Fault-tolerant interval estimation fusion by Dempster-Shafer theory

Fault tolerance is an important issue in network design because sensor networks must work in a dynamic, uncertain situation. In this paper, using Dempster-Shafer theory of evidence, we propose several new fault-tolerant interval integration functions, which give interval estimate fusion outputs and the corresponding belief levels, depending on prior information and practical requirements. Not only do these functions have a smaller output interval than that given by Marzullo function, but they also satisfy the local Lipschitz condition, which makes our algorithm locally stable.