A Novel Network Resilience Predicting Method Based on Machine Learning

DOI

10.2991/cecs-18.2018.35How to use a DOI?

Keywords

Network Resilience, Fault Tolerance, Machine Learning, ELM

Abstract

Network resilience is an important indicator for measuring the pros and cons of fault-tolerant strategies. For the resilience prediction of the network design stage, the traditional prediction methods can only be modeled and forecasted based on fault tolerance strategies based on a single network, and the impact of fault tolerant strategies deploying on multi-layer of network cannot be considered. For solving this problem, this paper proposed a network resilience prediction method based on network resilience theory and machine learning. First, a quantitative model of network fault tolerance strategy is established, and based on this model, the modeling method of the protocol stack is improved on the NS3 platform for simulation implement. A large number of structured data that can characterize network resilience are generated through simulation experiments. Then, a network resilience prediction model of single hidden layer feed-forward neural network (SLFN) is established based on the Extreme Learning Machine (ELM) theory. This model can better solve the overfit problem of traditional ELM. Finally, the prediction model was verified according to comparing with the traditional model-driven resilience prediction method. This method not only has a higher prediction accuracy, but also solves the problem that the traditional method cannot model the coupling relationship between fault-tolerant strategies, and thus cannot be based on the hybrid fault-tolerant strategy for resilience prediction. The case verification shows that this method can effectively predict the network resilience considering fault-tolerant strategies. Through cross-validation, the accuracy rate is more than 96%, which helps to consider the development of fault-tolerant design of hybrid fault-tolerance strategies.

Copyright

© 2018, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Weiqiang Wu
AU  - Qilin Hou
AU  - Lili Zhong
AU  - Hongbo Yan
AU  - Yiqiang Wan
AU  - Jianglei Wang
AU  - Jun Li
AU  - Feifei He
PY  - 2018/07
DA  - 2018/07
TI  - A Novel Network Resilience Predicting Method Based on Machine Learning
BT  - Proceedings of the 2018 International Symposium on Communication Engineering & Computer Science (CECS 2018)
PB  - Atlantis Press
SP  - 193
EP  - 203
SN  - 2352-538X
UR  - https://doi.org/10.2991/cecs-18.2018.35
DO  - 10.2991/cecs-18.2018.35
ID  - Wu2018/07
ER  -