Computer Sciences Commons^™

A Software Architecture For Reconstructability Analysis, Kenneth Willett, Martin Zwick

Complex Systems Faculty Publications and Presentations

Software packages for reconstructability analysis (RA), as well as for related log linear modeling, generally provide a fixed set of functions. Such packages are suitable for end‐users applying RA in various domains, but do not provide a platform for research into the RA methods themselves. A new software system, Occam3, is being developed which is intended to address three goals which often conflict with one another to provide: a general and flexible infrastructure for experimentation with RA methods and algorithms; an easily‐configured system allowing methods to be combined in novel ways, without requiring deep software expertise; and a system which …

Pperfgrid: A Grid Services-Based Tool For The Exchange Of Heterogeneous Parallel Performance Data, John Jared Hoffman

Dissertations and Theses

This thesis details the approach taken in developing PPerfGrid. Section 2 discusses other research related to this project. Section 3 provides general background on the technologies utilized in PPerfGrid, focusing on the components that make up the Grid services architecture. Section 4 provides a description of the architecture of PPerfGrid. Section 5 details the implementation of PPerfGrid. Section 6 presents tests designed to measure the overhead and scalability of the PPerfGrid application. Section 7 suggests future work, and Section 8 concludes the thesis.

Using Dynamic Optimization For Control Of Real Rate Cpu Resource Management Applications, Varin Vahia, Ashvin Goel, David Steere, Jonathan Walpole, Molly H. Shor

Computer Science Faculty Publications and Presentations

In this paper we design a proportional-period optimal controller for allocating CPU to real rate multimedia applications on a general-purpose computer system. We model this computer system problem in to state space form. We design a controller based on dynamic optimization LQR tracking techniques to minimize short term and long term time deviation from the current time stamp and also CPU usage. Preliminary results on an experimental set up are encouraging.

Infrastructure For Performance Tuning Mpi Applications, Kathryn Marie Mohror

Dissertations and Theses

Clusters of workstations are becoming increasingly popular as a low-budget alternative for supercomputing power. In these systems, message-passing is often used to allow the separate nodes to act as a single computing machine. Programmers of such systems face a daunting challenge in understanding the performance bottlenecks of their applications. This is largely due to the vast amount of performance data that is collected, and the time and expertise necessary to use traditional parallel performance tools to analyze that data.

The goal of this project is to increase the level of performance tool support for message-passing application programmers on clusters of …

A Browser For Incremental Programming, Andrew P. Black

Computer Science Faculty Publications and Presentations

Much of the elegance and power of Smalltalk comes from its programming environment and tools. First introduced more than 20 years ago, the Smalltalk browser enables programmers to “home in” on particular methods using a hierarchy of manually-defined classifications. By its nature, this classification scheme says a lot about the desired state of the code, but little about the actual state of the code as it is being developed. We have extended the Smalltalk browser with dynamically computed virtual categories that dramatically improve the browser’s support for incremental programming. We illustrate these improvements by example, and describe the algorithms used …

Adaptive Live Video Streaming By Priority Drop, Jie Huang, Charles Krasic, Jonathan Walpole

Computer Science Faculty Publications and Presentations

In this paper we explore the use of Priority-progress streaming (PPS) for video surveillance applications. PPS is an adaptive streaming technique for the delivery of continuous media over variable bit-rate channels. It is based on the simple idea of reordering media components within a time window into priority order before transmission. The main concern when using PPS for live video streaming is the time delay introduced by reordering. In this paper we describe how PPS can be extended to support live streaming and show that the delay inherent in the approach can be tuned to satisfy a wide range of …

Using Reconstructability Analysis To Select Input Variables For Artificial Neural Networks, Stephen Shervais, Martin Zwick

Complex Systems Faculty Publications and Presentations

We demonstrate the use of Reconstructability Analysis to reduce the number of input variables for a neural network. Using the heart disease dataset we reduce the number of independent variables from 13 to two, while providing results that are statistically indistinguishable from those of NNs using the full variable set. We also demonstrate that rule lookup tables obtained directly from the data for the RA models are almost as effective as NNs trained on model variables.

Under The Plastic: A Quantitative Look At Dvd Video Encoding And Its Impact On Video Modeling, Wu-Chi Feng, Jin Choi, Wu-Chang Feng, Jonathan Walpole

Computer Science Faculty Publications and Presentations

In this paper, we examine the DVD encoding process and the implications this process has video modeling and network traffic analysis. We have assembled a system that allows us to extract the video data from the DVDs as they were encoded for distribution. Analyzing the resulting video trace data, we describe how DVD encodings have evolved over time. In addition, our findings show that the underlying video content is fundamentally different than those produced by basic consumer video capture boards. We demonstrate how this affects current video modeling proposals and their affect on network traffic characterization. This research is based …

Ordering Genetic Algorithm Genomes With Reconstructability Analysis, Stephen Shervais, Martin Zwick

Complex Systems Faculty Publications and Presentations

The building block hypothesis implies that genetic algorithm effectiveness is influenced by the relative location of epistatic genes on the chromosome. We find that this influence exists, but depends on the generation in which it is measured. Early in the search process it may be more effective to have epistatic genes widely separated. Late in the search process, effectiveness is improved when they are close together. The early search effect is weak but still statistically significant; the late search effect is much stronger and plainly visible. We demonstrate both effects with a set of simple problems, and show that infonnation-theoretic …

Multi-Level Decomposition Of Probabilistic Relations, Stanislaw Grygiel, Martin Zwick, Marek Perkowski

Complex Systems Faculty Publications and Presentations

Two methods of decomposition of probabilistic relations are presented in this paper. They consist of splitting relations (blocks) into pairs of smaller blocks related to each other by new variables generated in such a way so as to minimize a cost function which depends on the size and structure of the result. The decomposition is repeated iteratively until a stopping criterion is met. Topology and contents of the resulting structure develop dynamically in the decomposition process and reflect relationships hidden in the data.

A Performance Study Of Lam And Mpich On An Smp Cluster, Brian Patrick Kearns

Dissertations and Theses

Many universities and research laboratories have developed low cost clusters, built from Commodity-Off-The-Shelf (COTS) components and running mostly free software. Research has shown that these types of systems are well-equipped to handle many problems requiring parallel processing. The primary components of clusters are hardware, networking, and system software. An important system software consideration for clusters is the choice of the message passing library.

MPI (Message Passing Interface) has arguably become the most widely used message passing library on clusters and other parallel architectures, due in part to its existence as a standard. As a standard, MPI is open for anyone …

Content Aware Request Distribution For High Performance Web Service: A Performance Study, Robert M. Jones

Dissertations and Theses

The World Wide Web is becoming a basic infrastructure for a variety of services, and the increases in audience size and client network bandwidth create service demands that are outpacing server capacity. Web clusters are one solution to this need for high performance, highly available web server systems. We are interested in load distribution techniques, specifically Layer-7 algorithms that are content-aware. Layer-7 algorithms allow distribution control based on the specific content requested, which is advantageous for a system that offers highly heterogenous services. We examine the performance of the Client Aware Policy (CAP) on a Linux/Apache web cluster consisting of …

Implementing Infopipes: The Sip/Xip Experiment, Calton Pu, Galen Swint, Charles Consel, Younggyun Koh, Ling Liu, Koichi Moriyama, Jonathan Walpole, Wenchang Yan

Computer Science Faculty Publications and Presentations

We describe an implementation of the Infopipe abstraction for information flow applications. We have implemented software tools that translate the SIP/XIP variant of Infopipe specification into executable code. These tools are evaluated through the rewriting of two realistic applications using Infopipes: a multimedia streaming program and a web source combination application. Measurements show that Infopipe-generated code has the same execution overhead as the manually written original version. Source code of Infopipe version is reduced by 36% to 85% compared to the original.

Gscope: A Visualization Tool For Time-Sensitive Software, Ashvin Goel, Jonathan Walpole

Computer Science Faculty Publications and Presentations

This paper describes gscope, a visualization tool for time-sensitive applications. Gscope provides an oscilloscope-like interface that can be integrated with applications. It focuses on software visualization and is thus designed to handle various types of signal waveforms, periodic or event-driven, in single or multithreaded environments as well as local or distributed applications. Gscope helps in visually verifying system correctness and modifying system parameters and thus can complement standard debugging techniques and be used to build compelling software demos. Initial experiments with using gscope show that the library has low overhead.

Provisioning On-Line Games: A Traffic Analysis Of A Busy Counter-Strike Server, Wu-Chang Feng, Francis Chang, Wu-Chi Feng, Jonathan Walpole

Computer Science Faculty Publications and Presentations

This paper describes the results of a 500 million packet trace of a popular on-line, multi-player, game server. The results show that the traffic behavior of this heavily loaded game server is highly predictable and can be attributed to the fact that current game designs target the saturation of the narrowest, last-mile link. Specifically, in order to maximize the interactivity of the game itself and to provide relatively uniform experiences between players playing over different network speeds, on-line games typically fix their usage requirements in such a way as to saturate the network link of their lowest speed players. While …

Supporting Low-Latency Tcp-Based Media Streams, Ashvin Goel, Charles Krasic, Kang Li, Jonathan Walpole

Computer Science Faculty Publications and Presentations

The dominance of the TCP protocol on the Internet and its success in maintaining Internet stability has led to several TCP-based stored media-streaming approaches. The success of these approaches raises the question whether TCP can be used for low-latency streaming. Low latency streaming allows responsive control operations for media streaming and can make interactive applications feasible. We examined adapting the TCP send buffer size based on TCP's congestion window to reduce application perceived network latency. Our results show that this simple idea significantly improves the number of packets that can be delivered within 200 ms and 500 ms thresholds.

Poster: Provisioning On-Line Games: A Traffic Analysis Of A Busy Counter-Strike Server, Francis Chang, Wu-Chang Feng, Wu-Chi Feng, Jonathan Walpole

Computer Science Faculty Publications and Presentations

A poster that illustrates the client/server model employed by an multiplayer online game, focusing on bandwidth usage.

Infopipes: An Abstraction For Multimedia Streaming, Andrew P. Black, Huang Jie, Rainer Koster, Jonathan Walpole, Calton Pu

Computer Science Faculty Publications and Presentations

To simplify the task of building distributed streaming applications, we propose a new abstraction for information flow – Infopipes. Infopipes make information flow primary, not an auxiliary mechanism that is hidden away. Systems are built by connecting predefined component Infopipes such as sources, sinks, buffers, filters, broadcasting pipes, and multiplexing pipes. The goal of Infopipes is not to hide communication, like an RPC system, but to reify it: to represent communication explicitly as objects that the program can interrogate and manipulate. Moreover, these objects represent communication in application-level terms, not in terms of network or process implementation.

A Comparison Of Evolutionary And Coevolutionary Search, Ludo Pagie, Melanie Mitchell

Computer Science Faculty Publications and Presentations

Previous work on coevolutionary search has demonstrated both successful and unsuccessful applications. As a step in explaining what factors lead to success or failure, we present a comparative study of an evolutionary and a coevolutionary search model. In the latter model, strategies for solving a problem coevolve with training cases. We find that the coevolutionary model has a relatively large efficacy: 86 out of 100 (86%) of the simulations produce high quality strategies. In contrast, the evolutionary model has a very low efficacy: a high quality strategy is found in only two out of 100 runs (2%). We show that …

Thread Transparency In Information Flow Middleware, Rainer Koster, Andrew P. Black, Jie Huang, Jonathan Walpole, Calton Pu

Computer Science Faculty Publications and Presentations

Existing middleware is based on control-flow centric interaction models such as remote method invocations, poorly matching the structure of applications that process continuous information flows. Difficulties cultiesin building this kind of application on conventional platforms include flow-specific concurrency and timing requirements, necessitating explicit management of threads, synchronization, and timing by the application programmer. We propose Infopipes as a high-level abstraction for information flows, and we are developing a middleware framework that supports this abstraction. Infopipes transparently handle complexities associated with control flow and multi-threading. From high-level configuration descriptions the platform determines what parts of a pipeline require separate threads or …

Querying Geographically Dispersed, Heterogeneous Data Stores: The Pperfxchange Approach, Matthew Edward Colgrove

Dissertations and Theses

This thesis details PPerfXchange’s approach for querying geographically dispersed heterogeneous data stores. While elements of PPerfXchange’s method have been implemented for other application areas, PPerfXchange shows how these elements can be applied to parallel performance analysis. The accomplishments of this thesis are:

• The design of an architecture for PPerfXchange, giving a uniform method to query heterogeneous data stores;
• A proof of concept prototype implementation of PPerfXchange including a partial implementation of an XQuery processor and a relational database virtual XML document; and
• Evaluation of PPerfXchange using example parallel performance analysis data.

Dual Heuristic Programming For Fuzzy Control, George G. Lendaris, Thaddeus T. Shannon, Larry J. Schultz, Steven Hutsell, Alec Rogers

Complex Systems Faculty Publications and Presentations

Overview material for the Special Session (Tuning Fuzzy Controllers Using Adaptive Critic Based Approximate Dynamic Programming) is provided. The Dual Heuristic Programming (DHP) method of Approximate Dynamic Programming is described and used to the design a fuzzy control system. DHP and related techniques have been developed in the neurocontrol context but can be equally productive when used with fuzzy controllers or neuro-fuzzy hybrids. This technique is demonstrated by designing a temperature controller for a simple water bath system. In this example, we take advantage of the TSK model framework to initialize the tunable parameters of our plant model with reasonable …

Reifying Communication At The Application Level, Andrew P. Black, Jie Huang, Jonathan Walpole

Computer Science Faculty Publications and Presentations

Middleware, from the earliest RPC systems to recent Object-Oriented Remote Message Sending (RMS) systems such as Java RMI and CORBA, claims transparency as one of its main attributes. Coulouris et al. define transparency as “the concealment from the … application programmer of the separation of components in a distributed system.” They go on to identify eight different kinds of transparency.

We considered titling this paper “Transparency Considered Harmful”, but that title is misleading because it implies that all kinds of transparency are bad. This is not our view. Rather, we believe that the choice of which transparencies should be offered …

An Information Theoretic Methodology For Prestructuring Neural Networks, Bjorn Chambless, George G. Lendaris, Martin Zwick

Complex Systems Faculty Publications and Presentations

Absence of a priori knowledge about a problem domain typically forces use of overly complex neural network structures. An information-theoretic method based on calculating information transmission is applied to training data to obtain a priori knowledge that is useful for prestructuring (reducing complexity) of neural networks. The method is applied to a continuous system, and it is shown that such prestructuring reduces training time, and enhances generalization capability.

Infopipes—An Abstraction For Information Flow, Jie Huang, Andrew P. Black, Jonathan Walpole, Calton Pu

Computer Science Faculty Publications and Presentations

Building Object-Oriented Distributed Systems has been facilitated by Remote Message Sending (RMS) systems like Java RMI and implementations of CORBA. However, RMS systems are designed to support request/response interactions. Streaming applications, in contrast, are characterized by high-bandwidth, long-duration communication with stringent performance requirements. Examples of streaming applications include video-on-demand, teleconferencing, on-line education, and environmental observation. These applications transfer huge amounts of data and focus on distributed information flow rather than request/response.

To simplify the task of building distributed streaming applications, we propose a new abstraction for information flow—Infopipes. Using Infopipes, information flow becomes the heart of the system, not an …

Globally Convergent Approximate Dynamic Programming Applied To An Autolander, J.J. Murray, Richard Saeks, C.J. Cox, George G. Lendaris

Complex Systems Faculty Publications and Presentations

A globally convergent nonlinear Approximate Dynamic Programming algorithm is described, and an implementation of the algorithm in the linear case is developed. The resultant linear Approximate Dynamic Programming algorithm is illustrated via the design of an autolander for the NASA X-43 research aircraft, without a priori knowledge of the X-43's flight dynamics.

Modeling The Transient Rate Behavior Of Bandwidth Sharing As A Hybrid Control System, Kang Li, Molly H. Shor, Jonathan Walpole, Calton Pu

Computer Science Faculty Publications and Presentations

This paper uses hybrid control to model a problem of computer network systems, the dynamic behavior of bandwidth sharing among competing TCP traffic. It has been well known in the computer network community that well-behaved (TCP-friendly) congestion control mechanisms are crucial to the robustness of the Internet. Congestion control determines the transmission rate for each flow. Right now, most TCP-friendly research focuses only on the average throughput behavior without considering how the data is sent out in the short-term (e.g. bursty or smooth). However, recent experimental results show that short-term rate adjustments can change the bandwidth sharing result. Therefore, it …

Infosphere Project: An Overview, Calton Pu, Jonathan Walpole

Computer Science Faculty Publications and Presentations

We describe the Infosphere project, which is building the systems software support for information-driven applications such as digital libraries and electronic commerce. The main technical contribution is the Infopipe abstraction to support information flow with quality of service. Using building blocks such as program specialization, software feedback, domain-specific languages, and personalized information filtering, the Infopipe software generates code and manage resources to provide the specified quality of service with support for composition and restructuring.

Moving Towards Massively Scalable Video-Based Sensor Networks, Wu-Chi Feng, Jonathan Walpole, Calton Pu, Wu-Chang Feng

Computer Science Faculty Publications and Presentations

Networking and computing technologies are becoming advanced enough to enable a wealth of diverse applications that will drastically change our everyday lives. Some past examples of these developments include the World Wide Web and wireless data networking infrastructures. As is quite obvious, the World Wide Web has enabled a fundamental change in the way many people deal with day-to-day tasks. Through the web, one can now make on-line reservations for travel, pay bills through on-line banking services, and view personalized on-line newscasts. More recently, developments in wireless technologies have enabled anywhere, anytime access to information over wireless medium. As wireless …

Rate-Matching Packet Scheduler For Real-Rate Applications, Kang Li, Jonathan Walpole, Dylan Mcnamee, Calton Pu, David Steere

Computer Science Faculty Publications and Presentations

A packet scheduler is an operating system component that controls the allocation of network interface bandwidth to outgoing network flows. By deciding which packet to send next, packet schedulers not only determine how bandwidth is shared among flows, but also play a key role in determining the rate and timing behavior of individual flows. The recent explosion of rate and timing-sensitive flows, particularly in the context of multimedia applications, has focused new interest on packet schedulers. Next generation packet schedulers must not only ensure separation among flows and meet real-time performance constraints, they must also support dynamic fine-grain reallocation of …