Edited from: https://www.eet-china.com/mp/a152690.html
Source: Softing Automotive Electronics
1. Automobile intelligence promotes the arrival of the era of software-defined automobiles
Looking back at the history of the development of the automobile industry, the automobile industry has experienced the development process from the mechanical age to the electronic age and now to the software age. Since the 1980s, ECUs have been continuously added to vehicles. The automotive industry has focused on Tier1 by adding ECUs to improve vehicle functions. During this process, automotive software has been developed in a way that is deeply coupled with hardware; The hardware configurations of different models have gradually converged, and the cost and function improvement space is limited. However, new energy and intelligence have gradually achieved success. Automotive software has begun to become the core element of differentiation for car companies. The automotive industry is gradually moving towards software-defined vehicles (Software Defined Vehicles). Vehicles, SDV) era.
The entire automotive industry is transforming towards intelligence. Different from traditional cars, smart cars can create a wealth of perceived value and a more comfortable driving experience for car owners through brand-new software technology. In recent years, more and more global vehicle manufacturers, component manufacturers, and technology companies such as Google, Apple, and Baidu have begun to invest in the research and development of smart cars, and smart cars are rapidly occupying the automotive market. Taking the autonomous driving function as an example, mainstream car companies around the world are intensively developing autonomous driving above L3 level. In the future, the deployment rate and level of autonomous driving will continue to increase.
Automobile intelligence will drive the explosive growth of automotive software development demand. A "digital" car (2015) can contain 100 million lines of software code, much higher than the code volume of high-tech products such as Facebook, fighter jets, and artificial satellites. With the development of intelligent modules such as smart cockpits and autonomous driving, the amount of automotive software code is still increasing at an annual growth rate of more than 20%. The code volume of a smart car produced in 2025 is expected to reach 700 million lines, an increase of 2.3 times compared to 2020. It can be seen that the technical barriers of automobile manufacturing have also changed from the integration capabilities of the traditional three major parts and components to the ability of code development. With the continuous upgrading of automobile intelligence and the gradual prosperity of software ecology, the demand for automobile software development will explode. , the proportion of vehicle software cost will increase significantly.
The scale of the automotive software market will continue to expand. Global automotive software market: Berylls Management Consulting expects the automotive software market to more than triple in size during 2020-2030, with an average annual growth rate of 13%, and the market size will grow from 76 billion euros to 252 billion euros. Specifically, the field of intelligent driving (ADAS/AD) will occupy the largest share of the growth of the automotive service market from 2020 to 2030, and the software platform, safety and integration test verification will also have a relatively high compound growth rate. The fastest will be high-performance computing platforms (HPC), expected to account for 37%. If the entire market increment is to be further split, the core increment (approximately 210 billion euros) comes from the increase in the complexity of intelligent functions, and at the same time, the improvement in efficiency due to software modularization and changes in development methods will also reduce development expenditures 62 billion euros.
1.1. The upgrade of EE architecture is the hardware foundation of software-defined cars.
Intelligentization and networking must be based on the computing power of the core of the electronic and electrical architecture. Without the hardware foundation, software-defined cars cannot be realized. The changes in the EE architecture of cars are mainly reflected in the following four Aspects:
Computing performance: Automotive chips shift from MCUs to SoCs. The MCU chip usually only includes a CPU processor unit, storage and interface unit, and the computing power is generally only a few hundred DMIPS; while the SoC is a system-level chip, which generally adopts the "CPU+AI chip (GPU\FPGA\ASIC)" architecture solution, such as NVIDIA Orin X has a computing power of up to 254TOPS. Smart cockpits and autonomous driving have brought an order of magnitude increase in the smart architecture and algorithm computing power of cars. MCU-based automotive chips will not be able to meet these needs, and will turn to SoC chips with stronger computing power; communication bandwidth: on-board
Ethernet Internet has become the backbone communication network of automobiles. In the traditional distributed architecture, ECUs mostly communicate through CAN communication, LIN communication, Flex Ray, etc., and the data transmission speed is very limited, generally only a few megabytes per second. With the increase in the number of sensors in the car, the data transmission volume and rate requirements are greatly increased. In the future, the automotive Ethernet will become the backbone network of the car, and the transmission rate of 100Mbit/s or even 1Gbit/s can be achieved on a single pair of unshielded twisted pair .
Software and hardware decoupling realizes OTA upgrade. The software is no longer developed based on a certain fixed hardware. The chimney vertical architecture of the original ECU software of the car is transformed into a horizontal layered architecture of general hardware platform + basic software platform + various application software to realize the decoupling of software and hardware. The hardware is pre-embedded, the software is deployed later, and the software function iteration is realized through continuous OTA to promote the vehicle function upgrade.
Better cost control. At present, the number of main ECUs in high-end models and highly intelligent models has reached more than 100, and the total number of ECUs with some simple functions can exceed 200. The increase in ECUs and the increase in wiring harnesses will increase costs. Through domain control integration, it can The number of ECUs is greatly reduced; in addition, ECUs are provided by different suppliers, and any function modification involves the redevelopment and verification of multiple controllers, which is time-consuming and labor-intensive, and the software logic is controlled by the supplier, and the OEM cannot implement the software functions efficient management.
Under the trend of intelligence and network transformation, software and hardware are decoupled at the component level, and software independently becomes the core component product. The multi-dimensional vehicle data and control rights obtained by automotive software products realize complex functions and task execution. Automotive software is becoming more and more complex, the number of lines is increasing rapidly, and the architecture of system OS and application software is gradually formed, making the development of automotive software more difficult.
1.2. SOA is the software trend of software-defined automobiles
. Under the traditional distributed EE architecture, the operation of automobile software is mainly based on the signal-oriented architecture (Signal-Oriented Architecture). This software architecture cannot meet the needs of intelligent automobiles: the lack of fixed
architecture flexibility. The codes of each function of ECU are pre-defined in the ECU sorting file in the architecture design stage, and are called in turn and run one by one during the running process. The signal sending and receiving relationship between ECUs is static, and the signal can only be forwarded by the gateway, which is not flexible. At the same time, this fixed software architecture limits the needs of users for personalized development. OTA external developers cannot define new functions by software, and cannot support online upgrades and iterative updates of software.
The signal-oriented architecture cannot realize human-vehicle interaction. The signal-oriented architecture only supports receiving and sending modes, not request and response modes, and cannot realize interaction, so the characteristics of smart cars cannot be brought into play.
Under the distributed architecture, software and hardware are highly coupled, and software operation depends on hardware. When the software is changed or upgraded, the integrated verification of the whole vehicle is required, which takes a long time and is difficult. In addition, when there is a problem with a certain controller, all corresponding functions may fail, which not only increases the cost, but also causes great safety problems under intelligent functions such as smart cockpit and automatic driving.
Changing software functions is costly and difficult. Under the traditional signal-oriented architecture, if a certain software function needs to be changed, the vehicle communication system and ECU must be changed, and the sharp increase in the number of ECUs has significantly increased the cost and complexity of this process.
SOA divides different functions and hardware capabilities of the vehicle into services, and splits the services into smaller interfaces according to the atomic capabilities of the vehicle. The interface of each service component is standardized and encapsulated, and can access each other and expand the combination through the established protocol; the core elements of SOA include loose coupling, standardized definition, software reuse, etc. SOA enables application layer functions to be reused on different models, and can quickly respond to new functional requirements of users based on standardized interfaces. When software engineers modify or add a certain software function, they only need to code the corresponding service components on the upper layer. It does not require recompilation and repeated development of the basic software layer, operating environment layer and other software components, which greatly reduces the complexity and cost of software upgrades and improves efficiency.
In the long run, automobile companies will introduce a large number of algorithm suppliers, software developers, and service manufacturers to jointly build SOA, provide high-quality operating platforms for intelligent automotive software, and provide customers with full-coverage software services. Therefore, major auto companies are gradually shifting their work focus to the cooperative development of SOA, and it is expected that the mass production peak of SOA will be ushered in in the next five years.
Of course, we must be soberly aware that the development cost of an ideal SOA is relatively high. Cross-ECU IPC (Inter-Process Communication) must be more complicated than IPC within an ECU. Additional interface packaging is required, which will increase additional scheduling and computing resources, and these Price and cost do not directly bring about the improvement of user experience. Therefore, an SOA-style architecture will not be achieved overnight.
2. Automotive Software Architecture
Smart car software is divided into three layers, including: 1. The underlying system software layer, including BSP, virtual machine, system kernel, middleware components, etc.; 2. Functional software layer: including library components, middleware, etc., located in the operating system, Above the network and database, the lower layer of the application software provides an operating and development environment for the application software, helping users develop and integrate complex application software flexibly and efficiently; 3. The upper application algorithm software layer, including intelligent cockpit HMI, ADAS/ AD algorithm, network algorithm, cloud platform, etc., are used to actually realize the control of the vehicle and various intelligent functions.
2.1. System software layer—operating system in a narrow sense
Automotive operating system is the bottom layer for managing and controlling hardware and software resources of smart cars, providing operating environment, communication mechanism and security mechanism, etc. According to the degree of transformation of the underlying operating system and the depth of capabilities, it can be mainly divided into the following types:
Basic operating systems: such as QNX, Linux, WinCE, etc., including a new underlying operating system and all system components, such as the system kernel, underlying drivers etc., and some also include virtual machines.
Customized operating system: refers to the in-depth customized development on top of the basic operating system (including modifying the kernel, hardware drivers, runtime environment, application framework, etc.), and finally realizes the cockpit system platform or automatic driving system platform, such as Volkswagen VW .OS, Tesla Version, Google Car Android, Huawei Hongmeng OS, AliOS, etc.
ROM-type automotive operating system: limited customized development based on basic operating systems such as Linux or Android, does not involve changes to the system kernel, and generally only modifies and updates the applications that come with the operating system. Most car companies generally choose to develop ROM-based operating systems.
Super APP: Also known as car-machine interconnection or mobile phone mapping system, it is not a complete car operating system. It uses the rich functions of the mobile phone to map to the car central control to meet the needs of car owners for entertainment. Representatives include Apple CarPlay, Baidu CarLife, Huawei Hicar et al.
2.1.1. Bottom OS: Determines system performance and is the key
operating system kernel for software-defined cars, also known as "bottom OS". It provides the most basic functions of the operating system and is responsible for managing the system's processes, memory, device drivers, files and The network system is the core of the system software layer. Due to the most difficult development and the highest security requirements, its market competition is relatively stable, mainly based on QNX, Linux, Android, and WinCE.
In the long run, the future market will be a situation where QNX, Linux, and Android stand on top of each other. According to statistics from IHS Automotive, the system kernel is currently dominated by QNX and open source Linux and Android, with a combined market share of nearly 90%. In terms of system performance, the three mainstream systems have their own advantages. At present, QNX firmly occupies the first place in the market share of automotive embedded operating systems by virtue of its high security, high stability and high real-time performance. Compared with QNX, the biggest advantage of Linux (including Android developed based on Linux) is that it is open source, has strong customization development flexibility, and strong scalability. In terms of applicable fields, the QNX system is more suitable for areas with higher safety requirements such as instrumentation systems and power systems, while Linux and Android have more advantages in the field of in-vehicle infotainment.
It is expected that the market share of the Android system will continue to increase. Compared with Linux, the Android system is more widely used in China and has a large development space in the field of in-vehicle infotainment, because Android is the best tool for docking mobile Internet content. Despite the problems of poor security and stability, it occupies a mainstream position in China by virtue of its advantages of open source, strong flexibility and rich ecology, especially in the field of in-vehicle infotainment with relatively small safety requirements. Domestic self-owned brands and new car-making forces are also mostly based on Android-customized ROM-type automotive operating systems.
2.1.2. Leading car companies and technology companies have entered the operating system, aiming to establish a leading edge.
The vehicle operating system is the key to the automotive ecology. Leading OEMs and third-party companies are actively deploying automotive operating systems. In car companies, similar to Tesla.OS, Volkswagen Group's VW.OS, Daimler MB.OS, BMW-OS, Geely GKUI, etc., all implement hardware abstraction based on Linux, QNX and other RTOS kernels, forming The middle layer operating system that supports application development, defines the interaction logic of developers, and builds the application layer.
The self-developed operating system helps simplify the vehicle software development process and increase OTA frequency. Taking Tesla as an example, due to the use of the open source Linux self-developed operating system, Tesla can no longer rely on software suppliers, but fully masters the stack by itself. Once a problem is found, it can be quickly corrected and upgraded through OTA to improve user experience. experience. Since the self-developed operating system was first used on the Model S in 2014, Tesla has made several major upgrades to its operating system through OTA technology.
Self-developed operating system and then open vehicle programming to industry chain enterprises can grasp the ecological resources of developers and form a certain monopoly advantage. With the operating system, an ecological monopoly can be established to fully control the components and applications of the upper layer. For example, German Volkswagen self-developed VW.OS, relying on its own annual sales of nearly 10 million cars, forced Tier1, software suppliers and even other OEMs to develop on the basis of VW.OS, making it the IOS in the field of smartphones, and finally formed the " OS licensing fee + Internet of Vehicles service + APP docking license fee + APP value-added service sharing" business model to obtain excess profits.
Third-party automotive operating system players include TINNOVE Wutong (Tencent), Banma Zhixing (Ali), China Automobile, Baidu and Huawei, etc. These companies mainly develop independent operating systems based on mainstream underlying OS. From a technical point of view, Internet and technology companies, relying on their own advantages in software research and development, have a high degree of system transformation and a rich product ecology. Therefore, the products of third-party companies have strong competitiveness, and they can cooperate with car companies with relatively simple ecology and lack of research and development capabilities to form a positive complementarity.
From the perspective of cooperation with car companies, third-party companies are actively cooperating with car companies with many partners. In 2016, Banma Zhixing cooperated with SAIC to launch more than ten models including Roewe and MG equipped with Alios system. In 2017, it cooperated with DPCA. Huawei's super APP system HiCar currently cooperates with more than 20 car companies, including Volvo, Changan, Geely, Dongfeng, GAC Trumpchi, BYD and other brands, and cooperates with more than 150 models.
2.2. System software layer——BSP layer
BSP (Board Support Package) is the interface layer between the kernel and hardware, and it is generally considered to be part of the operating system. BSP mainly includes Bootloader (loading the bootloader of the operating system with basic support code), HAL (hardware abstraction layer) code, driver, configuration document, etc. For a specific hardware platform, the code related to the hardware is encapsulated in the BSP, and the BSP provides a virtual hardware platform upwards, and interacts with the operating system through a defined interface, so that it can run better on the hardware motherboard. Its purpose is to provide a virtual hardware platform for the operating system, making it hardware-independent and portable on multiple platforms.
BSP is relative to the operating system, and different operating systems correspond to BSPs in different defined forms. For example, although the BSP of VxWorks and the BSP of Linux have the same function compared to a certain CPU, the writing method and interface definition are completely different. The OS maintains the correct interface.
Tier1, OEM, and Tier2 manufacturers all participate in the BSP market, and because the development of high-end chip BSP requires a deep understanding of the chip architecture, the current market is dominated by third-party companies that work closely with chip manufacturers, such as Thundersoft. ThunderSoft maintains in-depth cooperation with leading chip suppliers such as Qualcomm, Renesas, Texas Instruments, and NXP. It has a deep understanding of the chips of related companies and can provide BSP technical support for car manufacturers/Tier1 on behalf of chip manufacturers.
2.3. Virtualization layer (Hypervisor)
In order to allow different types of operating systems to run on a computing platform, the most direct technical path is virtualization (Hypervisor). The computer system in an isolated environment. At this time, the supplier no longer needs to design multiple hardware to achieve different functional requirements, but only needs to configure the virtualized software on the main chip of the vehicle to form multiple virtual machines. Running the corresponding software on the machine can meet the demand. Therefore, the vehicle virtualization operating system requires the following three technical requirements: (1) using resource partitioning technology to strictly isolate and allocate resources; (2) having a flexible and efficient real-time and non-real-time task scheduling mechanism; (3) inter-process communication, Implement message communication between virtual machines.
The current mainstream virtualization technology providers are QNX and ACRN. Common hypervisors include QNX of Blackberry, ACRN led by Intel and Linux, XEN represented by Mobica, COQOS of Open Synergy acquired by Panasonic, L4RE of Continental Motors of Germany, VOSySmonitor of VOSyS of France, etc., among which the most mainstream are QNX of Blackberry and ACRN dominated by Intel and Linux, and QNX, as the only virtualized operating system approved to reach the ASIL D level, has been mass-produced and applied to actual models. The entire operating system is a set of processes managed by microkernel scheduling, which guarantees security and real-time performance. At present, the system integrator partners of the BlackBerry VAI project in China mainly include Chuangda, Wuhan Kotei Information, Nanjing ArcherMind Technology, etc.
2.4. Middleware: the key "link" of software development
Before the emergence of middleware, system software was developed directly based on the operating system, resulting in a high degree of coupling between software and hardware. With the increase in the amount of code in the car, the complexity and cost of the system increase sharply. In order to improve the management, portability, tailoring and quality of the software, it is necessary to define a set of standard interfaces, high-quality seamless integration, and efficient Develop and manage complex systems through new models. Middleware separates software and hardware, abstracts and utilizes lower-layer hardware resources, drives chips and optimizes the operating system, provides service interfaces for upper-layer software, and provides different types of plug-ins for different algorithms. Middleware solves problems such as data transmission, application scheduling, system integration, and process management, and can greatly improve the development efficiency of application layer software.
Classic middleware design standard: AUTOSAR. Automotive electronic software standards mainly include AUTOSAR, OSEK/VDX, etc. Among them, the AUTOSAR standard has been developed for more than ten years and has formed a complex technical system and extensive development ecology. It is the mainstream design standard for automotive middleware. AUTOSAR stipulates the layered architecture, methodology and application interface specifications, enabling automotive embedded system control software developers to get rid of the dependence on the hardware system and realize the separation of software and hardware in the process of ECU software development and verification.
The entire AUTOSAR architecture is layered from top to bottom: Application Software Layer, Runtime Environment (RTE), Basic Software Layer (Basic Software Layer, BSW), and Microcontroller (Microcontroller). In order to maintain independence between each layer, each layer can only call the interface of the next layer and provide an interface for its upper layer.
The main advantages of AUTOSAR are: 1. It is beneficial to improve the reusability of software, so that software can be reused across platforms; 2. It is convenient for software exchange and update; 3. Software functions can be defined and verified at the architecture level in advance, thereby reducing Development errors; 4. Reduce the amount of manual code, reduce the burden of testing and verification, and improve software quality; 5. Use a standardized data exchange format (ARXML) to facilitate communication and cooperation between companies.
AUTOSAR is divided into two major platforms, Classic Platform and Adaptive Platform, among which Classic Platform is mainly oriented to distributed ECUs, and Adaptive AUTOSAR is mainly oriented to more complex domain controllers and electronic and electrical architectures of central computing platforms. Compared with Classic AUTOSAR, the advantages of Adaptive AUTOSAR are: strong real-time performance, high operating system portability and more flexible software upgrades.
AUTOSAR is the rules of the game formulated by giants in the traditional automotive industry, and there are currently more than 300 ecological partner companies. At present, there are very few companies in the world that can develop a complete underlying protocol stack based on the AUTOSAR architecture. Currently, world-renowned AUTOSAR solution manufacturers include ETAS (Bosch), EB (Continental), Mentor Graphics (Siemens), Wind River (TPG Capital ), as well as Vector, KPIT (US-India joint venture), etc., most Tier 1 and OEMs need to purchase underlying software from the above-mentioned suppliers. In China, the above-mentioned overseas suppliers of the development tool chain and basic software under the Classic AUTOSAR standard occupy a dominant position. The main domestic suppliers are Neusoft Reach, Huawei, Jingwei Hengrun, etc.; Adaptive AUTOSAR is still in its infancy. Mainland EB and Volkswagen Cooperate to apply AP AUTOSAR and SOA platforms to Volkswagen MEB platform ID series pure electric vehicles. Domestic manufacturers have focused on AP AUTOSAR, launched corresponding middleware and tool chain products, and seized market opportunities.
In the entire AUTOSAR framework, only the Application Layer is really effective for control. Other RTE and lower layers play an auxiliary role. The BSWs corresponding to the functions of different Application layers are basically the same, and AUTOSAR requires software engineers to strictly follow the standards. writing". In this way, under the distributed architecture, the greater the number of controllers, the greater the number of software lines, and the higher the cost of software development. Under the domain control architecture, since the ECU and actuator still follow the AUTOSAR standard, the amount of code cannot be greatly reduced. Therefore, Tesla does not use AUTOSAR, and relies on self-developed operating systems and basic software to achieve more efficient development.
On July 22, 2020, FAW, SAIC, GAC, Weilai, Geely, Great Wall, Changan, Beiqi Foton, Dongfeng, FAW Jiefang, Xiaopeng Motors, Neusoft Reach, Hengrun, Nasen, Horizon, Suzhou Zhitu, 20 companies including Wanxiang Qianchao, Weimax, Remodeling, and Zhongqi Chuangzhi formed the China Automotive Basic Software Ecological Committee (AUTOSEMO), aiming to form basic software architecture standards and interface specifications with independent intellectual property rights led by local companies , share knowledge achievements, and establish an industrial ecology.
2.5. Functional software layer
In the automotive software architecture, the functional software mainly includes the core common functional modules of autonomous driving. The core common functional modules include the general framework of automatic driving, network connection, cloud control, etc., combined with system software, together constitute a complete automatic driving operating system to support the realization of automatic driving technology.
At present, at the level of functional software, traditional Tier1, OEMs, technology giants, and third-party software suppliers all have a certain layout. All parties can provide solutions based on their own advantages and choose modules they are good at for development. For example, algorithm manufacturers with advantages in the field of sensors can focus on the development of sensor modules in functional software to achieve more effective division of labor and cooperation in the entire industry.
In the field of functional software, the model of cooperation with suppliers can help meet the needs of car companies for the development of smart car product functions. Compared with car companies, the software modules developed by major suppliers have been tested in different scenarios and on different products, and the quality is more guaranteed. And some software suppliers can provide more suitable solutions for car companies according to their own characteristics, so as to speed up research and development efficiency. For example, provide functional module solutions for car companies with relatively strong R&D capabilities, open and clean interfaces, and let car companies control the entire user experience and product definition by themselves. For car companies with incomplete software capabilities, they can provide software and hardware integrated delivery solutions, greatly shortening the R&D cycle of car companies.
2.6. Application software layer
The application layer software runs on the generalized operating system and is specifically responsible for the realization of functions. It mainly includes algorithms for autonomous driving, map navigation, in-vehicle voice, OTA and cloud services, infotainment, etc. A typical computing platform, after loading the operating system composed of operating system software and functional software, supports the development of application software, and finally realizes the overall function. The upper application software layer is an area where OEMs focus on research and development to create differentiation, such as cockpit HMI and automatic driving. Therefore, at present, OEMs, traditional Tier 1, start-ups, technology giants, and independent software companies are all actively making efforts in the field of upper-level software.
In the long run, upper-layer applications and algorithms have the greatest value. In the short term, system software such as virtualization technology, system kernel and middleware (AUTOSAR) are crucial if enterprises in the field of automotive software want to truly implement SOA software architecture. But in the long run, after the SOA architecture and extensive operating system framework are mature, the rich upper-layer application ecology and algorithms will have greater value space.
Typical upper-layer application example 1: upper-layer application algorithms in the autonomous driving domain The upper-layer application
algorithms of the autonomous driving domain controller include scene algorithms (covering data perception, decision planning, control execution, etc.), data maps, human-computer interaction (HMI), etc., among which the scene The algorithm is the most complex, typically including three-dimensional algorithms of perception, decision-making, and execution, and then realizes the automatic driving function in various scenarios.
Self-developed autonomous driving algorithms by OEMs have become a future trend. In the short and medium term, due to the lack of algorithm capabilities of most OEMs, OEMs will choose to cooperate with Tier 1 or software companies with obvious algorithm advantages. At this stage, companies such as Bosch, Continental, Desay SV, and Chuangda have obvious advantages . But in the long run, when OEMs have the advantages of talents and data, they will gradually self-develop key algorithms (such as fusion/decision-making algorithms) to eventually extend the full-stack algorithm capabilities. Under such a trend, suppliers should follow the current OEM needs to clarify their own positioning, or focus on a certain field of the autonomous driving application layer, so as to be proficient in algorithms and outstanding capabilities in a certain field in order to maintain sustainable competitiveness.
Typical upper-layer application example 2: digital map
In the field of autonomous driving, the support of high-precision positioning and maps (data maps) is the guarantee for the realization of advanced autonomous driving, and the market scale will continue to expand. High-precision maps can not only ensure that they still function under special weather conditions, but also effectively eliminate some sensor errors and make supplementary corrections to existing sensor systems. In addition, high-precision maps can also build a driving experience database, analyze dangerous areas, and provide drivers with new driving experience data sets.
At present, the main players in the field of high-precision maps are NavInfo, AutoNavi, and Baidu, and the three companies are in a tripartite confrontation. The three companies have their own advantages: Baidu is the first company in China to carry out high-precision map research, and launched the research and development of unmanned vehicle projects in 2013; AutoNavi has the full support of Alibaba and has made rapid progress; NavInfo is an old domestic map provider. In 2020, the three companies will occupy more than 65% of the market share, forming a "three pillars" situation.
3. SDV brings reshaping of the industry chain and analysis of competition pattern
3.1. Reshaping of the industry chain
3.1.1. OEMs try to dominate the software part of the vehicle, and software manufacturers advance to Tier1. OEMs
begin to deeply participate in software development. In the traditional automobile industry chain, the development of each system software is almost completed by Tier 1/2, and the black box is supplied to the OEM. The OEM is only the definer of the overall architecture, responsible for the definition of design and management system concepts, and finally completes system integration. And the work of inspection, that is, the yellow part in the figure below. As the importance of automotive software continues to increase, OEMs have begun to attach importance to defining software, deeply participating in system architecture, functional requirements analysis and other links, and even leading the design and development of software units, which are the green and red links in the figure below.
Software suppliers advance to Tier1. Regardless of whether OEMs establish cooperation with core software companies or conduct independent research and development, the traditional supply chain relationship will undergo fundamental changes. The cooperation between car companies and software suppliers will be further deepened. In order to gain control and reduce high R&D costs, OEMs will choose to directly cooperate with software suppliers with strong independent algorithm research and development capabilities. Leap to become a Tier1 manufacturer, breaking the tower-shaped supply model in which software manufacturers as Tier2 first supply to Tier1 and then to OEMs, and develop into a flat supply network model.
For software suppliers, with the continuous strengthening of OEM autonomy and software self-development capabilities, OEM OEMs have begun to seek direct cooperation with software suppliers. For example, OEMs will first seek to recover the functions of the cockpit HMI interactive system. Application software such as UI/UX design tools, speech recognition modules, sound effect modules, and face recognition modules directly purchase software authorization from software suppliers, thereby bypassing the traditional Tier1 and realizing independent development. For software suppliers, the more software IP product portfolios they can provide, the higher the value per vehicle may be. At the same time, software suppliers are also seeking to enter the traditional Tier1-controlled hardware design and manufacturing links, such as domain controllers (Thundertech), TBOX, etc., to provide diversified solutions.
3.1.2. Software development is evolving towards layering and modularization, and there are great market opportunities for intermediate module suppliers.
From traditional V-shaped development to layered development for software-defined vehicles, the basis for the transformation is the introduction of middleware. The complexity of the application algorithm under the domain control architecture is high, especially in the field of autonomous driving, almost no company can provide a complete set of software systems, so multiple suppliers are required to cooperate to complete the complete set of software packages. Due to the introduction of middleware, software and hardware The decoupling provides the possibility for the introduction of third-party software, and the third-party software needs to be carried on the functional interface provided by the middleware. This requires the automaker or supplier responsible for system integration to have very strong software system architecture capabilities and middleware platform design capabilities.
From the perspective of the development model of the software itself, the layering and modularization of software is an inevitable event when every industry is reshaped by software. Taking the Android and ios platforms in the mobile phone field as examples, each platform ecosystem has very rich development tools and basic software modules. There are many types of common SDK (Software Development Kit) in the Android ecosystem, and most of the modules are developed by third parties. Provided as a standalone module.
The industrial division of modularization in the mobile phone field has brought great market opportunities to standard module suppliers. Representative listed companies include Twilio and Salesforce. As a cloud communication company, Twilio provides an application programming interface that can help developers easily add SMS, voice and VoIP functions to other applications. According to the overall service framework of CPaaS (enterprise communication platform and service), CPaaS can be divided into five levels. Twilio can provide some of these content services, including basic modules such as SMS, Voice, and Phone Number that many users are concerned about. They are general APIs commonly requested by customers at present, and constitute more than 90% of the current CPaaS revenue. The demand for company webpage communication, RCS, email, video, etc. will achieve explosive growth in 2020. More and more enterprises and users accept email and omni-channel communication. In the future, IOT, AI, telemedicine, payment, biological The demand for modules such as identification security will continue to grow rapidly. The company's business revenue has grown with the growth of supported third-party applications.
Since its establishment, the company's revenue has continued to maintain rapid growth. In 2021, its revenue has exceeded 18 billion yuan, and its market value was once as high as 80 billion US dollars.
Under the trend of "software-defined cars", the automotive software industry will also usher in stratification and modularization, and a number of specialized intermediate module supply companies will be born. It is difficult for car companies to complete the research and development of the entire chain of software modules, while standard third-party modules will undergo more different scenarios and different product tests, have higher quality and vitality, and can also share and reduce the cost of each module. For example, in the field of cockpit, Meijia Technology has seized the development trend of industry division of labor, provided a digital cockpit solution for Li Auto, and opened a clean interface, allowing GAC NIO and Li Auto to control the entire user experience and product definition. The core functional modules in the smart car software functional system also involve multi-party cooperation, including the general framework module for autonomous driving, the network connection module, the cloud control module, the AI and vision module, and the sensor module. Using these common functional modules, developers can conduct research and development at the autonomous driving business level more efficiently. Suppliers can rely on their own advantages to provide solutions for car companies to achieve more efficient division of labor and cooperation.
3.2. Competitive landscape? ——Several dimensions to see the growth of software companies
3.2.1. Cutting into core ecosystems such as chips, operating systems, or middleware Auto
companies have shifted from building ecosystems around Ter1 in the past to building ecosystems around chips, operating systems, or middleware companies. In the era of traditional automobiles, the ecological chain of automobiles revolves around Tier1. Tier1 purchases chips and integrates software and supplies them to OEMs. In the era of smart cars, when car companies define a smart car, their core considerations will turn to judging how much computing power a chip is needed to support the expected intelligent functions. After finalizing the chip, peripheral components, operating systems, and software It will also choose among the related companies supported by the chip. So far, car companies will build an ecology around the Nvidia ecosystem, Qualcomm ecosystem, and Horizon ecosystem. Of course, there are also car companies that build ecosystems around operating systems, such as QNX, LINUX, Android, etc., or build ecosystems around middleware systems such as Apollo and AUTOSAR.
Therefore, if software suppliers can penetrate deeply into the above-mentioned core chips, operating systems, and middleware systems, they will be able to fully enjoy the explosion of the automotive software market. Let's take Chuangda as an example: (1) Chip field: deep binding with Qualcomm
Qualcomm is the leader in the field of smart cockpit chips. 1. In terms of automotive chips, Qualcomm's third-generation smart cockpit chip (Snapdragon 8155) is the first car-grade digital cockpit SoC built with a 7nm process technology. Its multi-core heterogeneous performance is twice that of other chips, and its CPU performance And GPU computing power is also much higher than other manufacturers' chips, and it is currently the most powerful cockpit SoC chip. At present, more than 20 of the world's Top25 car companies have produced models equipped with 8155 chips, and the market is very strong. 2. In terms of chip platforms, Qualcomm has also released products such as Ride, the fourth-generation Snapdragon platform, and 5G Internet of Vehicles. The fourth-generation chip SA8295P launched last year uses a 5nm process. Compared with the third-generation chips and The chips of other manufacturers have improved by an order of magnitude, and the GPU computing power has also increased by more than 50% compared with the third generation. The layout of the entire product line is planned, and the integration of the future cockpit domain and the self-driving domain is attempted.
Chuangda is deeply bound to Qualcomm and continues to benefit from Qualcomm's customer attraction. Since its establishment, Chuangda has formed a strategic cooperation with Qualcomm, which produces mobile phone terminal chips, in the mobile phone industry, specifically including the joint development of QRD (Qualcomm Reference Design) mobile phones, and the establishment of a mobile phone chip-system joint laboratory. Therefore, Chuangda is much more familiar with the Qualcomm chip platform than automotive software companies, which also laid the foundation for the priority cooperation between the two on the automotive side. In addition, Chuangda and Qualcomm established a joint venture company - Chongqing Chuangtong Lianda Intelligent Technology Co., Ltd. in 2016 to provide customers with a one-stop solution of "Qualcomm chip platform chip + operating system + core algorithm intelligent products or software" plan, and further carry out in-depth binding at the equity level. To sum up, Chuangda has integrated into Qualcomm's chip ecology, and Qualcomm customers have attracted customers for Chuangda, providing guarantee for the stable and continuous growth of Chuangda's smart cockpit business; at the same time, the long-term cooperation with Qualcomm has also deepened Chuangda's virtualization of chips , firmware drivers and other links of technology accumulation.
Thundersoft will optimize the middleware mitigation of the operating system, core SDK, and framework of Qualcomm's integrated hardware and software platform. Chuangda has a better understanding of car-level requirements at the middleware level and has accumulated mass-produced car-level operating system optimization technologies, which can help Qualcomm connect with downstream car companies, understand the customization needs of car companies, and provide car solutions based on Qualcomm's different generations of chips. Enterprises customize and develop OS, and make individual adaptations and improvements for different models. Therefore, deep binding with Qualcomm is Chuangda's core advantage over other software vendors.
The fusion of future cockpit and autonomous driving will bring Chuangda a broader display space. The integration of the cockpit domain and the self-driving domain will bring rapid growth in software complexity, and multiple systems, virtual machines, and various middleware will be used. The system level will integrate non-real-time operating systems and real-time operating systems. The complexity of the system has increased significantly, driving the value of software in the entire industry chain to further increase. The specific content provided by Chuangda in the RIDE platform mainly includes: chip virtualization; security middleware (need to encapsulate many underlying hardware resources, access services, access control, and provide unified service interfaces and permissions for the upper layer), Ensure real-time access to tasks, and access to algorithms or source data, etc.
(2) Operating system field: long-term deep cultivation of operating system
The company has been deeply involved in the underlying technology of the operating system for many years, and has a deep understanding of the tailoring and customized development of various operating systems. It is widely compatible with QNX, LINUX, and various virtual machine manufacturers, and has rich experience. 1. Chuangda is the solution provider of QNX (Blackberry) automatic driving system. BlackBerry has successively developed system platforms such as entertainment systems, smart cockpits, and assisted driving for automobile companies, providing developers with flexible tool choices. Thundersoft has successfully established an ecological partnership with QNX because of its strong core capabilities at the lowest level in the operating system, and has become one of QNX's solution providers in the field of autonomous driving. 2. Chuangda integrates into the Android ecosystem, further enhancing the advantages of the smart cockpit. Compared with QNX, the biggest advantage of Linux is that it is open source, and it has a strong flexibility in custom development. The Android system is an open source operating system developed by Google based on the Linux kernel. It is mainly used in the field of in-vehicle infotainment and navigation. Currently, the Android system occupies a dominant position in the field of in-vehicle infotainment systems in China. As the leader of the smart cockpit and the master of the Android ecosystem, the mutually beneficial cooperation between Chuangda and Android system providers has further expanded the advantages of Chuangda in the field of smart cockpits.
(3) Middleware field: Middleware capability is the underlying capability
of Chuangda. Chuangda has the ability to develop AutoSAR AP. Chuangda has built a "cloud-pipe-device" integrated SOA platform to provide a development tool chain for software development and vehicle integration. At the same time, Chuangda has also joined the Apollo ecosystem to carry out multi-dimensional in-depth cooperation. Chuangda provides a series of services and solutions at the software platform layer, such as customization and optimization of operating systems, development of intelligent driving algorithms, and customization and optimization of human-computer interaction interfaces.
3.2.2. Deep binding with downstream customers and formation of long-tail coverage The
automotive software market has long-tail market characteristics, that is, wider customer coverage means better performance sustainability and stronger product versatility. Therefore, Manufacturers with a wider distribution of downstream customers and lower customer concentration will have a certain advantage in the competition. At present, Zhongke Chuangda, Neusoft Reach, Wuhan Kotei, Jingwei Hengrun and other enterprises have rich domestic and foreign customer resources, and their products have covered most mainstream models.
Chuangda's customers cover almost all mainstream OEMs and Tier1, and the customer concentration is much lower than that of comparable companies. Compared with the other three manufacturers, Chuangda's top 5 customers' revenue accounted for 29.6% in 2020 and 26% in 2021, far lower than Kotei's 53% and Jingwei Hengrun's 52.7%. It can be seen that Chuangda The income structure of Chuangda is more dispersed and less dependent on a single customer, which means that Chuangda's operation is relatively more stable. In addition, since Chuangda is bound to Qualcomm, Chuangda has further absorbed Qualcomm's domestic and foreign customers. Therefore, Chuangda has better performance continuity and stronger product versatility.
3.2.3. Revenue model: "License+Royalty" has higher barriers and income is more stable than NRE. The
business model of automotive software generally adopts the model of "IP+solution+service". There are three main software charging modes: the first one is to receive outsourcing and provide a one-time quotation in advance. The NRE (Non-Recurring Engineering) mode is based on the number of developers and working hours required for the project. It is usually used for software and system development. business; the second is to sell IP and software development licenses for customers, and the specific charging standards include Royalty and License. The gross profit rate under this model is relatively high, usually exceeding 70%; the third is the package of "NRE+License" One-time fee is charged, and another fee is charged according to the number of bicycles.
According to the data of Zuosi Automotive Research Institute, the current IP authorization fee for a bicycle is 2,000-3,000 yuan. With the increase in the complexity of automobile functions and the rise of related software companies, the software authorization fee will continue to increase.
Chuangda sells IP and customer software development licenses, occupying a leading position in the industry. The vast majority of domestic automotive software vendors provide purely outsourced software development business and solution services, and business fees are usually charged in the form of projects or one-time NRE. The market barriers of the NRE model are relatively low, and automotive software vendors usually only need to provide technology and manpower. When the industry is in a stage of rapid growth, this model can achieve rapid growth through personnel expansion. Sustainability of outsourcing projects is doubtful, and income stability is low. Relatively speaking, the technical difficulty of IP licensing is more difficult. Only enterprises with leading technologies in subdivided fields can support the software licensing model. Their income mainly depends on the customer's car sales. This model is conducive to maintaining customer viscosity and income stability. At present, only Chuangda and Neusoft Group can rely on IP fees. Among them, Chuangda’s IP revenue accounts for a higher proportion, mainly from the licensing of IP such as Kanzi. With the continuous increase in the number of models equipped with 8155 cockpit chips, ROYALTY’s revenue in 2022 will increase. larger growth.
3.2.4. There will be differences in the market size between focusing on a single field and cross-segment fields
. Under the tide of intelligence, automotive software companies can have stable income and market size if they focus on a single field, but in the long run, companies that focus on a single field will grow In the future, with the product iteration and technological progress of various enterprises in the industry, it will be difficult for various enterprises to form differentiation, and there will be limitations in their competitive advantage. Therefore, compared with companies focusing on a single field, companies with a wider software business coverage and more fields have better growth potential in the long run.
Chuangda has acquired a number of upper-level companies, and the smart car business has achieved cross-field development. Chuangda has been on the market for many years, and has established a complete platform in the field of smart cockpits, forming a brand effect, but the company is not satisfied with the development of the underlying OS field, so it gradually expands to the upper (application layer) field. Since 2016, Chuangda has acquired many companies in different fields, such as Aipu Xinsi and Rightware, and the company's business has also been fully expanded and strongly improved. As a result, Chuangda has become a rare operating system technology company at home and abroad that fully covers the chip layer, system layer, application layer, and cloud. In high-growth fields such as IVI technology, HMI interface design, image processing algorithms, and automatic parking Occupy a place, but other companies in the industry have not yet expanded into these fields or lack corresponding customers, which is less of a threat to Chuangda. Therefore, through mergers and acquisitions, Chuangda leads the industry in the field of cross-automobile segments, which will help Chuangda to expand the market and have unique advantages in multi-field markets.
Rightware's Kanzi software is a key addition to the smart cockpit technology that enabled ThunderTech's breakthrough. Kanzi has a very strong technology in the car infotainment system, and there are many advantages in Kanzi products. (1) Kanzi's special effects such as particles and PBR are excellent, reaching the special effect level of end-game games, and can better restore the real world; (2) Support millisecond-level quick start and MCU chip support, support for multiple development languages, and Android The system is seamlessly connected, and at the same time, the memory usage is extremely low; (3) It can be used quickly, and part of the technical design can be completed without writing code, which greatly improves efficiency; (4) It meets the C language development standard and meets ASPICE and other car-level quality Safety verification; (5) Kanzi currently uses more than 100 models, has completed the delivery of one million car software, and has more than 100 original factories around the world to provide technical support and have mass production capabilities. Kanzi has become the preferred tool for HMI design of more than 80% of domestic car companies. With the help of Kanzi's technical strength, Chuangda has packaged Kanzi and other core capabilities, and implemented an IP authorization model to charge software license fees. Therefore, through the acquisition of Rightware, Chuangda has given priority to occupying the smart car segment market, and at the same time broke through the "License+Royalty" barrier, improving its own bargaining power in Tier1.
The company's business structure is driven by three wheels. In addition to cars, there are also "mobile phones + IoT". In 2021, the company's mobile phone/q car/IoT revenue will be 1.055 billion/1.2 billion 1.27 billion respectively. The mobile phone business has maintained a steady growth rate all year round under the background of declining terminal shipments. The IoT business has shown explosive growth. The revenue in 2021 will increase year-on-year 82.87%. The main markets of Chuangda's IoT business are robot market, drone market, video conference (video conference), smart camera/AI camera, VR and other emerging markets, and it is expected to show explosive growth in 2022.
3.2.5. Attachment: Financial data of key software companies compared with
per capita income: the number of employees of Chuangda has increased significantly, and per capita income in the next few years is not the goal of Chuangda. Chuangda’s per capita income is lower than that of Neusoft Group. The per capita income is declining year by year, and the income in 2021 will be about 360,000 yuan, which has not yet exceeded 400,000 yuan. The main reason is that Chuangda is currently focusing on the research and development of automotive software, and the number of employees will continue to increase significantly in the next five years. By attracting more excellent product managers and ecological builders, the company's business channels will be broadened. Therefore, per capita income will not be the core KPI of Chuangda in the next five years.
Profit: Chuangda’s profit margin is in a leading position in the industry. In 2021, the profit rate will fluctuate due to foreign exchange. In the vertical comparison, the gross profit rate of Chuangda has stabilized at around 40% in recent years, and the net profit rate has basically remained in the range of 12-18%. From 2018 to 2020, the gross profit rate and net profit rate will increase steadily, and in 2021, it will be reduced to a certain extent due to the impact of exchange rate fluctuations; compared horizontally with peers, Chuangda's gross profit rate will be 44.22% in 2020, and the net profit rate will be 17.11%, with a slightly lower profit margin. In Kotei information, but still far higher than the average level of automotive software manufacturers. Chuangda's high profit margin is mainly due to its superior performance in terms of SDV, ecological cooperation, customer breadth, and IP licensing analyzed above.
R&D expenses: Chuangda's R&D expense rate is in a leading position.
Zhongke Chuangda invests more in R&D, which has the potential to build software technical barriers. Chuangda's R&D expense ratio (R&D expenses/total revenue) has been maintained at more than 15% in recent years, and R&D expenses accounted for 150% of comparable companies such as Neusoft, Kotei, and ArcherMind. The number of R&D personnel in the company is increasing year by year. In 2020, the proportion of R&D personnel will be as high as 92%, ranking in the first echelon in the industry. With the development of automobile intelligence, Chuangda has invested more energy in the high-tech research and development of intelligent cockpit, automatic driving and other modules and the application of the latest technology, which is conducive to the formation of product differentiation and the construction of automotive software technical barriers. In the long run, Chuangda focuses on the research and development of new technologies and products to form technical barriers, which will greatly help the company's future market expansion and further revenue growth.
3.3. How do you view self-research by car companies?
Software-defined cars have become an industry consensus, and car companies will definitely increase their investment in the software field. When making a decision on self-research or outsourcing, there are many fields involved, such as which stages of the development process are self-research and which are outsourced? Is the operating system self-developed in layered technology? Or which domain is self-developed? Every company is different and can make choices based on its own resource endowment and strategic goals.
Leading new power car companies develop their own research, while other car companies tend to purchase basic software and self-develop application software.
Top new power companies choose software for self-development. At this stage, the full-stack self-developed companies are only leading new power companies. Such car companies have strong demands for differentiation, and at the same time have sufficient financial strength and software talents to support them. Wei Xiaoli adopts self-developed methods in the fields of autonomous driving perception layer, decision-making layer algorithm and basic software, in order to pursue leading in automatic driving functions and create differentiated selling points. In the field of cockpit, Xiaopeng P7 (Qualcomm 820A) and Xiaopeng P5 (Qualcomm 8155) rely on Chuangda to develop the underlying software, but when it comes to the Qualcomm 8295 generation of cockpit chips, Xiaopeng began to directly connect with Qualcomm to improve efficiency. Strive for market leadership.
Mainstream domestic and foreign car companies are very cost-sensitive, and tend to purchase general-purpose, standardized hardware platforms plus basic software, and self-develop upper-layer application software to seek differentiation. Generally speaking, automakers are better at defining user scenarios and using user data than system-oriented capabilities. Therefore, as the underlying hardware becomes more and more concentrated, more and more complex software is needed to make up for it, the value of the software will increase, and the market opportunities for manufacturers such as Thundersoft will continue to increase. For long-tail OEMs with insufficient resources, packaging services for hardware, underlying software, and application-layer software are required.
The top new force models are generally above B-level, and the price is above 250,000 yuan, which is not the main sales force
At present, the prices of models of new power companies that choose full-stack self-developed vehicles are generally above 250,000 yuan, and the grades are generally B-level and above. Looking at the entire domestic passenger car market, only 24% of the models in 2021 will be priced at more than 250,000 yuan. The sales volume of B-class and above models accounted for only 34%, and the main domestic sales force were B-class and below models below 250,000 yuan. Looking at the domestic market shares of Tesla and Wei Xiaoli alone, they will only account for 2.9% in 2021.
In addition, according to the information disclosed in Chuangda’s 2021 financial report performance briefing, the number of major customers (with revenue of more than 2kw) will increase from 4 in 2021 to more than 22 in 2022, covering GM, Ford, Great Wall, SAIC, International and domestic mainstream car companies such as Geely, Volkswagen, Toyota, BYD, etc., reflect that Chuangda's market share among these customers is constantly increasing, and these customers are increasingly dependent on Chuangda. Therefore, we believe that in addition to leading new power companies that have the ultimate pursuit of differentiation, mainstream car companies will use the form of industry division of labor to promote software-defined cars.
It is the general trend for automotive software to evolve towards layering and modularization.
Referring to the era of PCs and mobile phones, automotive software will also form a layered and modular division of labor. In the early stage of the industry, various technologies are not mature, and car companies will choose to invest, but after investing to a certain limit, their investment will decline. In addition, full-stack self-research is very uneconomical. Standard third-party modules will be tested in more different scenarios and different products, and their own vitality and quality will become higher and higher. By apportioning them under many scenarios, The cost will also be lower. From a business point of view, suppliers will definitely be selected, and the industry will have a process of division of labor.
