Distributed Channel Access Control System over the Controller Area Network Bus

Abstract

This work studies the problem in which multiple subsystems equipped with smart sensors sharing a capacity-limited communication network, try to perform their control tasks. The networked control systems follow time-division multiplexing, in which at most one subsystem is allowed to transmit data over the network at each time step. We propose a hybrid priority assignment scheme to use the limited communication resources efficiently with respect to a quadratic cost which reflects the control performance. More specifically, each subsystem is assigned with a dynamic priority at each time step, and a static priority, which ensures that that the quadratic cost is minimized and that data transmission is collision-free, respectively. These priorities are then utilized for contention resolution by binary count down methods typically used in controller area network (CAN) protocols. To validate the efficacy of our scheme and demonstrate its compatibility with the mature hardware technology of CAN, we consider its hardware implementation and discuss how the practical issues such as synchronization can be addressed. We propose one broadcast and synchronize method, in which one dedicated subsystem CANoe broadcasts one periodic trigger message to synchronize the subsystems, and a real-time operating system is introduced to guarantee the time requirements. Finally, we demonstrate how the channel access decisions are determined in the experimental setup and evaluate its performance in large scale networks via simulation.