Dependability in Zonal E/E Architectures with Central Compute for Autonomous Driving
This article explores how dependable electronics like sensors and MCUs within zonal E/E architectures are important for reliable, safe, and secure autonomous vehicles (AVs).
According to Statista, there were 31 million automobiles worldwide in 2019 with at least some level of automation—that number is expected to reach over 54 million by 2024.
For AVs to be successful in such a competitive market, both drivers and consumers must view them as trustworthy. Developing this trust is a function of both reductions in accidents and how these vehicles behave daily on the streets. If the vehicles do not perform and behave as consumers expect them to, they will be rejected. Due to this, there is an increased demand for electronics and systems architectures that provide mission-critical dependability supporting the type of decision-making and control processes illustrated in Figure 1.
Figure 1. Automated cars must be able to reliably acquire sensor data and process it in a dependable, safe way to ensure the safety of both the passengers in the vehicle and nearby vehicles. Image used courtesy of Infineon
Dependable electronics are foundational elements necessary to realize autonomous features in vehicles. The performance required to meet autonomy must be combined with reliability, availability, safety, and security—these are key to consumer acceptance and proliferation of this growing technology. The electronics that form the building blocks of autonomous designs, ranging from basic angle sensors to powerful MCUs, must also be engineered not just for performance but for reliability.
In this article, we'll explore zonal architectures for AVs, what dependable electronics mean within the scope of autonomous driving, and finally go over some examples of zonal architecture and AV electronics.
Zonal E/E Architecture for Autonomous Vehicles
AV architectures are rapidly progressing toward a more software-designed vehicle, with some industry leaders working to achieve seamless integration between vehicles and the cloud.
As the functionality of AVs continues to increase, the architecture is changing to a combination of zonal E/E design and central compute to simplify engineering complexity and enable compute scalability. This approach differs significantly from the more traditional and mixed domain/zone architecture, as illustrated in Figure 2.
Figure 2. Domain and mixed domain/zone architecture are compared to zonal architecture. Image used courtesy of Infineon
Domain architectures often utilize modules known as ECUs (electronic control units) that provide a specific function, such as telematics or vehicle motion control. Each ECU is highly optimized for the task it is intended to perform. However, the number of ECUs on a car can easily reach 150, which leads to complexity and potential issues with dependability, security, and safety.
Zonal architecture approaches AV systems by viewing the car as a computer with zones where various actuation functionality is managed. Central compute manages all internal algorithms and complex computations and is connected to the various sensors and devices via a networked gateway.
Overall, the zonal approach reduces the system's complexity and increases reliability and dependability. Hardware and software scalability is enhanced because pooled computational resources are used. This software-focused approach also makes it easier to safeguard system security and protect against future threats.
However, the successful introduction of zones requires highly dependable electronics to enable the E/E architectures for autonomous vehicles.
The official definition of dependability is the quality of being trustworthy. However, in the context of autonomous vehicles, this definition becomes far more specific to reliability and availability. Even the most advanced automated driving technology is unusable if it does not meet the highest standards of safety and security.
Functional safety has long been an integral part of vehicle design with an increasing presence as vehicle designs become more complex. However, automated driving brings a logarithmic increase in functional safety requirements. AVs must meet ASIL D-level safety combined with SOTIF requirements to (a) identify and avoid risks and (b) allow both fail-safe and fail-operational methods to bring the vehicle to a safe state.
With ongoing cyber security issues, electronics must also support protection against malicious attacks that can include unauthorized access, electronic systems, control algorithms, and communication networks. High availability, failure operational performance, and fail-safe functionality must also be considered. Above all, an automotive architecture is only as dependable as the electronics used to implement it.
Potential Solutions for Zonal and Central Compute Architectures
One company that focuses on zonal and central compute architectures is Infineon. Functional safety, high availability, fail-operational performance, and fail-safe functionality with the ability to integrate into zonal architecture are key features of Infineon’s portfolio of Pro-SIL products. Infineon's Pro-SIL products are either ISO26262-compliant or ISO26262-ready, with many different ISO26262 compliant products readily available. In addition, Infineon also supports the advanced security needs of zonal E/E architecture with a focus on crypto-agility and has a zero defect mentality for all its Pro-SIL vehicle automation products. This includes ppm levels that have been reduced to sub-ppm levels, and 90% of their products already meet zero defect requirements.
An excellent example of Infineon’s dependable zonal architecture solutions is the 32-bit AURIX™ TriCore™ microcontroller product line. The Tri-Core microcontrollers combine a RISC processor core, microcontroller, and a DSP into a single MCU. This MCU can be used in several automotive applications, including:
- Transmission control units
- Advanced driver assistance
- Power steering systems
- Chassis domains
However, this is far from the only Infineon product that supports zonal architecture solutions. With that in mind, we can look at a few sensors and central compute examples as well.
Sensor Fusion and Decision Making in a Central Compute System
Consider the central compute system, a car computer that centralizes high-performance computation for AVs. The central compute system, as shown in Figure 3, receives sensor data and then performs perception, fusion, and decision-making functions based on that data.
Figure 3. Examples of autonomous system design include the central car compute system. Image used courtesy of Infineon
In a central compute module, high-performance neural network accelerators and processors are used for running perception on data received from a myriad of sensors that include cameras, radars, ultrasonic, etc. A simplified view of the pipeline is that after perception, the fusion of perceived data from various sensors is performed to assess the complete driving scenario. Once fusion is complete, decision-making algorithms are run on the fused data to determine the needed vehicle motion commands.
The sensor fusion and motion control command tasks must conform to the highest level of safety and require high availability. Continuing with the example of Infineon's Aurix microcontrollers with ASIL D-level safety and performance features, these devices can be well-suited to run these types of post-perception tasks. Additionally, these microcontrollers offer safe boot-up and run diverse algorithms as needed to achieve functional safety. They are designed for low power consumption, especially during boot-up, when running fail operational modes, and when handling power management of the ECU.
If further safety is a concern, Infineon also offers the industry’s first Functional Safety (ASIL D) NOR Flash memory products. These products have been designed for very high reliability to serve flash memory requirements for the central compute ECU.
To complete the full system offering, Infineon provides companion PMICs, MOSFETs, load guards, CAN/LIN transceivers, and many other products to support a full system solution for a central compute module.
Shifting Towards Zonal E/E and Central Compute Architectures
Zonal E/E and central compute architecture combined with highly dependable electronics can help provide a way to simplify designs while offering key safety and security features necessary to increase consumer and regulatory confidence.
This article was authored by:
Priya Muralidharan and Bill Stewart at Infineon Technologies.
Priya Muralidharan has joined from Renesas, where she spent seven years working in ADAS and Autonomous marketing. She was responsible for North American business acquisition for ADAS/AV products and for setting the go-to-market strategy for this business line. She has previously worked at Magna in their EV/HEV and Electronics division and also at Nexteer in various embedded software engineering roles. Priya also has a background in functional safety for both MCUs and SoCs and certification in ISO26262. Overall, she has 15 years of automotive experience. Priya holds a Bachelor’s degree in Physics and dual masters in Electronics and Computer Science.
Bill has worked in the Semiconductor Industry for over 15 years in various business development roles across applications for the entire vehicle. He joined the company in 2017 and is responsible for enabling customers to develop systems that focus on Safety Critical and Active Safety/Autonomous Applications. His role includes engaging global teams to drive new product and application strategies for future automotive innovations.
Industry Articles are a form of content that allows industry partners to share useful news, messages, and technology with All About Circuits readers in a way editorial content is not well suited to. All Industry Articles are subject to strict editorial guidelines with the intention of offering readers useful news, technical expertise, or stories. The viewpoints and opinions expressed in Industry Articles are those of the partner and not necessarily those of All About Circuits or its writers.