An edge-cloud collaboration architecture for pattern anomaly detection of time series in wireless sensor networks

Abstract With large deployment of wireless sensor networks, anomaly detection for data is becoming increasingly important in various fields. As a vital form data, time series has three main types anomaly: point anomaly, pattern and sequence anomaly. In production environments, the analysis most rewarding one. However, traditional processing model cloud computing crippled front amount widely distributed data. This paper presents an edge-cloud collaboration architecture series. A task migration algorithm developed to alleviate problem backlogged tasks at edge node. Besides, related long-term correlation short-term are allocated node, respectively. multi-dimensional feature representation scheme devised conduct efficient dimension reduction. Two key components trend identification extraction elaborated. Based on result representation, performed with improved kernel density estimation method. Finally, extensive experiments conducted synthetic sets real-world sets.