A Data-Centric Framework for Cyber-Physical-Social Systems

C yber-physical systems (CPSs) enable the physical world to merge with the virtual world by integrating computation and physical processes. A CPS facilitates tight integration between computa tion, communication, and con trol in its operation and interactions with the environment in which it’s deployed. Now, in addition to embracing cyber and physical features, interest is growing in harnessing human and social factors in CPSs. For instance, one recently proposed cyber-physicalsocial system (CPSS) embraces the cyberspace-enabled parallelism: a real system and its artificial counterparts run in parallel and interactively through cyberspace.1 Another proposal is a human-inthe-loop CPS, which infers users’ intent by measuring human cognitive activity through body and brain sensors.2 These studies take human and social features as important elements in CPSs mainly from the system automation and control perspective. However, to grasp the fuller potential of a CPSS, data-centric realization is necessary. Consider an urban big data system that’s a typical CPSS system. It integrates and derives information from cyberspace (for example, data from governments and institutions), physical space (surveillance cameras, smartphones, and so on), and social space (such as mobile crowd sensing3 and mobile social networks). This kind of data-driven CPSS (D-CPSS), however, has yet to be fully studied and investigated. In this article, we study CPSS from a data-centric perspective, characterize its core features, and present a layered framework (architecture). In particular, we focus on multispace collaborative sensing and cross-space data fusion, and identify areas that require further research and development.