What are the advantages of domain-centralized architecture?

With the need for electronic and intelligent development, the traditional distributed architecture has gradually evolved into a domain-centralized architecture, and "domains" and "domain controllers" have emerged. The domain controller was first proposed by Tier 1 manufacturers such as Bosch, Continental, and Delphi. By using multi-core CPU/GPU chips with stronger processing power, introducing Ethernet and integrating scattered ECUs into domain controllers with stronger computing power to relatively centrally control each domain, it solves the limitations of distributed architecture such as cost and computing power.

The advantages of domain-centralized architecture mainly include:

1) Domain-centralized architecture can save costs and reduce assembly difficulty. In a distributed architecture, the large number of internal communication requirements generated by the increase in the number of ECUs leads to increased wiring harness costs and increased assembly difficulty; while the domain-centralized architecture separates sensing from processing, and sensors and ECUs are no longer one-to-one, making management more convenient, effectively reducing the number of ECUs and wiring harnesses, thereby reducing hardware costs and labor installation costs, and is more conducive to component layout.

2) The domain-centralized architecture can improve communication efficiency, achieve software and hardware decoupling, and facilitate vehicle OTA upgrades. In the distributed architecture, the software development frameworks and underlying codes of ECUs from different suppliers are different, resulting in redundancy and increasing the difficulty of maintenance and OTA unified upgrades; while the domain-centralized architecture achieves unified management and information interaction of each ECU, and a unified software underlying development framework, thus facilitating future OTA upgrades and the realization of expanded functions.

3) The domain-centralized architecture can further concentrate computing power and reduce redundancy. The computing power of each ECU in the distributed architecture cannot be coordinated and is mutually redundant, resulting in great waste. The domain control architecture centralizes the computing power of the originally dispersed ECUs, processes data in a unified manner, reduces computing power redundancy, and can better meet the high computing power requirements of advanced autonomous driving.

Based on the centralized functional zoning, traditional Tier 1 companies such as Bosch divide automotive electronic control systems into five domains: power domain (safety), chassis domain (vehicle motion), cockpit domain (entertainment information), autonomous driving domain (driving assistance) and body domain (body electronics).

The power domain is used for the optimization and control of the powertrain, and also has functions such as electrical intelligent fault diagnosis, intelligent power saving, and bus communication. The power domain controller is an intelligent powertrain management unit that uses CAN/FLEXRAY to achieve transmission management, lead-in management battery monitoring and alternator regulation. Its advantages are that it calculates and distributes torque for a variety of power system units (internal combustion engine, motor generator, battery, gearbox), achieves CO2 emission reduction through predictive driving strategy, and communication gateway. It is mainly used for the optimization and control of the powertrain, and also has functions such as electrical intelligent fault diagnosis, intelligent power saving, and bus communication. The chassis domain will integrate the vehicle's lateral, longitudinal, and vertical control functions such as vehicle braking, steering, and suspension to achieve integrated control.

The transmission system is responsible for transmitting the power of the engine to the drive wheels, and can be divided into mechanical, hydraulic and electric types; the driving system connects the various parts of the car into a whole and supports the entire car; the steering system ensures that the car can drive straight or turn according to the driver's wishes; the braking system forces the road surface to exert a certain external force on the car wheels that is opposite to the direction of the car's travel, and performs a certain degree of forced braking on the car. Its function is to slow down and stop the car, and to park the car. The chassis domain can integrate multiple functions in the transmission system, driving system, and braking system. The more common ones are air spring control, suspension damper control, rear wheel steering function, electronic stabilizer bar function, steering column position control function, etc. If enough computing power is reserved in advance, the chassis domain will integrate the vehicle's lateral, longitudinal, and vertical control functions such as braking, steering, and suspension to achieve integrated control. To realize the functions of the chassis domain, it is necessary to integrate the chassis domain drive, braking, and steering algorithms.

The smart cockpit domain integrates cockpit electronics such as HUD (head-up display), instrumentation, and in-vehicle infotainment to achieve "one chip, multiple screens". The components of the smart cockpit mainly include full LCD instrumentation, large-screen central control system, in-vehicle infotainment system, head-up display system, streaming media rearview mirror, etc. The cockpit domain controller realizes the integration of head-up display, instrument panel, navigation and other components through Ethernet/MOST/CAN. It not only has traditional cockpit electronic components, but also further integrates the smart driving ADAS system and the vehicle networking V2X system, thereby further optimizing smart driving, vehicle interconnection, infotainment and other functions. The smart cockpit domain can achieve "independent perception" and "interaction mode upgrade". On the one hand, the vehicle has the ability to "perceive" people. On the other hand, the in-vehicle interaction mode has been upgraded from only "physical button interaction" to the coexistence of "touch screen interaction", "voice interaction" and "gesture interaction", which provides a better experience.

The autonomous driving domain enables vehicles to have the capabilities of multi-sensor fusion, positioning, path planning, decision control, image recognition, high-speed communication, and data processing. The autonomous driving domain usually requires multiple external cameras, millimeter-wave radars, lidars, and other on-board sensors to perceive the surrounding environment, and formulate corresponding strategies through sensor data processing and multi-sensor information fusion, as well as appropriate working models, to make decisions and plans. The input of the domain controller is the data of various sensors, and the algorithm processing performed covers the three levels of perception, decision-making, and control. Finally, the output is transmitted to the actuator to control the vehicle's lateral and longitudinal directions. The functions integrated in the autonomous driving domain basically do not involve mechanical components, and interact closely with the cockpit domain. Like the smart cockpit domain, it needs to process a large amount of data and has high requirements for computing power. Therefore, it is necessary to match chips with strong core computing power to meet the computing power requirements of autonomous driving, simplify equipment, and greatly improve the integration of the system.