CEP for Analysis and Pattern Recognition in Event Streams

The recognition of relevant event patterns is carried out by Event Processing Agents (EPA), which perform three tasks:

• Filtering relevant events from a large event stream
• Aggregating events into complex events based on functional context
• Monitoring the absence of particular expected events

Loosely Coupled

EPAs are loosely coupled, and can be easily rearranged or expanded in relation to  context within a component-oriented system, such as by use of publish-subscribe communication. 

Main Components of a distributed system architecture for event processing and decisioning

BRM for Rules-Based Decisioning

The task of the BRM system is to make decisions based on the collected and aggregated information. It processes the appropriate rules for the specific complex event.

Stateful Execution

Depending on the specific application, it may be necessary to keep track of the current state of rule processing in order to guarantee full and complete execution of business rules. 

Direct Interaction

The result of rules processing – the decision – is implemented in the target application. For example, the result can lead to a new event that is directed into the event stream. So data mutually interact between the CEP engine and the BRM system. 

Automation of Intelligent Decisions Based on High Event Volumes

With the CEP & BRM based system, companies are able to transparently automate intelligent, complex decisions despite high event volumes.

Optimal Combination

A component-oriented system architecture allows both parts to perfectly combine their specific individual capabilities and responsibilities:

• The CEP engine is responsible for analyzing large quantities of events and recognizing patterns in real time.
• The business rules system applies rules based on particular events in order to make decisions and, if necessary, to apply them without delay.

In a similar way, IT and business departments can add their specialized expertise:

• IT experts are responsible for all tasks required for event analysis.
• Defining the rules for decisions to be implemented is a task for the respective business experts (e.g. rating analysts).
• IT and business experts cooperate to define the event pattern to be recognized.

Easy System Expansion  

The Visual Rules business rules system makes it easy to expand and adapt even complex business logic for decisioning. It is also used for integrated rule management.

In the component-based system, Event Processing Agents can also be easily expanded and, if necessary, rebundled. To this end, the EPAs must be structured accordingly and arranged into an Event Processing Network (EPN).

Efficient Implementation of Business Logic  

It’s the business experts who know decisioning logic in event-based systems. It is the core task of Visual Rules to support these experts to intuitively define the business logic, to test it, update it, and get it into production.

How do companies deploy the combination of CEP and BRM today? And what benefits do they derive from them? Here are two examples:

Real-Time Securities Trading

One typical application is in real-time securities trading. Share prices change, companies release financial reports, central banks announce new interest rates, political events occur, and even the weather can affect market activity. Traders must quickly pick relevant information from a flood of data, recognize connections, and make the right decisions. An application that consistently and continuously condenses information and automatically makes decisions on the basis of defined rules greatly eases traders’ day-to-day challenges.

Traffic Control

CEP and BRM are also combined and employed in traffic control and management. The goal is early recognition of causal factors for traffic problems (e.g. increased traffic due to large public events) followed by a quick response (such as by extending the green phase of a given traffic light).

Other applications are:

• Recognizing and preventing the formation of Flash Crowds
• Decentralized network management using the Smart Grid
• Early recognition and containment of incipient epidemics (see Google Flu Trends), evacuation before predicted natural disasters (see tsunami warning)
• Runtime optimization of plants, motors, and machines using wireless sensor networks 
• Product tracking and tracing, and logistics management using RFID