A Machine Learning Approach for Quantifying the Design Error Propagation in Safety Critical Software System

Abstract

In general, the safety critical systems are zero error tolerance systems, designed with the high precision approach and with maximum perfection. Hence the authors attempted to create a flawless design by analyzing the various components including error occurrence at low-level design of making software. In view of this, the migration of design defects is quantified from origin to multiple states through hidden markov model approach. Here the probabilistic natures of selected defects by observing the operation of anti-lock braking system in various scenarios are modeled. It is observed that this model supports in identifying and quantifying the behavioral properties of selected errors while interacting with subsystems. The behavior of software is determined in terms of hidden state sequence. The sensitivity and precision quotient are measured for goodness-of-fit. This approach of early analysis of software hidden design errors will enhance the precision in producing any of the safety critical systems in practice.