NXP accelerates the development of software defined vehicles with new family of Ethernet switches

A new S32J family of Ethernet switches from NXP will assist OEMs in their transition to software-defined vehicles by simplifying their network architecture and reducing the complexity of the software and hardware integration.

The S32J series shares a common switch core, NXP NETC, with NXP’s latest S32 microcontrollers and processors, allowing them to operate as one expanded virtual switch. The common networking switch core simplifies integration and software reuse with other solutions within the recently announced NXP CoreRide platform and offers OEMs more efficient and re-configurable networking choices.

The S32J provides 80Gbps bandwidth with ports ranging from 10Mb to 10Gb, and powerful dual Arm® Cortex-R52 cores to address diverse requirements of new vehicle architectures. The S32J devices meet TSN automotive standards for managing mixed-criticality traffic, and provide robust ASIL-D safety, hardware security engine (HSE) and MACsec ports.

According to NXP, the combination of the S32J family with the NXP CoreRide platform provides production-grade networking solutions with pre-integrated software and tooling. The solutions include a complete software enablement kit for HSE and MACsec security, TSN stacks and remote configuration and monitoring capabilities. A virtual development kit for the S32J family will be available by the end of 2024. The solution will be available to OEMs and Tier-1 suppliers in 2025.

The NXP CoreRide platform marks a major step forward in helping automakers overcome software and hardware integration barriers, while scaling development efforts for new architectures in software-defined vehicles (SDVs), according to the company.

The platform integrates NXP’s S32 compute, networking, system power management with middleware, OSes and other software from the world’s leading automotive software providers, including Accenture ESR Labs, ArcherMind, Blackberry QNX, Elektrobit, ETAS, Green Hills Software, Sonatus, Synopsys, TTTech Auto, Vector Informatik, and Wind River, Tier-1 suppliers like Valeo as well as integration service providers like Foxconn.

Automotive Industries (AI) asked Günter Sporer, Head of Marketing, Systems and Applications Ethernet Networking Solutions (ENS), how the S32 CoreRide platform addresses the complexity, scalability, cost-efficiency and reduced development efforts required for next-generation vehicles.

Sporer: We have expanded the S32 platform by focusing on networking. Hardware isolation mechanisms allow us to integrate many functions that typically live in individual ECUs (electronic control units). They now live on a single system-on-chip or SOC. With that come isolated execution environments which isolate the different functions in partitions and allows individual suppliers to provide their own functions.

From a software and ecosystem perspective, NXP is integrating more of these components into the SOC than has been done traditionally. Our approach is that we know the capabilities of NXPs S32 SOC hardware better than anyone.

When we are involved in the software development for our platform, we take care of integration and pre validation and can maximize the overall system performance. It also reduces development time for Tier 1s and OEMs as the software is pre-integrated.

There are also cost savings as replacing multiple ECUs with a single SOC reduces the bill of material as well as the overall total cost of ownership.

AI: Is the NXP CoreRide Platform paving the way to a new era of SDVs?

Sporer: There are many industry definitions of a SDV, but if you look at it from a consumer perspective it is a vehicle to which you can add through over the air updates. The necessary hardware is already built in.

This requires a new approach to the physical architecture of control systems and data flows in a car. You need to centralize the computing function because it is easier to manage if you have the software in one place.

To do this you need to aggregate input and output functions, so-called Zonal ECUs, which are physically closer to the actuators and sensors. And this is where you need to have the right semiconductor components.

NXP has a scalable compute solutions, ranging from low end microcontrollers all the way to the SOC interconnected through our end-to-end networking solutions. NXP is a leader in networking technologies such as LIN, CAN, Flexray and Ethernet.

For the development of a software defined vehicle, you need agility. A piece of software that does a certain function may live in an ECU at one point, but needs to be moved for whatever reason later in the development cycle. You may also want to re-use the function and move it to a new architecture.

Our high-speed Ethernet backbone and compute platform enables flexibility during the design phase and subsequent

updates independent of the device.

AI: What demand do SDVs place on automotive networks?

Sporer: The SDV architectures are evolving into tightly integrated Central ECUs, which has functions which may need to be updated through the lifecycle of the car, and the network needs to enable that. Signals are aggregated in Zonal ECUs and edge nodes can be found throughout the car, building the vehicle infrastructure.

While the placement may differ between models, the high-speed Ethernet backbone needs to deliver a certain quality of service, while at the same time it needs to be safe and functional. This, of course, includes cyber security from external attacks. They could come through a radar sensor which is accessible from outside the vehicle and is connected to the Ethernet network.

In order to meet the demands, the network must be designed to ensure it enables critical applications or functions to receive and transmit data in a timely and reliable manner, which relies on the time synchronization and security of the network.

What is perhaps not so obvious is that vehicles go through different states, when they are parked, when they are being charged, or to detect when you approach and unlock once your identity has been verified.

That means certain microcontrollers have to be activated, with the network having the ability to wake up the car. With SDVs you want to be able to dynamically re-configure the network meeting the demands of the evolving software.

AI: Can the NETC network foundation be used across multiple complementary NXP device families?

Sporer: The short answer is yes. We use the NETC networking foundation across the S32 platform. What we have added is an Ethernet management software stack. From a networking perspective, you virtually connect all the NXP processor families, building one large virtual switch.

What is also important is that we are not locking anyone into only using our devices. They system will accommodate any microcontroller through standard Ethernet and PCIe interfaces that connect with our high-performance Ethernet switch.

Of course, if you use NXP processors, you see the benefits of how the two network foundations can be combined to form a single virtual switch. Developers will see only one large virtual switch with many scalable ports.

AI: How is your end-to-end Cortex-A network solution leveraging modern Ethernet and legacy CAN networking portfolios?

Sporer: From a networking perspective, it’s very important that you offer a seamless end-to-end portfolio. The Controller Area Network (CAN) will not go away.

Ethernet is what is now essential, offering speeds from 10 megabits to 10 gigabits in the future. What we are adding to the solution is the virtual switch concept which enables us different entities to share functions through coupling.

Extra features we have added couple them tightly through, for example, a PCI Express link that enables function sharing between the embedded Arm® Cortex-R52 controller cores on the Ethernet switch and S32N processor.

The PCI Express provides memory map sharing and virtualization of networking functions.

This means that the R52 core on the Ethernet switch can use the resources of the S32N, as its own peripheral and vice-versa, which creates a powerful compute subsystem.

It turns an Ethernet switch into a powerful network controller by using the capabilities of the microcontroller subsystem to run a variety of software stacks that can even include vehicle functions.

If the resources on the host processors are busy, you can use the compute resources on the Ethernet switch. In addition, the Ethernet management stack has very powerful network software which, when embedded in the Ethernet switch, turns it into a network controller.

AI: How do you ensure that your dynamic architectures evolve with every new software update?

Sporer: The hardware has certain technical constraints, but if we focus on the workflow, we then think about the needs of the development team. You may have a large group of people spread across the world, working on network software.

NXP offers our customers a network exploration simulation environment service which can be used to analyze all kinds of different topologies. Planning and designing a network is one thing, deploying the configuration is another challenge.

When you deploy a new or updated car network configuration you do not want to physically connect to the hardware. It needs to connect through a cloud using principles from software defined networking, which separates the control plane from data plane.

Developers and designers also need to be able to monitor and debug the network remotely. And, of course, this cycle repeats. Our workflow model supports continuous integration and delivery/deployment (CI/CD) through virtual development kit, which is part of our NXP cloud studio which can be found on nxp.com.

It includes virtual models of our processes to enable agile, dynamic engineering workflow where software updates that may include network configuration can be deployed much easier and faster.

AI: Does your pre-integrated, pre-validated networking platform remove complexity, and accelerate time-to-market?

Sporer: We have a selection of vehicle architecture reference models to start the process with. It is not a “take it or leave it” approach, but should be seen as a good starting point. What you get out of the box is not just hardware for which you download some software examples. It is already pre-integrated and pre-validated to cover the basic functionality of a network, allowing the developer to focus on adapting it to the needs of your application.

OEMs and Tier 1s do not have to do all the development themselves. NXP together with its partner ecosystem ensures that any new system we develop is pre-validated and pre-integrated before it is made available to our customers, accelerating their time-to-market.

AI: What is next for NXP?

Sporer:  We invite the industry to join forces with NXP. We are ready to drive the industry forward with
our NXP CoreRide Networking approach, help OEMs and Tier 1s navigate complex network challenges associated with software-defined vehicles. Our goal is to make the lives of the developers of software defined vehicles easier. We want to help them to accelerate the overall transition to SDVs.