Dependability of complex software systems with component upgrading

Some very large and complex systems, such as telecommunication systems, must (and typically do) exhibit exceptional dependability. These systems are seldom totally replaced with a new system because of the increased likelihood of a lapse in service. Rather, systems are upgraded incrementally while operational, albeit that this often involves large-scale software changes. It is especially important then to ensure that new or replacement components are ready for online installation before they are incorporated into an operational system. It is often costly and time-consuming to determine the readiness of new components for installation. Even then, the result may be unpredictable. Hence, we have developed effective and economical methods for software component verification that ensure and increase the overall system dependability. We tested our technologies using a telecommunication application, an Internet call-agent. Our experimental results show that our dynamic design analysis approach reduces computational costs and detects more errors than conventional approaches. The more frequently that changes are made, the greater the savings in the time required for model analysis and property prediction.