Using sensorranks for in-network detection of faulty readings in wireless sensor networks

Abstract

In this paper, the problem of determining faulty readings in a wireless sensor network without compromising detection of important events is studied. By exploring correlations between readings of sensors, a correlation network is built based on similarity between readings of two sensors. By exploring Markov Chain in the network, a mechanism for rating sensors in terms of the correlation, called SensorRank, is developed. In light of SensorRank, an efficient in-network voting algorithm, called TrustVoting, is proposed to determine faulty sensor readings. Performance studies are conducted via simulation. Experimental results show that the proposed algorithm outperforms majority voting and distance weighted voting, two state-of-the-art approaches for in-network faulty reading detection.

Outlier-Aware Data Aggregation in Sensor Networks

“…However, the robust aggregate obtained by a minimum support of 3 (depicted with the blue line) is significantly more accurate and manages to eliminate the spurious readings and the readings of nodes that fail-dirty in all but a few cases. We also examined an alternative technique were we perform the witness test by using the extended Jaccard coefficient [11]. Because the extended Jaccard coefficient is sensitive to the relative difference in the magnitude of the values, in Figure 4 we notice that it performs significantly better, as the readings of nodes that fail-dirty and have reached a large value cannot witness those of functional nodes.…”

Trust Models in Wireless Sensor Networks: A Survey

“…The model does not use second-hand information, and how to refresh the reputation value is an issue. Xiao et al, in [59] developed a mechanism called SensorRank for rating sensors in terms of correlation by exploring Markov Chains in the network. A network voting algorithm called TrustVoting was also proposed to determine faulty sensor readings.…”

Another Outlier Bites the Dust: Computing Meaningful Aggregates in Sensor Networks

“…Another localized voting scheme is proposed in [23] that ranks sensor readings according to their validity. Ranking explores a correlation network that requires several epochs to be finalized and is based on readings from a starting epoch.…”

“…7 we show the resulting reported aggregate for this very challenging data set. In this experiment we examine an alternative technique for computing similarity, namely the extended Jaccard coefficient [23]. We note that this change requires only the modification of the canWitness() function in our framework.…”

“…Using a single, monolithic collection tree constructed in advance, as in [15], [16], [23], does not take into account the existing readings and can lead to suboptimal decisions when computing and communicating outliers in the network. We overhaul the aggregation and outlier detection processes and periodically determine proper routing paths, based on simple statistics collected during query processing.…”