Edge Computing-Based Anomaly Detection for Multi-Source Monitoring in Industrial Wireless Sensor Networks

Abstract

Industrial wireless sensor network (IWSN) is a large-scale system commonly vulnerable to various types of failures due to some anomalies. Thus, the detection of anomalies in IWSN is a major challenge for tasks such as fault diagnosis and application monitoring. Previous solutions are primarily concerned with single source or cloud network processing, with limited consideration for the connection of time and space. This paper introduces an edge computing-based multi-source monitoring of anomaly detection in IWSN which focusing on detection accuracy and its computation time. The simulation results indicate the reliability and applicability of the proposed scheme for IWSN.