Resilient Architectures Through Dynamic Management of Software Components
Abstract
The architecture of software-intensive systems is determined by their functionality and the environment they operate in. For Cyber-Physical Systems (CPS), the environment can vary in complexity and change with time. These systems operate independently, with little or no supervision, and with little maintenance. They are expected to last for long periods, that can extend to even decades. Since their operating conditions can change over time and the requirements for the systems can evolve, it is necessary to have a non-intrusive way to instantiate newer software components as the system is operating. This can extend the usability, adaptability, and longevity of such systems. The ability to replace software components as systems are operating can also be a way to improve software diversity, safety and security. Techniques that include redundancy and diversification of the software can become a reality only if a dedicated approach for software components update is available during run-time. The practicality of this can be achieved if component interfaces, state, and dynamic management are handled appropriately. This study explores the mechanisms, and the challenges of dynamic changes through the manipulation of software components while the system is in running. The focus is on safety-critical embedded systems, which may be resource-constrained and have real-time requirements. Improving security and reliability through fault-tolerant mechanisms is the primary goal of this work. The study of dynamic architecture is not solely restricted to individual systems. The principles that are presented are discussed in the scope of large scale systems and system of systems. Run-time assurance is also receiving attention as part of the life of dynamic systems. Techniques such modeling and validation are explored in the same context. The work presents some experimental results and discusses possible practical approaches for designing dynamic system of systems.
Degree
Ph.D.
Advisors
Dietz, Purdue University.
Subject Area
Computer Engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.