An Improved Outlier Detection Scheme In Wireless Sensor Networks
Kanishka Garg, Tarun Kumar
SNs, BS, CH, Non- CH
In the field of Wireless Sensor Networks (WSNs), the measurements that significantly deviate from the normal pattern or values of sensed data are considered as outliers. The possible sources of outliers can be noise and errors, events, and malicious attacks on the network. Traditional outlier detection techniques are not directly applicable to WSNs due to the nature of sensor data and specific requirements and limitations of the WSN.
In this Dissertation, the problem of determining faulty readings in a WSN will be studied. A correlation network will be there which will be based on similarity between readings of two sensors. Rank of the each sensor on the basis of correlation will be calculated. In light of this SensorRank, an efficient in-network voting algorithm will be used to determine faulty sensor readings.
To make outlier detection more energy efficient, we will use clustering in which CH collect the outlier data from its cluster and send it to the Base Station. Cluster and cluster head will be more important part and CH will be elected base on fuzzy rules considering different membership functions. Performance studies are conducted via simulation.