Research on network time reliability evaluation method based on uncertainty theory

Abstract

Addressing the shortcomings of current network time reliability assessment methods, which often focus solely on inherent uncertainty and overlook the effects of epistemic uncertainty due to insufficient fault information, this study proposes a new method based on uncertainty theory. This approach innovatively designs measurement parameters for both single node pair and multi-node pair time reliability, accommodating a broad spectrum of network reliability scenarios. The development of the extended uncertain network model is a cornerstone of this research, specifically crafted to incorporate the epistemic uncertainty attributes of nodes and links. Building on this model, we introduce algorithms to calculate time reliability for single and multi-node pairs, utilizing the most reliable path and the most reliable extended uncertain subnetwork to refine assessment precision. The validity of the proposed method is demonstrated through application to a six-node network and a major academic backbone network, evaluating two key time reliability indicators. The findings confirm the method's accuracy and efficacy, underscoring its potential to enhance network reliability assessments significantly.