Browsing by Author "Mihail Devetsikiotis, Committee Member"

Now showing 1 - 4 of 4

Active Queue Management and Scheduling Methods for Packet-Switched Networks
(2005-03-13) Akin, Ozdemir; Yannis Viniotis, Committee Member; J. Keith Townsend, Committee Chair; George Rouskas, Committee Member; Mihail Devetsikiotis, Committee Member
To support the myriad of envisioned communication products of the future, there is a need to develop a network infrastructure that can provide larger bandwidth, with better control of quality of service (QoS). However, with increasing demand for applications running over packet networks, congestion at the intermediate nodes (e.g., routers and switches) can be a serious problem. Consequences include long delays, large delay variation and high packet loss rates. Different solutions requiring varying levels of modification to the currently used algorithms have been proposed both for responsive (e.g., TCP) and unresponsive (e.g., UDP) protocols. However, most of the solutions are either too complicated to implement in real life or not general enough to be applicable to an arbitrary network topology. In this thesis, we investigate two mechanisms - active queue management (AQM), and scheduling - that can improve QoS in the packet networks. AQM techniques attempt to prevent congestion and regulate the queue length by sending congestion signals (i.e., dropping and/or marking packets) in a proactive manner, which would eventually cause the senders to decrease their sending rates. We use an analytic model derived for TCP in the literature to develop an AQM scheme that not only controls the queue length at the intermediate nodes but also distributes the resources fairly between the users. We present two different schemes that have different levels of complexity and performance. We also propose a distributed networking scheme that improves the performance of our new AQM algorithms. Although AQM schemes work well with responsive protocols such as TCP, the performance degrades for unresponsive protocols since unresponsive protocols do not change their packet sending rate in response to the congestion notifications sent by the network. Scheduling algorithms can regulate both responsive and unresponsive flows and can also provide guarantees on some QoS parameters, such as latency and fairness. We propose a new scheduling algorithm that can guarantee a better latency and fairness bound than the previous schemes with a similar complexity. This new scheme uses different queues for each round to simplify the common sorting problem with the scheduling schemes. Also a new notion, virtual round, is used to time-stamp incoming packets.
Joint Optimization in Supply Chain
(2006-07-27) Zhou, Sean Xiang; Henry Nuttle, Committee Member; Xiuli Chao, Committee Chair; Mihail Devetsikiotis, Committee Member; Shu-Cherng Fang, Committee Co-Chair
With increased globalization and competition in the current market, supply chain has become longer and more complicated than ever before. An effective and efficient supply chain is crucial and essential to a successful firm. In a supply chain, inventories are a very important component as the investment in inventories is enormous. Inventory management is always coupled with other functions, for example purchasing, production and marketing. In this dissertation, we study inventory management for both single-stage and multi-echelon systems. Two main streams of research work are summarized. The first is the joint optimization of pricing and inventory control for continuous/periodic review single-stage inventory/production systems. We characterize the optimal policies and further develop efficient computational algorithms to find the optimal control parameters. We also provide insights on the pricing and inventory relationship. The second is the analysis of multi-echelon inventory systems, in which we derive the optimal inventory control policies for several different systems that have not yet been studied in the current literature. Moreover, simple bounds and heuristics for the optimal policies are developed for the serial systems with and without expedited shipping so that the implementability of the optimal policies is improved.
Performance analysis of power management in WLAN and UMTS
(2006-08-15) Lei, Hongyan; Arne A. Nilsson, Committee Chair; Alexandra Duel-Hallen, Committee Member; Wenye Wang, Committee Member; Mihail Devetsikiotis, Committee Member
Wireless networks have enjoyed the exponential development, and wireless communication has become an essential part of modern life. Many new wireless applications demand higher speed and consume more energy. However, wireless devices are always powered by batteries, which have limited life time and constrain the use of wireless devices and the growth of wireless networks. Energy efficiency becomes an important issue in wireless networks. We study the energy efficiency in the IEEE 802.11 based WLAN (Wireless Local Area Network) and the third generation cellular system UMTS (Universal Mobile Telecommunication System), in which the basic mechanism is to put a mobile device into a low power consumption state when it is idle and wake it up periodically to transmit/receive traffic. In WLANs, the study is focused on the MAC (Media Access Control) sublayer. Two queueing models for the power management mechanisms in an infrastructure network are proposed: the M/G/1 queue with bulk service and the D/G/1 queue. Based on the analytical and simulation results, suggestions are given about how to optimally configure the power management parameters. We also propose the enhanced power management schemes for both infrastructure and independent networks, which outperform the original schemes based on our analysis and simulation. In UMTS, the impacts of Discontinuous Reception (DRX) mechanism and inactivity timer are studied. The simulation of the performance of power saving mechanism is carried out by inputting several typical traffic models specified by 3GPP (third generation partnership project). From the results that different traffic models demand different optimal parameters, we propose to adaptively configure the DRX cycle and inactivity timer parameters based on real-time measurements.
Resource Optimization and Security in Distributed Computing
(2009-12-11) Xiong, Kaiqi; Mihail Devetsikiotis, Committee Member; Matthias Stallmann, Committee Member; Dr. David Thuente, Committee Member; Harry Perros, Committee Chair
With the number of e-Business applications dramatically increasing, service level agreements (SLA) will play an important part in distributed service computing. An SLA is a combination of several quality of service (QoS) metrics, such as security, performance, and availability, agreed between a customer and a service provider. Due to the complexity of these metrics, most existing research typically addresses only one of these QoS metrics. In the case of the response time as a performance metric, the average time to process and complete a job is typically used in the literature. However, this may not be of real interest to a customer. A statistically bounded metric, that is, a percentile response time, is more realistic than the average response time. Moreover, in enterprise service computing, customer requests are typically distinguished by different request characteristics and service requirements. This dissertation includes a study of trustworthiness, percentile response time, service availability, and authentication among service stations or sites that may be owned by different service providers. The first part of this dissertation contains an analysis of percentile response time, which is one of the most important SLA metrics. Effective and accurate numerical solutions for the calculation of the percentile response time in single-class and multi-class queueing networks are obtained. Then, the numerical solution is incorporated in a resource allocation problem. Specifically, we present an approach for the resource optimization that minimizes the total cost of computer resources required while preserving a given percentile of the response time. In the second part of this dissertation, we extend the approach to consider trustworthiness, service availability, and the percentile of response time in Web services. We clearly define these QoS metrics and provide their quantitative analysis. Then, we take into account these QoS metrics in a trust-based resource allocation problem in which a set of computer resources is used by a service provider to host a typical Web services application for single-class customer services and multiple-class customer services respectively. We formulate the trust-based resource allocation problem as an optimization problem under SLA constraints in which we calculate the number of servers in each service site that minimize a cost function that reflects operational costs for single-class customer services and multiple-class customer services respectively. We solve this problem using an efficient numerical procedure. Experimental results show the applicability of the procedure and validate its accuracy. Finally, in the third part of this dissertation we first present a thorough performance evaluation of two notable public key cryptography-based authentication techniques, Public-Key Cross Realm Authentication in Kerberos (PKCROSS) and Public Key Utilizing Tickets for Application Servers (PKTAPP, a.k.a. KX.509/KCA), in terms of computational and communication times. We then demonstrate their performance difference using queueing networks. PKTAPP was proposed to address the scalability issue of PKCROSS. However, our in-depth analysis of these two techniques shows that PKTAPP does not perform better than PKCROSS in a large scale system. Thus, we propose a new public key cryptography-based group authentication technique. Our performance analysis demonstrates that the new technique can scale better than PKCORSS and PKTAPP.