David A. Fisher
Software Engineering Institute, Carnegie Mellon University
4500 Fifth Avenue, Pittsburgh PA 15213
Phone: 412-268-8995
Email: dafisher@ieee.org
DVP term expires December 2013
David A. Fisher is currently a Senior Research Scientist in the Software Engineering Institute (SEI) at Carnegie Mellon University where he conducts research on next generation information security. Dr. Fisher has held technical and executive positions in academia, industry, and government. His research interests include modeling and simulation, emergent behavior, and automated reasoning especially as they relate to security, HPC, and socio-technical systems. He has degrees in computer science (Ph.D. Carnegie Mellon 1970), electrical engineering (M.S.E. Univ. of Pennsylvania), and mathematics (B.S. Carnegie Mellon), and is a Senior Life Member of the IEEE.
Accurate Agent-Based Modeling and Simulation
Inaccurate models limit the utility, reuse, and performance of modeling and simulation applications, but are inherent in object-based computation. This presentation describes a property-based modeling approach that overcomes many traditional limitations. The approach employes declarative model specifications, abstract agent models, and simulation driven compilation. Fundamental aspects of the modeling and simulation process, limitations of the object-oriented paradigm, abstract specifications using property-based types, separation of domains of expertise in software development, and automated methods for performance optimization are discussed.
Design of A Very High Level Programming Language
Omega is a very high level programming language with potential to dramatically increase productivity, accuracy, and performance in software development and use. Although it is intended specifically for accurate modeling and simulation and research in emergent behavior, it is a general purpose, declarative, property-oriented, and agent-based language that can be used in a broad spectrum of applications. It enables accurate abstract descriptions of anything and provides specialized notations for function and process descriptions. Omega is derived from and incorporates lessons learned from Easel, an experimental agent-based modeling and simulation language developed and used from 1998 to 2006. The presentation includes design, implementation strategy, and example programs.
Emergent Behavior in Software and Networked Systems
Emergent behaviors are pervasive, but poorly understood, in physical, biological, and social sciences. Until recently, they were seldom seen in software and automated systems and their existence is still sometimes denied. Emergent properties are irreducible characteristics of a system. They cannot be explained in terms of their constituent parts. This presentation discusses domain independent principles of emergence and gives examples of emergent effects as the source of software failures and security vulnerabilities. It discusses emergent effects in the form of cascade effects, epidemics, phase transformations, dissipative structures, spontaneous synchronization, emergent structure, and laws of new emergent domains, and explores the possibility of exploiting emergence to achieve previously unavailable beneficial capabilities.
