Date of Award

Spring 2002

Project Type


Program or Major

Engineering - System s Design

Degree Name

Doctor of Philosophy

First Advisor

Eugene C Freuder


Modeling, Diagnosis, and Model Debugging are the three main areas presented in this dissertation to automate the process of Interoperability Testing of networking protocols. The dissertation proposes a framework that uses the Constraint Satisfaction Problem (CSP) paradigm to define a modeling language and problem solving mechanism for interoperability testing, and uses Case-Based Reasoning (CBR) for debugging interoperability test cases.

The dissertation makes three primary contributions: (1) Definition of a new modeling language using CSP and Object-Oriented Programming. This language is simple, declarative, and transparent. It provides a tool for testers to implement models of interoperability test cases. The dissertation introduces the notions of metavariables, metavalues and optional metavariables to improve the modeling language capabilities. It proposes modeling of test cases from test suite specifications that are usually used in interoperability testing performed manually by testers. Test suite specifications are written by organizations or individuals and break down the testing into modules of test cases that make diagnosis of problems more meaningful to testers. (2) Diagnosis of interoperability problems using search supplemented by consistency inference methods in a CSP context to support explanations of the problem solving behavior. These methods are adapted to the OO-based CSP context. Testers can then generate reports for individual test cases and for test groups from a test suite specification. (3) Detection and debugging of incompleteness and incorrectness in CSP models of interoperability test cases. This is done through the integration of two modes of reasoning, namely CBR and CSP. CBR manages cases that store information about updating models as well as cases that are related to interoperability problems where diagnosis fails to generate a useful explanation. For the latter cases, CBR recalls previous similar useful explanations.