Software Safety Based on Information Entropy and Mamdani Fuzzy System
Yao Li, Jin Guo, Lingjing Kong, Haiquan Song · 2013
Safety is highly important to safety critical software systems. To quantifiably measure the safety of software, the definition of risk is used for reference; the safety is analyzed based on probability and severity. To obtain assessment data, the expert evaluation method is used. Combining its features, information entropy theory is introduced to determine if the data is within the optimum range and to reduce the subjectivity of the experts. To get the risk level, an assessment process is presented based on Mamdani fuzzy inference system which principle is researched. The risk matrix is given as the inference rule base; the membership function is presented and the inference rules are put forward. Finally, an example of analyzing interlocking software is elaborated, which indicates the solution is effective and feasible.