MENU

   C4I Home

   Center Overview
   C4I Brief
   C4I People

   Objectives
   Programs
   Funding & Support
   Industry Partners

   Publications
   Recordings

   C4I Events
   News

   Internet Conference

GMU C4I Center Seminar

---

Measurement Science for Complex Information Systems

Kevin Mills, Ph. D.

Friday, March 27, 2009 at 2pm

Science & Technology II building, Room 320

ABSTRACT

Modern distributed systems, such as the Internet, consist of millions of independent components whose collective behavior exhibits macroscopic properties, which feedback to the components, altering their behavior and exerting additional influence on observed macroscopic properties. These feedback loops cause difficulty in predicting global behavior, which emerges in a large system. As a result, injecting new technology into such systems risks unexpected, undesirable behaviors. To overcome this problem, NIST researchers are defining methods to measure, understand and predict macroscopic behavior in large, distributed systems; aiming toward a measurement science for complex information systems. This lecture introduces the motivation underlying the research, identifies the major research challenges and outlines project organization. The bulk of the lecture investigates a specific challenge problem: predicting effects on macroscopic network behavior from adopting proposed replacement congestion-control algorithms for the Internet. First, we introduce a reduced-scale system model and illustrate how we establish confidence that the model provides reasonable representation of an Internet-like network. Second, we describe how we model individual congestion-control algorithms and how we validate those models. Third, we explain and demonstrate several data analysis techniques that reveal similarities and differences among macroscopic behaviors caused by varying congestion-control algorithms.

BIO

Kevin Mills has held various research and management positions at NIST since 1982, except for three years as a DARPA program manager. Mills, who received his PhD in information technology from George Mason University (GMU), worked for five years in industry and the USMC, where he developed real-time systems. Concurrent with his NIST career, Dr. Mills served for ten years on the adjunct faculty of the GMU School for Information Technology and Engineering. He is a senior member of the IEEE. For more information about Dr. Mills, please consult: http://www.antd.nist.gov/~mills.