What are the interesting problems and questions in event processing research?
Participants:
- Alan Colman
- Gabriel Hermosillo
- Amirhosein Taherkordi
- Young Yoon
Distributed Event Dissemination and Processing - A Research Agenda
Slides from the presentation.
We consider applications to cut across all these areas and cluster related research problems which can be attacked within a single project.
Understand and partition the problem space
- What is the appropriate terminology for describing and understanding the area?
- What are the relevant applications and use-cases?
- What are the relevant research and engineering problems and how do they interrelate?
- What are the design dimensions for event processing?
- What are the well-defined and self-contained abstractions, components, services, etc that can be used by others as building blocks?
- What are the appropriate architectures for organizing these building blocks?
- How can (ultra large scale) systems-of-systems be composed from these building blocks?
Define a unified formal event processing language
- Define the SQL / relational algebra for event processing.
Identify the appropriate consistency and completeness models for event dissemination and understand their consequences and usages
- What are the possible consistency and completeness models for event dissemination?
- Which of these are useful in practice?
- What are the consequences for implementation, performance, reliability, etc for each of the models?
- How can these models be used in practice, what programming support is required?
Explore and exploit the spectrum from moving the data to the computation to moving the computation to the data
- What computations can be done within the event processing system? How?
- What are the criteria and trade-offs for determining where the computation is best made?
- How are logical specifications best deployed on a physical infrastructure?
Understand how to control and adapt open and dynamic event processing systems
- Understand the available options for controlling open and dynamic event processing systems
- Develop a policy framework for describing QoS/SLA within the system and between the system and its users
Develop autonomic event processing systems with self-* properties (self-configuring, self-optimizing, self-adapting, self-understanding, self-healing, self-maintaining, etc)
Integrate event processing in cyber-physical systems (external responses)
- How can event processing systems interact with external physical system?
"Cyber-Physical Systems (CPS): Orchestrating networked computational resources with physical systems. The Vision: Reliable and Evolvable Networked Time-Sensitive Computational Systems, Integrated with Physical Processes." Cyber-Physical Systems, Brooks and Lee, 2009.
Knowledge-based (semantic) event processing
- How can event processing be integrated with semantic web technologies? Is it a good idea?
- How to take context into consideration in event processing systems?
Uncertain event processing including hypothetical event detection
- What is an appropriate uncertainty formalism for representing uncertain events?
- What are the consequences for composite event detection when events are uncertain?
- How can approximate queries be incorporated with event processing?
- How can belief revision be incorporated in event processing (the system believed that A happened, but later finds out that A did not happen, how to update the system with that knowledge (what needs to be retracted?))
Automatic discovery and adaptation of event models, event mining
- Can existing event pattern languages be reused to define meta event patterns (if the event is detected, it means that a new event pattern has been found)?
- How to adapt your current set of event models to the actual input of your system? (Related to Pascal Felbers talk about estimating the number of false positives in approximate queries.)
- What type of event models can be discovered?
- Is event discovery orthogonal to event processing?
Support security, trust and privacy in event processing
- How to provide safe and secure event dissemination infrastructure?
- What privacy models are appropriate and acceptable to the users of the system?
- How can sociological and technical issues related to privacy be managed?
- What are the trade-offs between security and performance?
- How can providence be incorporated in event models?
- How can you process events without brokers knowing about the content?
Guiding Questions
Here are some questions to get you started, but feel free to define your own.
- What are exciting research challenges that the community should look at?
- What research questions have been neglected in the past?
- Unification
- What important synergies with other areas need to be explored and developed?
- One-size-fits-all
- Federation, integration, and interoperation of systems designed for constrained environments (e.g. sensor networks) with systems designed without these constraints
Session-1