Towards machine learning of time task scheduling in cyber-physical systems

Cyber-Physical Systems (CPS) are complex systems with tight composition of computation, communications and control technologies. The modeling and analysis that act an important part of the model-driven system of systems (SoS) development play also a great significant role in CPS. Scheduling algorithms are an important part of CPS model design. With the increasing number of system service tasks, CPS needs to complete computing, control, and communication in a limited amount of time. The newly added physical devices and newly generated system services will impose higher time requirements on task scheduling calculations. To adapt to such conditions, CPS application system often adopt machine learning techniques to eliminate the need for unnecessary redesign. In this paper, we present machine learning method for time task scheduling based on the Simple and Proximate Time Model (SPTimo) framework for to solve the problem of efficient scheduling when the system scale is expanded