain the user satisfaction levels and fulfill their expectations of the serviceoffered. Numerous work has been established in finding the relationship between the networklayer and QoE. But, it is fact that the transport layer is much closer to the end-user than thenetwork layer in the TCP/IP communication protocol suite. Thus, any changes in the degreeof satisfaction or degree of annoyance are directly reflected onto transport layer than on thenetwork layer. So, it becomes more significant to study the behavior of user satisfaction inrelation to transport layer than the network layer. Objectives: This research is to relate the behavior of TCP to QoE. The main considerations tobridge the gap between them are: (a) Analyzing the effects of using different versions of TCPon server and client side, (b) Monitoring and analyzing the intensity of TCP traffic in thereverse direction and (c) Investigating TCP control flags from client to server. Methods: QoE related parameters used in this research are: (a) Quality of video i.e., MOS, (b)Degree of disturbance caused by initial delay, (c) Degree of disturbance caused by jerkinessand (d) Degree of disturbance caused by freezes. Effects of network impairments like delay,jitter and packet loss are considered in this research. NetEm is used as the traffic emulationsoftware to shape the traffic. The packet capture analysis of traffic exchange is implementedusing tcpdump. Results and Conclusions: The aim of this research is to provide a passive-estimation methodto assess the user perceived performance. The results of this research provide valuablecontribution to service providers/operators to note the early warning signals from TCP reversetraffic to evaluate the decrease of user satisfaction level and try to cope or/and recover fromimpairments in the network. This research also provides a scope for future researchers toinvestigate other protocols both in transport and application layers.

Le cloud computing permet d'offrir un accès à la demande à des ressources de calcul et de stockage. Le succès du cloud computing nécessite la maîtrise d'aspects système et réseau. Dans cette thèse, nous nous sommes intéressés aux performances du protocole TCP Cubic, qui est la version par défaut de TCP sous Linux et donc présent dans de nombreux serveurs opérationnels dans les data centers actuels. Afin de comprendre les performances d'un environnement cloud, qui offre un faible produit bande passante-délai pour le cas intra-data center, et un fort produit dans le cas inter-data center, nous avons développé des modèles analytiques pour les cas d'une ou plusieurs connexions TCP Cubic. Nos modèles se sont révélés précis dans le cas intra-datacenter, mais ne capturaient pas la synchronisation des pertes indiquée par les simulations ns-2 dans le cas inter-datacenter. Nous avons complété les simulations par des tests en environnements réels avec (i) un réseau expérimental à l'I3S ; et (ii) une solution cloud interne à Orange : Cube. Les études dans Cube nous ont démontré la forte corrélation qui pouvait exister entre performances réseau et système, et la complexité d'analyser les performances des applications dans des contextes cloud. Les études dans l'environnement I3S ont confirmé la forte synchronisation qui peut exister entre connexions TCP Cubic et nous ont permis de définir les conditions d'apparition de cette synchronisation. Nous avons étudié deux types de solution pour lutter contre la synchronisation: des solutions niveau client, avec des modifications de TCP Cubic, et des solutions réseau avec l'utilisation de politiques de gestion de tampon, notamment PIE et Codel. / Cloud computing enables a flexible access to computation and storage services. This requires, for the cloud service provider, mastering network and system issues. During this PhD thesis, we focused on the performance of TCP Cubic, which is the default version of TCP in Linux and thus widely used in today's data centers. Cloud environments feature low bandwidth-delay products (BPD) in the case of intra data center communications and high BDP in the case of inter data center communications. We have developed analytical models to study the performance of a Cubic connection in isolation or a set of competing Cubic connections. Our models turn out to be precise in the low BDP case but fail at capturing the synchronization of losses that ns-2 simulations reveal in the high BDP case. We have complemented our simulations studies with tests in real environments: (i) an experimental network at I3S and (ii) a cloud solution available internally at Orange: Cube. Studies performed in Cube have highlighted the high correlation that might exist between network and system performance and the complexity to analyze the performance of applications in a cloud context. Studies in the controlled environment of I3S have confirmed the existence of synchronization and enabled us to identify its condition of appearance. We further investigated two types of solution to combat synchronization: client level solutions that entail modifications of TCP and network level solutions based on queue management solutions, in particular PIE and Codel.

No / People, systems, and things in the cities generate large amount of data which is considered to be the most scalable asset of any smart city. Linux users are rapidly increased in last few years, and many large multinational organizations are deploying long distance high bandwidth (LDHB) cloud networks for centralizing the data from various smart cities on a central location. TCP is responsible for reliable communication of data in these cloud networks. For reliability communication among various smart cities, a number of TCP congestion control mechanisms have been developed in the past. TCP Compound, TCP Fusion, and TCP CUBIC are the default TCP congestion control mechanisms for Microsoft Windows, Sun Solaris, and Linux operating systems respectively. The response function of TCP CUBIC is higher than the response function of Standard TCP, which is a trademark congestion control mechanism. As a result, TCP CUBIC does not behave friendly with Standard TCP in LDHB cloud networks. The Congestion Window (cwnd) reduction and growth of TCP CUBIC is very aggressive, which causes high packet loss rate and unfair share of available link bandwidth among competing flows from various smart cities. The aim of this research is to design a new TCP congestion control mechanism for Linux operating system to achieve maximum performance in LDHB cloud networks being used by smart cities. In this paper, congestion control module for slow start (CCM-SS) is designed by increasing the lower boundary limit of cwnd size in slow start phase of communication. Congestion control module for loss event (CCM-LE) is designed by increasing the cwnd reduction rate at each packet loss event and finally Advance Response Function for TCP CUBIC (ARFC) is proposed to design a new congestion control mechanism for Linux operating system. NS-2 is used to compare the performance of TCP CUBIC* with TCP CUBIC in short distance high bandwidth (SDHB) and long distance high bandwidth (LDHB) cloud networks. Results show that TCP CUBIC* has outperformed in LDHB networks, at least by a factor of 18% as compared to TCP CUBIC.

No description available.