General Outline

Software architectures involving control systems have long been used to abstract away some of the complexity of managing a set of interconnected software components. However, the complexity of the control systems themselves has recently been approaching the complexity of the systems they manage, and the costs of developing and maintaining these systems has risen accordingly.

In recognition of this growing problem, a new branch of software engineering was created which has come to be known as autonomic software. Autonomic software systems are designed so that they can make modifications to their own execution, adapting to changes in situation and learning from the outcomes of those decisions. Eventually, this will allow the development of autonomic systems which are entirely self-managed and require a minimum of human interaction.

Autonomic software systems will often take a multitude of contextual information into account when making decisions. This information is gleaned from environmental sensors, and sensors fitted as part of the autonomic system itself, which is the case for the systems built into modern automobiles. The system extracts relevant sensor data from the environment, filter and distill them, and transform them into relevant situational information so that decisions can be made in context. This allows the system to adapt in response to changes internal and external to the system.

The provision of a control system which manages and regulates the behaviour of subordinate components can be thought of as the introduction of a distinct layers in the architecture. We suggest that software with autonomic behaviours can be built in at any level of the architecture. Rather than write entirely new systems, in which autonomic capabilites are programmed at every level, we suggest the establishment of autonomic control systems, that have the capability to manage systems that are themselves not inherently autonomic. This allows them to be applied to systems that are already deployed, and could bridge the divide between traditional static systems, and fully adaptive systems.

Who Is Eoin Bailey (academically speaking)?

Currently I can be found in UCD, University College Dublin, in the Computer Science and Informatics department. I have been here since May 2006 (my first week I was a student volunteer at Pervasive 2006, my fourth week I think it was, I was a volunteer for ICAC 2006 too!). This section and it's sub-sections outline some of the work I am currently involved in.

More information on me can be found here: About Eoin