Browsing by Author "Dr. Michael Devetsikiotis, Committee Chair"

Now showing 1 - 7 of 7

An Autonomic Service Delivery Platform for Service-Oriented Network Environments
(2008-03-24) Callaway, Robert David; Dr. Michael Devetsikiotis, Committee Chair; Dr. Yannis Viniotis, Committee Co-Chair; Dr. Mihail L. Sichitiu, Committee Member; Dr. Adolfo F. Rodriguez, Committee Member; Dr. Andrew J. Rindos, Committee Member
Service-oriented architectures offer a more effective and flexible approach to integrating technology with business processes than traditional information technology (IT) architectures. Service-oriented architectures are the foundation for both next-generation telecommunications and middleware architectures, which are rapidly converging on top of commodity transport services. Services such as triple/quadruple play, multimedia messaging, and presence are enabled by the emerging service-oriented IP Multimedia Subsystem, and allow telecommunications service providers to maintain, if not improve, their position in the marketplace. Service-oriented architectures are aggressively leveraged in next-generation middleware systems as the system model of choice to interconnect service consumers and providers within and between enterprises. We leverage previous research in active, overlay, and peer-to-peer networking technologies, along with recent advances in XML and Web Services, to create the paradigm of service-oriented networking (SON). SON is an emerging architecture that enables network devices to operate at the application layer to provide functions such as service-based routing, content transformation, and protocol integration to consumers and providers. By adding application-awareness into the network fabric, SON can act as a next-generation federated enterprise service bus that provides vast gains in overall performance and efficiency, and enables the integration of heterogeneous environments. The contributions of this research are threefold: first, we formalize SON as an architecture and discuss the challenges in building SON devices. Second, we discuss issues in interconnecting SON devices to create large-scale service-oriented middleware and telecommunications systems; in particular, we discuss the concept of federations of enterprise service buses, and present two protocols that enable a distributed service registry to support the federation. Finally, we propose an autonomic service delivery platform for service-oriented network environments. The platform enables a self-optimizing infrastructure that balances the goals of maximizing the business value derived from processing service requests and the optimal utilization of IT resources.
Modeling and Performance Prediction of IP Multimedia Subsystem Networks
(2006-05-04) Rajagopal, Nisha; Dr. Khaled Harfoush, Committee Member; Dr. Yannis Viniotis, Committee Member; Dr. Michael Devetsikiotis, Committee Chair
IP Multimedia Subsystem (IMS) is envisioned as the solution for the next generation multimedia rich communication. Based on an open IP infrastructure, IMS enables access independent convergence of data, speech, video and mobile network technologies. Session Initiation Protocol (SIP) is the signaling protocol of choice for IMS. In this thesis, we propose service models for IMS networks with utility functions. These service models act as control mechanisms to optimize network properties. We analyze the IMS network based on the SIP signaling delay and predict performance trends of the network. Our focus is on the formulation of queueing models for the IMS network and characterization of the SIP server workload. This approach of theoretical evaluation combined with realistic performance characterization can be used for designing IMS networks with optimal performance. Our analysis is based on a careful study of real-life SIP network traffic.
On-line Measurement-based Capacity Allocation Schemes
(2003-06-30) Haciomeroglu, Fatih; Dr. Michael Devetsikiotis, Committee Chair; Dr. J. Keith Townsend, Committee Member; Dr. George N. Rouskas, Committee Member
Today's high-speed packet-switched networks are faced with the task of handling an increasing amount and variety of services, requiring different QoS constraints. To cope with this demand, the networks need dynamic and measurement-based resource allocation algorithms. For this task, the choice of appropriately accurate but also "practically implementable" algorithms is crucial. In this thesis, we perform a comparative study of alternative on-line algorithms, we analyze their complexity, and perform comparisons via simulation experiments. Our motivation is to use these algorithms in the data plane of "self-sizing" frameworks, and make use of their output in taking control plane decisions either locally or globally, in an "on-line" fashion. Due to the dynamic characteristics of the algorithms, we encounter the choice of time resolution, namely the setting of measurement time scale and window. After numerous simulations, we gain insight on the critical effect of these choices on the performance of the algorithms. We deduce that the time scale parameter itself is to be determined "dynamically" so that measurement-based algorithms can perform successfully independent from the varying traffic conditions. Finally, we demonstrate the effectiveness of this new approach over the static one, in our measurement-based capacity allocation algorithms.
Performance Evaluation of iSCSI Protocol for Mirroring Application.
(2008-11-25) Godbole, Chaitanya Umesh; Dr. Michael Devetsikiotis, Committee Chair; Dr. Wenye Wang, Committee Member; Dr. Khaled Harfoush, Committee Member
FC SAN has been the work-horse of the storage industry for quiet sometime now. It has prevailed and prospered in the enterprise-level storage environment due to its high performance and reliability. Data mirroring for disaster management and data recovery is gaining immense importance as the amount of data being stored increases exponentially. FC has been the default transport protocol for mirroring due to its performance advantages. But now with the demand for mirroring solutions for small & medium sized businesses being on the rise, the high acquisition and maintenance costs of FC have propelled iSCSI to be a cost-effective and viable alternative to FC. Traditionally iSCSI has been deemed unfit for delay sensitive applications due to its slow nature and lower throughput. In this thesis, we attempt to evaluate the performance of mirroring over iSCSI and show that it can be performed satisfactorily and economically, without requiring the costlier FC option. This study involves the use of a customized caching algorithm deployed on a SAN in order to exploit any performance enhancements when it comes to the response time for an application running on iSCSI. The cache tries to reduce the response-time for a write request for mirroring by deploying a two-level primary cache with a faster, smaller backup cache or a faster primary cache with a two-level backup cache. A comparative study of the results obtained for mirroring after deploying these caches over FC and iSCSI show that iSCSI provides adequate performance and reliability for a successful deployment of the mirroring application over it.
QoS Provisioning and Pricing in Multiservice Networks: Optimal and Adaptive Control over Measurement-based Scheduling
(2005-08-14) Xu, Peng; Dr. Michael Devetsikiotis, Committee Chair; Dr. George Michailidis, Committee Member; Dr. Peng Ning, Committee Member; Dr. Wenye Wang, Committee Member; Dr. Ioannis Viniotis, Committee Member
In order to ensure efficient performance under inherently and highly variable traffic in multiservice networks, we propose a generalized adaptive and optimal control framework to handle the resource allocation. Even though this framework addresses rigid Quality of Service concerns for the deterministic delay-bound classes by reserving part of the link capacity and employing appropriate admission control and traffic shaping schemes, our research actually emphasizes the adaptive and optimal control of the shared resources for the flexible delay-bound classes. Therefore, the resource allocation is delivered by a subsystem of this generalized framework, the measurement-based optimal resource allocation (MBORA) system. By applying a simple threshold policy, we first validate the advantages of the adaptivity of our proposed framework through extensive simulation results. Then we introduce a generalized profit-oriented formulation inside decision module of MBORA system, that supplies the network provider with criteria in terms of profit, by leveraging the utility charge revenue and delay-incurred cost. The optimal resource allocation will be affected by the various types of pricing models together with the different levels of service guarantee constraints. As a case study, we investigate this generalized profit-oriented formulation under generalized service models. Combining further with a linear pricing model subject to average queue delay constraints, we propose a fast algorithm for online dynamic and optimal resource allocation under this specific scenario. Finally, we propose a delay-sensitive nonlinear pricing model for the generalized profit-oriented formulation, that realizes two-tier delay differentiation. By better understanding the fluid queueing model, we propose a generalized solution strategy for linear, nonlinear or mixed pricing models that is free of the dimensionality problem and amenable to online implementation.
QoS Provisioning in Wi-Fi Networks: Capacity Modeling and Resource Control
(2005-11-09) Hui, Jie; Dr. Arne Nilsson, Committee Member; Dr. Wenye Wang, Committee Member; Dr. Michael Devetsikiotis, Committee Chair; Dr. Steve Roberts , Committee Member
The ubiquitous Wireless Fidelity (Wi-Fi) networks, and their increasing quality of service (QoS) requirements for emerging applications, motivate extensive studies of QoS provisioning in such networks. Two tasks, namely, capacity modeling and resource control, are crucial in solving the problem: A performance model is first needed to predict the network QoS metrics from the network settings. We propose a new unified analytical model to study the saturation throughput and delay performance of 802.11e Enhanced Distributed Coordination Function (EDCA), which is easier to apply than the most current ones. In order to find usable mathematical models for most cases where analytical models are not feasible, we first advocate the application of metamodeling techniques to Wi-Fi performance studies and formulate a general metamodeling framework for such purpose. The results in three case studies support the validity of our methodology: our saturation capacity metamodel for 802.11 Distributed Coordination Function (DCF) displays an interesting log-linear relationship between capacity and number of users; our voice over Wi-Fi admission capacity metamodel gives a much tighter bound than bounds existing in the literature; and, finally, our throughput metamodel for a simple ad-hoc network, for the first time, characterizes the cross-layer effects between MAC and network layer. Our work, therefore, points out a new direction for future performance studies of Wi-Fi networks. Then, based on the performance models we derive, control schemes of input parameters can be designed to achieve certain level of QoS outputs in some cases. For example, we are able to design a Weighted Round Robin (WRR) scheduler at the MAC layer to control the share of the radio resources, by applying our analytical model to a special case of EDCA configuration. Furthermore, based on our fitted metamodel for the capacity of voice over Wi-Fi, a more practical admission control scheme is composed.
Resource Pricing for Connection-Oriented Networks
(2004-12-01) Ninan, Bobby M; Dr. Michael Devetsikiotis, Committee Chair; Dr. Stephen Campbell, Committee Member; Dr. Arne Nilsson, Committee Member; Dr. Yannis Viniotis, Committee Member
Network pricing has important implications in the revenue generation, resource management, system optimization and congestion control of computer networks. We depart from the prevalent idea of marginal cost pricing and provide a holistic, bi-level optimization framework to model the interaction between network entities in a connection oriented network. Users are treated as utility maximizing entities who allocate the available bandwidth among themselves by playing a distributed, noncooperative rate game. The ensuing Nash equilibrium is analyzed for the single link Erlang network and the multi-link product form networks. Variants based on the upper bound of the blocking are also studied owing to their role in reducing computational complexity. Theoretical results are then validated using numerical simulation for varying network scenarios. An extension of the rate adaptation game based on Recursive Least Squares is proposed for dealing with the imperfect information scenario. These exhibited favorable convergence, accuracy and scalability properties. Gradient-free schemes are then developed for revenue maximization. These are based on novel stochastic approximation techniques such as Finite Difference Stochastic Approximation (FDSA) and Simultaneous Perturbation Stochastic Approximation (SPSA). It is observed that the network employed price discrimination for optimizing its objective function and partitioning its available capacity among competing users.