Contents

In-vehicle network basics

How vehicle electronics exchange information, why cars use several kinds of network, and what matters when choosing between them.

Why a car needs networks

A wheel-speed sensor, brake controller and dashboard may all need the same information. Giving every sender a separate wire to every receiver would create a complicated harness. Networks let electronic control units (ECUs) share communication paths.

Different jobs need different paths. A mirror adjustment sends little data; a camera produces a continuous stream. A brake function needs predictable timing; a software download needs capacity. Cars therefore combine networks with direct electrical connections.

Core ideas in plain words

Signal versus message/frame. A signal is a meaningful value, such as vehicle speed, with a unit and interpretation. Several signals can travel together in a message. A frame is the structured unit sent over a particular network, including addressing or identification, control information and error checks. Engineers sometimes use “message” and “frame” loosely; agree on terminology.

Bit rate versus payload. Bit rate describes how quickly bits travel. Payload is the useful data carried inside a frame. Headers, checks, idle gaps, arbitration and retries consume capacity, so application throughput is lower than the advertised rate.

Bus versus switched network. On a shared bus, connected nodes share transmission capacity. A switch forwards Ethernet frames between ports, allowing separate links to carry traffic simultaneously. Traffic still competes when it needs the same outgoing port.

Arbitration and determinism. Arbitration decides who transmits when several nodes want access. Controller Area Network (CAN) uses priority-based arbitration; Local Interconnect Network (LIN) uses a controller-managed schedule. Determinism means timing can be bounded under stated conditions. Neither priority nor a fast link alone guarantees every deadline.

Topology. This is the arrangement of nodes and cables: a line with short branches, point-to-point links, a star around a switch, or combinations. Cable length, branches and connector placement affect electrical behavior and timing.

Physical layer. A transceiver converts controller logic into electrical signals and back. Twisted pairs help differential links resist interference and limit emissions. Termination controls signal reflections; its placement and value depend on the interface. High-speed CAN typically terminates the two bus ends. LIN uses a single communication wire with a shared ground reference.

The networks in one table

Rates below describe common configurations or technology capabilities, not guaranteed application throughput.

Technology Medium Typical rate Access method Typical use
Hardwired I/O Dedicated conductor(s) No universal bit rate Direct electrical input/output Switches, analog sensors, actuator drive
LIN Single communication wire plus ground reference Up to 20 kbit/s Scheduled headers and responses Seats, mirrors, simple actuators
Classical CAN Usually twisted pair Often 125–500 kbit/s; up to 1 Mbit/s Priority arbitration Body and control signals
CAN FD Usually twisted pair Often 500 kbit/s arbitration, 2–5 Mbit/s data phase Priority arbitration Control traffic, diagnostics
CAN XL Twisted pair Data phase up to about 20 Mbit/s with suitable hardware Priority arbitration Larger messages and aggregated data
10BASE-T1S Single twisted pair; supports multidrop 10 Mbit/s Shared-medium access, optionally PLCA Local Ethernet edge nodes
Automotive Ethernet 100M/1G/multi-gig Usually single twisted pair per link 100 Mbit/s, 1 Gbit/s, 2.5/5/10 Gbit/s Switched full-duplex links Backbone, compute and sensor connections
SerDes camera links Coaxial cable or shielded pair Commonly multi-Gbit/s; technology-dependent Dedicated serialized link Camera video to processing units

SerDes means serializer/deserializer. These links often carry video in one direction with a smaller control channel back. Physical Layer Collision Avoidance (PLCA) coordinates transmission opportunities on a shared 10BASE-T1S segment; application deadlines still require traffic planning.

How they fit together in a zonal car

A zone controller connects nearby sensors and actuators, using direct inputs, LIN, CAN or Ethernet as appropriate. An Ethernet backbone connects zones to central computers. A gateway moves selected information between networks; a zone controller may also perform this role.

Consider a door switch. Its state reaches a zone controller, which checks and forwards it to a function elsewhere. That function may return an actuator command. Each hop adds processing and possible waiting time.

A useful layering picture is signals → protocol data units (PDUs) → frames. Software packs values into PDUs; communication layers carry those units in network frames. This is a simplified model: PDUs can contain other PDUs, and large units may span multiple frames.

Gateways may repack signals or route packets. They must preserve meaning, freshness and timing expectations.

SOME/IP, short for Scalable service-Oriented MiddlewarE over IP, supports service communication over Internet Protocol (IP), usually carried by Ethernet. It sits above the Ethernet frame layer.

Example door-switch information path
Door switchProvides switch state
Zone controllerChecks and forwards state
Ethernet backboneCarries information between zones and computers
Remote functionUses switch state

This example shows the outward information path; processing and waiting along the route contribute to latency.

How to choose

  • Cost: Include controllers, transceivers, connectors, software and validation.
  • Bandwidth: Count payload, protocol overhead, bursts, diagnostics and growth.
  • Latency: Define deadlines and maximum acceptable variation; include every gateway and queue.
  • Safety: Identify communication failures and required detection and recovery behavior.
  • Security: Define permitted senders, receivers and routes; select authentication and confidentiality where needed.
  • Wiring: Evaluate harness length, branches, shielding, connector pins and repairability.

Time-Sensitive Networking (TSN) provides Ethernet mechanisms for controlled timing and reliability. Their selection and configuration matter; IEEE 802.1DG profiles them for automotive networks.

Best practices

  • Define each signal’s unit, encoding, owner, update interval and timeout.
  • Budget capacity and end-to-end latency under peak traffic and fault conditions.
  • Validate cables, termination and signal quality on representative harnesses.
  • Plan sleep, wake-up, startup and diagnostic access alongside normal operation.
  • Define receiver behavior for missing, stale or implausible data.

Common pitfalls

  • Treating nominal bit rate as available payload bandwidth.
  • Assuming higher priority guarantees timely delivery under any load.
  • Expecting a gateway to translate protocols without mapping data semantics.
  • Assuming frame error checks provide sender authentication or complete safety protection.
  • Choosing topology after selecting the highest possible data rate.
  • Forgetting that a failed shared segment can affect several functions.

Where to go next

  • Classical CAN — identifiers, priority arbitration, frames and shared-bus behavior.
  • CAN FD — larger payloads and separate arbitration and data-phase rates.
  • LIN — scheduled communication for simple, inexpensive devices.
  • Automotive Ethernet — point-to-point links, switches and IP communication.
  • 10BASE-T1S — shared single-pair Ethernet and PLCA access coordination.
  • TSN — Ethernet time synchronization, traffic scheduling and bounded latency.
  • CAN XL — the next CAN generation, bridging towards Ethernet.
  • SOME/IP & DDS — service-oriented communication on top of Ethernet.

What it means for your discipline

  • Networks & communication Core

    Chooses network technologies, topology and communication timing; checks capacity and delivery deadlines across connected segments.

  • E/E architecture Core

    Allocates functions to controllers and defines how local networks connect to the backbone; weighs wiring, gateway dependencies and fault containment.

  • Base software Partial

    Configures communication drivers, message layouts, routing and sleep/wake behavior so applications exchange the intended data.

  • Hardware & silicon Partial

    Selects transceivers, cables, connectors and termination; validates signal quality and electromagnetic compatibility on the real harness.

  • Cybersecurity Partial

    Defines trust boundaries, gateway filtering and communication protection; checks which nodes can inject or receive sensitive traffic.

  • Diagnostics & lifecycle Partial

    Plans tester access, diagnostic routing and software-download paths through multiple network segments.

  • Functional safety Partial

    Defines communication deadlines and fault responses; checks how missing, delayed or corrupted information affects vehicle functions.

  • Software update & OTA Partial

    Plans software-download routes through gateways and network segments; accounts for usable throughput, competing traffic and connectivity when estimating flashing time.

No direct stake: HPC & application software

Where it sits in the software stack

Generic layered software of a central HPC (application processor with hypervisor, plus a safety island) and of a classic MCU-based ECU (AUTOSAR Classic). Highlighted modules matter for this topic. Click any module to open its page.

HPC · application processor (MPU) and safety island (MCU)
  • SOME/IP · DDS — Carries service messages over IP networks using SOME/IP or DDS.
  • BSP · drivers — Provides network interface drivers that send and receive frames on the Ethernet backbone.
  • Classic RTE & COM — Connects local control signals to Classic communication stacks for CAN and LIN networks.
MCU-based ECU · AUTOSAR Classic
  • Communication — Packs signals into PDUs, routes traffic and manages network communication states.
  • ECU abstraction — Provides interfaces between communication protocols and bus-specific drivers.
  • Microcontroller abstraction (MCAL) — Operates CAN, LIN and Ethernet controllers with the configured timing and hardware settings.