Technology trends in automotive millimeter wave radar are primarily moving towards higher resolution, cost reduction, and integration.

Higher resolution enables more detailed recognition of the vehicle's surrounding environment, improving the detection accuracy of vehicles, pedestrians, and obstacles. This enhances vehicle safety and enables responses to complex traffic situations.

Achieving autonomous driving requires installing numerous sensors to recognize the surrounding environment. Reducing the cost of these sensor systems is a crucial point in promoting the adoption of autonomous driving. Semiconductor manufacturers producing millimeter wave radar chips are developing products that integrate components such as radar transmit and receive functions, high-frequency analog circuits, ADCs, signal processing DSPs, and processors onto a single chip. Integrating millimeter wave radar functions onto a single chip reduces the system implementation area, enabling miniaturization of millimeter wave modules. Also, because processing from the high-frequency front end to final processing can be handled seamlessly within the package, radar system development costs can be reduced. This seamless integration of processing, from the high-frequency front end to final processing, reduces complexity and contributes to lower radar system development costs.

As a crucial technology for enhancing vehicle safety and convenience, we can expect automotive millimeter wave radar to continue to evolve.

Automotive millimeter wave radar has two types of architectures: edge radar and distributed radar.

Each of these types is described below.

Edge Radar Architecture

With this architecture, radar sensor modules are placed in various locations on the vehicle. In edge radar, each module independently handles tasks like object clustering (grouping) and tracking, sending only object-level data to the central processing unit. An advantage of edge radar is the reduction in data traffic between the sensor modules and the central processing unit. However, this approach increases costs because individual radar sensors require features such as high-performance CPUs and dedicated signal processing DSPs (SPT: Signal Processing Toolbox). Additionally, the central processing unit must integrate the object data received from each sensor.

Edge Radar Architecture

Distributed Radar Architecture

With this architecture, radar sensor modules are placed in various locations on the vehicle. In distributed radar (distributed cooperative type), individual radar sensors do not perform processing. Instead, preprocessed data is gathered from the sensors by the central processing unit for collective processing. Because all sensor data is processed together in the central processing unit, distributed radar allows for efficient signal processing and enhances the safety of ADAS functions. Costs can also be reduced because individual sensors do not require features such as high-performance CPUs or signal processing DSPs (SPT). By collectively processing the preprocessed data received from individual sensors, the central processing unit achieves low latency and improves both resolution and object recognition performance. However, this method increases the volume of communication data, as sensors send large amounts of preprocessed data to the central processing unit. This necessitates high-speed data communication interfaces, such as Gigabit Ethernet, capable of managing the increased data traffic between the sensors and the central processing unit. For this reason, the central processing unit requires extremely high processing power to handle the vast amounts of data simultaneously.

Distributed Radar Architecture

Today, edge radar is the predominant architecture for automotive millimeter wave radar systems.

Semiconductor manufacturers are currently advancing the development of products like single-chip radar ICs and high-performance, radar specific processors compatible with Gigabit Ethernet. As these radar chipset technologies evolve, it is anticipated that future automotive millimeter wave radar architectures will shift towards distributed systems.

• Reference

• Four reasons to consider using distributed radar architecture

SAF85xx: high-performance 77 GHz RFCMOS automotive radar single chip SoC [under development]

The SAF85xx is a radar sensor compatible with centralized radar architectures.

It integrates essential functions for radar processing onto one chip, including the RF front end, a multicore processor, and an SPT (radar signal processing DSP).

The SAF85xx RFCMOS automotive radar system on chip (SoC) is a high-performance single-chip product for automotive FMCW radar applications, optimized for fast chirp modulation. (excerpt) "...this 3rd generation RFCMOS radar device supports short, mid, and long-range radar applications in a small form factor, enabling everything from entry-level NCAP sensors to high-performance front radar sensors.

This fully integrated RFCMOS chip includes four transmitters, four receivers, ADC conversion, phase rotators, a low phase noise VCO, SPT radar accelerator, BBE32 Vector DSP, Arm Cortex A53, Arm Cortex M7 cores, and SRAM. It supports a wide range of applications and various radar data outputs, such as object data, point cloud data, and FFT output."

SAF85xx Block Diagram

Source:

• High-Performance 77 GHz RFCMOS Automotive Radar Single-Chip SoC (NXP Official)

Programs are stored in external flash memory. It is equipped with two CAN FD channels and one Gigabit Ethernet channel for external communication.

For more detailed information, please visit NXP's official website: High-Performance 77 GHz RFCMOS Automotive Radar Single-Chip SoC.

• High-Performance 77 GHz RFCMOS Automotive Radar Single-Chip SoC