Analysis of key technologies of new vehicle controller

From the perspective of vehicle electrification, intelligence, networking, and sharing, the key technical requirements for new vehicle controllers are elaborated , including high computing performance, high communication bandwidth, high functional safety, and continuous software updates. In response to the above needs, this paper summarizes the current status of the key technology industries of Ethernet, CANFD, multi-core chips, dual core, and OTA, and looks forward to future development trends.

1 Introduction 

Electrification, intelligence, connectivity and sharing are recognized future development directions of the automobile industry. As a core component of electric vehicles, the vehicle controller must be able to support the "four modernizations" of the vehicle. It must meet the requirements of high computing performance, high communication bandwidth, high functional security, and continuous software updates. The current vehicle electronic and electrical architecture and vehicle controller technology are generally unable to meet the above needs. In order to cover the above needs, future automotive products will gradually adopt centralized electronic and electrical architecture. At the same time, the vehicle controller must include key technologies such as Ethernet, CANFD, multi-core chips, dual core, and OTA.

This article will first introduce the relationship between the vehicle controller and the two electronic and electrical architectures of distributed and centralized, then introduce the key technologies of the new vehicle controller, analyze the technical content, propose future development trends and conduct Outlook.

2. Vehicle controller and electronic and electrical architecture

(1) Vehicle controller and distributed electronic and electrical architecture

Under the premise of previous chip capabilities, limited by computing power and communication capabilities, the vehicle controller cannot integrate all vehicle control software, even software related to new energy component control cannot be integrated. This determines that the vehicle controller can only be used as a member of the distributed electronic and electrical architecture, but this relationship limits functional changes and expansion.

In the distributed electronic and electrical architecture, a vehicle-level function is completed by multiple controllers. The implementation of a certain function may require the cooperation of several or a dozen controllers, and these controllers may be distributed in different networks throughout the vehicle (Figure 1). The entire interaction process is extremely complex with time coordination. A vehicle generally has more than 100 controllers and hundreds of vehicle-level functions. The physical connections between the functions and the controllers themselves are intertwined into a huge and complex network, which is very unfavorable for modular design and expansion. In this case, adding a new function requires considering the correlation of various parts on the above-mentioned complex function network, and modifying and testing a large amount of controller software.

Figure 1 The position of the vehicle controller in the distributed electronic and electrical architecture

(2) Vehicle controller and centralized electronic and electrical architecture

With the development of chips and vehicle Ethernet, vehicle controllers have the ability to integrate most vehicle control software. The distributed electronic and electrical architecture is gradually developing towards a high degree of integration and intelligence. The position of the vehicle controller in the electronic and electrical architecture has also changed accordingly, truly realizing a vehicle-level integrated controller, whose control covers power, chassis and Some gateway functions. The relationship between the vehicle controller and the centralized electronic and electrical architecture is shown in Figure 2. Integrating most functions into the vehicle controller will greatly reduce the length of the vehicle's wiring harness and the number of controllers.

Figure 2 The position of the vehicle controller in the centralized electronic and electrical architecture

3. Key technologies for new vehicle controllers

In order to support the "four modernizations" of automobiles, the vehicle controller must meet many requirements such as high communication bandwidth, high computing performance, high functional safety, and continuous software updates. Among them, high communication bandwidth has given rise to the development of automotive Ethernet and CANFD technologies; high computing performance has given rise to the development of multi-core chips and dual-core control architecture technologies; and continuous software updates have given rise to the development of OTA technology. These technologies will be widely used in new vehicle controllers. These technologies will be introduced separately below.

(1)Vehicle Ethernet

Communication bandwidth issues have plagued the automotive industry over the past 20 years. During this period, CAN bus was the mainstream vehicle network technology. Its nominal speed of 1 Mbit/s was sufficient margin for automotive bandwidth requirements in the early days of the technology. However, in recent years, as vehicle control logic has become more and more complex, the number of required controllers and sensors has increased dramatically. Although the centralized electronic and electrical architecture can reduce the number of controllers to a certain extent, the computing power of domain controllers is much higher than The original vehicle controller, so the CAN communication bandwidth of 1 Mbit/s is obviously unable to meet the data interaction needs.

Higher communication bandwidth requirements have accelerated the convergence of Ethernet and the automotive industry. Ethernet was born in the 1970s, and its earliest prototypes are completely different from the Ethernet currently running in homes, offices, server rooms, and data warehouses. Although Ethernet develops with the times, there are still some problems when applied to automobiles, the most important one being electromagnetic compatibility. These limitations were broken with the advent of BroadR-Reach technology, which provides 100 Mbit/s bandwidth over a single unshielded twisted pair. This transmission method has never been used in previous Ethernet networks. Even if the physical layer changes, this technology can still achieve seamless integration with Ethernet at the upper layer and operate unchanged. Currently, this technology has been used in mass-produced models. At the same time, RTPGE technology that supports faster speeds is under development. While retaining software compatibility, its bandwidth is expected to be increased to 1 Gbit/s.

Although communication bandwidth has obvious advantages, due to cost and power consumption factors, automotive Ethernet is mainly used in backbone networks. Used for communication between the vehicle controller and other domain controllers, as shown in Figure 3. For smart actuators and sensors within the domain, other low-cost solutions are used, such as CANFD, CAN, LIN.

Figure 3 The vehicle controller uses Ethernet to communicate with other domain controllers

Of course, adding automotive Ethernet to the vehicle controller will face huge changes: a larger software protocol stack compared to CAN communication; greater controller power consumption; greater quiescent current, all of which need to be considered during system design. be considered.

(2) CANFD

Considering cost and power consumption, only the backbone network on the entire vehicle uses high-bandwidth Ethernet communication. But for other subnets, the nominal 1 Mbit/s CAN communication also urgently needs to increase the communication speed. The currently mature CANFD technology is a good solution.

The CANFD bus is a high-bandwidth solution for the CAN bus. Bosch first proposed the CANFD concept in 2011 and first released the CANFD1.0 version in 2012. While retaining the main characteristics of the CAN bus, it improves the error frame miss detection rate and ensures that most of the software and hardware in the network, especially the physical layer, remain unchanged. Increase the maximum transmission rate of the bus to more than 5 Mbit/s (the maximum transmission rate of CAN communication is 1 Mbit/s, and the actual usage rate is up to 500 kbit/s).

More importantly, the CANFD data length is up to 64 bytes, which makes the CANFD data field account for nearly 85%. CAN's data field accounts for only about 50%. This means that even with the same communication bandwidth, CANFD can transmit about 70% more effective data. The CANFD frame format is shown in Figure 4.

Figure 4 CANFD frame format

More importantly, since CANFD retains most of the key characteristics of CAN, all CANFD chips are compatible with CAN. This allows the controller that chooses the CANFD chip to adapt to the CAN communication network by only modifying the software without changing the hardware. CANFD technology has multiple advantages. For a long time to come, automotive Ethernet and CANFD will coexist for a long time, each performing its own duties and developing together.

(3)Multi-core chip

As in the early days of traditional consumer electronics, in order to obtain faster processing speeds, the automotive industry adopted the method of increasing core frequencies to increase processing speeds. However, in order to take stability into account, the core frequency increase has encountered a bottleneck. A small increase in the core frequency in the future will no longer be able to meet the growing demand for software execution speed. In this case, the automotive industry has chosen the same technical route as consumer electronics, using multi-core chips.

Multi-core chips have greatly improved the computing power of the chip. This is a parallel approach. Therefore, if you want to obtain the same effect in an application, you need to reasonably allocate each part of the software to each core during software design. The principle is to make all software as parallel as possible. The acceleration ratio of the computing power of a multi-core chip to that of a single-core chip with the same frequency can be evaluated using Amdahl's law. The formula is as formula (1):