Research on the Development of Multi-domain Electronic and Electrical Architecture Technology for Intelligent Connected Vehicles

Abstract

With the rapid development of vehicle intelligence and connectivity technologies, the traditional electronic and electrical architecture can no longer meet the new demands of the integrated development of vehicles, roads, clouds, and networks in the future. This paper focuses on the multi-domain electronic and electrical architecture of future intelligent connected vehicles, providing a detailed review of existing technologies from four aspects: overall design, hardware systems, communication systems, and software systems, and looks forward to the development prospects of electronic and electrical architecture in China.

Introduction

With the rapid development of vehicle electrification and intelligence, the deep integration of the automotive industry with mobile computing, ubiquitous vehicle networking, artificial intelligence, and other ICT technologies has accelerated, triggering a new wave of automotive digitization and software-defined vehicles. This has given birth to a new intelligent transportation system capable of achieving integrated operation of “human-vehicle-road-cloud-network”, which is expected to greatly enhance the capacity, energy efficiency, safety, and driving experience of future transportation systems. Intelligent connected vehicles (ICVs) have become an inevitable trend in the automotive industry’s upgrade within the integrated system of “human-vehicle-road-cloud-network”. ICVs are equipped with intelligent perception systems, intelligent decision-making and control systems, and intelligent execution systems, closely integrated with communication networks and artificial intelligence, enabling information exchange between vehicles and various domains (vehicles, roads, pedestrians, clouds, etc.). ICVs serve as the physical carriers for the transformation of automobiles from traditional transport tools to the next generation of intelligent terminals, posing new challenges and requirements for the foundational design theories and methods of automotive electronic and electrical architecture (referred to as E/E architecture). The E/E architecture technology, as one of the design technologies for ICV systems, has a decisive impact on the integration of the entire vehicle’s software and hardware systems, functional realization, development costs, and overall vehicle performance.

The E/E architecture of vehicles is defined as the organizational structure of automotive electronic and electrical components that realize the vehicle’s functions and its hardware and software systems. It emphasizes the interaction and interdependence between components and the overall vehicle environment, as well as the principles that guide design and evolution. As the top-level design of the ICV system and its functional composition, existing E/E architectures face some shortcomings. How future E/E architecture designs should meet the complex needs of ICVs and adapt to new technological trends is an important issue of concern in the automotive field.

The design technology of ICV’s E/E architecture faces the following challenges:

(1) In overall architecture design, the existing experience-based design processes are difficult to support high-precision design throughout the entire development cycle. There is a need to establish model-based design theories and evaluation systems, guided by diversified requirements, to strengthen the comprehensive matching of architecture’s software and hardware, functional safety, data security, and information security design.

(2) In hardware system design, ICV-specific intelligent controllers should be designed in conjunction with vehicle function design to achieve high computing power and low energy consumption; optimizing power supply systems and wiring system design concepts to reduce overall vehicle costs and weight.

(3) In communication system design, the current communication mechanisms are inadequate to meet the rapidly growing data transmission demands, necessitating the design of high-bandwidth, strong real-time, and low-latency jitter vehicle communication mechanisms, enhancing the configurability of communication networks and the scalability of multiple communication protocols.

(4) In software system design, the differentiation of software functions and rapid iterations will become core competitive advantages. Software-defined vehicles (SDVs) and service-based software design concepts will become the cornerstones of system software design, where designing decoupled, upgradeable, easily configurable, highly secure, and personalized software will become the main battlefield for the competition among vehicle enterprises.

The above challenges pose significant demands on the development of E/E architecture technology, and how to guide the further development of E/E architecture technology is a major issue that needs to be addressed in ICV architecture design.

This paper conducts an in-depth analysis of the key technologies in the study of multi-domain E/E architecture for ICVs from the perspectives of overall architecture design, hardware systems, communication systems, and software systems, and looks forward to future development trends.

Status of Multi-domain Electronic and Electrical Architecture Technology

Based on the degree of computing power concentration, this paper divides E/E architecture into distributed architecture, domain-centric architecture, and centralized architecture, discussing the characteristics of each architecture in detail.

1.1 Distributed Architecture

Distributed E/E architecture is divided according to the different functions of the vehicle, with each electronic control unit (ECU) designed based on specific functional requirements. In this architecture, various ECUs communicate via CAN bus to achieve the functions of the entire vehicle. A typical hardware topology is shown in Figure 1. In this architecture, each ECU is responsible for the implementation of a single function, and a vehicle typically has hundreds of ECUs, which not only directly drive actuators and sensors but also undertake complex business function control logic. The hardware and software of this architecture are tightly coupled, and every time a new function is added, corresponding ECUs and communication signals must also be added. However, due to limited computing power of ECUs, restricted communication bandwidth, and difficulties in function upgrades, this architecture faces bottlenecks that hinder architectural upgrades and affect vehicle safety performance. Moreover, as the number of ECUs increases, the wiring harness inside the vehicle also becomes longer, increasing the overall weight and cost of the vehicle, while causing significant difficulties for the layout and assembly of the entire vehicle.

Research on the Development of Multi-domain Electronic and Electrical Architecture Technology for Intelligent Connected Vehicles

Figure 1 Distributed Architecture

1.2 Domain-Centric Architecture

With the widespread application of on-board Ethernet and the low-cost high-performance chips, domain-centric architecture has gradually overcome the challenges of distributed architecture in terms of safety, scalability, and more. The basic idea of domain-centric architecture is to cluster the functions of multiple electronic control units (ECUs) based on functionality, deploying only a few domain controllers (DCUs) as the main control units in the vehicle. A typical centralized domain architecture based on a central gateway is shown in Figure 2. Each DCU is responsible for data processing and functional decision-making within its domain and manages the control of sensors and actuators subordinate to that domain. Domains exchange necessary data through a central gateway, ensuring communication and interoperability between domains while achieving information security and functional safety.

Figure 2 Domain-Centric Architecture

Compared to traditional ECUs, DCUs possess powerful hardware computing capabilities and rich software interface support, allowing more core functional modules to be centralized within the DCU, thus improving the system’s functional integration. The role of individual ECUs is weakened, and complex data processing and control functions are unified within the DCU, with ECUs gradually evolving into executors of DCU commands. In terms of communication, Ethernet becomes the backbone of inter-domain communication, significantly improving communication rates. Thanks to the decoupling of hardware and software, interface standardization, and the upgrading of signal performance, domain-centric architecture represents a watershed in architectural design thinking, shifting from a signal-driven model to a service-oriented architecture (SOA). In domain-centric architecture, software and hardware can be feasibly decoupled, reducing system coupling. Remote software upgrades (OTA) and hardware deployments become more convenient, while standardized interfaces allow sensor and actuator modules to be produced without being tied to specific ECUs, thus supporting the standardized production of components.

1.3 Centralized Architecture

To reduce the complexity of structural connections within the vehicle, improve the utilization of computing power, lower costs, and enhance safety, centralized architecture further integrates multiple DCUs from domain-centric architecture into one or more central computing platforms (CCPs) with stronger computing power, combining multiple operating systems. On-board sensors and actuators are no longer deployed based on functionality but are divided by physical location for nearby access to zone controllers (ZCUs). The typical topology of centralized architecture is shown in Figures 3 and 4. In this architecture, each data acquisition and execution node transmits raw data to one or more CCPs for processing via ZCUs, with all data processing and decision-making completed in the CCP. ZCUs are primarily responsible for data acquisition, communication protocol conversion, and data transmission functions. Multiple ZCUs form a ring network through Ethernet, further enhancing communication redundancy and reliability. By allowing sensors and actuators to access nearby zones, the configuration layout is simplified, and wiring harness length is shortened. As shown in Figure 4, this architecture centralizes all vehicle control computing functions into a single CCP. However, from the current technological capabilities, the multi-CCP architecture represented in Figure 3 demands higher standards in hardware design, software development, and safety redundancy, thus single CCP architecture is the mainstream solution at present.

Figure 3 Multi-Central Computing Unit Centralized Architecture

Figure 4 Single Central Computing Unit Centralized Architecture

In summary, the development of E/E architecture from distributed architecture to domain-centric architecture and then to centralized architecture brings the following advantages:

(1) Centralization of computing power improves utilization. During actual operation, most of the time, only a portion of the chips are executing computing tasks, leading to the idle state of the computing power of dispersed independent function ECUs. Adopting a centralized computing architecture can maximize the utilization of processor computing power under comprehensive circumstances.

(2) Unified interaction achieves vehicle functional collaboration. In traditional distributed architectures, actuators, sensors, controllers, software algorithms, etc., are all tightly coupled designs, resulting in low efficiency in design and development across components and ECUs, and difficulties in upgrades. Centralized architecture provides a basis for decoupling software and hardware, reducing the number of ECUs, achieving true vehicle-level characteristic development, facilitating rapid iteration and market launch, and significantly lowering development and upgrade costs.

(3) Shortens the length and weight of the vehicle wiring harness, reducing failure rates. Traditional distributed ECUs lead to long and complex wiring harnesses, causing electromagnetic interference and high failure rates. Centralized architecture, by enabling regional access for actuators and sensors, shortens wiring harness lengths and reduces overall vehicle weight.

(4) Lays the foundation for decoupling software and hardware, supporting software-defined vehicles. Distributed architectures tightly couple software and hardware, making decoupling difficult, while centralized architecture centralizes functionality and computing power, providing conditions for software-hardware decoupling and layered software.

(5) Vehicles are easier to platform, with enhanced scalability. Under centralized architecture, the functionality of ECUs is weakened, and the interfaces of sensors and actuators become standardized and generalized. Domain controllers and zone controllers can be configured and adjusted according to needs to accommodate different sensor and actuator schemes.

Overall Design Technology of Multi-domain Electronic and Electrical Architecture

2.1 Main Tasks of Overall Architecture Design

Traditional automotive electronic and electrical architecture design mainly focuses on the reasonable layout of components to achieve optimal performance and minimum cost. However, multi-domain electronic and electrical architecture must not only meet traditional goals but also serve as the infrastructure for the software and hardware of intelligent connected vehicles, supporting the functions and performance of automotive systems. The main tasks of ICV multi-domain electronic and electrical architecture design include:

(1) Reasonably dividing the functions of each subsystem according to vehicle functional requirements, clarifying the logical connection relationships between functions, and achieving software-hardware mapping.

(2) Weighing factors such as functional interaction, cost, and power supply and distribution, designing hardware spatial topology, connection topology, and communication topology.

(3) Forming a multi-dimensional overall system design scheme integrating controllers, sensors, processors, wiring harnesses, functional software, etc.

(4) Ultimately reducing system redundancy and improving system verifiability, high integration, high safety, and scalability.

2.2 Overall Architecture Design and Evaluation Methods

The complexity and diversity of ICV function configuration have triggered changes in the theories and methods of electronic and electrical architecture design. Currently, model-based systems engineering (MBSE) methods for automotive electronic and electrical architecture design and development are gradually gaining attention. MBSE expresses the design of electronic and electrical architecture in the form of models from the initial stages, providing unambiguous graphical descriptions, analyses, and designs of various complex system requirements, structures, and behaviors, thereby establishing a unified communication platform among relevant designers. The MBSE method can address issues of engineering data inconsistency, verifiability, and traceability in the development process of the vehicle’s electronic and electrical architecture, reduce the difficulty of overall vehicle product development, identify and avoid potential risks early, enhance development efficiency, and lower development costs and later maintenance costs. Figure 5 demonstrates the V-shaped design development process based on MBSE.

Figure 5 V-shaped Design Development Process

The design of electronic and electrical architecture is one of the core tasks of overall vehicle design, while the evaluation of electronic and electrical architecture serves as a direct reference for optimizing architectural schemes. Based on the current mainstream development design processes of electronic and electrical architecture and the requirements of electronic and electrical architecture for ICVs, the key contents of the overall design of multi-domain electronic and electrical architecture primarily include the following five aspects: architecture requirement definition, architecture functional design, architecture topology design, architecture system design, and architecture analysis and evaluation.

2.2.1 Architecture Requirement Definition

Whether for traditional or multi-domain electronic and electrical architecture development, a comprehensive requirement analysis must be conducted from the perspective of market demand. Based on analysis and evaluation, the architecture requirement definition needs to determine the goals for functional scheme realization, formulate overall vehicle requirements for the developed vehicle model, clarify the requirements of the overall vehicle system and each subsystem, and simultaneously develop the overall vehicle verification testing specifications. By conducting requirement analysis, development goals and constraints are identified, serving as the starting point for the entire architecture design.

2.2.2 Architecture Functional Design

Based on the defined requirements for the architecture, the overall functional design of the architecture is completed. To reduce the complexity of the electronic and electrical architecture, the overall functions need to be subdivided and segmented, and software and hardware should be decoupled. A commonly used functional design method is to first divide the overall vehicle functions into primary functional domain levels, and then conduct detailed secondary functional division of the functional domains to transfer controller functions from the secondary network to the domain controllers, providing a foundation for the realization of higher-level functions. In the functional architecture design phase, preliminary network topology, electronic and electrical schemes, subsystem technical specifications, and functional scheme design work need to be completed.

2.2.3 Architecture Topology Design

Based on the functional architecture, the basic topology structure of the architecture is extracted, including hardware topology architecture, connection topology architecture, and communication topology architecture. Through the refinement and optimization of the topology architecture, the optimal topology scheme is output, providing design specifications for software and hardware development for other design departments. The hardware topology architecture mainly involves the overall installation layout of hardware components, internal composition, and detailed information on external interfaces, including the combination relationships between components and internal details of components. The connection topology architecture describes the logical connection methods and implementation status between components, including specific wiring and cable connection methods, as well as the internal structure of fuse relay boxes. The communication topology completes the networking and protocol determination of the communication network based on different communication needs between domains and within domains.

2.2.4 Architecture System Design

Based on the previously established power distribution diagram, grounding points, overall vehicle layout, and interface control files provided by suppliers, architecture system design needs to complete the design of overall vehicle principles, interface definitions, and functional specifications, and establish an overall architecture model. With the support of topology layer information, existing development databases, and experiential input, correct logic and algorithm definitions can be achieved. The solution for the system-level electronic and electrical architecture and the system-level verification testing specifications are formulated. Ultimately, the functions are issued and updated to the product component design for implementation and verification.

2.2.5 Architecture Analysis and Evaluation

Traditional vehicle electronic and electrical architectures find it difficult to achieve vehicle-level simulation before installation, often only completing component-level simulations. However, with the development of commercial architecture evaluation software such as RTaW, CANoe, and VEOS, the industry has gradually adopted more comprehensive architecture simulation evaluation software for iterative verification and optimization in aspects such as functionality, communication, and safety. In the analysis and evaluation of multi-domain vehicle electronic and electrical architectures, in addition to traditional objectives such as hardware costs, development costs, production costs, warranty costs, vehicle performance, and fuel economy, the following new issues also need to be addressed:

(1) Whether it can meet personalized user needs and potential changes in future demand, especially whether it can meet the demand changes for L3-level and above autonomous vehicle architectures;

(2) Whether the platform has good reusability and commonality, whether it can meet the basic technical requirements for high-level autonomous driving and intelligent connectivity, and whether it possesses advanced technological features.

Hardware Systems of Multi-domain Electronic and Electrical Architecture

3.1 Domain Control Units and Key Technologies

To reduce bus lengths and the number of ECUs, achieving the goals of reducing electronic component weight and lowering vehicle manufacturing costs, dispersed ECUs are classified by functionality and integrated into domain control units (DCUs) with stronger computing power and richer interfaces. Existing technical solutions typically divide the vehicle into vehicle control domains, intelligent driving domains, and cabin domains. The vehicle control domain controllers are responsible for controlling the vehicle’s power system, chassis system, and body system. Intelligent driving domain controllers are equipped with rich interfaces to accommodate various sensor signal acquisitions, integrating high-performance heterogeneous computing platforms to support complex sensor data fusion algorithms, combining high-precision maps and navigation information for environmental recognition and path planning, and outputting vehicle control instructions to realize higher-level intelligent driving functions. A typical intelligent driving domain controller is shown in Figure 6, where the computing platform integrates general computing units, AI computing units, real-time control units, and various interfaces.

Figure 6 Function Diagram of Intelligent Driving Domain Controller

The cabin domain controller typically integrates functions such as full LCD instruments, head-up displays, streaming media rearview mirrors, cabin entertainment systems, vehicle networking, and remote information, while also serving as a human-machine interaction interface. The intelligent cabin domain controller needs to possess strong processing capabilities and complex operating systems, consisting of main control chips, real-time microprocessors, digital signal processors, CAN and Ethernet ports, among others, as shown in Figure 7.

Figure 7 Function Diagram of Intelligent Cabin Domain Controller

3.2 Zone Controllers and Key Technologies

Zone controllers (ZCUs) mainly include three major functions: regional data center, regional I/O center, and regional power distribution center, as shown in Figure 8. As the regional data center, ZCUs are equipped with rich network interfaces, such as ETH, CAN, and LIN, playing the roles of regional gateways and switches to achieve network communication and routing. The regional I/O center supports various types of sensor, actuator, and display interfaces. As the regional power distribution center, ZCUs are responsible for transmitting power to controllers and actuators. Currently, there is a trend to use electronic fuses (eFuses) instead of traditional relay and fuse solutions for intelligent management. Meanwhile, ZCUs are capable of absorbing the functions of other ECUs within the region, abstracting the functions at the service layer, and achieving virtualization of control I/O. Due to the high safety, real-time, and reliability requirements involved in vehicle control functions, the main control chips of ZCUs are typically equipped with ASIL-D level MCUs, and the future development trend is to introduce high-performance computing units.

Figure 8 Function Diagram of Zone Controllers

3.3 Central Computing Units and Key Technologies

The core positioning of the central computing unit is to provide sufficient computing power to support the business logic related to intelligent driving and intelligent cabin features. At the same time, it needs to possess high bandwidth and low latency communication capabilities to support data exchange with zone controllers and to realize vehicle networking functions, connecting to both vehicle and cloud. At the hardware level, central computing units typically use multiple heterogeneous multi-core SoC chips, with high-speed serial communication or PCIe connections between chips. The architecture of SoC chips is mainly divided into hardware-isolated and software-isolated forms, both adopting virtualization schemes to run multiple operating systems simultaneously. In hardware isolation, the operating systems running on each core are determined during the software design phase, and are isolated at the hardware level, possessing dedicated hardware resources. In software isolation, the operating systems do not have dedicated hardware resources, and hardware resources are dynamically allocated by the hypervisor layer.

3.4 Power Supply Systems and Key Technologies

With the increasing electrical load of the entire vehicle, the development of electrical architecture, and breakthroughs in semiconductor technology, the design of power supply systems has shifted from the combination of power components to the system design of power networks and the control design of power networks. Traditional vehicle power supply systems typically adopt a central electrical box scheme, using relays and fuses for circuit control and protection, but there are issues such as relay burning and the inability to reuse fuses after they are blown. Currently, the main technical route for power supply systems is the integration of protection and control, using MOSFET-based eFuses for power distribution. A single chip integrates various diagnostic functions such as driving, current detection, thermal protection, overvoltage protection, overcurrent protection, EMC, and open/short circuit detection.

3.5 Wiring Harness Systems and Key Technologies

The wiring harness plays a crucial role in the realization of the electrical and electronic functions of the entire vehicle and is a research hotspot for architectural optimization design. In the overall design of wiring harness layout, various relevant boundary conditions need to be fully considered, and the potential impact of relevant components on wiring harness layout must be taken into account, along with reasonable requirements for the design of relevant components. Currently, the design of wiring harness systems is trending towards maturation and comprehensiveness, with multi-dimensional, multi-objective wiring harness modeling, design, evaluation, and optimization methods based on PREEvision software greatly simplifying the design process of wiring harness systems, improving design efficiency and effectiveness.

Communication Systems of Multi-domain Electronic and Electrical Architecture

4.1 Development and Status of On-board Communication Systems

The electronic and electrical architecture of vehicles relies on communication systems to enable information transmission between hardware. Currently, there are five main communication technologies: Controller Area Network (CAN), Local Interconnect Network (LIN), Media Oriented Systems Transport (MOST), FlexRay bus, and on-board Ethernet (ETH). The main characteristics of these five communication technologies are shown in Table 1.

Table 1 Characteristics of Various Communication Technologies

In addition to these communication technologies, there are also some new on-board communication technologies currently in the experimental stage. For example, the third-generation CAN communication technology, CAN XL, reduces the transmission speed and coupling gap between CAN and Ethernet, providing connectivity between signal-based communication and service-oriented communication. In the future, the security and confidentiality of on-board communication systems will be emphasized. Fiber optic communication, with its advantages of resistance to electromagnetic interference, no radiation, and difficulty in eavesdropping, will find widespread application in vehicle communication security, fault diagnosis, and high-precision control.

As the level of intelligent driving continues to increase, the number of on-board components is growing exponentially, and the volume of information data is increasing, imposing higher requirements on the transmission rate, real-time performance, fault tolerance, and cost of vehicle bus networks. Although CAN bus is limited by low data transmission volume and asynchronous timing, its technological maturity remains high, and it is still the backbone of vehicle bus technology. LIN bus, MOST bus, and FlexRay bus typically serve as local network access based on their respective characteristics. With its high bandwidth and low cost advantages, Ethernet will become the backbone network of communication systems, leading the development of the next generation of on-board networks in the future. Currently, forming a unified vehicle bus protocol standard will still take a considerable amount of time. Therefore, before that, vehicle network systems will still need to adopt a coexistence of multiple buses to meet different transmission needs, further improving the compatibility and interoperability of various vehicle bus standards to achieve better data exchange and system integration remains one of the key issues that multi-domain electronic and electrical architecture needs to address.

4.2 Analysis and Research of Time-Sensitive Network Communication Protocols

With the widespread deployment of high-precision sensors and the continuous enhancement of infotainment system functions, the data volume inside vehicles has surged, and traditional on-board networks struggle to effectively support and process the growing demands for high-speed, high-bandwidth communication. Time-Sensitive Networks (TSN) are considered a key solution to these issues, enabling deterministic, real-time, low-latency, and secure data transmission over Ethernet.

TSN can achieve low-cost, high-bandwidth transmission, with transmission rates ranging from 10 Mb/s to 10 Gb/s, utilizing unshielded twisted pairs for full-duplex communication, reducing costs by 80% compared to traditional shielded cables, and decreasing weight by 30%. Furthermore, TSN also offers good scalability and versatility, supporting various configurations of vehicle network topologies to facilitate the transmission of different application data.

TSN protocols that significantly impact vehicle communication can be categorized into four types: time synchronization, traffic control, reliability, and resource management. The following sections will provide detailed introductions to these protocols.

4.2.1 Time Synchronization Protocols

In the communication system of a multi-domain electronic and electrical architecture deployed with TSN, a unified time scale is necessary to ensure the precision of time synchronization. The TSN IEEE 802.1AS-2020 protocol defines and explains the time synchronization methods and processes for TSN streams. Through a timestamp mechanism, all components are controlled by the same global clock while allowing the existence of different time domains within the network. Research on this protocol mainly includes factors influencing synchronization accuracy, local clock calibration, and synchronization quality assessment.

In multi-domain electronic and electrical architectures, clock synchronization accuracy is fundamental to ensuring that various sensors achieve high-precision responses and locate external environments. Although there is substantial research on clock synchronization for industrial TSN, there is a lack of dedicated studies focusing on the clock synchronization characteristics of in-vehicle TSN. The in-vehicle communication environment differs significantly from industrial automation systems, as factors such as vehicle vibration, temperature variations, and electromagnetic interference can disrupt clock synchronization accuracy. Therefore, further research is needed to explore the factors influencing clock synchronization accuracy in in-vehicle TSN, ensuring high reliability and efficiency of the in-vehicle communication system.

4.2.2 Traffic Control Protocols

Traffic control is one of the key technologies for achieving low-latency transmission and flow determinism in TSN. The traffic control process of TSN includes traffic classification, traffic shaping, traffic scheduling, and traffic preemption, with corresponding TSN protocols listed in Table 2.

Table 2 Traffic Control Protocols

The current hot research areas in traffic control protocols mainly include analysis of maximum end-to-end latency for various flows, research on TSN traffic shaping methods, and studies on traffic scheduling methods for time-critical flows. Current research primarily focuses on single protocols, while the next stage needs to explore the synergistic mechanisms between protocols and their applications in practical vehicle network scenarios.

4.2.3 Reliability Protocols

The reliability of TSN refers to the network’s ability to prevent and recover from failures, primarily including IEEE802.1CB and IEEE802.1Qci protocols. IEEE802.1CB establishes a Frame Replication and Elimination (FRER) mechanism to mitigate the impacts of frame congestion or failures during flow transmission, mainly targeting control frames, strictly limiting packet loss rates to ensure transmission reliability. IEEE802.1Qci implements a Frame Filtering and Policing (PSFP) mechanism to address flow handling issues during network failures, avoiding flow overload and erroneous delivery, thereby enhancing system robustness. Research on TSN reliability issues primarily includes redundancy mechanisms, fault detection, and reliability under synchronous failures. Future research should focus on the reliability of vehicle TSN networks under various failure conditions, ensuring safety and stability during vehicle operation.

4.2.4 Resource Management Protocols

Resource management primarily includes managing and configuring network resources, as well as monitoring and analyzing performance data. The IEEE802.1Qat flow reservation protocol addresses the registration and reservation of flows, serving as a prerequisite for shaping, scheduling, and transmission processes. The IEEE802.1Qcc protocol resolves centralized control issues in TSN networks, proposing three TSN network control models: distributed, centralized, and centralized network-distributed user models. Current research mainly focuses on the implementation and deployment of architectural models. These research findings provide important technical support and references for the realization of vehicle TSN network resource management. Future research should emphasize how to manage and configure vehicle TSN, focusing on overcoming key challenges such as the management of event-triggered flows and vehicle-cloud secure interaction management.

TSN, as an important component of multi-domain E/E architecture, has received significant attention. However, current research on TSN mainly focuses on industrial internet domains, while research on vehicle TSN networks is still insufficiently deep. During the technological migration process, several urgent challenges need to be addressed:

(1) Scene Construction Issues: Constructing complex models of big data and various types of vehicle TSN networks is challenging, making it difficult to model event-triggered random signal flows.

(2) Function Matching Issues: How to design software to implement TSN-related standards, and how the TSN protocols perform and their effects in vehicle scenarios need experimental validation.

(3) Hardware Support Issues: Currently, there are relatively few chips that support TSN Ethernet, and there are no specialized testing devices for vehicle TSN, making it difficult to build hardware experimental platforms. Despite these challenges, the potential of TSN in vehicle real-time communication applications cannot be denied. In the future, the bandwidth advantages of TSN are expected to further increase; the combination of vehicle TSN and IP protocols will enable more complex vehicle safety and multimedia applications; as the level of autonomous driving improves, the reliability of TSN will increase with the enhancement of vehicle network information security; the openness of TSN protocols will also provide broader space for academic research and industrial deployment.

4.3 Service-Oriented Software-Defined Networks

Traditional vehicle networks face the following issues: uneven traffic load distribution, high message sending delays, low network throughput, poor compatibility of network modules, and low openness. These issues hinder further development and innovation and do not facilitate the interconnectivity of intelligent on-board systems in future vehicle models. To address this, the concept of Software-Defined Vehicle Networks (SDVN) has been proposed. SDVN applies Software-Defined Networking (SDN) technology to vehicle networks, transforming the architecture of vehicle networks with the idea of software-defined networking. Firstly, SDVN separates the data forwarding plane from the control plane in vehicle network devices, then centralizes all control planes into a logically centralized controller, and finally uses this centralized controller to control the forwarding behavior of all data forwarding plane messages within the vehicle network. SDVN can effectively improve network performance, reduce the costs of network service updates, simplify network management, and accelerate network innovation. Currently, service-based SDVN is still in its infancy, with many critical technical issues regarding security, mobility, service efficiency, deployment, and standardization yet to be resolved. However, as a programmable and highly flexible network architecture, SDVN still holds great development potential, applicable to efficient bandwidth allocation, vehicle-road-cloud elastic computing allocation, and various other scenarios.

In summary, future vehicle communication networks will exhibit the following characteristics:

(1) Future vehicle communication protocols will develop towards high bandwidth, low cost, and high security, with vehicle TSN becoming the backbone network, providing deterministic, high-bandwidth, and highly secure connections, while existing bus forms will still be retained in certain specific scenarios.

(2) To address the challenges posed by intelligent driving, vehicle networks will implement more security features, and the application of SDVN will further enhance network configurability and flexibility.

(3) The interfaces between different communication software components will be further standardized, significantly improving software interchangeability.

Software Systems of Multi-domain Electronic and Electrical Architecture

5.1 Software-Defined Vehicles

5.1.1 Basic Concept of SDVs

With the continuous increase in functions, the core of vehicle design has gradually shifted from hardware design to software development. Software has become a core element shaping the competitiveness of automakers. The concept of SDVs has become a consensus in the industry, with software development and upgrades becoming key components throughout the entire lifecycle of the vehicle, from design to sales and services. The vehicle development process based on SDVs will form a dual closed-loop model where user interaction evaluation information guides new vehicle development, and OTA technology enables continuous software updates and iterations. The service-based software architecture is shown in Figure 9. This software architecture is generally divided into four layers.

Figure 9 Service-Based Software Architecture

The significant advantage of SDVs lies in reducing the impact of hardware differences on software, pursuing multi-vehicle reuse and minimizing differentiation through software design at the device abstraction layer and atomic service layer. By standardizing API interfaces, redundant work is minimized, software complexity is reduced, and software design and development efficiency is improved. The application layer design focuses on creating differentiated and customized functions, ultimately achieving high added value and personalized services for software components. At the same time, the emergence of SDVs and OTA technology has also brought new changes to the vehicle development process.

5.1.2 Decoupling and Mapping of Software and Hardware

A crucial prerequisite for achieving SDVs is the decoupling of software and hardware, which means that the design of software systems is entirely independent of hardware. In the software framework, hardware interfaces are abstracted to accommodate different hardware devices. The key to decoupling software and hardware lies in the standardization of interface definitions, which requires the entire automotive industry to have a reasonable division of labor and work together to form unified technical specifications for software and hardware interface definitions. Achieving software and hardware decoupling will have significant implications for future vehicle development, verification, and after-sales. Firstly, the decoupling of software and hardware allows data to be liberated from individual subsystems, enhancing the automaker’s control over function realization, which will have important implications for industrial division of labor. Secondly, software can be independently verified apart from hardware, allowing functions that previously required hardware-in-the-loop testing to be validated through software-in-the-loop testing in an integrated hardware environment, significantly accelerating vehicle development and testing speed and reducing verification costs. Additionally, the upgradability of vehicles throughout their lifecycle will effectively enhance the maintainability and safety of vehicles; through remote upgrades (OTA), functions can be gradually liberated, effectively enhancing user experience and improving vehicle resale value. However, the decoupling of software and hardware still falls short of the ideal state due to the influence of traditional R&D models, difficulties in enterprise transformation, and contradictions in industrial division of labor.

Accompanying the decoupling of software and hardware comes the issue of software and hardware mapping. Since DCUs and CCPs need to integrate numerous functions, including sensor data processing, intelligent human-machine interaction, and high-precision control decision-making, the complexity of data processing increases sharply. How to map functions with different data computation characteristics to matching processors and achieve optimal collaboration between software and hardware is the core issue to be resolved in software and hardware mapping. Multi-domain E/E architecture introduces various microprocessors, a large number of heterogeneous computing resources, and combinations of communication links, making the factors to be considered even more complex. Early research typically focused on single-core heterogeneous systems based on task communication relationships and attributes, considering factors such as time, cost, and power consumption. With the development of multi-core embedded chips, a wealth of research has proposed optimization design methods for software and hardware mapping issues in multi-core distributed heterogeneous systems, with optimization objectives including energy consumption and hardware cost optimization. Vehicle multi-core heterogeneous chips are extremely sensitive to factors such as cost, power consumption, safety, computing power, and real-time performance, making it a key challenge for future vehicle main control chip design to comprehensively consider these factors, design proprietary chip structures based on functional design, and achieve easy-to-decouple software and hardware mapping.

5.2 Service-Oriented Software Design

The service-oriented architecture (SOA) is an advanced concept introduced into the automotive industry from the IT industry, recognized as an important direction for the software development of ICVs due to its characteristics of reusability, easy upgrades, easy deployment, and loose coupling. The SOA concept aims to make services no longer limited to specific functional environments through flexible interfaces, enabling service sharing. In this concept, the definition of interfaces needs to follow SOA standards, independent of operating systems and hardware platforms. This complements the previously mentioned concepts of the atomic service layer and device abstraction layer in SDVs. The introduction of SOA breaks the traditional rigid and closed ecosystem of automotive software, gradually making it open and open-source.

Currently, the automotive industry has conducted practices related to SOA software design and proposed SOA-based software development models, verifying that SOA can significantly reduce system complexity and simplify the reuse of software components across different generations of vehicles.

To ensure the information interoperability between system services and the extensibility of combined forms, each service module communicates through service-based middleware, altering the communication methods within vehicles. The traditional signal-based communication approach defines the communication matrix during vehicle design, where the data volume, sending cycles, and routing paths of signals are all fixed and static. In contrast, service-based middleware establishes network connections between services and applications through a certain level of abstraction between applications and networks. This communication process is typically dynamic and can be configured at runtime, without the need for it to be fixed during design.

The mainstream service-based middleware currently includes DDS (Data Distribution Service) and SOME/IP (Scalable service-oriented middleware over IP). Both are integrated as standardized modules in AutoSAR, thus regarded as top-tier solutions in the industry. The following Table 3 compares SOME/IP, DDS, and signal-driven communication mechanisms.

Table 3 Comparison of Communication Mechanisms

5.3 Vehicle Operating Systems

Vehicle operating systems are collections of system programs within vehicles, primarily responsible for managing hardware resources, hiding internal logic to provide software platforms, providing user programs with system interaction interfaces, and offering basic services for upper-layer applications. They include vehicle control operating systems and on-board operating systems.

5.3.1 Vehicle Control Operating Systems

Vehicle control operating systems are mainly divided into safety vehicle control and intelligent driving subcategories, as shown in Figure 10. Safety vehicle control operating systems are primarily aimed at traditional vehicle chassis, power, body, and other functional areas, with extremely high real-time and ASIL-D safety requirements. Currently, most mainstream safety vehicle control operating systems are compatible with OSEK and AUTOSAR Classic Platform (AUTOSAR CP) standard software architectures, and related technologies are relatively mature. The development of operating system software based on AUTOSAR CP has achieved decoupling between application layer and lower-layer software, as well as between software and hardware, enhancing the portability, reusability, extensibility, upgradeability, safety, and maintenance capabilities of software to a certain extent, thereby benefiting the reduction of software development cycles and costs.

Figure 10 Basic Architecture of Vehicle Control Operating Systems

Intelligent driving operating systems, on the other hand, are aimed at the upgrade background of the new generation of centralized E/E architecture, requiring high computing power, high performance, high safety, and high reliability for intelligent driving functions. This type of operating system is currently at a developmental opportunity stage, with various countries exploring preliminary stages. Since AUTOSAR CP is difficult to fully accommodate the needs of intelligent driving operating systems, the AUTOSAR organization released the AUTOSAR Adaptive Platform (AUTOSAR AP) in 2017, which defines standard programming interface specifications between operating systems and applications based on the POSIX PSE51 subset to meet the demands of vehicle intelligent driving services on heterogeneous chip platforms.

In the field of vehicle control operating systems, most domestic and foreign enterprises have developed their systems based on AUTOSAR, indicating that the AUTOSAR software architecture standard plays a crucial leading and reference role in the field of vehicle control operating systems, currently being the mainstream automotive standard software architecture internationally. The implementation of software architectures based on AUTOSAR standards relies on corresponding configuration toolchain solutions. Currently, mainstream toolchains include Vector’s DaVinci series tools for AUTOSAR CP and MICROSAR Adaptive for AP; Bosch’s ETAS with RTACAR and RTA-VRTE for CP and AP; as well as ElektroBit’s EB tresos and EB corbos series CP and AP configuration tools; Siemens’ Capital VSTAR; and KPIT’s KSAR Classic and KSAR Adaptive. Domestic companies are also actively laying out AUTOSAR, with PwC’s basic software and Neusoft Ruichi successively launching their AUTOSAR solutions, aiding the implementation of domestic toolchain practices.

5.3.2 On-board Operating Systems

On-board operating systems are mainly applied to infotainment functions within vehicles, with relatively low requirements for safety and real-time performance, leading to rapid development in this area. Currently, mainstream on-board operating systems have been compared in terms of real-time performance, safety, and application scenarios, as shown in Table 4.

Table 4 Comparison of Functional Attributes of Various On-board Operating Systems

As the intelligent and interconnected development continues to deepen, a single on-board operating system is no longer sufficient to meet the increasingly rich infotainment function demands within vehicles. Therefore, on-board operating systems are gradually transitioning to multi-operating system architectures. Multi-operating system architectures can be realized in two ways: one based on hardware isolation and the other based on virtualization management technology (Hypervisor). The hardware isolation architecture simplifies the corresponding resource allocation management issues by physically partitioning resources, making it easy to develop. However, fixed hardware partitions may lead to lower flexibility and can result in some degree of resource waste. In contrast, the architecture based on Hypervisor for isolating and managing multiple operating systems can avoid the fixed allocation of system resources, improving resource utilization. Additionally, it uses host memory as a medium for data exchange, significantly enhancing data sharing capabilities. However, it also increases the complexity of system development and security risks.

Research Outlook

Currently, research on the multi-domain electronic/electrical (E/E) architecture of intelligent connected vehicles (ICVs) is increasing. Academic and industrial sectors in various countries are conducting extensive research, and some large automotive manufacturers have already implemented preliminary deployments on advanced models. However, due to the comprehensive and complex nature of the elements involved in E/E architecture, there is still no complete design theory, engineering methods, or tool software established. Therefore, it is recommended to further strengthen the following research directions.

(1) Strengthen research on overall architecture design theories and methods.

Currently, the industry’s architecture development still mainly relies on engineering experience. However, with the increasing complexity of functions, diversification of demands, and rapid iterations, relying solely on experience makes it difficult to achieve optimal design outcomes. Therefore, it is necessary to form a complete design theory and methods as soon as possible, from overall design theory to engineering practical applications, providing guidance for overall architecture design. Future research should start from the essence of the design issues of ICV’s E/E architecture, studying the design mechanisms that achieve safety, economy, and scalability. Through theoretical analysis and experimental validation, the intrinsic relationships between automotive functional requirements, safety requirements, and architecture design can be clarified, completing the standardized modeling of requirements and accurate segmentation of functions. Based on existing mainstream architectures and technological levels, research can be conducted on architecture modeling, system optimization, and analysis to form theories and methods for architecture design.

(2) Establish a standard system for software, hardware, and communication interfaces.

Architecture design involves the software, hardware, and communication systems within the vehicle, as well as the interoperability with external vehicle, road, and cloud systems. The various interfaces are complex and diverse, making it difficult for a single manufacturer to complete end-to-end designs for all interfaces. Only by establishing a standard system for software, hardware, and communication interfaces can all parties in the industrial chain fully leverage their advantages, allowing automakers to integrate and flexibly configure based on the overall architecture design framework, promoting the rapid implementation of ICVs. In top-down service design, standardized interfaces should allow application layer and communication layer development to focus on business logic without being limited by hardware implementation; in bottom-up abstract design, lower-level hardware devices should be able to focus on differences between vehicle models, with the capability to reduce code differences through flexible changes in configuration.

(3) Develop a simulation testing and verification system for E/E architecture.

The simulation evaluation technology of E/E architecture is the foundation for verifying design rationality and achieving rapid iterative updates. Therefore, it is necessary to establish a multi-level, integrated, virtual-real combined testing and verification system for E/E architecture. Research can be conducted on multi-environment interactive technologies that integrate virtual simulation, closed scenarios, and open road testing, developing E/E simulation scenario libraries suitable for failure analysis and risk assessment, and creating real-time evaluation simulation analysis platforms to realize the platformization and standardization of architecture evaluation and simulation testing. Additionally, it is necessary to develop high-fidelity and real-time simulation technologies for physical signals in hardware-in-the-loop and vehicle-in-the-loop testing, creating communication signal simulation devices under connected scenarios, gradually building a multi-level testing and verification system for E/E architecture, and forming multi-tier testing and evaluation methods at the component, system, and vehicle levels to achieve integrated design of the E/E architecture testing and verification system.

(4) Strengthen the design of multi-dimensional redundancy architecture systems and research on information security depth defense technologies.

To address the hidden and sudden challenges of ICV architecture failures, research is needed on fault detection methods and proactive restructuring control theories at the sensor, controller, and actuator levels under redundancy architecture systems, exploring efficient and precise fault detection methods, and establishing a comprehensive proactive restructuring control mechanism to ensure that ICVs maintain normal driving capabilities under certain faults. To ensure the network security, data security, and information security of high-level autonomous driving systems, a multi-layered depth defense system should be constructed from multiple dimensions, including external network security, inter-domain control security, on-board network communication security, and controller safety, establishing depth defense technology theories that ensure system safety while reducing redundancy and system complexity.

(5) Accelerate the localization process of the core component industrial chain of ICVs.

China has already established a first-mover advantage in the ICV field, but there are still certain gaps compared to developed countries like Europe and the United States in high-performance chips, vehicle operating systems, and architecture design tool software. Although many domestic solutions have emerged, their functional completeness and industrial support are relatively weak, and a complete localization industrial chain has not yet formed. Therefore, China needs to further accelerate the localization research and development of key technologies, transforming its first-mover advantage into leading capabilities, striving to develop a Chinese automotive industry with independent characteristics, enhancing the competitiveness of independent brands, and promoting the high-quality development of China’s automotive industry.

Conclusion

The multi-domain electronic/electrical (E/E) architecture is of great significance for the popularization of intelligent connected vehicles (ICVs) and realizing their expected functions. However, at this stage, the field still lacks a complete methodology, technical theory system, and toolchain; the industry remains in a stage of exploration and research, necessitating extensive research and practice.

Related posts

Leave a Comment Cancel reply