DATA-DRIVEN FAULTY NODE DETECTION SCHEME FOR WIRELESS SENSOR NETWORKS

DATA-DRIVEN FAULTY NODE DETECTION SCHEME FOR WIRELESS SENSOR NETWORKS

 

ABSTRACT

In this paper, a faulty node detection scheme witha hybrid algorithm using a Markov chain model that performscollective monitoring of wireless sensor networks is proposed.Mostly wireless sensor networks are large-scale systems, heavilynoised, and the system workload is unfairly distributed amongthe master node and slave nodes. Hence, the master node may noteasily detect a faulty slave node. In this paper, a more accuratefaulty node detection scheme using a Markov chain model isinvestigated. Each slave node’s condition can be divided into threestates by probability calculation: Good-, Warning-, and Bad-state.Using this information, the master node can predicts the areain which an error frequently occurs. Simulation results showthat the proposed method can improve the reliability of faultynode detection and the miss detection rate for a Wireless SensorNetworks.

PROPOSED SYSTEM:

The main contribution of the scheme is enhanced bythe reliable determination of faulty node detection in thedecoding process. The simulation results clearly show that theperformance of our proposed scheme is better than that that ofprevious methods in terms of the faulty node detection timeand miss detection miss-detection rate.

EXISTING SYSTEM:

Agi et al. proposed a faulty node detection scheme thatperforms the collective monitoring of a distributed system byusing an insertion paradigm of the BCH code. This schemeutilizes the distributed algorithm with a single BCH codeinsertion. This study, the decision phase is the most crucialone. The master node analyzes the behavior of each of theBCH fragments continuously at all times. In this study, for thereal-time observation of the slave node condition a Markovchain was used with BCH code.We propose a new faulty decision method by using aMarkov chain process, which is a probabilistic method. Eachslave node uses the Markov chain process to examine thenode’s state in real time. Then, the Markov chain information,containing the Good-, Warning-, and Bad-states of the slavenodes, is sent to the master node.Basically, each slave node condition can be divided intothree states; Good-, Warning-, and Bad-state. Initially, allnodes are assigned to Good-state. During observation, eachnode state may be changed dynamically, depending on theBCH sequence that comes from each node. The change usuallyoccurs during certain intervals between two errors.The basic idea is that if the interval between two errors isless than a threshold, then the node state should be recalculated.If the error sequence from a certain node is not foundattn, and the error occurred at tn+1, where tis lessthan the threshold time tiihn+1, then the node state based on theMarkov chain will be changed. A similar approach is usedwhen a Warning-state should be changed to a Bad-state, inwhichcase, the node is confirmed as a faulty node.

CONCLUSION

The proposed scheme improved the performance in termsof the detection reliability and miss detection rate by using aMarkov chain algorithm. The results show that the detectionability is good regardless of channel noise levels and Inlarge-scale system is detected in real time, which means thatsurvivability and reliability are guaranted. It is also shown thatthis scheme are more robust compared to previous works.

REFERENCES

[1] S. Keerthana and J. C. M. J. Pamila, “A survey on fault node detectionand recovery mechanisms in wireless sensor network,” in 2015 InternationalConference on Advanced Computing and Communication Systems,Jan 2015, pp. 1–5.

[2] M. Winkler, K. Tuchs, K. Hughes, and G. Barclay, “”Theoretical andpractical aspects of military wireless sensor networks”,” Journal ofTelecommunications and Information Technology, pp 37-45, 2008.

[3] B. Jie, L. Hou, and M. Yongguang, “Machine fault diagnosis usingindustrial wireless sensor networks and support vector machine,” in2015 12th IEEE International Conference on Electronic MeasurementInstruments (ICEMI), vol. 01, July 2015, pp. 153–158.

[4] M. Ding, D. Chen, K. Xing, and X. Cheng, “Localized fault-tolerantevent boundary detection in sensor networks,” in Proceedings IEEE 24thAnnual Joint Conference of the IEEE Computer and CommunicationsSocieties., vol. 2, March 2005, pp. 902–913 vol. 2.

[5] P. Jiang, “”A new method for node fault detection in wireless sensornetworks”,” Sensors 9, no. 2, pp 1282-1294. 2009.

[6] G. Jian-Liang, X. Yong-Jun, and L. Xiao-Wei, “”Weighted-median baseddistributed fault detection for wireless sensor networks”,” Journal ofSoftware, pp 1208-1217, 2007.

[7] A. Prasetiadi and D. S. Kim, “Faulty node detection in distributedsystems using bch code,” IEEE Communications Letters, vol. 17, no. 3,pp. 620–623, March 2013.