Out of range error probability assessment for any computer program

Abstract

Software fails due to execution of software fault. A fault exists in software due to errors. There are various approaches presented in literature for locating faults and estimating number of faults in any computer program. In general, a computer program cannot be tested for its entire input space due to huge resource requirements. One critical error, which is difficult to locate and can lead to catastrophic consequences, due to unexpected software response, is setting any variable value out of its storage capacity during execution. This paper proposes an approach to evaluate probability of such errors. This approach applies a Discrete Probability Distribution method for assessing probability density functions of intermediate and output variables and then evaluating Out of Range Error probability considering the fixed permissible range for the variable. This approach is expected to become useful tool for locating Out of Range Error possibilities in computer programs and make them robust against such errors.