A Distributed Anomaly Detection Scheme Based on Correlation Awareness in WSN

Zhongmin Wang,Rui Gao,Cong Gao,Yanping Chen,Fengwei Wang,Gao, Rui
DOI: https://doi.org/10.1007/s11277-024-10930-w
IF: 2.017
2024-03-21
Wireless Personal Communications
Abstract:Wireless sensor devices are affected by internal constraints and the external environment, generating abnormal data. Currently, many anomaly detection schemes ignore the correlation between screening nodes, wasting resources due to excessive communication. Therefore, this paper proposes a distributed anomaly detection scheme based on adaptive grouping using the correlation between nodes in wireless sensor networks. Limiting the scope of collaboration between nodes can reduce the waste of resources due to excessive communication. Since the computing resources of sensor nodes are limited, an edge-cloud framework is established. The scheme uses Spatio-temporal correlation and graph theory for wireless sensor networks to determine node groups with solid correlations on the cloud server. Based on the grouping results, anomaly detection is implemented locally. A Bayesian network model is constructed at the node within the group, and outlier detection is realized by inference on nodes. A correlation consistency evaluation method is proposed to improve anomaly detection accuracy to check the data consistency on the cluster head. The proposed scheme is verified by a generated data set and the real data of Intel Berkeley Research Lab. The effectiveness of the proposed method is verified by comparing it with three existing algorithms. Experimental results show that the method improves detection accuracy and reduces false detection.
telecommunications
What problem does this paper attempt to address?