Yeung, Fei-Fei. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2003. / Includes bibliographical references (leaves 144-146). / Abstracts in English and Chinese. / Introduction --- p.1 / Chapter Part I: --- A New Socket API for Enhancing Server Efficiency --- p.5 / Chapter Chapter 1 --- Introduction --- p.6 / Chapter 1.1 --- Brief Background --- p.6 / Chapter 1.2 --- Deficiencies of Nagle's Algorithm and Goals and Objectives of this Research --- p.7 / Chapter 1.2.1 --- Effectiveness of Nagle's Algorithm --- p.7 / Chapter 1.2.2 --- Preventing Small Packets via Application Layer --- p.9 / Chapter 1.2.3 --- Minimum Delay in TCP Buffer --- p.10 / Chapter 1.2.4 --- Maximum Delay in TCP Buffer --- p.11 / Chapter 1.2.5 --- New Socket API --- p.12 / Chapter 1.3 --- Scope of Research and Summary of Contributions --- p.12 / Chapter 1.4 --- Organization of Part 1 --- p.13 / Chapter Chapter 2 --- Background --- p.14 / Chapter 2.1 --- Review of Nagle's Algorithm --- p.14 / Chapter 2.2 --- Additional Problems Inherent in Nagle's Algorithm --- p.17 / Chapter 2.3 --- Previous Proposed Modifications on Nagle's Algorithm --- p.22 / Chapter 2.3.1 --- The Minshall Modification --- p.22 / Chapter 2.3.1.1 --- The Minshall Modification --- p.22 / Chapter 2.3.1.2 --- The Minshall et al. Modification --- p.23 / Chapter 2.3.2 --- The Borman Modification --- p.23 / Chapter 2.3.3 --- The Jeffrey et al. Modification --- p.25 / Chapter 2.3.3.1 --- The EOM and MORE Variants --- p.25 / Chapter 2.3.3.2 --- The DLDET Variant --- p.26 / Chapter 2.3.4 --- Comparison Between Our Proposal and Related Works --- p.26 / Chapter Chapter 3 --- Min-Delay-Max-Delay TCP Buffering --- p.28 / Chapter 3.1 --- Minimum Delay --- p.29 / Chapter 3.1.1 --- Why Enabling Nagle's Algorithm Alone is Not a Solution? --- p.29 / Chapter 3.1.2 --- Advantages of Min-Delay TCP-layer Buffering versus Application-layer Buffering --- p.30 / Chapter 3.2 --- Maximum Delay --- p.32 / Chapter 3.2.1 --- Why Enabling Nagle's Algorithm Alone is Not a Solution? --- p.32 / Chapter 3.2.2 --- Advantages of Max-delay TCP Buffering versus Nagle's Algorithm --- p.33 / Chapter 3.3 --- Interaction with Nagle's Algorithm --- p.34 / Chapter 3.4 --- When to Apply Our Proposed Scheme? --- p.36 / Chapter 3.5 --- New Socket Option Description --- p.38 / Chapter 3.6 --- Implementation --- p.40 / Chapter 3.6.1 --- Small Packet Transmission Decision Logic --- p.42 / Chapter 3.6.2 --- Modified API --- p.44 / Chapter Chapter 4 --- Experiments --- p.46 / Chapter 4.1 --- The Effect of Kernel Buffering Mechanism on the Service Time --- p.47 / Chapter 4.1.1 --- Aims and Methodology --- p.47 / Chapter 4.1.2 --- Comparison of Transmission Time Required --- p.49 / Chapter 4.2 --- Performance of Min-Delay-Max-Delay Scheme --- p.56 / Chapter 4.2.1 --- Methodology --- p.56 / Chapter 4.2.1.1 --- Network Setup --- p.56 / Chapter 4.2.1.2 --- Traffic Model --- p.58 / Chapter 4.2.1.3 --- Delay Measurement --- p.60 / Chapter 4.2.2 --- Efficiency of Busy Server --- p.62 / Chapter 4.2.2.1 --- Performance of Nagle's algorithm --- p.62 / Chapter 4.2.2.2 --- Performance of Min-Delay TCP Buffering Scheme --- p.67 / Chapter 4.2.3 --- Limiting Delay by Setting TCP´ؤMAXDELAY --- p.70 / Chapter 4.3 --- Performance Sensitivity Discussion --- p.77 / Chapter 4.3.1 --- Sensitivity to Data Size per Invocation of send() --- p.77 / Chapter 4.3.2 --- Sensitivity to Minimum Delay --- p.83 / Chapter 4.3.3 --- Sensitivity to Round Trip Time --- p.85 / Chapter Chapter 5 --- Conclusion --- p.88 / Chapter Part II: --- Two Analytical Models for a Refined TCP Algorithm (TCP Veno) for Wired/Wireless Networks --- p.91 / Chapter Chapter 1 --- Introduction --- p.92 / Chapter 1.1 --- Brief Background --- p.92 / Chapter 1.2 --- Motivation and Two Analytical Models --- p.95 / Chapter 1.3 --- Organization of Part II --- p.96 / Chapter Chapter 2 --- Background --- p.97 / Chapter 2.1 --- TCP Veno Algorithm --- p.97 / Chapter 2.1.1 --- Packet Loss Type Identification --- p.97 / Chapter 2.1.2 --- Refined AIMD Algorithm --- p.99 / Chapter 2.1.2.1 --- Random Loss Management --- p.99 / Chapter 2.1.2.2 --- Congestion Management --- p.100 / Chapter 2.2 --- A Simple Model of TCP Reno --- p.101 / Chapter 2.3 --- Stochastic Modeling of TCP Reno over Lossy Channels --- p.103 / Chapter Chapter 3 --- Two Analytical Models --- p.104 / Chapter 3.1 --- Simple Model --- p.104 / Chapter 3.1.1 --- Random-loss Only Case --- p.105 / Chapter 3.1.2 --- Congestion-loss Only Case --- p.108 / Chapter 3.1.3 --- The General Case (Random + Congestion Loss) --- p.110 / Chapter 3.2 --- Markov Model --- p.115 / Chapter 3.2.1 --- Congestion Window Evolution --- p.115 / Chapter 3.2.2 --- Average Throughput Formulating --- p.119 / Chapter 3.2.2.1 --- Random-loss Only Case --- p.120 / Chapter 3.2.2.2 --- Congestion-loss Only Case --- p.122 / Chapter 3.2.2.3 --- The General Case (Random + Congestion Loss) --- p.123 / Chapter Chapter 4 --- Comparison with Experimental Results and Discussions --- p.127 / Chapter 4.1 --- Throughput versus Random Loss Probability --- p.127 / Chapter 4.2 --- Throughput versus Normalized Buffer Size --- p.132 / Chapter 4.3 --- Throughput versus Bandwidth in Asymmetric Networks --- p.135 / Chapter 4.3 --- Summary --- p.136 / Chapter Chapter 5 --- Sensitivity of TCP Veno Throughput to Various Parameters --- p.137 / Chapter 5.1 --- Multiplicative Decrease Factor (α) --- p.137 / Chapter 5.2 --- Number of Backlogs (β) and Fractional Increase Factor (γ) --- p.139 / Chapter Chapter 6 --- Conclusions --- p.142 / Bibliography --- p.144

Ma Ke. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2003. / Includes bibliographical references (leaves 64-68). / Abstracts in English and Chinese. / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Background --- p.1 / Chapter 1.2 --- Motivation --- p.2 / Chapter 1.3 --- Contribution --- p.3 / Chapter 1.4 --- Organization --- p.4 / Chapter 2 --- Literature Review --- p.5 / Chapter 2.1 --- Overview --- p.5 / Chapter 2.2 --- Multi-Path Routing --- p.6 / Chapter 2.2.1 --- OSPF-ECMP --- p.7 / Chapter 2.2.2 --- LFI --- p.7 / Chapter 2.2.3 --- QSMP and QDMP --- p.9 / Chapter 2.2.4 --- WDP --- p.10 / Chapter 2.2.5 --- DMPR --- p.11 / Chapter 2.2.6 --- Cidon's Analysis --- p.13 / Chapter 3 --- LSLF and SLSLF Conditions --- p.15 / Chapter 3.1 --- Problem Formulation --- p.15 / Chapter 3.2 --- LFI Conditions --- p.16 / Chapter 3.3 --- LSLF Conditions --- p.17 / Chapter 3.4 --- SLSLF Conditions --- p.20 / Chapter 4 --- Performance of LSLF and SLSLF --- p.24 / Chapter 4.1 --- Overview --- p.24 / Chapter 4.2 --- Numerical Results --- p.26 / Chapter 5 --- Analysis of Multi-path Routing --- p.42 / Chapter 5.1 --- Assumptions --- p.43 / Chapter 5.2 --- M/M/C/C Queueing System --- p.44 / Chapter 5.3 --- Performance Analysis --- p.48 / Chapter 5.3.1 --- "Case 1 Only QoS flows between (s, d) exist" --- p.48 / Chapter 5.3.2 --- Case 2 QoS flows between other SD pairs also exist --- p.50 / Chapter 5.3.3 --- Case 3 A QoS flow can try m times before it is dropped --- p.53 / Chapter 5.4 --- Numerical Results --- p.56 / Chapter 6 --- Conclusion --- p.62

Tan, Min. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2008. / Includes bibliographical references (p. 35-36). / Abstracts in English and Chinese. / Abstract --- p.i / Acknowledgement --- p.iii / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Previous Work --- p.1 / Chapter 1.2 --- Motivation --- p.2 / Chapter 1.3 --- Contributions --- p.2 / Chapter 1.4 --- Thesis Organization --- p.3 / Chapter 2 --- Linear Network Coding Basics --- p.5 / Chapter 2.1 --- Formulation and Example --- p.5 / Chapter 2.2 --- Some Notations --- p.9 / Chapter 3 --- A Unified Framework --- p.13 / Chapter 3.1 --- Generic Network Codes Revisited --- p.13 / Chapter 3.2 --- A Unified Framework --- p.24 / Chapter 3.3 --- Simplified Proofs --- p.29 / Chapter 4 --- Conclusion --- p.33 / Bibliography --- p.35

Fong, Lik Hang Silas. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2007. / Includes bibliographical references (leaves 40). / Abstracts in English and Chinese. / Abstract --- p.i / Acknowledgement --- p.iii / Chapter 1 --- Introduction --- p.1 / Chapter 2 --- Linear Network Code --- p.4 / Chapter 2.1 --- Linear Network Code without Link Failures --- p.4 / Chapter 2.1.1 --- Linear Multicast and Linear Broadcast --- p.6 / Chapter 2.2 --- Linear Network Code with Link Failures --- p.8 / Chapter 2.2.1 --- Static Linear Multicast and Static Linear Broadcast --- p.9 / Chapter 3 --- Variable-Rate Linear Network Coding --- p.11 / Chapter 3.1 --- Variable-Rate Linear Network Coding without Link Failures --- p.11 / Chapter 3.1.1 --- Problem Formulation --- p.11 / Chapter 3.1.2 --- Algorithm and Analysis --- p.12 / Chapter 3.2 --- Variable-Rate Linear Network Coding with Link Failures --- p.23 / Chapter 3.2.1 --- Problem Formulation --- p.23 / Chapter 3.2.2 --- Algorithm and Analysis --- p.23 / Chapter 3.3 --- The Maximum Broadcast Rate of Linear Network Code --- p.28 / Chapter 4 --- Conclusion --- p.38 / Bibliography --- p.40

Thesis (M.S.)--Rochester Institute of Technology, 2007. / Typescript. Includes bibliographical references.

Inference, measurement and estimation of network path properties is a fundamental problem in distributed systems and networking. We consider a specific subclass of problems which do not require special support from the hardware or software, deployment of special devices or data from the network. Network inference is a challenging problem since Internet paths can have complex and heterogeneous configurations. Inference enables end users to understand and troubleshoot their connectivity and verify their service agreements; it has policy implications from network neutrality to broadband performance; and it empowers applications and services to adapt to network paths to improve user quality of experience. In this dissertation we develop end-to-end user-level methods, tools and services for network inference. Our contributions are as follows. We show that domain knowledge-based methods can be used to infer performance of different types of networks, containing wired and wireless links, and ranging from local area to inter-domain networks. We develop methods to infer network properties: 1. Traffic discrimination (DiffProbe), 2. Traffic shapers and policers (ShaperProbe), and 3. Shared links among multiple paths (Spectral Probing). We develop methods to understand network performance: 1. Diagnose wireless performance pathologies (WLAN-probe), and 2. Diagnose wide-area performance pathologies (Pythia). Among our contributions: We have provided ShaperProbe as a public service and it has received over 1.5 million runs from residential and commercial users, and is used to check service level agreements by thousands of residential broadband users a day. The Federal Communications Commission (FCC) has recognized DiffProbe and ShaperProbe with the best research award in the Open Internet Apps Challenge in 2011. We have written an open source performance diagnosis system, Pythia, and it is being deployed in ISPs such as the US Department of Energy ESnet in wide-area inter-domain settings. The contributions of this dissertation enable Internet transparency, performance troubleshooting and improving distributed systems performance.

When a network packet is formed by a computer's protocol stack, there are many components (e.g., Memory, CPU, etc.) of the computer that are involved in the process. The objective of this research is to identify, characterize and analyze the effects of the various components of a device (e.g., Memory, CPU, etc.) on the device's network traffic by measuring the changes in its network traffic with changes in its components. We also show how this characterization can be used to effectively perform counterfeit detection of devices which have counterfeit components (e.g., Memory, CPU, etc.). To obtain this characterization, we measure and apply statistical analyses like probability distribution fucntions (PDFs) on the interarrival times (IATs) of the device's network packets (e.g., ICMP, UDP, TCP, etc.). The device is then modified by changing just one component (e.g., Memory, CPU, etc.) at a time while holding the rest constant and acquiring the IATs again. This, over many such iterations provides an understanding of the effect of each component on the overall device IAT statistics. Such statistics are captured for devices (e.g., field-programmable gate arrays (FPGAs) and personal computers (PCs)) of different types. Some of these statistics remain stable across different IAT captures for the same device and differ for different devices (completely different devices or even the same device with its components changed). Hence, these statistical variations can be used to detect changes in a device's composition, which lends itself well to counterfeit detection. Counterfeit devices are abundant in today's world and cause billions of dollars of loss in revenue. Device components are substituted with inferior quality components or are replaced by lower capacity components. Armed with our understanding of the effects of various device components on the device's network traffic, we show how such substitutions or alterations of legitimate device components can be detected and hence perform effective counterfeit detection by statistically analyzing the deviation of the device's IATs from that of the original legitimate device. We perform such counterfeit detection experiments on various types of device configurations (e.g., PC with changed CPU, RAM, etc.) to prove the technique's efficacy. Since this technique is a fully network-based solution, it is also a non-destructive technique which can quickly, inexpensively and easily verify the device's legitimacy. This research also discusses the limitations of network-based counterfeit detection.

Thesis (Ph. D.)--School of Electrical and Computer Engineering, Georgia Institute of Technology, 2004. Directed by Nikil Jayant. / Vita. Includes bibliographical references (leaves 150-160).

This thesis presents several decentralized, learning-based algorithms for on-line routing of bandwidth guaranteed paths. The presented routing algorithms do not need any a-priori knowledge of traffic demand; they use only their locally observed events and update their routing policy using learning schemes. The employed learning algorithms are either learning automata or the multi-armed bandit algorithms. We investigate the asymptotic behavior of the proposed routing algorithms and prove the convergence of one of them to the user equilibrium. Discrete event simulation results show the merit of these algorithms in terms of improving the resource utilization and increasing the network admissibility compared with shortest path routing. / We investigate the performance degradation due to decentralized routing as opposed to centralized optimal routing policies in practical scenarios. The system optimal and the Nash bargaining solutions are two centralized benchmarks used in this study. We provide nonlinear programming formulations of these problems along with a distributed recursive approach to compute the solutions. An on-line partially-decentralized control architecture is also proposed to achieve the system optimal and the Nash bargaining solution performances. Numerical results in some practical scenarios with well engineered networks, where the network resources and traffic demand are well matched, indicate that decentralized learning techniques provide efficient, stable and scalable approaches for routing the bandwidth guaranteed paths. / In the context of on-line learning, we propose a new algorithm to track the best action-selection policy when it abruptly changes over time. The proposed algorithm employs change detection mechanisms to detect the sudden changes and restarts the learning process on the detection of an abrupt change. The performance analysis of this study reveals that when all the changes are detectable by the change detection mechanism, the proposed tracking the best action-selection policy algorithm is rate optimal. On-line routing of bandwidth guaranteed paths with the potential occurrence of network shocks such as significant changes in the traffic demand is one of the applications of the devised algorithm. Simulation results show the merit of the proposed algorithm in tracking the optimal routing policy when it abruptly changes.

Despite the persistent change and growth that characterizes the Internet, the Transmission Control Protocol (TCP) still dominates at the transport layer, carrying more than 90\% of the global traffic. Despite its astonishing success, it has been observed that TCP can cause poor end-to-end performance, especially for large transfers and in network paths with high bandwidth-delay product. In this thesis, we focus on mechanisms that can address key problems in TCP performance, without any modification in the protocol itself. This evolutionary approach is important in practice, as the deployment of clean-slate transport protocols in the Internet has been proved to be extremely difficult. Specifically, we identify a number of TCP-related problems that can cause poor end-to-end performance. These problems include poorly dimensioned socket buffer sizes at the end-hosts, suboptimal buffer sizing at routers and switches, and congestion unresponsive TCP traffic aggregates. We propose solutions that can address these issues, without any modification to TCP. <br> <br> In network paths with significant available bandwidth, increasing the TCP window till observing loss can result in much lower throughput than the path's available bandwidth. We show that changes in TCP are {em not required} to utilize all the available bandwidth, and propose the application-layer SOcket Buffer Auto-Sizing (SOBAS) mechanism to achieve this goal. SOBAS relies on run-time estimation of the round trip time (RTT) and receive rate, and limits its socket buffer size when the receive rate approaches the path's available bandwidth. In a congested network, SOBAS does not limit its socket buffer size. Our experiment results show that SOBAS improves TCP throughput in uncongested network without hurting TCP performance in congested networks. <br> <br> Improper router buffer sizing can also result in poor TCP throughput. Previous research in router buffer sizing focused on network performance metrics such as link utilization or loss rate. Instead, we focus on the impact of buffer sizing on end-to-end TCP performance. We find that the router buffer size that optimizes TCP throughput is largely determined by the link's output to input capacity ratio. If that ratio is larger than one, the loss rate drops exponentially with the buffer size and the optimal buffer size is close to zero. Otherwise, if the output to input capacity ratio is lower than one, the loss rate follows a power-law reduction with the buffer size and significant buffering is needed. The amount of buffering required in this case depends on whether most flows end in the slow-start phase or in the congestion avoidance phase. <br> <br> TCP throughput also depends on whether the cross-traffic reduces its send rate upon congestion. We define this cross-traffic property as {em congestion responsiveness}. Since the majority of Internet traffic uses TCP, which reduces its send rate upon congestion, an aggregate of many TCP flows is believed to be congestion responsive. Here, we show that the congestion responsiveness of aggregate traffic also depends on the flow arrival process. If the flow arrival process follows an open-loop model, then even if the traffic consists exclusively of TCP transfers, the aggregate traffic can still be unresponsive to congestion. TCP flows that arrive in the network in a closed-loop manner are always congestion responsive, on the other hand. We also propose a scheme to estimate the fraction of traffic that follows the closed-loop model in a given link, and give practical guidelines to increase that fraction with simple application-layer modifications.