A Classification Problem Formulation for the Reliability Assessment of End-To-End Communication Among Cyber Devices in Cyber-Physical Systems

Telecommunication Networks (TLCNs) enable the control of Cyber-Physical Systems (CPSs). Reliable End-to-End (E2E) communication among TLCN devices is fundamental to prevent CPS failures. Simulating the complex underlying phenomena that cause E2E communication failure, such as transmission delays and package dropouts, might be computationally expensive, challenging the reliability assessment of E2E communication. To tackle this issue, we reframe the E2E communication as a classification problem to streamline its reliability assessment. We use Gradient Boosting (GB) as an example of an off-the-shelf classifier that can be tailored to the purpose of estimating the E2E communication success or failure, without relying on the long-running simulation of the information processing through the nodes of a TLCN embedded into a CPS. The CPS considered as case study is an Integrated Power and Telecommunication Network (IP&TLCN) system.