The Research Of Fault-tolerance Mechanism On AFDX

With the rapid development of high-performance aircraft and the increasing complexity of air environment, the requirement about avionics electronic systems is more and more, and the dependency about the avionics electronic systems becomes more and more strong. The traditional avionics data communication bus, such as ARINC429 and MIL-STD-1553, can't meet those requirements of the next generation of avionics communication system, so the Avionics Full Duplex Switched Ethernet (AFDX) has been designed.In order to meet the critical reliable real-time communication demands of the avionic application, AFDX uses the network mechanism of redundant communication link, so there are two separate network, A and B in AFDX network. Normally, the AFDX end system will transmit data paralleling on the network A and network B. So, every end system will receive two pieces of same packets, and the end system need to filter the packets and merge the two data flow into one non-redundant stream in the redundancy management module. The redundancy management module must ensure that the network can accept the correct packets when there are some unexpected errors such as abnormal transmitted frame, loss of a frame, reset of end system, switch babbling, and so on. In this article, our main work is to do some research about the AFDX network and its redundancy management to increase the safety of the AFDX network.The main content and result of the subject is as follows: (1) The jitter-EDD mechanism has been added into the AFDX network to control the end-to-end delay jitter and solve the problem of the loss of valid packets and increase the data safety.(2) The architecture of the AFDX end system and switch has been changed to meet the requirement of the jitter-EDD mechanism. In the original jitter-EDD scheme, there is no valid deadline distribute algorithm, so a method for deadline distribution which based on weight has been proposed.(3) The existing AFDX network can't meet the demands for critical fly-control in reliability and safety Therefore, the study of the schedulability of the task set on the network nodes becomes very necessary. In this paper, a necessary and sufficient condition has been given out.(4) A new redundancy management algorithm is designed to management the packet and its redundant packet to assure the reliability of the network.(5) SimEvents is used to establish AFDX network simulation platform according to the topology of A380. And, we use fault injection methods to verify the redundancy management algorithm. It has shown that the new redundancy management algorithm was useful.