A Simulator for Time-Sensitive In-Vehicle Networks in Autonomous Driving

Abstract

Recently, large amount of data traffic from various sensors, image, and navigation systems within vehicles is generated for autonomous driving. Broadband communication networks within vehicles have become necessary. New autonomous Ethernet networks are being considered as alternatives. The Ethernet-based in-vehicle network (IVN) has been standardized in the IEEE 802.1 time-sensitive networking (TSN) group since 2006. TSN-related ongoing projects are many, approaching about 20, and new projects are still being built as needed. The detailed standards progressing into the project add the necessary functions to the TSN one by one and present the necessary technical requirements. However, each standard is handled independently, and the standard has not been established in consideration of the environment in which functions of various standards work together. At this point, in order to confirm the maximum performance and maximum efficiency of TSN and to reliably define parameter values to be defined as recommendations in the new standard, an environment in which the main functions of TSN can operate at the same time should be configured. In this dissertation, the main function of Audio/Video Bridging (AVB) transmission protocol specialized for the transmission of electronic control unit (ECU) updates of the vehicle was developed, and performance test was conducted in conjunction with HW. Since the existing AVB system was developed for streaming audio and video, there is no function for loss prevention and recovery for updating the firmware of the ECU. To this end, a sliding window-based AVB file transfer method is proposed to provide a fast frame transfer and a recovery solution for frame loss. Then, an integrated TSN simulator is developed to implement the significant functions of the networks in the TSN standards that are being developed. This effort would minimize the performance gaps that can occur when the functions of these standards operate in an integrated environment. As part of this purpose, I analyzed the simulator to verify that the traffic for autonomous driving satisfies the TSN transmission requirements in IVN and the preemption (which is one of the main TSN functions) and reduces the overall End-to-End delay. An IVN topology for autonomous vehicles was designed and the performance evaluation was conducted by configuring the traffic to be used for various sensors and ECUs. In conclusion, through the TSN integration performance evaluation, the optimal guard band size was derived as 128 bytes, and the transmission efficiency for critical control data is about 43.8% improved when transmitted using the time-triggered TSN transmission method compared to the existing event-triggered AVB transmission method.