Open Source Summit Japan 2023

In the CASE era of automotives, it's very important to have end-to-end observability through vehicles, mobile/backbone networks, and the service for controlling a fleet of vehicles. So far, the monitoring of the mobile/backbone network side and the vehicle/service side has been done independently, and this made Root Cause Analysis of troubles difficult and time consuming. There are several approaches to this problem, and we shared an effort to address them from C-Plane before. This time, as an example of other approaches, we share our effort to address and ease the difficulty of RCA for troubles in the 5G network U-Plane and/or Service Center side using distributed tracing technology. In this presentation, we will first explain the end-to-end overall architecture of a connected vehicle service. Next, we will explain the challenges in trouble analysis of the system and ideas to address those difficulties using distributed tracing. As a case study, we will present a POC. Starting from the detailed system configuration, we show how easily we can identify the root cause of pseudo troubles that we induced in the POC system. Finally, we will walk through future works, including those for applying to production systems. This work is a part of joint work between TOYOTA MOTOR CORPORATION and KDDI Corporation. Keywords: Observability, OpenTelemetry, Kubernetes, Free5GC、UERANSIM