차량 네트워크를 위한 CAN-to-Ethernet 게이트웨이

Abstract

Vehicle diagnostic communication has evolved over the resent years. Controller Area Network (CAN) is an only communication network to provide traditional On-Board Diagnostic (OBD) services. The OBD could not be connected to Internet because of the physical differences and separation between CAN and Ethernet. In order to connect ECUs from remote controllers through Ethernet, vehicles should be equipped with a CAN-to-Ethernet In-Vehicle Network (IVN) gateway. If the most widely installed Ethernet applies to vehicle diagnostics, drivers or engineers will easily access Electric Control Units (ECUs) to check and reprogram the latest firmware from remote sites. However, the IVN gateway should resolve the inefficiency in the frame forwarding and the ECU programming completion time delay so that the remote vehicle diagnostics would be provided in the heterogeneous networks. The first problem is the inefficiency of the CAN frame forwarding through Ethernet. A single Ethernet frame should contain a mere 8-bytes CAN message although the header size of an Ethernet frame is 72-bytes. This inefficiency problem increases the frame overhead in the Ethernet network. The other problem is a delay of ECU programming completion time induced by burst frames and bus arbitrations. Each CAN frame includes a priority value inside the header to prevent collisions during bus arbitrations. When the priority value is low, the frame should be delayed until a lower priority frame is met. This dissertation presents a multiple CAN Transport layer-Protocol Data Unit (TP-PDUs) multiplexing method that improves transmission efficiency for the remote vehicle diagnostics between CAN and Ethernet networks. Also, it proposes a time-scheduling architecture for IVN gateways to minimize the ECU programming completion time by preventing burst frames and bus arbitrations. The performance results show that the proposed gateway decreases the average number of transmitted packets by about 48% and the total programming completion time for the multiple ECUs by a maximum of 41.99% when the total data size is 1.5MB and the Minimum Separation Time (STMin) is 1ms.