A Quantitative Analysis Method of Model Reliability

Requirement analysis is an essential step in software development. Its accuracy and reliability often determine the viability of entire process of development. But with no exceptions, all devices, resources, and transmissions possess imperfection in operational reliability. Therefore, deviations are inevitable in actual operation even for models that theoretically meet customers’ needs flawlessly. It is thus imperative to assess the reliability of the systematic models, so that customers can choose appropriately based on their own needs. In this paper, we propose a Markov Decision Process (MDP) and PRISM-based approach to analyze the model reliability quantitatively. It has been shown experimentally that the proposed method can assist one in decision making and early assessment of systematic model’s reliability.