This research focuses on how society uses, values, and protects citizens? personal information. From the perspective of system design, software engineers need methods and tools to enable them to design systems that reflect those values and protect personal information, accordingly. This research examines how privacy considerations and value systems influence the design, deployment and consequences of IT. The goal is to develop concepts, tools and techniques that help IT professionals and policy makers bring policies and system requirements into better alignment. An action-oriented set of conceptual tools, including guidelines and privacy- relevant policy templates will be constructed and validated.

This proposal focuses on efforts to integrate core research and educational objectives. The research addresses important issues in the discovery, elaboration and management of system use scenarios for the specification of software requirements. The ultimate goal is to develop viable solutions for supporting the early stages of the software lifecycle by ensuring requirements coverage.

This research focuses on how society uses, values, and protects citizens' personal information. From the perspective of system design, software engineers need methods and tools to enable them to design systems that reflect those values and protect personal information, accordingly. This research examines how privacy considerations and value systems influence the design, deployment and consequences of IT. The goal is to develop concepts, tools and techniques that help IT professionals and policy makers bring policies and system requirements into better alignment. An action-oriented set of conceptual tools, including guidelines and privacy-relevant policy templates will be constructed and validated.

The goal of the project is to develop a rendering program that will produce computer generated images of a three dimensional human model to illustrate the effects of ultraviolet radiation on the human anatomy. This model will determine exposure by using a ray-tracing algorithm integrated with a geodesic sun-tracking calculation and a broadbank radiative transfer equation. A three dimensional representation of the human form will illustrate the results. This program will produce images to be used to build an Internet web site for dissemination of real-time surface solar radiation measurements

The centroid decomposition, an approximation for the singular value decomposition, had a long but early history within the statistics/psychometrics community for factor analysis research. We revisit the centroid method first in its original context and then generalize and modernize it to arbitrary matrices. We show the centroid method can be cast as an n-step (linear) ascent method on a hypercube. Furthermore, we have shown empirically that the centroid decomposition is statistically sound. A major purpose of this work is to show fundamental relationships between the singular value, centroid and semi-discrete decompositions. This unifies an entire class of truncated SVD approximations.

This project will investigate methods for navigating complex information spaces. Work will focus on a system designed to help viewers visualize, explore, and analyze large, multidimensional datasets. Detailed local displays will be combined with a high-level global overview of areas of interest within a dataset. Local views will use perceptual cues to harness the low-level human visual system. Global overviews will identify and cluster elements of interest to produce an underlying graph that: (1) support efficient navigation via graph traversal, and (2) provide an effective visualization of the areas of interest and their relationships to one another.

This project will address three issues: (1) can we harness and apply low-level human perception to the problem of visualizing large, complex, multidimensional datasets? (2) can we embed this knowledge in an AI-based system that will assist viewers in constructing perceptually-optimal visualizations in a general way to address a wide range of problem environments? (3) can results from perception be bound to stylistic properties in Impressionist painting, thereby creating a system that allows a viewer to "paint" an expressive visual representation of their data, while at the same time ensuring the result accurately portrays the underlying data values being displayed?

This project studies methods for interactive exploration of complex data spaces through the combination of textual and graphical discourse engines, a plan recognition system, and an interaction manager. Users begin by asking questions about their data. The system responds using text and graphics. Text responses are built by a discourse engine; graphical images are constructed using a perceptual visualization assistant. Plan recognition algorithms analyze queries and users' reactions to the responses they receive. This allows the system to anticipate future queries, cache relevant statistics, and guide the discourse and visualization systems during evaluation of new user queries.

Symbolic representations are used in analysis of finite and infinite state concurrent system. However, they could be subjected to constraint explosion much like state explosion in analysis of finite state designs of concurrent systems. The reason for both of these explosions is the consideration of all interleavings, of a concurrent system, during their analysis. Partial-order techniques depend upon the notion of independence among actions to avoid considering all possible interleavings. The proposed research will investigate the notion of unfolding, which aids both in discovery of independent actions and in succinctly representing the state space of systems.

The project will explore semantic theories of systems that have both non-determinism and probabilistic choice. In particular, notions of equality and approximate equality of system behaviors will be investigated. Furthermore, the effect of these notions on compositional reasoning will also be studied. The second topic of the investigation will be a thorough comparison of the semantic theories developed in this project against traditional approaches to dealing with non-determinism and probabilistic choice. Finally, practical algorithms for process minimization and for checking equality (and approximate equality) of processes will be designed and implemented in the Concurrency Workbench of New Century.

The workshop titled ``Radical Innovations of software and systems engineering of the future'' to be held October 7-11, 2002 at Universita' Ca'Foscari di Venezia in Venice, Italy, will bring together leading researchers in all aspects of Software and System Engineering with a view towards discussing potential research topics of tomorrow. To accomplish this goal, the workshop will have several talks and several open discussions, with the hope that the latter would lead to cross-fertilization of ideas from several disparate sub-areas of Software and Systems Engineering.

Concurrent systems are difficult to build and debug. Those that have to contend with unreliable components are even harder to work with. We propose to investigate how formal methods can be used to address both logical correctness and reliability/performance constraints of concurrent systems. To that end, we propose to investigate: (a) Design languages for complex systems that contain non-deterministic and probabilistic behavior, (b) Design of a temporal-logic based requirements language and design of model-checking algorithms, and (c) Implementation of probabilistic model-checking in Concurrency Workbench.

The Center for Advanced Computing and Communication (CACC) is a membership-based industry/university cooperative research center co-located at North Carolina State University and Duke University. North Carolina State University was selected by the National Science Foundation in 1981 as a site for an industry/university cooperative research center in communications and signal processing. The center was named the Center for Communications and Signal Processing until 1994 when a second center site at Duke University was added. The CACC research goal is to create concepts, methods and tools for use in the analysis, design and implementation of advanced computer and communication systems.

It provides support for a graduate student for one year and the project will study whether and how information gathered by a to improve the detection accuracy. The main tasks of the project include network management system can be utilized by an intrusion detection system developing algorithms for anomaly detection and early sensing of intrusions.

This project proposes a novel system for rapid development and deployment of effective and cost-sensitive IDSs. Our system automates feature construction by analyzing the patterns of normal and intrusion activities computed from audit data. Detection models are constructed automatically using cost-sensitive machine learning algorithms to achieve optimal performance on the given cost metrics. Our system finds the cluster of attack signatures and normal profiles and constructs one light model for each cluster to maximize the utility of each model. A dynamically configurable group of such light models can be very effective and efficient, and resilient to IDS-related attacks

Given the centrality of explanation in science, one of the most intriguing forms of intelligent multimedia learning environments revolves around explanation. Self-explaining learning environments will enable learners to pose questions to objects and processes in 3D worlds that can dynamically generate clear cinematic and narrative explanations of their own structure, function, and causality.

This work puts forth a two-tier approach to reduce the processor frequency of complex embedded systems. First, tight worst-case timing analysis reduces the perceived upper bound on the number of cycles consumed by tasks. Second, architecture simulation and processors with dual frequency/voltage modes enable significant additional power savings. Architecture simulation produces an approximate worst-case timing estimate. A higher recovery frequency is utilized as a fall-back mode to ensure safe operation bounded by tight worst-case timing analysis. These two approaches complement each other. They initially reduce the power requirements by significant amount when compared with naive approach.

This proposal addresses problems in exploiting the memory bandwidth of shared-memory multiprocessors (SMPs) for scientific applications. For contemporary high-performance clusters of SMPs, it has been found that a number of scientific applications utilizing a mixed mode of MPI+OpenMP are performing worse than when relying on MPI, only. Considering that the architectural model of SMPs seems to be a close fit to the OpenMP threading model, this performance gap seems particularly surprising. The objective of this proposal is to determine the sources of inefficiencies in utilizing memory hierarchies for threaded programs vs. parallel processes and to assist the programmer in alleviating these problems. The methodology to perform this analysis relies on binary rewriting.

This project is motivated by current limitations of intrusion detection systems, which are generally unable to fully detect unknown attacks, or even unknown variations of known attacks, without generating a large number of false alarms. The focus of this project is to integrate intrusion detection with visualization techniques and human computer interaction strategies to address these limitations. Our system will include interactive intrusion detection algorithms that capitalize on human knowledge and judgment, novel visualization and interaction techniques to monitor for potential attacks, and semi-automated tools for constructing and evaluating attack profiles to extend the capabilities of an intrusion detection system.

Current intrusion detection systems (IDSs) usually generate a large amount of false alerts, and often do not detect novel attacks or variations of known attacks. Moreover, most of the existing IDSs focus on low-level attacks or anomalies; none of them can capture the logical steps or strategies behind these attacks. This project will exploit alert correlation techniques to reduce false alerts, discover attackers¹ high-level strategies, and predict possible future attacks based on the detected attacks in progress.

Current intrusion detection systems (IDSs) usually generate a large amount of false alerts, and often do not detect novel attacks or variations of known attacks. Moreover, most of the existing IDSs focus on low-level attacks or anomalies; none of them can capture the logical steps or strategies behind these attacks. As a result, it is difficult for human users or intrusion response systems to understand the nature of the attack and to take appropriate actions. To address these issues, this project will investigate techniques to correlate intrusion alerts on the basis of the prerequisites and consequences of attacks.

The principle investigators will undertake an analysis of the performance and alternate routing issues that arise in optical burst switching (OBS) networks implementing the Just In Time signaling protocol. he objective of this study is to estimate the throughput and the burst dropping probability in a network of OBS nodes. We will first develop a queueing model for a single OBS node, and subsequently we will use it to construct a model of a network of OBS nodes. At the same time, we will also develop a simulation model of a single OBS node and subsequently of a network of OBS nodes. The simulation model will be used to validate the accuracy of the queueing model. It will also be used to investigate various scenarios and assumptions that may not be possible to capture in the queueing model.

We will study the concept of Compartment in a networking system to handle different types of network infrastructure attacks. We particularly will develop a compartment-based network system to protect BGP (Border Gateway Protocol) routing protocol. Furthermore, we will combine Compartments with the Deciduous model to more effectively identify the intrusion sources.

The next generation Internet will have mechanisms for providing Quality of Service (QoS). The standard mechanisms do not address the issues of security, or prevention of misuse. In this project, we address three new vulnerability questions with regard to QoS: (1) how to prevent "killer reservation" denial-of-service (DoS) attacks; (2) how to detect and respond to DoS attacks that are directed at the data flow; (3) how to reduce the vulnerability of resource reservation and allocation mechanisms. Our objective is to improve the survivability of the QoS mechanisms.

We propose to study a new class of error recovery techniques that focuses on eliminating error propagation. The approach is to isolate errors when they occur by preventing them from propagating. The delays in repairing data losses affect only the duration of error propagation. Our Recovery from Error Spread using Continuous Updates (RESCU) does not introduce any delay in video playout, and has potential to achieve good error resilience. Our proposed techniques, in contrast, have potential to work well with or without feedback channels, and to be scalable for multicast. Encouraging preliminary results indicate that such potential is highly realizable.

As the Internet becomes more diversified in its capabilities, it becomes more feasible to offer services that were not possible under earlier generations of Internet technologies. Realtime multimedia streaming and IP multicast are two such emerging technologies. The objective of this proposed work is to develop, verify analytically and experimentally, and implement a suite of end-to-end flow control protocols for unicast and multicast real-time streaming applications. The developed protocols are evaluated based on fairness, TCP friendliness, stability and scalability. These properties must hold regardless of the types of networks, or more specifically whether networks are symmetric or asymmetric in bandwidth and delays.

The career development plan outlined in this proposal focuses on the integration of the research and teaching objectives of the principal investigator. The research project addresses a number of important issues in the design and interconnection of optical networks. The ultimate goal of the proposed work is to develop viable solutions for a seamless nationwide all-optical network infrastructure. The educational component includes development of graduate-level courses on all-optical networking and on internetwork architectures and protocols, integration of topics on lightwave technology in existing graduate and undergraduate introductory network courses, and active involvement of undergraduate students in network projects.

The proposed research will develop a regional optical testbed network carrying IP traffic between MCNC, UNC, NCSU, and Duke. The testbed architecture will consist of interconnected access networks. The access networks and the interconnecting network will utilize an optical single-hop architecture. Network elements consist of access nodes interconnected to a passive coupler to form a broadcast star topology. Access node implementations will utilize an extended version of the HiPer-1 optical Media Access Control protocol developed and analyzed at NCSU.

Wavelength routed wide area networks will be an integral part of the future all-optical network infrastructure. It is envisioned that these networks will act as the backbone that provides the interconnection for local area lightwave sub-networks attached to them.. However, the deployment of an infrastructure that will enable large scale interconnection of optical networks raises a number of fundamental and challenging problems that require novel and innovative solutions. The research plan outlined in this proposal identifies a number of significant but open problems arising in wavelength routed environments, and presents a comprehensive and novel approach to address them.

The research is a continuing investigation of fundamental questions involving the structure and size of combinatorial families and various methods for counting and generating them. The algorithmic component of the research concerns the design of fast methods for exhaustive generation of combinatorial families as well as schemes for listing the members according to special criteria which are dictated by applications in which they arise. The mathematical component investigates open questions about classical combinatorial structures such as integer partitions, compositions, and the Boolean lattice. The interaction between the two components is significant: the algorithms are applied to study the mathematical questions; the mathematical insights lead to more efficient algorithms.

Successful interaction relies heavily upon trust. This applies equally to electronic commerce and virtual social communities. However, figuring out who to trust and to what extent is extremely difficult in open networked information environments. Trust is a complex concept and involves aspects of competence and good nature (of the trusted party) and the risk tolerance and urgency (of the trusting party). This project studies distributed, scalable computational approaches for trust management, especially with regard to aggregate phenomena such as the emergence of subcommunities, pivots (which link different subcommunities), and the sensitivity of a community to invasion by nontrustworthy players.

Business protocols structure and streamline interactions among autonomous business partners. Traditional representations of protocols specify legal sequences of actions but not their meaning. Thus they cannot adequately support flexible interactions, e.g., to handle exceptions and exploit opportunities. This project is developing a declarative model of protocols that gives meaning to, and reasons about, states and actions based on the participants' commitments. This approach improves flexibility while maintaining rigor. This project is studying practical protocols from real-life domains such as transactions among financial institutions and other varieties of electronic business.

This research focuses on the concept of Ibots, interface agents that interact with software applications through the graphical user interface, in essentially the same way that human users do. The proposal will produce an environment for agent exploration and evaluation, and a model of tool use in the interface. The project will produce tools that facilitate research in A1 planning, user interface agents, and cognitive modeling. It will also give results of general interest to the agents community.

Providing cognitive models with access to human/robot interfaces can help explore the design space, provide a way to support the user, and support the development of an engineering-style iterative approach to interface design based on automatic evaluation. The most important new features we can identify are the ability to work with existing interfaces, the role of interaction, and the ability to include user models that learn. We propose to explore each of these areas, by combining ongoing work in interface design, cognitive models, and interface agents.

The theory and application of Markov chain techniques constitute a unifying theme in the application of mathematics to many problems in engineering, economics, physical science, and social science, and the numerical computation of stationary probabilities associated with large scale applications is a fundamental concern. This is particularly true in applications such as the modeling of flexible manufacturing systems, telecommunication systems, computer performance evaluation, biological systems, economic modeling and forecasting, and more generally, in applications where discrete models are used to understand and analyze the dynamics of large systems comprised of a collection of loosely coupled subsystems. This project emphasizes both the practical and theoretical aspects surrounding the application and implementation of algorithms for computing stationary probabilities.

Due to the popularity of object-oriented technology and network applications, the use of concurrent object-oriented programs is increasing significantly. The goal of the project is to develop efficient and effective techniques for analyzing and testing concurrent object-oriented programs. This project will investigate the following problems: (a) How to efficiently and effectively analyze and test a class that defines data to be shared by multiple processes? (b) How to efficiently and effectively analyze and test a set of classes that define processes communicating with each other? (c) How to perform incremental analysis and testing of a concurrent object-oriented program according to the relationships between classes in the program?

NCSU graduate student will develop a machine learning approach to linking extreme atmospheric ridging events with extreme surface temperatures, employing a Gaussian Process (GP)-based predictive analysis tool that leverages the predictive value of spatial and temporal correlations and builds on ORNL’s past success in spatial classification, temporal change prediction, and parallelizing GP for large spatiotemporal extents.

Scientific Data Management Center is a SciDAC funded center with a goal to establish an Enabling Technology Center that will provide a coordinated framework for the unification, development, deployment, and reuse of scientific data management software, including scientific workflow technologies, specifically through SDM?s Scientific Process Automation (SPA) focus area. The goal of this technology is to allow for easy and accurate interactions and flows among distributed computational, storage and application resources used in scientific discovery.

The principle goals of this project are to develop an advanced but affordable network-centric collaborative educational paradigm and environment which will increase the effectiveness of the undergraduate and graduate learning experience by facilitating and promoting rapid integration of research results into undergraduate and graduate courses and curricula. The paradigm operates in the context of an efficient state-of-the-art network-based engine that supports advanced virtual laboratory spaces, and collaborative content capture, development and delivery mechanisms. It is also very much an "appliance" along the lines white-boards and overhead projectors are.

The purpose of the Girls on Track (GoT) project is to increase Middle Grade Girls' interest in math-related careers by engaging them in computer-based mathematical explorations of urban problems in their communities. It is a joint project between North Carolina State University (Center for Research in Mathematics and Science Education and Department of Computer Science), Meredith College, Wake County Public Schools, and North Carolina Department of Public Instruction.

This project is a seven-year longitudinal research study of young women who were identified as talented in mathematics in middle school. The purpose of this research is to address the current low participation of young women in IT careers. The project will develop and test a model of the factors associated with young women's decisions to persist in advanced mathematics and computer science courses so as to prepare themselves for, and decide to make information technology [IT] their career. IT careers are defined in this proposal as those requiring an electrical engineering, computer science, or computer engineering degree.

Initial experimental results indicate that pair-learning, whereby two students work together at one computer, improves the success and morale of the students. Students working in pairs are able to answer each other’s questions. They no longer look to the teaching staff as their sole source of technical advice; educators are no longer burdened by an onslaught of questions. Grading can be significantly reduced when two students submit one assignment. The classes are calmer; the students are more satisfied and self-sufficient. We also believe that women and minority students would especially benefit from this pedagogical approach.

This project will investigate approaches to building a strategy generation engine as a component of a flexible trading agent that converts user preferences, auction rules, and a model of the other agents into a decisionable format. The first strategy generation engine will produce game-theoretic representations of the decision problem. For small problems, the game can be solved and an equilibrium bidding strategy selected. However, for intractable larger problems, alternate strategy generation engines will be constructed which use other decision technologies. Ideally, the agent will be able to make this decision by assessing the structure of the problem instance.

The use of virtual environments has shown success in applications ranging from education to entertainment. One limitation of these systems is that users' activities within them are over- or under-constrained. In this project, I will develop new models for the structure of user interactions within virtual worlds. Because a user's understanding of the activity in a world provides scaffolding for her own exploration, presenting the user with an environment in which action can be readily understood encourages the user to acquire and employ knowledge of the environment. This activity leads to an increased understanding of the world the environment models.