Self-adaptation for cyber-physical systems: a systematic literature review

Context: Cyber-physical systems (CPS) seamlessly integrate computational and physical components. Adaptability, realized through feedback loops, is a key requirement to deal with uncertain operating conditions in CPS.

Objective: We aim at assessing state-of-art approaches to handle self-adaptation in CPS at the architectural level.

Method: We conducted a systematic literature review by searching four major scientific data bases, resulting in 1103 candidate studies and eventually retaining 42 primary studies included for data collection after applying inclusion and exclusion criteria.

Results: The primary concerns of adaptation in CPS are performance, flexibility, and reliability. 64% of the studies apply adaptation at the application layer and 24% at the middleware layer. MAPE (Monitor-Analyze-Plan-Execute) is the dominant adaptation mechanism (60%), followed by agents and self-organization (both 29%). Remarkably, 36% of the studies combine different mechanisms to realize adaptation; 17% combine MAPE with agents. The dominating application domain is energy (24%).

Conclusions: Our findings show that adaptation in CPS is a cross-layer concern, where solutions combine different adaptation mechanisms within and across layers. This raises challenges for future research both in the field of CPS and self-adaptation, including: how to map concerns to layers and adaptation mechanisms, how to coordinate adaptation mechanisms within and across layers, and how to ensure system-wide consistency of adaptation.