Long used in automated trading applications in the financial industry, complex event processing (CEP) coupled with machine learning (ML) correlates low-level events into actionable high-level events based on predefined rules. With the advent of large streaming (big) data from sources as diverse as environment sensors, RFID, and mobile accelerometers, CEP is an important component in an increasingly intelligent environment driven by the internet of things (IoT.)
What distinguishes CEP from business rule management systems (BRMS) is the ability to operate in real-time on multiple data streams and trigger actions based on event patterns. With applications in the healthcare, industrial production, military, environmental and social sensing domains, CEP and IoT form a basis for ambient intelligence (AmI) applications.
CEP and Machine Learning
While the rule patterns in CEP systems are usually defined manually, the volume and velocity of today’s data streams require the use of machine learning (ML) to derive them. Research in this field (see “References” below) has found the PART rule-based classifier algorithm (among others) to be particularly effective at this. The model creation is performed offline and an interesting area of further research is the application of ML rule-pattern induction techniques to real-time data streams.
CEP Applications
CEP applications include logistics where transportation containers are monitored for event correlations in their sensor data streams, such as accelerometer events (speed, temporal, and positional) that detect abnormal conditions (similarly applicable to mobile handsets), and assisted living applications where data from embedded wireless devices and sensors is uploaded to the cloud or processed nearer to the source in edge devices. The Ambient Assisted Living Joint Programme (AAL JP) and their partner Bdigital’s eKauri Smart Home and Assisted Living Platform (see “References” below) is one such example.
CEP Tools
CEP tools include:
- WSO2 Stream Processor (which includes their CEP engine Siddhi)
- Apache Flink
- Drools Fusion
- Esper
- WSO2 also provides their open source IoT Server.
References
Determination of Rule Patterns in Complex Event Processing Using Machine Learning Techniques.
Case Study: BDigital Delivers E-Health and Smart Home Platform Using the WSO2 Carbon Platform.