Higher Automotive MCU Performance With Interface IP

By Ron DiGuiseppe and Hezi Saar

AI is making waves across many industries, and automotive is no exception. Today’s vehicles are smarter and more connected than ever, and AI is at the heart of it all. Many new advanced driver assistance system (ADAS) applications, such as automatic emergency braking, adaptive cruise control, and lane-keeping assistance, are built using the latest AI algorithms. In addition to ADAS, AI is increasingly used in additional applications such as battery management, domain/zone control, and electric vehicle (EV) motor control.

Central to bringing AI into vehicle subsystems are microcontrollers designed with parallel processing capabilities to support AI workloads. This is where companies like Infineon come into play. The Infineon AURIX TC4x family of microcontrollers (MCUs) is the company’s latest in automotive MCUs for next-generation e-mobility, ADAS, automotive electrical/electronic (E/E) architectures, and affordable AI applications. To enable the connectivity needed to move data between in-vehicle sensors and the automotive MCUs for processing by AI algorithms, Infineon uses Synopsys Interface IP. The two companies have enjoyed a long history of collaboration to support high-performing automotive applications.

“Our increasingly AI-driven automotive landscape will help reduce system complexity, minimize potential points of failure, reduce emissions by increased efficiency, and improve reliability while lowering costs,” said Dr. Jörg Schepers, VP, Automotive Microcontrollers at Infineon. “With our longstanding collaboration with Synopsys and our integration of Synopsys Controller and PHY IP, including PCIe, MIPI, and Ethernet, into our automotive chips, we are delivering the intelligence and connectivity that are defining the automotive world.”

The impact of vehicle electrification and autonomous features

The automotive landscape is continuing to evolve, with vehicle electrification, autonomous capabilities, and changing automotive E/E architectures dramatically shaping its future. How well applications such as object detection and automatic steering perform relies on the ability of AI algorithms to deliver real-time insights based on vast amounts of data collected by in-vehicle sensors. None of this can happen without high-speed, low-latency connectivity.

These capabilities bring to life a variety of AI-driven automotive applications, including:

• Motor control: Rather than relying on expensive hardware sensors to monitor and control a vehicle’s electric motor, AI can use predictive analytics to control the rotation of the rotor inside the motor.
• ADAS: AI algorithms can help identify what the radar or LiDAR in an ADAS are seeing in the proximity and distance of the car, enabling more accurate performance of systems such as autonomous driving, lane keeping, object detection, and automatic braking.
• Battery management: Machine learning algorithms can help monitor battery health. And, based on this insight as well as the driving profile, AI can help to accurately predict the battery’s remaining useful life. For fast charging, AI can control the state-of-charge for each single cell and in combination with low-latency clusters, achieve optimized cell balancing for the entire battery pack in real time.
• Navigation: AI can help guide drivers on the fastest or most efficient routes to their destinations, saving time as well as fuel/electricity.

Collaborating to deliver quality, reliability, and safety

Together, Infineon and Synopsys are providing the technology foundation for today’s and tomorrow’s vehicles. Scalable and ASIL-D-compliant, Infineon AURIX TC4x MCUs feature the company’s Tri-Core 1.8 architecture and its AURIX accelerator suite, including a new parallel processing unit (PPU) and multiple smart accelerators.

To support the 5Gbps Ethernet, 10BASE T1S Ethernet, PCI Express, and MIPI D-PHY interfaces in its MCU family, Infineon relies on Synopsys Interface IP, which provide leading power, performance, area (PPA), and security for the most widely used protocols. The automotive-grade IP are designed and tested for AEC-Q100 quality standards to ensure additional reliability and are compliant to the ISO 26262 functional safety standard to help with SoC-level assessment and qualification for target ASILs. In addition, the PPU in its AURIX TC4x family, which accelerates AI algorithms such as recurrent neural networks (RNNs), radial basis function neural networks (RBFs), convolutional neural networks (CNNs), and multi-layer perceptron, is powered by Synopsys ARC EV Processor IP. Mutual customers can get a head start on software development by using the Synopsys ARC MetaWare Toolkit for AURIX TC4x, which provides a complete suite of tools, runtime software, and libraries to program the PPU, and Synopsys Virtualizer Development Kit for virtual testing and evaluation of automotive systems.

Synopsys and Infineon are both committed to enabling high-performing automotive systems. For Infineon, it was important to work with a vendor who could support its automotive IP requirements now and into the future, delivering on reliability, quality, and safety. Synopsys, in turn, stays up to date with automotive standards as they evolve to ensure that our broad breadth of IP meets the latest requirements.

As AI algorithms become increasingly sophisticated, automotive OEMs, Tier 1s, and semiconductor vendors will need more powerful processing capabilities and super-fast connectivity to infuse greater intelligence into their vehicles. Fortunately, companies like Synopsys and Infineon are continuing to team up to define what is possible for the next generation of automotive designs.

Hezi Saar is senior director of product line management at Synopsys.