The objective of this paper is to present a novelmethod for achievingmaximumreliability in fault-tolerant optimal network design when networks have variable size. Reliability calculation is a most important and critical component when fault-tolerant optimal network design is required. A network must be supplied with certain parameters that guarantee proper functionality and maintainability in worse-case situations. Many alternative methods for measuring reliability have been stated in the literature for optimal network design. Most of these methods, mentioned in the literature for evaluating reliability, may be analytical and simulation-based. These methods provide significant ways for computing reliability when a network has a limited size. Significant computational effort is also required for growing variable-sized networks. A novel neural network method is therefore presented to achieve significant high reliability in fault-tolerant optimal network design in highly growing variable networks. This paper compares simulation-based analytical methods with improved learning rate gradient descent-based neural network methods. Results show that improved optimal network design with maximum reliability is achievable by a novel neural network at a manageable computational cost.
Fault tolerant packet-switched network design and its sensitivity
