Handling New and Changing Requirements with Guarantees in Self-Adaptive Systems Using SimCA

摘要

Self-adaptation provides a principled way to deal with change during operation. As more systems with strict goals require self-adaptation, the need for guarantees in self-adaptive systems is becoming a high-priority concern. Designing adaptive software using principles from control theory has been identified as one of the approaches to provide guarantees. However, current solutions can only handle pre-specified requirements either in the form of setpoint values (S-reqs) or values to be optimized (O-reqs). This paper presents SimCA* that makes two contributions to control-based self-adaptation: (a) it allows the user to specify a third type of requirement that keeps a value above/below a threshold (T-reqs), and (b) it can deal with requirement sets that change at runtime (i.e., requirements can be adjusted, activated, and deactivated on the fly). SimCA* offers robustness to disturbances and provides adaptation guarantees. We evaluate SimCA* for two systems with strict goals from different domains: an underwater vehicle system used for oceanic surveillance, and a tele-assistance system for health care support. The test results demonstrate that SimCA* can deal with the three types of requirements (STO-reqs) operating under various types of dynamics and the set of requirements can be changed on the fly.