RunMore · Run-time Model Projections for Software Failure Prediction

The scope of this project is to contribute to the development of conceptual foundations, engineering techniques, and computing infrastructure for the systematic development of dynamically adaptive software systems.Current Software Engineering aims at designing self-adaptive systems which are able to react and reconfigure themselves minimizing human intervention and ideally guaranteeing a lifelong requirement fulfillment. Current software engineering paradigms, systems do not anticipate events which may lead to failures, but only react accordingly to them.The RunMore project introduces the novel concept of Run-Time Model Projection for Failure Prediction. By this we mean the ability of a software system at run-time to automatically forecast potentially dangerous events by reasoning on models which represent the expected future behavior of the system (i.e., models projections) and thus work around predicted failures before their occurrence. This approach empowers self-adaptation capabilities of software systems obtaining an increased degree of dependability and availability. The project focuses on software self-adaptation in terms of performance and reliability and relies on statistical algorithms to predict the future behavior of the system and run-time model-checking to verify such behavior with respect to desired requirements.