Generating Effective Test Suites for Reactive Systems using Specification Mining

Abstract

Failures in reactive embedded systems are often unacceptable. Effective testing of embedded systems to detect such unacceptable failures is a difficult task.We present an automated black box test suite generation technique for embedded systems. The technique is based on dynamic mining of specifications,in the form of a finite state machine (FSM), from initial runs. The set of test cases thus produced may contain several redundant test cases. Many of the redundant test cases are then eliminated by an aggressive greedy test suite reduction algorithm to yield the final test suite. The tests generated by our technique were evaluated for their effectiveness on five case studies from the embedded domain. The evaluation of the results indicate that a test suite generated by our technique is promising in terms of effectiveness and scales easily. Further, the test suite reduction algorithm may sometimes remove non-redundant test cases too. Therefore, in our experimentation, we have also evaluated the change in the effectiveness of test suites due to this reduction.In this thesis, we describe the test suite generation and reduction technique in detail and present the results of the case studies.

Date of Award	14 Jun 2014
Original language	English
Supervisor	Padmanabhan Krishnan (Supervisor) & Marcus Randall (Supervisor)