Kramarenko K.E.   Moldovanova O.V.  

Neural Network algorithm of Fault Diagnosis in distributed computer systems

Reporter: Kramarenko K.E.


К.Е. Кramarenko1, О.V. Moldovanova1
1Siberian State University of Telecommunications and Information Sciences, Novosibirsk,

In the work neural network algorithm applicability for self-diagnosis of distributed computer systems is researched. A distributed computer system consisting of N elementary machines (referred below as nodes), connected with communication links is considered. Each node can be in a fault-free or faulty state. A task is assigned to each pair of nodes. A comparison of obtained results of its solving gives an information about nodes’ states. If results coincide, both nodes are fault-free, otherwise one or both nodes are faulty. A set of all comparison results is called a syndrome. To obtain states of all nodes in the system a syndrome must be decoded. In the work artificial neural networks are used for this. A neural network receives the diagnostic syndrome as input, and outputs a set of fault-free and faulty nodes. The diagnostic algorithm was implemented and researched in the interactive simulation environment MATLAB.      

The work was supported by the Presidential Council for Grants (MD-2620.2014.9) and the Russian Foundation for Basic Research (15-07-00048-а, 15-07-00653-а).

1. Хорошевский В.Г. Архитектура вычислительных систем: Учеб. пособие. – 2-е изд., перераб. и доп. – М.: Изд-во МГТУ им. Н.Э. Баумана, 2008. – 520 с.
2. Duarte Jr., E. P., Ziwich, R. P., Albini, L. C. P. A survey of comparison-based system-level diagnosis // ACM Comput. Surv. 43, 3, Article 22, 2011. – 56 p.
3. Elhadef M., Nayak A. Comparison-based system-level fault diagnosis: a neural network approach // IEEE Transactions on parallel and distributed systems, Vol. 23, No. 6, June 2012. – P. 1047-1059.
4. Осовский С. Нейронные сети для обработки информации / Пер. с польского И.Д. Рудинского. – М.: Финансы и статистика, 2004. – 344 с.

To reports list