John Georgas

Broad Research Interests

My broad research interests revolve around software architecture and the use of architectural models in all aspects of software engineering. Rather than being limited to design-time artifacts, architectural models can be useful in implementing, maintaining, and evolving software systems.

Building toward this pervasive vision for the use of architecture, my current interests include: self-adaptive software architectures; architectural description languages and their accompanying tool support; self-adaptive robotic control systems; domain-specific architectural styles for robotics; decentralized peering architectures; and, systems-of-systems modeling.

Students: If you're interested in research opportunities and any or all of this sounds interesting to you, just drop me an email or come by my office and visit. We'll chat and work out a research project for you!

Current Projects

PBAAM - Policy-Based Architectural Adaptation Management

The PBAAM project is focused on supporting the development of self-adaptive systems through the application of an architecture- and policy-based approach. Systems are first constructed through architecture-based means, where each behavior is encapsulated in independent and decoupled components. Then, self-adaptive behaviors are added to the system by encapsulating them in independent policies that can be modified at runtime. These systems and policies are managed at runtime by the PBAAM tool infrastructure. This approach has been successfully applied and demonstrated in the robotics domain, through the construction of Robocode and Mindstorms NXT robots based on the ArchStudio open-source architecture-based development environment from the University of California, Irvine.

In the near future, I plan on exploring the integration of the existing PBAAM toolset with other robotic platforms and development environments, such as Microsoft Robotics Studio (MSR), to develop and experiment with more robust robotic platforms.

ARCM - Architectural Runtime Configuration Management

The ARCM work is focused on providing support for the runtime maintenance of configuration information related to autonomous self-adaptive systems. For policy-based self-adaptive systems, the adaptation process is unpredictable and non-deterministic, which means that architectural states of these systems cannot be predicted at design-time. This mandates the use of runtime facilities to maintain cognitive control over such systems and provide support for advanced capabilities, such as the maintenance of reliability bounds during runtime adaptation.

In the near future, the ARCM codebase is receiving a much-needed face-lift, and I am exploring the addition of support for the preservation of reliability characteristics in collaboration with Dr. Roshanak Roshandel of Seattle University.