Distributed embedded systems which are employed in many applications (e.g., automotive, avionics, military, and industrial control systems), generally consist of several processing units that are connected to each other through a communication network. Some of these applications are safety-critical and real-time, therefore, a system or network failure in these applications can result in catastrophic consequences for humans or the environment. This means that in these applications, systems should perform their tasks correctly and moreover, should meet the specified deadlines. Consequently, in a safety-critical real-time distributed embedded system, fault tolerance and network efficiency improvement should be considered, simultaneously. One of the most common communication protocols employed in distributed embedded systems is the Controller Area Network (CAN) communication protocol. Although this communication protocol has a maximum bit rate of 1Mbps and is based on broadcasting on a bus, it has some inherent limitations. In this paper, a technique is proposed which employs bus redundancy and a scheduling algorithm to improve the fault tolerance and the efficiency in CAN communication networks. To evaluate the proposed technique, a CAN-based distributed embedded system composed of four nodes (using ARM microcontrollers, i.e. STM32F4 and STM32F2) has been implemented. The proposed technique has been compared with the other related techniques in CAN network protocol. To perform this comparison, four experiments have been performed in two forms of fault-free and faulty buses. The evaluation results show that the proposed technique improves throughput, response time, and energy consumption parameters compared to a standard CAN, 29%, 37% and 19%, and compared to a redundancy-based technique, 77%, 41% and 21%, respectively.

Keywords