SDV Track 2 About Me Masanori Itoh Affiliation TOYOTA MOTOR CORPORATION Digital Information and Communication Dept. InfoTech Div. Open Source Program Group (Toyota OSPO) Works R&D for Connected Vehicle Systems • E2E Observability, Standardization, Diagnostics, … OSPO Operations Keywords Operating System, Cloud Infrastructure, etc. URLs https://github.com/thatsdone https://www.linkedin.com/in/masanori-itoh-6401603/ me (by my daughter)
SDV Track 3 Agenda Overview of Connected Vehicle Data Processing In-Vehicle OS – Automotive Grade Linux and SDV Reference Platform 5G Cellular Network Multi-Slicing and its Use Cases Full OSS 5G Cellular Network Stack AGL and 5G Network Stack Integration (and others) POC at AECC Summary
SDV Track 4 Overview of Connected Vehicle Data Processing In-Vehicle OS – Automotive Grade Linux and SDV Reference Platform 5G Cellular Network AGL and 5G Network Stack Integration (and others) POC at AECC Summary
SDV Track 6 Reality of In-Vehicle Network Citation: “TOYOTA Developers Night 〜ソフトウェアエンジニアが革新するクルマ開発の伝統〜 Short ver.” TECHPLAY 2021/04/02 https://techplay.jp/video/16 Roughly means “Reinventing Automotive Traditions: The Role of Software Engineers in Driving Innovation.” “Computer Network with in a Vehicle”
SDV Track 7 Overview of Connected Vehicle Data Processing In-Vehicle OS – Automotive Grade Linux and SDV Reference Platform 5G Cellular Network AGL and 5G Network Stack Integration (and others) POC at AECC Summary
SDV Track 8 AGL – Automotive Grade Linux A Linux distro for vehicles supporting IVI (In-Vehicle Infotainment) and Instrument Cluster (Meter) based on Yocto Linux Highly customizable for various kinds of SOCs and purposes AGL Boot Image Yocto Linux (core) External Layers (e.g., open-embedded) AGL Common Layers Feature Dependent Layers RefHW (R-Car H3) RasPi5 qemu kernel user software hardware … IVI Screen (Flutter) Instrument Cluster Screen (Flutter) https://www.automotivelinux.org/
SDV Track 9 Automotive Grade Linux and SDV Reference Platform SDV – Software Defined Vehicle Huge Hype Customer side view Quick and Uniform New Feature Offering (Update & Upgrade) Manufacture side view Rapid Development Consolidation of ECUs AGL SoDeV Full OSS SDV Stack based on AGL, Xen, Zephyr etc. Deep dive to AGL reference SDV platform -SoDev- (Kurokawa-san) Software Defined Vehicles: Innovations Through Open Source Collaboration (Jerry san) Using Zephyr in SDV (Tokita san)
SDV Track 10 Overview of Connected Vehicle Data Processing In-Vehicle OS – Automotive Grade Linux and SDV Reference Platform 5G Cellular Network AGL and 5G Network Stack Integration (and others) POC at AECC Summary
SDV Track 12 3GPP 5G Cellular Network Architecture – Abbreviations So many components… C-Plane NRF Network Repository Function C-Plane AMF Access and Mobility management Function C-Plane SMF Session Management Function C-Plane PCF Policy Control Function C-Plane UDM Unified Data Management C-Plane UDR Unified Data Repository C-Plane NSSF Network Slice Selection Function C-Plane AUSF Authentication Server Function C-Plane CHF Charging Function C-Plane NEF Network Exposure Function U-Plane UPF User Plane Function (Relaying/switching/routing data packets) U-Plane gNB Next Generation Node B (Base station) U-Plane UE User Equipment (e.g. Smartphones, IoT Devices, Connected Vehicles) Mostly focusing on U-Plane components
SDV Track 13 Use Cases of 5G Slices A Simplified View(*1) of 5G multi-slice communication A ‘Slice’ of 5G mobile Radio Access Network (RAN) is like a VLAN 5G Cellular Network Internet Slice#1 (Normal) Slice#2 (B/W) I-UPF PSA-UPF I-UPF PSA-UPF gNodeB gNodeB Service#1 Service#2 (*1) Simplified view w.r.t. Network Functions Slice#2 For Bandwidth Sensitive Applications (e.g. Premier Service) Slice#1 For Normal Applications IoT Devices (very light traffic) Chatting with friends/parents (light traffic) Watching a movie at home (heavy traffic) Connected Vehicles (mixture of light & heavy traffic) Adaptive switching
SDV Track 14 Overview of Connected Vehicle Data Processing In-Vehicle OS – Automotive Grade Linux and SDV Reference Platform 5G Cellular Network AGL and 5G Network Stack Integration (and others) POC at AECC Summary
SDV Track 15 Emulated 5G Network using OSS Major OSS Implementations free5GC https://free5gc.org/ A work of NYCU (National Yang Ming Chiao Tung University) in Taiwan !! free5GC joined to the Linux Foundation! – “Advancing 6G Technology Development: NYCU’s Pioneering 5G Core Network Software ‘free5GC’ Joins the International Open-Source Platform” Open5GS OAI (Open Air Interface) UERANSIM RAN(Radio) side emulation contains UE + gNodeB https://github.com/aligungr/UERANSIM Nephio Network Orchestrator on top of Kubernetes Our POC system is composed of free5GC and UERANSIM
SDV Track 19 Slice/Interface Selection w.r.t. Destinations Problem#1 How can an application request particular network slice depending on its requirement? Resolution#1 Use URSP (UE Route Selection Policy) – Android and iOS implements Problem#2 How can an application select the appropriate network device w.r.t the selected slice? Resolution#2 Android/iPhone provides the mechanism transparently Other Linux – No Silver Bullet (for now) – Use static/dynamic routing mechanism – Use bind(2) before calling connect(2) » e.g. https://github.com/thatsdone/mslice Slice#1 (Normal Slice) AGL 5G Network Stack Integration – Additional Considerations Linux based OS (e.g. AGL) Network dev. for Slice#1 (useimtun0) Application #1 (w/o bandwidth needs) Slice selection mechanism Application #2 (w/ special bandwidth needs) Network dev. for Slice#2 (uesimtun1) Slice#2 (Premier Slice for B/W) Network dev. (default) Service#1 e.g. VoIP Service#2 e.g. Data Collection Essentially, we need Linux(or POSIX) generic URSP implementa tion
SDV Track 20 Overview of Connected Vehicle Data Processing In-Vehicle OS – Automotive Grade Linux and SDV Reference Platform 5G Cellular Network AGL and 5G Network Stack Integration (and others) POC at AECC Summary
SDV Track 21 AECC – Connectivity, Edge Computing & Collaboration AECC : Automotive Edge Computing Consortium (https://aecc.org) An Open Collaborative Activity for promoting Edge Computing focusing on Automotive Use Cases, with POCs Vehicle System Connected Car Vehicle System Connected Car Vehicle System Connected Car Vehicle System Connected Car Cellular Network Edge Server Pre- processing Cloud Edge Server Pre- processing Moving logs Biometric Sensors Cameras Cruising Data Sensor Data Control Data Intelligent Driving Enterprise Network CAN
SDV Track 22 Overall AECC Activities White Papers https://aecc.org/resources/publications/ POCs https://aecc.org/proof-of-concepts/ SIGs based on particular solution areas (now under forming process) (1) Data-first SIG Starting from V2V enabling more flexible/reliable vehicle data utilization (2) Telco API SIG CAMARA API + Remote SIM Provisioning (3) Green Computing SIG How to utilize green energy distributing workload by optimizing edge/route selection
SDV Track 23 Motivation Exploring enhancements of an existing POC POC10 : “Optimal Edge Selection for Realizing Digital Twins and Green Energy Utilization” Shooting for Carbon Neutral by selecting optimal compute resources spread across regions by choosing optimal routes/Edges Points of enhancements Evolution from Single-Slice to Multi-Slice configuration and show feasibility and benefits of utilizing multiple slices Improvement of the vehicle side software stack from just a workload generator to more realistic virtual vehicles based on multiple AGL instances
SDV Track 24 A Big-Picture of the POC A POC System of Connected Vehicles Service Data Processing gNB#1 W A N Location #1 (e.g. Tokyo) Location #2 (e.g. Osaka) Public Cloud #1 CAN Camera DynamicMap Others… LB Ops Subsystem Pseudo Vehicle Virtual Vehicle (SoDeV) Edge#1(Region #2) Auth GW Dispatcher Offload Process Slice#1 (Lat.) Auth GW For Fallback W A N Data Accumulation Edge#2(Region #2) Auth GW Dispatcher Offload Process UE#1 Slice#2 (B/W) UPF#2 UPF#1 Dedicated Line Virtual Vehicle (SoDeV) UE#1 Slice#1 (Lat.) Slice#2 (B/W) OCI /etc. Public Cloud#2 UPF#2 UPF#1 Fail Over gNB#2 gNB#2 gNB#1 Orchestr ation gNB#1 Dedicated Line/VPN Dedicated Line/VPN Dedicated Line/VPN
SDV Track 27 Fleet of Virtual Vehicles POC System on top of a Single Rack-mount Server Using Servers with ARM CPUs, (almost) no performance impact Virtual Vehicle #3 (Ricefish) Virtual Vehicle #2 (Salmon) Host OS Linux (Ubuntu 24.04, arm64) Virtual Vehicle #1 (Trout) KVM VM#1-1 AGL ivi VM#2 AGL cluster VVF Manager (Python) Windows etc. VVF Dashboard GTK+ 3.0 (Python on WSL2) Linux kernel VM#3 AGL gateway VM#3 AGL telematics libvirtd Dnsmasq (DHCP/DHCP)
SDV Track 28 Overview of Connected Vehicle Data Processing In-Vehicle OS – Automotive Grade Linux and SDV Reference Platform 5G Cellular Network AGL and 5G Network Stack Integration (and others) POC at AECC Summary
SDV Track 29 Summary SoDeV – AGL SDV Reference Platform Full OSS SDV Reference Platform 5G Cellular Network (w/Multi-Slice) and its Full Open Source Stack Full OSS 5G Mobile Network and Vehicle Integration POC based on AGL and Full OSS 5G Cellular Network Stack Cross Community Collaboration
SDV Track 30 Appendix (1) – Related Talks Related Talks “E2E Observability for Connected Vehicle Service via Distributed Tracing” at KubeCon NA 2023 https://kccncna2023.sched.com/event/1R2oh “E2E Observability for Connected Vehicle Services Including 5G Cellular Network U-Plane Troubles” at OSSJ 2023 https://ossjapan2023.sched.com/event/1Tyrm “Virtual Vehicle Fleet in your Lab - A Use Case of AGL as a High-Level R&D Testbed Component” at AGL AMM Spring (Feb. 27, 2025) https://sched.co/1vePR “Advancing the Virtual Vehicle Fleet; E2E Vehicle Data Processing and SDV Reference Platform Integration” at AGL AMM Summer (Jul. 10, 2025) https://sched.co/2412C
SDV Track 33 Appendix (2) – AGL OSPO-EG Activity – Survey Current Status Responses 58 responses so far (= 33(ja) + 25(en)) Random Initial Impression (w/o systematic/deep analysis) Japanese responses tend to complain show more issues/problems Perhaps, Japan is the worst place to work on OSS? ((((；ﾟДﾟ))))) More than 40% of Japanese responses are from OEMs (w/ > 20K employees) On the contrary, non-Japanese (mainly from Europe) responses are from Tier-N suppliers (w/smaller number of employees and (could be) more light weight business process/decisions) Call for Participation Do please respond to the survey !!
thatsdone
sudataka
oracle4engineer
infraplumber0
maguroalternative
endohizumi
ojima_h
biwashi
tak0x00
kakehashi
bengo4com
mugi_uno
maggiecrowley
danielanewman
michaelherold
addyosmani
jmmastey
thekraken
mza
dougneiner
tammielis
rocio
hannesfritz
