SD-CPS: software-defined cyber-physical systems. Taming the challenges of CPS with workflows at the edge
(2019) Cluster Computing — Vol. 22, n° 3, p. 661-677 (2019)
Files
Kathiravelu2019-SD-CPS.pdf
Open Access - Adobe PDF
- 1.11 MB
Details
- Authors
Kathiravelu, Pradeeban 
- Abstract
- A cyber-physical system (CPS) is a smart mechanical environment, developed by an amalgamation of computation, networking, and physical dimensions. Each CPS consists of a network of devices, often limited in computing, storage, or bandwidth resources. Moreover, the frequent small-scale communications between the various counterparts of CPS require data and computation of CPS to be deployed close to each other, with the ability to support micro-executions. Due to these operational requirements, CPS faces several inherent challenges, uncommon to a traditional computational environment. In this paper, we describe software-defined cyber-physical systems (SD-CPS), a CPS framework built by extending and adapting the design principles of software-defined networking (SDN) into CPS. We realize the support for CPS operation as a workflow of microservices, possibly in continuous or cyclic execution. SD-CPS coordinates each CPS execution step, performed by a microservice, through an extended SDN controller architecture. By creating, placing, deploying, migrating, and managing the computation processes of CPS as service workflows at the edge, SD-CPS orchestrates the entire lifecycle of the CPS effectively and efficiently. SD-CPS thus addresses the general challenges of CPS, concerning modeling, development, performance, management, communication and coordination, scalability, and fault-tolerance, through its software-defined approach. Our evaluations highlight the efficiency of the SD-CPS framework and the scalability of its SDN controller to manage the complex CPS environments.
- Affiliations
Citations
Kathiravelu, P., Van Roy, P., & Veiga, L. (2019). SD-CPS: software-defined cyber-physical systems. Taming the challenges of CPS with workflows at the edge. Cluster Computing, 22(3), 661-677. https://doi.org/10.1007/s10586-018-2874-8 (Original work published 2019)