Optimal Priority Assignment for Scheduling Mixed CAN and CAN-FD Frames

Controller Area Network with Flexible Data-rate (CAN-FD) has been drawing considerable attention as the most promising substitute of Controller Area Network (CAN), thanks to its higher bandwidth, larger payload size, and physical-layer compatibility with CAN. In particular, the physical-layer compatibility allows legacy CAN-based Electronic Control Units (ECUs) to share the same communication bus with CAN-FD-based ECUs. However, CAN-based ECUs always treat a CAN-FD frame as an erroneous frame due to the difference in frame format. This, in turn, makes CAN-FD-based ECUs unable to communicate with each other via CAN-FD frames. A straightforward solution to this problem is to utilize the silent mode of the current CAN controller, in which a CAN node does not transmit any frame including error frames, but can receive CAN frames. However, a non-negligible time overhead is incurred by each mode transition, thus degrading the schedulability of mixed CAN and CAN-FD frame sets significantly. We propose a new algorithm, called Priority Assignment with Mode Transition (PAMT), that minimizes the required number of mode transitions by clustering frame instances based on their frame type. Our evaluation results show that PAMT can schedule about 17% more mixed CAN and CAN-FD frame sets than existing optimal priority assignment algorithms for CAN.

I. Introduction

Controller Area Network (CAN) [28] is the de facto standard of current in-vehicle networks because of its robustness, wide deployment, low resource requirement, and real-time support. However, the advent of new functions to improve the driver’s safety and comfort will make CAN unlikely to meet in-vehicle communication requirements in the near future [33]. To overcome the shortcomings of CAN, a new protocol, Controller Area Network with Flexible Data-rate (CAN-FD), has recently been proposed [29]. CAN-FD not only overcomes the drawbacks of CAN but also allows use of existing/legacy CAN infrastructures — e.g., Electronic Control Units (ECUs) developed with CAN controllers and transceivers, CAN wires, etc. — thanks to its physical-layer compatibility with CAN. As a result, CAN-FD has been attracting significant attention as the most promising substitute of CAN [34].