Accelerating the shift to next-gen SDVs with zonal MCUs

As the automotive industry shifts toward software-defined vehicles (SDVs), with features and updates delivered via software, vehicle electrical/electronic (E/E) architectures are evolving. These architectures are moving away from traditional flat and domain-based designs, in which separate control units for each function were the norm, to zonal architectures that group vehicle functions based on various criteria, such as physical location, application group, and other functional criteria in zone electronic control units (ECUs).

In a zonal architecture, each zone of a vehicle, such as front, rear, or cabin, has a high-performance controller managing local devices and communicating with other controllers over high-speed networks. This approach is faster to update, easier to scale, less complex, and even cheaper, but it also presents new technical challenges for automakers to overcome. The good news is that innovative microcontroller (MCU) solutions can help tackle those challenges.

Key challenges in the shift to zonal architectures

Zonal architectures offer clear benefits of reducing wiring complexity and consolidating compute resources.

However, each zone controller in an SDV must handle a wide range of tasks, from low-level real-time control of sensors and actuators (such as sensing braking pressure, steering angle, or controlling window-lift motors) to high-level data processing tasks (including controlling advanced safety systems or vehicle dynamics or running local AI models to process sensor data). This means the ECUs in a zonal design require compute capabilities that can handle both time-sensitive and compute-intensive workloads concurrently.

Maintaining freedom from interference among functions is a key challenge. Zonal designs consolidate diverse functions, from safety-critical functions such as braking to non-critical functions such as body control on a single ECU with shared compute and communication resources. Without proper safeguards, a fault in one function could disrupt others.

To prevent this, robust isolation mechanisms such as hardware-enforced separation are needed. Ideally, these features should be embedded in the hardware design rather than software implementation to minimize the impact on performance, behavior, or safety.

In-vehicle networks also need rethinking. Modern SDVs generate massive amounts of data from sensors and devices, including cameras, LiDAR, and radar, to support advanced driver-assistance and autonomous-driving features. The vehicle network must ensure that this data is delivered in a predictable and time-sensitive manner.

Traditional in-vehicle networks such as CAN, LIN, and FlexRay struggle to meet these demands at the scale of the modern SDV. Moreover, Ethernet is increasingly used in vehicles for its advantages in performance, reliability, and connectivity options, and at the same time, many legacy devices in the vehicle still use CAN or LIN.

To ensure network performance and avoid communication bottlenecks, next-gen vehicle controllers should include native support for Ethernet and CAN networking on-chip. A well-architected MCU will also enable sufficient control and acceleration features to reduce latency and ensure predictable message delivery, ultimately resulting in deterministic communications at the vehicle level.

Making updates secure, easy, and fast

A hallmark of SDVs is the ability and the need to receive secure over-the-air (OTA) updates throughout the vehicle’s life. This contrasts with traditional vehicle designs with dozens of ECUs for which updates are complex, time-consuming, and often performed only in service environments.

For an SDV built on a zonal E/E architecture, software updates can be pushed to a selected number of controllers that can be responsible for the software of an entire region of the vehicle, such as the front zone or the rear chassis zone. To support these frequent updates with minimal downtime, embedded nonvolatile memory (eNVM) needs to be upgraded as well to a more versatile option with much faster write speeds and higher endurance.

How modern MCUs lay the foundation for next-gen SDVs

Next-generation automotive MCUs need to adopt improved, low-latency architectures and design philosophies to overcome these challenges. Rather than relying on a homogeneous architecture, the implementation of heterogeneous cores optimized by functions is required.

The NXP S32K5 MCU, for example, integrates different kinds of CPU cores complemented by various accelerators for dedicated functions such as AI/ML, signal processing, and network acceleration, enabling designers to assign the right function to the right core. Furthermore, a dedicated low-power engine is available for simple tasks, such as periodic wakeups and sensor monitoring, which can help optimize battery life in electric vehicles.

The integration of multiple functions of mixed criticality warrants robust mechanisms to ensure freedom from interference. The S32K5 lives up to this challenge by implementing features such as XRDC, an in-house NXP IP for isolation and resource allocation in hardware from pin to core.

Moreover, dedicated network accelerators need to be implemented for CAN2CAN (communication between CAN networks) and CAN2ETH (communication between CAN and Ethernet networks) acceleration to ensure deterministic and low-latency architectures.

In addition, integrating an Ethernet switch with fast data-routing capabilities ensures that the latest-generation Ethernet vehicle architectures can achieve latency requirements with minimal CPU load.

With an embedded MRAM as the eNVM, the S32K5 offers write speeds that are 15× faster than conventional eFlash. Write without erase and endurance of 1M write cycles of MRAM helps optimize costs, simplify software development, and minimize downtime for the frequent OTA updates expected in an SDV architecture.

Building the foundation for next-gen SDVs

Zonal architectures are emerging as a necessity for next-gen vehicles because they align with the needs of SDVs. They concentrate compute, reduce complexity, and enable new, software-driven experiences. With the latest generation of MCUs as the building blocks, automakers now have more tools to develop zonal vehicle architectures to deliver the high performance, reliability, and flexibility required by next-gen vehicles.

The post Accelerating the shift to next-gen SDVs with zonal MCUs appeared first on EDN.