Advanced Driver Assistance Systems (ADAS) use SerDes links to move large amounts of sensor data. ADAS SerDes systems connect remote sensors to an ADAS domain controller through serializer ICs and deserializer ICs. ADAS SerDes systems use coaxial or STP cables to carry forward data, a back-channel, and power over coax.

High-speed serializer/deserializer links carry real-time sensor data from remote nodes to central processors. Sensors might include:
  • Cameras
  • Radar
  • Lidar
  • Ultrasound
ADAS moves the data in real time. SerDes multiplexes video, control, and power onto a single coaxial or shielded twisted-pair (STP) cable. The SerDes link is used in place of a multi-wire harness.

Why SerDes is used in ADAS

Automotive sensors generate data volumes that you cannot route through conventional wiring. A single high‑resolution camera can produce several Gbps of uncompressed pixel data. A vehicle might include a dozen or more sensors. SerDes addresses these challenges by:

  • Reducing the cable count and weight - SerDes replaces wide parallel buses with a single cable per sensor.
  • Delivering power over the same cable - Power over Coax (PoC) eliminates the need for a separate power harness for each remote node.
  • Supporting bidirectional control - Bidirectional control embeds a back-channel in the serial link. The back-channel lets I2C configuration commands and GPIO signals travel from the central processing unit to the remote sensor without additional wiring.
  • Enabling precise multi-sensor synchronization - Serializer ICs distribute a synchronization trigger to all cameras simultaneously. The synchronization trigger captures all frames at the same instant.

Common ADAS Sensing Applications

Table 8. Common ADAS Sensing Applications
ADAS Sensing Application Description
Surround View and Parking Assist
A surround view system combines four or more wide-angle cameras to construct a 360-degree view around the vehicle. Each camera captures one view of the vehicle:
  • Front of vehicle
  • Rear of vehicle
  • Left side of vehicle
  • Right side of vehicle

Each camera connects to a serializer. All serializers transmit over coax to a multi-port deserializer hub. The hub forwards the aggregated streams to an ADAS SoC (System on Chip) for stitching and rendering. PoC supplies power to each camera module over the same coax cable.
Front Camera – Forward Collision Warning, Lane Departure, Traffic Sign Recognition Front-facing cameras are typically high-resolution cameras with a narrow field of view. High resolution is 2 megapixels (MP) to 8 MP and higher. The cameras require high forward-channel data rates to carry uncompressed high-resolution imagery to the ADAS domain controller. The controller processes the imagery in real time.
Rear View and Trailer Assist Rear-view cameras provide a live feed for backup assistance and trailer guidance. These cameras are physically located far from the head unit or domain controller. A serial link over coax is the standard approach to run a cable through the vehicle chassis without excessive signal degradation.
Driver and Occupant Monitoring – In-Cabin Sensing
Driver monitoring systems (DMS) and occupant monitoring systems (OMS) use near-infrared cameras inside the cabin. The cameras track numerous factors.
  • Gaze
  • Head pose
  • Fatigue indicators
  • Seat occupancy
These cameras operate in low-light conditions and transmit continuous video to a dedicated SoC. The back-channel of the SerDes link sends illuminator control commands back to an IR LED driver at the camera module.
Automotive Radar Modern radar modules output processed or raw radar data through high-speed digital interfaces (CSI-2 or LVDS). A serializer near the radar module combines the data. The serializer then sends the data over a coax cable to a deserializer at the controller.
LiDAR LiDAR sensors generate dense 3-dimensional point clouds at high data rates. SerDes links carry the data from the sensor housing to a central high-performance compute platform. The sensor housing is often mounted on the roof or in the bumper. The long cable runs involved, combined with the high data bandwidth requirements, make a high-speed serial link over coax the preferred physical layer.
Ultrasound – Parking Sensors Short-range ultrasonic sensors detect nearby objects during low-speed maneuvers. The data rate of individual sensors is low. Multiple sensors might share a serializer node with other sensor types. The node transmits the aggregated data over a single cable to the central processor.

SerDes Standards used in ADAS

Table 9. SerDes Standards used in ADAS
Standard Forward Rate, Maximum Physical Medium Key Features
FPD-Link III ~4.16 Gbps Coax or STP

PoC

Back-channel I2C/GPIO

CSI-2 input
FPD-Link IV ~15.2 Gbps Coax or STP

PoC

Bidirectional control

CSI-2 input

ASIL-capable
GMSL2 ~6 Gbps Coax or STP

PoC

Back-channel

I2C/SPI/UART
GMSL3 ~12 Gbps Coax or STP

PoC

Enhanced back-channel bandwidth
Note Refer to the datasheet for your serializer or deserializer for part-specific features and the maximum forward rate.

ADAS SerDes System Components

A typical ADAS SerDes system includes the following components:
  • Remote sensor module: - An image sensor (or radar/lidar output stage) connects to a serializer IC via a digital interface such as MIPI CSI-2. The serializer encodes the data stream and drives the coax cable forward-channel while receiving the back-channel.
  • Coaxial or STP cable: - A single cable carries the forward serial data, the back-channel, and the PoC supply voltage. The system transmits these signals between the sensor module and the vehicle chassis.
  • Deserializer or deserializer hub - The deserializer is located near the domain controller. The deserializer performs two tasks:
    • Recovers the forward-channel data.
    • Forwards the data stream as a digital interface to the SoC. The digital interface is typically CSI-2.
    Multi-port hubs aggregate several sensor links into one or two CSI-2 outputs.
  • ADAS domain controller / SoC - Receives CSI-2 streams. Performs real-time image signal processing (ISP), object detection, and sensor fusion.