One of the most common applications of Internet of Things (IoT) devices is sensing, followed by transmission of raw data back to a base station or server for storage and further analysis.
The most common of these sensing tasks seems to be environmental sensing, such as monitoring gas concentrations, temperature, humidity, etc.
These devices are so ubiquitous that they have become ultra-optimized for small footprint and low power consumption, even being available in module formats that can be attached to existing systems.
New mmWave components are pushing IoT systems into new application areas, namely sensor fusion, as well as various types of non-contact sensing and measurement.
Hardware that used to require complex circuit designs or bespoke RFIC designs can now be accomplished with off-the-shelf mmWave ASICs and some embedded software.
In this article, we’ll look at some of the emergent applications of mmWave sensing in IoT platforms.
THREE MMWAVE SENSING APPLICATIONS FOR IOT
Around 2018, some of the first automotive radar ICs from major semiconductor started to hit the market.
These chirped radar ICs were operating in the 24 GHz or 77 GHz bands, and soon after 60 GHz versions of these components would reach the market. Other frequencies are accessible with a range of RFICs and a moderately powerful processor (typically a moderately-sized FPGA).
With digital control over RFICs or other circuitry, it’s possible to implement capabilities such as:
- Beamforming
- MIMO sensing
- Time-of-flight measurements
- Proximity measurements
- Phase measurements
- Chirped pulse generation
Collected data can then be sent back to a central location for further processing, although in many cases the data can be processed on the device as long as there is enough compute.
mmWave Radar + Computer Vision
The most popular application for mmWave sensing is radar, particularly in automotive.
The commercially available automotive radar components largely operate in the 77 GHz band, but there is also an unlicensed 60 GHz band that can be used with off-the-shelf transceivers. The 60 GHz band is useful for radar systems outside of automotive, such as sensor fusion between radar and computer vision.
Radar plus computer vision is a powerful combination of mmWave sensing because radar allows direct measurement of:
- Object distance
- Object size or height
- Speed of an object
- Heading of an object
Why would we need radar in a system that is already equipped with cameras? The reason has to do with the onboard processing power that would be required to use AI inference to estimate an object’s size, distance, speed, and heading from a video image.
In dynamic scenes without a proper reference object used for calibration, video streams require huge amounts of compute to perform inference on high-resolution images captured at a modest frame rate on a frame-by-frame basis.
Radar allows these values to be measured directly. When coupled with a camera image or video stream, a small IoT system with an on-chip radar can identify and track objects with much more accurate determination of velocity and distance compared to a vision-only system.
Respiratory Rate and Heart Rate Detection
Another use of the chirped RF signals used in mmWave radar is in vital signs monitoring. In particular, high-resolution mmWave sensors can use FMCW to detect chest movements associated with breathing without requiring any contact with the patient.
A related measurement, but at even higher repetition rate and resolution, is detecting minute chest movements associated with heartbeats. Both measurements take advantage of Doppler shifts in reflected waves from a patient’s moving chest, just like in a radar system, but at much higher resolution.
The rate of oscillation between red-shift and blue-shift due to the Doppler effect can be extracted using standard DSP algorithms. As the Doppler measurements are captured over the sensor’s field of view, the resulting motion is mapped at each scan point until a DSP routine identifies a respiratory or heart rate oscillation.
Because these are high-resolution measurements, the PCB for an IoT platform with these capabilities would require a large antenna array which scans a scene with very fine angular resolution. This is challenging in compact IoT systems as these antennas are typically printed and can take up most of the available board space on one of the PCB surface layers.
Industrial IoT Systems
Industrial settings demand a range of automated measurements running at all times during production. As factory assets become more connected and share more data, sensor systems start to look a lot more like commercial or consumer IoT products.
In mmWave-enabled industrial sensors, many different measurements are possible:
- Material thickness measurements
- Level sensing
- Proximity sensing
- Vibration sensing and analysis
These sensor measurements can be simple time-of-flight measurements, Doppler measurements, or absorption/transmission measurements.
Depending on the required resolution and the sensor’s field of view, these systems may not require the same array size as used in health monitoring applications or in advanced radars. This helps keep the form factor smaller and reduces system complexity.
As with any industrial sensor, reliability of the design is a persistent challenge. This includes ensuring devices can withstand large temperature swings, ESD, exposure to noxious substances, exposure to water, and system vibration.
While many of the mmWave RFICs target automotive applications, the reliability standards for automotive semiconductors support the use of these components in industrial settings.
WHAT’S NEXT FOR MMWAVE SENSORS?
mmWave sensing capabilities offer new directions for IoT sensing platforms that complement many other sensing systems, particularly ultrasonic and vision. For example, a mmWave + vision system has significant advantages over all-optical computer vision systems based only on cameras or LiDAR.
While the advantages of mmWave sensors for direct measurements are quite clear, a major challenge for many designers is in signal processing of mmWave signals to extract useful information about the surrounding world.
Despite the challenges, designers should expect greater integration and more developer resources from semiconductor vendors to enable integration of mmWave capabilities into IoT products.
