A methodology for assessing the correctness of control programs

Abstract

Control programs occur in a number of critical systems, such as process control systems, ballistic missile defense systems, air-traffic control systems, etc. It is important to accurately determine the correctness of a control program prior to using it, since any failures can be disastrous. Ideally it is preferable to prove that the program is correct. However, program proofs are not adequate since the specification itself can be incorrect or incomplete due to the complexity of the requirements. Hence, they have to be supplemented with systematic testing. In this paper we first analyze the role of program proofs in estimating the correctness of control programs. Then we develop a fuzzy-set-based measure of the correctness possibility of these programs. The input domain is partitioned into a set of hierarchical fuzzy equivalence classes. We estimate the computational and control flow correctness of each equivalence class. The computational correctness possibility is derived from an estimate of the complexity of the computations performed by the equivalence class. The control flow correctness possibility is based on the boundary value aspects and cohesiveness of the equivalence class. The model also incorporates the development history of the program. This is used to minimize the number of test cases.