Application Study on Defect Prediction of Radar System Software Based on System-theoretic Accident Modeling Process

Software is playing an important role in radar products. Software quality has become key factor of radar quality and efficiency. Predicting the trend of radar software defects is of great importance for radar software quality control and security. In order to show the mechanism of radar software defects, a radar software defect prediction model based on system-theoretic accident modeling process (STAMP) is proposed. Firstly, based on STAMP and historical defect data, a radar software control process model is built. Then, a Bayesian network learning model is constructed, and a training process is conducted on historical defect data to get radar software defect prediction rules. Finally, the rules are applied on real radar software testing projects to predict possible defects. The prediction results are analyzed to verify the effectiveness and applicability of the proposed approach.