CAN-Ethernet 1,a) 1 1 2 2 2 CAN-Ethernet CAN CAN CAN OMNeT++ CAN Ethernet CAN-Ethernet protocol convert algorithm for automotive networks Jun Matsumura 1,a) Yutaka Matsubara 1 Hiroaki Takada 1 Masaya Oi 2 Masumi Toyoshima 2 Akihito Iwai 2 Abstract: CAN-Ethernet protocol convert algorithms for gateways in automotive control networks have been studied so far. These algorithms can reduce either delay time of CAN messages or the processing amount of a gateway by changing parameters such as conversion timing and buffer size. Due to a tradeoff between them, determination of proper parameters for each network is difficult. To meet deadline constraints of CAN messages, we proposed a new algorithm, called deadline algorithm. The deadline algorithm adjusts the wait time at the gateway according to transmission period of each CAN message. To evaluate the proposed algorithm, we have developed a simulation environment based on OMNeT++ which is an open-source discreate network simulator. The results of evaluations present that the deadline algorithm can save the network resources with satisfying the deadline constraint. Keywords: CAN, Ethernet, protocol convert algorithm, gateway, automotive network 1. DoIP Diagnostics on Internet Protocol 1 Nagoya Univercity, 464 8601, Japan 2 DENSO CORPORATION, Japan a) j matsu@ertl.jp CAN Controllar Area Network CAN 1Mbps Eth- 1
ernet ECU Ethernet CAN Etheret CAN-Ethernet CAN-Ethernet CAN-Ethernet CAN Ethernet PDU Protocol Data Unit CPU [1] CPU CAN OMNeT++ 2 3 4 5 6 2. Kern CAN-Ethernet 4 [2] 2 CPU Fig. 1 1 Network configuration and conversion direction ADFweb HD67048[3] CAN CAN 2 Ethernet OPNET[6] CAN Lund TrueTime[7] Matlab/Simulink 3. 1 CAN Ethernet CAN CAN ECU 4. 4.1 [2] 4 1 CAN 1 1 Ethernet 2 CAN Ethernet 3 CAN 2
1 Table 1 Parameters of convert algorithms CAN basic 1 - - - - buffer - - - time - - urgency - deadline - - 2 Fig. 2 Words of the deadline algorithm CAN 4 Ethernet CPU 4.2 1 CAN CAN CAN CAN CAN CAN Ethernet 2 2 T CAN 4 t create t GW CAN T latency(can) T CAN GW GW t deadline(gw ) t deadline(gw ) t GW Ethernet t deadline(gw ) t nextconv t deadline(gw ) t nextconv t deadline(gw ) t nextconv CAN 1 1 4.3 3 4 CAN 4 CAN CAN GW GW 4 CAN GW 6 CAN 4 CAN Ethernet 4.4 CAN CAN 2 1 5ms 2 20% CAN CAN CAN CAN 3
3 Fig. 3 Behavior of the deadline algorithm 5 Fig. 5 Simulation model OMNeT++ 4.2.2 Ethernet INETFramework[8] 1.99.4 CAN CAN Venet[9] 0.2% 0.2% CAN 4 Fig. 4 Overview of the simulation environment 5. 5.1 OMNeT++ [4] [5] 4 OMNeT++ Ethernet CAN CAN-Ethernet ECU 5 XLS XML 5.2 3 CAN CAN Ethernet CAN ECU 1 CAN Ethernet CAN Ethernet Ethernet Ethernet CPU 60% 70% 4
Fig. 6 6 Network configuration for the evaluation 3 Table 3 Parameters of the experiment CAN [ms] [ms] [%] basic 1 - - - - buffer 40 20 - - - time 40 20 20 - - urgency 40 20 20 20 - deadline (t=5ms) 20 5 - - 5ms deadline (r=0.5) 20 5 - - 50% 2 CAN-Ethernet Table 2 Parameters of CAN-Ethernet network CAN 500kbps Ethernet 100Mbps CAN 90 Ethernet 1 CAN-EthernetGW 2 CAN 4 CAN 216 194 1% 5.3 6 4 CAN 1 Ethernet CAN-Ethernet 2 CAN-Ethernet 3 CAN-Ethernet 2 CAN 5ms deadline(t=5ms) CAN 50% deadline(r=0.5) 7200 5.4 7 1 CAN Ethernet 4 CAN Ethernet empty CAN Ethernet not empty CAN Ethernet Ethernet GW1 5.5 7 CAN 6. CAN-Ethernet CAN CAN OMNeT++ Ethernet CAN 5
7 Fig. 7 Maximum delay ratio Table 4 4 CAN Ethernet Average workload of each bus and the number of conversions to Ethernet packets [%] CAN Ethernet GW1 GW2 CAN1 CAN2 CAN3 CAN4 Ethernet empty not empty empty not empty basic 32.78 69.61 45.31 50.81 2.14 1 1753806 1 11453219 buffer 32.78 69.61 45.31 50.81 0.34 1 359999 1 363375 time 32.78 69.61 45.31 50.81 0.35 1 366819 1 426005 urgency 32.78 69.61 45.31 50.81 2.01 1 1667816 1 10637389 deadline(t=5ms) 32.78 69.61 45.31 50.81 0.45 1 718992 1 824190 deadline(r=0.5) 32.78 69.61 45.31 50.81 0.48 1 718992 1 978810 Ethernet CAN CAN GW CAN CAN CAN Ethernet Ethernete CAN CAN JSPS 24700027 [1] J. Sommer and R. Blind, Optimized Resource Dimensioning in an embedded CAN-CAN Gateway, in Industrial Embedded Systems, 2007. SIES 07. International Symposium on, pp. 55 62, july 2007. [2] A. Kern, D. Reinhard, T. Streichert, and J. Teich, Gateway Strategies for Embedding of Automotive CAN- Frames into Ethernet-Packets and Vice Versa, in Architecture of Computing Systems - ARCS 2011, vol. 6566 of Lecture Notes in Computer Science, pp. 259 270, Springer Berlin / Heidelberg, 2011. [3] ADFweb, http://www.adfweb.com,. [4] A. Varga and D. History, OMNeT++ Discrete Event Simulation System, version 3.1 edition,, 2005. [5] A. Varga and R. Hornig, An overview of the OMNeT++ simulation environment, in Proceedings of the 1st international conference on Simulation tools and techniques for communications, networks and systems & workshops, Simutools 08, ICST, Brussels, Belgium, pp. 60:1 60:10. [6] X. Chang, Network simulations with OPNET, in Simulation Conference Proceedings, 1999 Winter, vol. 1, pp. 307 314 vol.1, 1999. [7] D. Henriksson, A. Cervin, and K.-E. Arzen, True- Time: Real-time Control System Simulation with MAT- LAB/Simulink, in Proceedings of the Nordic MATLAB Conference, 2003. [8] K. Wehrle, J. Reber, D. Holzhausen, V. Boehm, V. Kahmann, and U. Kaage, INET Framework for OMNeT++. [9] Venet, http://www.kumikomi.net/archives/2004/11/ 31et04/interd/vs.pdf,. 6