A cooperative distributed problem-solving framework

This dissertation describes a framework for cooperative distributed problems solving (CDPS) and how this framework can be used to solve the class of problems comprised of semi-independent sub-problems. The CDPS framework provides speedup over traditional sequential problem solving methods, increased reliability and flexibility, reduced static consistency checking, graceful degradation owing to the increased spectral nature of knowledge, and efficient knowledge acquisition. In addition, the CDPS framework facilitates the coupling of disparate knowledge sources by means of a robust communication protocol, this facilitates the coupling of artificial intelligence (AI) techniques with more traditional problem solving techniques such as decision theory, operations research, etc. The CDPS framework also promotes the preservation of multiple perspectives and as a result multiple solutions. The dissertation describes the verification of the CDPS framework via the successful implementations of two prototypes, namely, the Computer Systems Manager and the Supplier-side Risk Analyzer. The dissertation also describes design experiments carried out on the CDPS framework that serve as useful guidelines for future developers of such systems. The design experiments involve comparisons with a Sequential Model and a Distributed Model in order to get a real feel for the performance of the CDPS framework.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Business Administration

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2011

Author dc:creator

• Sundararajan, Srikanth

• Business Administration, Management
• Artificial Intelligence
• Computer Science

Statement dc:rights

• Copyright 1990 Sundararajan, Srikanth

Language dc:language

eng

Identifier

AAI9026329
(UMI)AAI9026329

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/19185