High-performance complex event processing to detect anomalies in streaming RDF data

Abstract

A lot of sensors nowadays are embedded in smart factories which generate massive real-time data about the functional conditions of the manufacturing equipments. Complex Event Processing(CEP) systems are involved to analyze continuous behavior of these machines, detect undesired patterns and give alerts in case of anomalies. In this thesis, we introduce an architectural design and concrete implementation of high-performance system which is able to solve this problem raised by DEBS Grand Challenge 2017. The thesis goes through the details of analyzing RDF streaming events to detect potential anomalies using Markov Model technique. In addition, we conducted experiments that showed promising results regarding low-latency anomaly detection and an ability to scale up and out the system.