mmWave Sensors

Introduction, Integration, and Implementation

Image Source: AndSus/Stock.adobe.com

By Tenner Lee for Mouser Electronics

Published March 20, 2024
Originally published May 22, 2023

Millimeter wave (mmWave) sensors use a specific frequency band within the electromagnetic spectrum: between the frequencies of 30GHz and 300GHz, or between the corresponding wavelengths of 10mm and 1mm. The wavelengths in which these sensors operate give rise to their name and the nomenclature used to reference them. Due to their operating frequencies, mmWave sensors are sometimes synonymous with radio detection and ranging (radar).

The use of mmWave sensors in devices has increased rapidly in the past few years due to the development of autonomous vehicles, Internet of Things (IoT), smart buildings, and industrial automation, which often incorporate the technology in object detection and ranging systems. Because of this increased use, the cost of mmWave sensors has dropped, fueling further development and adoption. Considering their performance, mmWave sensors are versatile and cost-effective, and they play a key role in multiple technology trends.

To leverage mmWave sensors fully, engineers need a deep understanding of the application being designed and the relative benefits and drawbacks of mmWave sensors. A properly operated, well-designed device in a well-understood environment allows for scalable, low-cost, and effective systems.

The following overview will aid in the understanding of how mmWave sensors work and how to develop them for various applications.

Overview

In almost all cases, mmWave sensors operate as active sensors and thus transmit energy to sense the surrounding environment. Because of their extremely high frequency (EHF) range, mmWave sensors offer multiple benefits, including smaller component sizes, higher resolution, and better accuracy. However, the higher frequencies also create drawbacks such as higher cost, higher attenuation in weather, and higher scattering. Table 1 presents some of the most common benefits and drawbacks of mmWave sensors.

Table 1: Benefits and drawbacks of mmWave sensors versus lower-frequency sensors. (Source: Author)

Optical time-of-flight (ToF) sensors—and more specifically light detection and ranging (lidar) sensors—are often compared with mmWave sensors and have also seen an incredible rise in use. These comparisons are important in determining which type of sensor to use because of the overlap in functionality and general use cases. For example, a mmWave sensor may make more sense than lidar for foreign object debris (FOD) detection at sufficient ranges; lidar point clouds may miss small, thin objects due to the relative spacing between lidar points. Table 2 presents the advantages of mmWave sensors versus the advantages of lidar.

Table 2: Benefits of both mmWave sensors and lidar. (Source: Author)

Components

mmWave sensors are made up of three major subcomponents: An antenna or radiating element, a transmitter, and a receiver. Each of these subcomponents can be divided further depending on the design goals. All three subcomponents are equally critical and require considerable expertise to design and integrate into a working application. 

The antenna array is the radiating element that provides angular resolution. The antenna array also allows the sensor to perform beam steering, null sources of interference, and improve the beam pattern of the sensor. A downside to an antenna array is the large footprint that must be dedicated to the sensor as more antennas are added.

The transmitter and receiver of a mmWave sensor dictate the waveform of the sensor and how well the sensor can process returns from the antenna. Range resolution for radar, for example, is tied directly to the transmitter and receiver through the waveform bandwidth. Transmitter and receiver design (Figure 1) is critical to addressing in-phase and quadrature imbalance, with proper design helping to overcome mmWave sensor performance limits.

Figure 1: Schematic representing the complicated design and integration of subcomponents in a mmWave sensor. (Source: Author)

The radar range equation offers a succinct way to describe how these three subcomponents work together and affect mmWave sensor (radar) performance. Assuming that the mmWave sensor is designed well and ambiguities are handled properly, the relative maximum detection range is

$$R=^4\sqrt{{P_tG_pG_tG_Rλ^2σL}\over{(4π)^3 P_R k_B T_s B_n}}$$

Where P_t is transmit power, G_P is process gain, G_t is transmit gain, G_R is receive gain, P_R is receive power, σ is the target RCS, L is other losses in the system, k_B is Boltzmann's constant, T_s is the system noise temperature, and B_n is the noise bandwidth.

Key Parameters

When discussing mmWave sensors, several key parameters need to be considered. Table 3 presents a basic list of these parameters:

Table 3: Key mmWave sensor parameters. (Source: Author)

In most cases, only a transceiver (i.e., transmitter and receiver) will be provided, omitting the antenna. Unless the mmWave sensor has an integrated antenna array, the design of the antenna is excluded and will need to be designed and integrated later. Generally, this may be the better option as it allows engineers to specify design parameters and tailor the mmWave sensor to the specific application rather than being confined to a set of performance values that may not be applicable.

Measurement and Tracking

With the radar range equation in hand, the following steps offer a very brief overview of how mmWave sensors detect objects:

1. The transmitter sends a signal, usually a linear frequency-modulated signal.
2. The receiver receives the reflected signal, which is mixed with the transmitted signal.
3. The signal passes through a bandpass filter to remove artifacts.
4. The ADC samples the signal.
5. Pulse compression occurs.
6. Range processing occurs.
7. Doppler processing occurs.
8. Angle processing occurs.
9. Detections are formed (e.g., through constant false alarm rate detection).
10. Tracks are formed and objects are identified (e.g., through m-of-n detection, constant-acceleration Kalman filter) (Figure 2).

Figure 2: This schematic displays the signal processing chain for mmWave sensors. Steps show high-level processes and order in which objects are detected and tracked. (Source: Author)

During the process of forming detections, the range resolution, range accuracy, doppler resolution, doppler accuracy, and angle estimation accuracy are all key metrics (Table 4).

Table 4: mmWave sensor basic equations. (Source: Author)

Integration and Implementation

Size, Weight, Power, and Cost

mmWave sensor size and weight should be considered when comparing with other sensors. In terms of size, the antenna is the limiting factor in most cases, as the array may be quite large to meet gain, side lobe, or angular resolution requirements for high-performance applications. The size of mmWave sensors is unlikely to change in the future because several components are tied strictly to the associated wavelengths.

The power consumption of mmWave sensors varies depending on the application. For automotive applications, mmWave sensors are tightly regulated to a maximum equivalent isotropic radiated power (55dBm) by the U.S. Federal Communications Commission. In other applications that require high power, mmWave sensors can scale to meet those demands. As noted in the radar range equation, higher power generally means higher performance. As power consumption (i.e., transmit power) increases, the cost, size, and weight will increase as well.

As noted, the cost of mmWave sensors has decreased substantially in the past few years. mmWave sensors may be more cost-effective than alternative sensors for a given application, but the choice of sensor depends on the desired output from the sensor.

Noise and Artifacts

Sources of noise or artifacts that degrade the performance of mmWave sensors include typical culprits such as thermal noise and phase noise. Thermal noise is the dominant contributing factor in the noise figure of the sensor. Phase noise, on the other hand, occurs in mmWave sensors due to imperfect oscillators or noise on the clock used to generate the transmitted wave. Phase noise contributes to sidebands or a general degradation in the response of the sensor. If the phase noise is bad enough, targets can be masked by the elevated sidebands and degrade the side lobe level of the sensor.

Even with a well-designed mmWave sensor, engineers must consider and mitigate other sources of noise or artifacts, such as clutter, multipath, or interference caused by other mmWave sensors.

Signal Processing Integration

Integration and design of mmWave sensors into any application require special attention. The amount of data at the output of the mmWave sensor—the data cube—may be extremely large, depending on the ADC samples and IF bandwidth. To handle and process the data appropriately, engineers must design a proper signal processing chain. Suppressing artifacts (e.g., clutter, interference) and optimizing tracking performance are prerequisites; thus, a proper compute resource is needed as part of the integration. This compute source could include a field-programmable gate array (FPGA). Graphics processing unit (GPU) radar processing could become an alternative compute source if the signal processing timing meets latency requirements. (GPUs might present significant barriers in tracking applications as compared to FPGAs due to latency and how the digital signal processing chain can be parallelized.)

Communication with a mmWave sensor is generally achieved through a microcontroller that offers SPI, I2C, debug UARTs, and other interfaces. An integrated DSP module is responsible for front-end configuration, control, and calibration. In general, proper integration of waveform design and waveform control into the mmWave sensor is needed. Timing considerations and how waveforms are used in certain cases must be carefully planned out. Modes of operation will need to be defined through the DSP module and the microcontroller in order to influence how the sensor performs.

Mechanical Integration

During mechanical integration of mmWave sensors, engineers should pay attention to any clutter in front of the antenna face as well as proper orientation to the sensor’s frame of reference. Improper orientation degrades performance, introduces unneeded artifacts, and can introduce multipath and false tracks. Extrinsic and intrinsic calibration removes and suppresses artifacts that may appear in the sensor.

Applications

Functionally, mmWave sensors are tied to three broad categories of operation: object detection, characterization, and tracking. Applications of mmWave sensors are found in industrial, robotics, automotive, and other settings.

Industrial Use Cases

mmWave sensors play an important role across a variety of industrial tasks, such as characterizing objects for defects, assuring quality, and tracking inventory in a production line. Importantly, mmWave sensors can penetrate thin materials and characterize materials by the relative reflection within the frequency range of operation. In the industrial vertical, trust in the reliability and performance of mmWave sensors is critical; performing tasks such as FOD detection with low reliability can have detrimental effects.

Robotics/Automotive Use Cases

In the automotive setting, mmWave sensors have become critical in vehicle autonomy up to Level 4. Use cases of mmWave sensors in the automotive industry include object detection, tracking, and characterization. mmWave sensors are crucial in the autonomous automotive sector due to their performance robustness in different types of weather, their ability to scale based on the application, and the reliability of the sensors themselves.

Special Use Cases

Many reading this will have encountered mmWave sensors in airport scanners, which play a critical role across the world. Through the use of mmWave sensors in airports, security teams can identify objects obscured by clothing, preventing invasive pat-down processes. mmWave scanners also provide an alternative to backscatter X-ray systems.

The use of mmWave sensors can also be extended to human tracking and detection, enabling systems to detect heartbeats and track people behind obstructions.

Conclusion

mmWave sensors are derived from a mature and rich technical foundation; a well-designed device in a well-understood environment allows for scalable, low-cost, and effective systems. This brief introductory article presents mmWave sensors and some of the considerations needed to integrate them.