Applications for Short-Range Radar in IoT and Embedded Designs

Why a crossover technology deserves serious consideration

The horizon for Internet of Things (IoT) applications is broad and perpetually expanding. Many, if not most, of these applications rely on embedded sensors to perform task-critical measurements, or as an integral part of a control loop. Often an IoT end-point’s capabilities and limitations are ultimately coupled to the characteristics of its sensors since they provide the “last mile” to the physical world.

Novel sensor developments can enable new applications. This also holds true when existing sensor technology used in one industry is optimized for another. For example, much of the original work to commercialize high-reliability, low-cost MEMS accelerometers was driven by the automotive industry in the 1990s. Today this sensor technology is an essential element in fitness tracking watches.

Smart Streetlights to Drone Altimeters
One of the latest “crossover” sensor technologies for IoT and Smart Lighting is embedded radar. When a product benefits from measuring or detecting a remote object’s presence, velocity, direction of travel and distance, short-range radar may be worth serious consideration. Some of the most obvious applications include perimeter and area security, room occupancy detection for lighting and HVAC control, automatic door and gate controls, robotics, smart streetlights (streetlights that increase brightness along the path of an approaching vehicle), UAV drone altimeters, industrial fluid/solid level sensing, and speed measurement.

The operating principles of radar aren’t new, of course. Radar (RAdio Direction And Ranging) has roots stretching back to Heinrich Hertz’s experiments with radio waves in the 1880s. It was developed into a practical air defense tool by Sir Robert Watson-Watt and others prior to World War II. All radar systems are based on the principle of bouncing a radio wave off a remote target. In practice, however, various implementations affect cost, complexity and capability. Most IoT, automotive, and embedded applications typically use radar architectures commonly referred to as “short-range,” Unlike conventional high-power radar systems intended for navigation and military/defense, short-range radar systems are implemented differently due to the relatively close distance to their targets (i.e., 10’s of meters vs. 100’s of kilometers), potential for increased signal “clutter” in their near-field target sensing area, low-power RF transmission levels, miniature size, etc.

Furthermore, there are several types of short-range radar, including continuous wave (CW) Doppler, Frequency Modulated Continuous Wave (FMCW), Pulse Doppler, and ultra-wideband (UWB). For the purposes of this article I’ll review the first two. Both CW Doppler and FMCW are widely available as RF chipsets and operate at low power levels, often important in IoT and embedded designs.

A Quick Refresher on Doppler Shift
A Doppler shift is the observed change in frequency of a wave reflected from a target back to its source when the source and target are moving closer together or further apart. As in Figure 1(a), when the target moves toward the source, each successive wave crest is reflected from a position closer to the source than the previous wave, effectively compressing them and causing an increase in frequency. Conversely, as the target is moving away from the source, each wave is reflected from a position farther from the source than the previous wave, so the arrival time between successive waves is increased, reducing the frequency.

Figure 1: CW Doppler Concepts

In certain applications, a target’s direction and velocity are important, but not its absolute distance from the radar. Police radar guns, security (detecting movement in a secure area), smart streetlights, door and gate controls, room occupancy detection, speed monitors, and sports equipment might fall into this category. Low-cost short-range CW Doppler radars are well suited for these systems. A typical CW Doppler radar arrangement using quadrature mixing can be implemented in a few ICs, similar to Figure 1(b). Like most small CW Doppler systems, this example is based on a coherent architecture, in that the same oscillator is used for both receive and transmit functions. This allows the radar to measure small phase changes in the reflected carrier. When the target’s velocity is positive, the I channel will lead the Q channel; and when the target’s velocity is negative, Q will lead I. These signals are digitized and fed to a microcontroller for processing. Firmware can derive the amplitude of the reflection and compare it against a trigger threshold (to suppress small or false triggers) and also compare the received phase change and direction against stored parameters for velocity threshold, etc. In this manner the microcontroller can be used to trigger an alarm only when a person attempts to enter an airport security checkpoint from the “exit” side, and so on. Note that when the target is not moving, there is no Doppler shift, thus no phase change for it to detect. As a result, basic phase coherent CW radars cannot “see” static objects, and therefore they do not provide distance (ranging) information.

Short-range FMCW radar uses a different approach; it’s a better choice for applications that require target ranging data, such as altimeters, industrial tank level sensors, and “smart” IoT parking garages that track and tally which parking spots are occupied. The tradeoff is that FMCW is somewhat more expensive to implement due to added components, and it generally requires a faster microprocessor.

FMCW uses a continuous carrier wave like CW Doppler. However, the carrier is frequency modulated. The modulation scheme is often a linear sawtooth or triangular waveform swept across a set bandwidth using a voltage controlled oscillator (VCO) or in steps. The ramp repeats at a given rate, called the chirp rate. When the radar’s FM carrier is amplified and transmitted, a portion of it is reflected back by the target. Received signals will encounter a delay as they travel to and from the target (its time of flight). And the frequency of the carrier is being varied over time, so the instantaneous frequency of the delayed signal appearing at the radar’s receiver will be shifted—and proportional to distance—because the waveform is now at a different point along its ramp, per Figure 2(a). The greater the distance to the target, the greater the shift in frequency. The FMCW architecture mixes this frequency-shifted reflected signal with the original modulated carrier, amplifies it, and passes it through a filter. The result for a non-moving target is a constant frequency offset known as the beat frequency. With a bit of math, it reveals distance to the target. Interestingly, if there are multiple targets in range, multiple beat frequencies will be superimposed on each other. The distance and phase of each target can be determined individually by using Fourier analysis.

Figure 2: FMCW Concepts

Why Automotive is Looking to 77-81GHz
Beyond radar architecture, other considerations are important, such as frequency selection, antenna design, and commercial availability of development tools. Most short-range radar systems operate in the gigahertz range. Others, particularly in defense applications, are optimized for extremely long distances and may operate as low as the HF range (10’s of MHz). As you might expect, performance, resolution, and antenna size are interrelated. In general, radars that operate at high frequencies are smaller, require less power, are more directional and offer better resolution, but they also have increasingly limited range.

Many CW Doppler and FMCW radars operate in the 24GHz industrial, scientific and medical (ISM) band, since it is unlicensed and offers sufficiently high bandwidth for most applications. Previous generations of automotive short-range radar favored 24GHz technology, and semiconductor companies were encouraged to make significant investments in monolithic microwave integrated circuit (MMIC) design and fabrication based on that band. More recently, however, the automotive industry has been transitioning from 24GHz to 77-81GHz to facilitate improvements in blind-spot detection resolution, lane-change assist, autonomous emergency braking, and adaptive cruise control.

Design Help
Radar antenna design at any frequency range is not trivial, particularly at the upper end of the spectrum. To make this easier, there are specialized radar antenna design firms in the Americas, Europe, and Asia. MMIC manufacturers may also provide reference antenna layouts as Gerber files and/or source code for their evaluation modules.  I found that very helpful a few weeks ago when developing a miniature CW Doppler test unit to compare against incumbent passive infrared (PIR) occupancy switches used in lighting applications.

In theory, short-range radar would be a significant improvement; not only is it more sensitive to small movements in a room (keeping the lights on while you’re working), the velocity and direction data could be used to illuminate footpaths and roadways in advance of approaching pedestrians or vehicles based on their speed. To build my prototype I used an Infineon Sense2GoL evaluation board (Figure 3). It’s a complete 24GHz CW Doppler radar with antenna, assembled on PCB about 2.5 cm square. The evaluation package included source code, design examples, and a USB-hosted debugger. The unit’s transmit and receive antennas are traces on the underside of the PCB.

Figure 3: The 24GHz CW Doppler board I used in my miniature CW Doppler test unit prototype

I didn’t choose FMCW for this application because the specifications didn’t require measuring distance. (If distance or angle information are needed, FMCW-based evaluation boards are available as well.) The results were impressive. In our tests, the prototype could detect motion from at least 30 meters away, which is about 2-3x further than a typical PIR switch. This means fewer would be needed to cover a large area. By comparing the detected object’s trending velocity against a set of window values, it can also differentiate whether the target is a pedestrian or vehicle with reasonable confidence. And finally, since 24GHz signals can pass through most plastic materials, it’s possible to mount the entire sensor inside an IP67-rated or other sealed enclosure. This is challenging for PIR switches—they require an IR-transmissive Fresnel lens.

Low-cost, short-range radar technology is available today, along with accessible design tools. In many cases it can extend the performance and capabilities of IoT, embedded and Smart Lighting applications. It’s also likely to foster the development of entirely new and unanticipated IoT product categories, just as low-cost MEMS accelerometers have done in the past.

Cary Eskow is Vice President of LightSpeed at Avnet. He has published over 80 articles and papers on various aspects of optical, laser and LED-based system design, and is a frequent speaker at international LED, green photonics, photobiology and medical electronics conferences. His special interests include photophysiology, IoT security, closed-loop sensing and high-power UV. Eskow is a member of UL-8750 Standards Technical Committee for LEDs at Underwriters Laboratories, advisory board member of Strategies in Light and of LIGHTFAIR International. His prior development work includes systems for pulmonary delivery of micronized insulin and LED-based optical image classifiers. Eskow’s first patent for optical sensing was issued two decades ago.

Share and Enjoy:
  • Digg
  • Sphinn
  • Facebook
  • Mixx
  • Google
  • TwitThis


Extension Media websites place cookies on your device to give you the best user experience. By using our websites, you agree to placement of these cookies and to our Privacy Policy. Please click here to accept.