North Carolina State University

Header Navigation:

Main Navigation:

 

Cyberinfrastructure

Project Leader: John Baugh, NCSU


Project Description 

Computational aspects of the proposed projects include the need for working with diverse data sources covering large geographic regions, representing and visualizing evolving landforms and infrastructure systems, monitoring physical systems with wireless sensor networks in real time, providing novel design approaches for protective structures using risk-based analysis, and integrating system-wide analysis and modeling approaches in a computational framework for decision support.  While existing tools and approaches can be utilized for various parts of the project activities, the collective aims of this research are ambitious and forward-looking, and as such there exist no off-the-shelf solutions for addressing its overall computational requirements.  Indeed, for concurrent real-time systems, high performance numerical optimization, and software architectures supporting a suite of analysis tools, as examples, contributions in the area of software and computer systems development are needed.
This project addresses computational aspects of the Engineering to Enhance the Resilience of the Built and Natural Environments focus area by:

  1. Eliciting and defining requirements for computational approaches to support overall research project activities
  2. Modeling software and computer architectures to enable assessment of viability before development and deployment
  3. Developing software that implements needed components and, where appropriate, builds upon and integrates existing tools and approaches identified by other projects.

Tasks designed to meet these crosscutting objectives are coordinated with other research project areas and are carried out in an iterative manner that allows for feedback and assessment.

Analytical Approach

Like any engineering activity, the development of software is a matter of experience and judgment, and the use of any particular technology cannot guarantee success.  However, some fundamental principles, including incremental development, iterative extraction and refinement of requirements, and periodic evaluation by clients and external reviewers, have all been shown to reduce the chance that the software development process will go awry; these principles appear sound and appropriate for research-oriented prototype development as well.  This project will draw on these and on our own experience in designing cyberinfrastructure components.

While general software engineering principles apply throughout, some aspects of the cyberinfrastructure warrant special consideration, particularly those involving real-time and high performance requirements, since they inevitably give rise to issues of concurrency, distribution, and multiprocessor communication, and in the case of systems modeling the need to accommodate and integrate diverse analytical tools. For instance, remote user interfaces, locality and distribution of data, communication and sampling rates, and network management are among the design considerations in developing a system for health monitoring and condition assessment.  Beneath the veneer of a user interface lies an inherently distributed system of heterogeneous processors and sensors, which monitor and assess the state of geographically remote structures.  Such systems must manage persistent data, perhaps through database systems or serialized objects, and offer them through web- or server-based technologies such as servlets. In addition, given the distances required for covering civil infrastructure components, such as levees, dams, and other protective structures, wireless technology is envisioned with the determination of spatial density, number of hops, and aggregator station placement based on bandwidth considerations and transmission ranges for appropriate commercially available nodes.

As with physical systems, the conception and development of this type of computer architecture can benefit from modeling approaches that foster insight.  Where appropriate, quantitative approaches such as communicating finite state machines may be combined with simulation or perhaps model checking to allow proposed architectures to be conceptualized and assessed, enabling predeployment validation. Designing the cyberinfrastructure with these considerations is particularly important if this system were to be extended for real-time decision-making during an event, which requires the integration of data from networks of diverse physical sensors and devices to enable system-wide observation across differing time scales. By explicitly modeling system concurrency and real-time aspects, performance measures and dependability concerns can be assessed, for instance, for configurations with variations in the sampling, persistence, and locus of data being collected.

With respect to integrated systems modeling, major activities include the scoping of needs and identification of requirements for modeling and computational support.  For instance, functional requirements will be defined for integrating analytical tools and modeling and optimization approaches through the use of data exchange standards, coordination languages, and component technology to facilitate interoperability. Such an effort requires the implementation of wrappers that enable the execution of software systems within software systems so that component-level engineering analysis can be performed within the context of system-wide optimization, as an example.  While the cyberinfrastructure system will be designed with extensibility in mind, special consideration will be given for integration with and building upon existing tools of relevance (e.g., building upon the existing Community Vulnerability Assessment Tool for New Hanover County [http://www.lib.ncsu.edu/gis/commvuln.html] and considering the next-generation of multi-hazard tools such as HAZUS-MH).

Because the integrated systems models are expected to be compute-intensive, additional non-functional requirements dictate a high performance computing environment that employs task-level parallelism over a local area network of processors and perhaps intermittent use of special-purpose machines or grid-level computing. Potential speedup in execution time through distributed computing can be achieved in optimization and other decision support capabilities, such as Monte Carlo simulation, which include non-sequential steps. Because of the relative amount of work that may be performed independently among these components without synchronizing, a coarse-grained distributed solution can be developed that uses this to advantage.  In addition, search procedures that allow conjunctive use of analysis models may be implemented in a distributed setting in some cases, allowing a more accurate model to screen and correct the search direction.

We will incrementally develop these capabilities of the cyberinfrastructure using real case studies in the first two years. We intend to refine the requirements of the cyberinfrastructure through development, implementation, and testing of a series of prototypes for civil infrastructure design and analysis of at least two real case studies. The case studies to be considered will be coordinated with the case studies chosen for consideration in the other projects, especially the case studies for which the integrated systems models are to be developed and investigated. As these real case studies shed light on the necessary capabilities of the envisioned cyberinfrastructure, the design and implementation of the prototypes will shape the system specifications and definitions for the eventual implementation that will continue into the planned work in years 3 to 6.

Data Collection

These efforts are coordinated with other area research both within and at the interfaces of the Engineering to Enhance the Resilience of the Built and Natural Environments research projects.

Measurable Goals

Project goals include the development and implementation of computational tools and prototypes that are meant to demonstrate concepts, and also to support the research goals of other research projects.  This will include contributions in the areas of visualizing landforms and infrastructure systems, health monitoring and condition assessment, computational support for risk-based design, and integrating system-wide analysis and modeling approaches in a high performance framework for decision support.

Timeline with Milestones and Deliverables

Through our investigations and experience with cyberinfrastructure prototypes for the case studies, we expect to have the work in the first two years accomplish the following milestones and deliverables:

  1. Identification and scoping of real civil infrastructure hazard assessment case studies;
  2. Description/specification and demonstration of a representative set of analytical and optimization approaches for solving the integrated systems model;
  3. Design and implementation of prototypes; and
  4. Testing of the prototypes using different case study scenarios

The following timeline for activities is anticipated to accomplish the milestones described above.

  1. August 2008 to February 2009
  2. January 2009 to December 2009
  3. Mach 2009 to December 2009
  4. Octobert 2009 to August 2010

Years 3 and 4 will include further development of prototypes based on the specification and modeling enhancements produced in the other research projects.  These will be implemented, validated, and then demonstrated to the broader research group as well as to DHS personnel at regular project meetings and at annual center conferences.  Because the cyberinfrastructure framework will be designed with modularity and computational performance in mind, refinements in years 5 and 6 will focus on enabling optimization and decision support for large scale integrated systems.

 

 

 



College of Engineering