Creation of a Deep Learning-Based Model to Predict Reliability of a Distributed Network Owning Stochastic Capabilities  
Author Ding-Hsiang Huang


Co-Author(s) Ping-Chen Chang


Abstract The study presents the concept of a distributed network with stochastic capacity (DNSC). Besides, the cloud servers and edge computing are considered in the distributed network. distributed network reliability (DNR) is defined the probability of satisfying demands for an DNSC so as to is serve as a performance indicator. Existing algorithms generated minimal requirement states in every component to calculate DNR; but became inadequate for complex DNSC in the real-world environment with complicated transmission demands. To address this challenge, an artificial neural network (ANN) architecture is proposed to create a prediction model for DNSC reliability, enabling quick learning of DNSC capabilities with varying data. DNSC information is transformed into an appropriate format for training the DNR prediction model, and related information for ANN setting is defined using reasonable approaches with suitable hyperparameters. The model can be used for the future management of the distributed network.


Keywords Distributed network, cloud server, edge server, artificial neural network, prediction model
    Article #:  RQD28-216

Proceedings of 28th ISSAT International Conference on Reliability & Quality in Design
August 3-5, 2023