Testing automotive radar for ADAS and autonomous driving

Tomohide Yamazaki discusses testing millimetre-wave automotive radar for ADAS and autonomous driving.

Automotive millimetre-wave radar is used to detect people and objects in Advanced Driver Assistance Systems (ADAS) and Autonomous Driving (AD) applications. Typically, sensors to monitor the environment outside the vehicle include radar, cameras and Light Detection And Ranging (LiDAR). ADAS assists the driver in a wide range of driving operations, such as collision warning, automatic braking and parking assist. In autonomous driving, the sensor data is used to control the vehicle automatically.

Automotive radar typically operates in the 24 and 76GHz bands. The 24GHz band is used for short and middle-range perimeter monitoring for pre-crash and collision damage reduction, as well as front and rear monitoring. Due to its high resolution, 76-77GHz radar is used primarily to detect obstacles 100 to 200 metres ahead of the vehicle. Radar is also used for interior or ‘in-cabin’ sensing applications. For example, systems that prevent children or animals from being left in cars operate in the 60-77GHz bands. Table 1 shows the automotive radar frequency allocations used in each region.

RADAR METHODS

Automotive radar measurements use either the pulse or Frequency-Modulated Continuous Wave (FMCW) method. The pulse method measures the round-trip time the radio wave takes from the transceiver to the object and its reflection to come back.

The FMCW method transmits periodic radio signals with an increased frequency in a certain period to an object. Here, the transmitted and reflected radio waves interfere to generate a beat signal. By analysing the difference in frequency and other factors obtained from the beat signal, the distance to an object and its velocity can be determined. Figure 1 shows the signal modulation schemes for automotive radar.

Unlike pulse radar, FMCW offers benefits such as low transmission power and high signal-to-noise ratio. Furthermore, the relatively low response frequency of the transceiver circuit enables a simple design, which reduces costs. Consequently, the FMCW method is widely used in automotive radar.

FMCW BASICS

A signal for which the frequency increases linearly with time is known as a chirp, shown in Figure 2 (a), and is key to the performance of the FMCW signal. From the chirp signal shown in Figure 2 (b), with the vertical axis replaced by frequency, the range or distance resolution and maximum distance range are obtained. These are the main performance characteristics of the FMCW radar. The range resolution Dres is expressed by Dres = c/2B = c/2STc, where c is speed of light, B is the chirp bandwidth (end stop frequency - start frequency), S is the chirp slope, and Tc is the chirp duration.

From the equation, the wider the chirp bandwidth, the higher the resolution that can be detected. For example, the range resolution is about 7.cm for a chirp bandwidth of 2GHz, and the range resolution is about 3.8cm for a chirp bandwidth of 4GHz. The maximum detection range is inversely proportional to the chirp slope S, which represents the rate of increase in frequency. This means that the smaller the chirp slope the greater the maximum detection range. For a fixed chirp duration, a wider bandwidth B will result in a higher resolution. However, this results in a trade-off as the maximum detection range is reduced because the chirp slope increases with bandwidth. This trade-off requires caution when designing automotive radar systems.

Automotive millimetre-wave radar will prioritise detection range or resolution depending on the application. For example, in adaptive cruise control the ability to detect a vehicle at long range is important, while high resolution is not so necessary. On the other hand, collision avoidance requires high resolution as the vehicle needs to respond rapidly to sudden changes at close range.

From the chirp signal, the radar velocity resolution Vres and the maximum detection velocity Vmax can also be calculated using Vres = /2Tc and Vmax = /4Tc where radar wavelength  = c/f.

The maximum detection velocity Vmax is inversely proportional to the chirp duration Tc. Reducing chirp duration increases maximum detection velocity. However, shortening chirp duration adversely affects the range resolution. A radar frame comprises a few to several hundred chirps. A chirp frame is depicted in Figure 3.

The frame time Tf is calculated by multiplying the number of chirps by the sum of the chirp duration and the waiting (idle) time until the next chirp signal is sent out: Tf = (Tc + Twait) x N where Twait is the wait or idle time until the next chirp is sent, and N is the number of chirps. The reason for using multiple chirps within a frame is to obtain Doppler information from the object to accurately ascertain its velocity. There is also a variable off time between each frame that can be used to improve the power efficiency of the chipset.

FMCW SIGNAL MEASUREMENT

Factors that make measuring chirp signals difficult include chirp frequency changes in an extremely short time, ultra-wideband modulation, and higher chirp frequency. Often all three need to be taken into account simultaneously.

Chirp signals are usually measured with a spectrum analyser, which is typically used to evaluate transmission characteristics for wireless communications equipment. There are two types of spectrum analysers — swept and real time. The swept spectrum analyser is based on superheterodyne technology where sampling and signal processing are sequentially repeated operations. Where the chirp frequency changes over an extremely short period, the swept spectrum analyser often cannot keep up due to sequential processing, failing to capture some of the chirps. Sections where the chirp is not captured are referred to as blind spots.

The real-time spectrum analyser, which leverages fast Fourier transforms, performs sampling and signal processing in parallel, enabling it to capture short-time changes in the chirp signal. However, the measurement frequency range or analysis bandwidth of the real-time spectrum analyser is limited to the bandwidth of the instrument, which is typically a few tens to a few hundred MHz. This is inadequate compared to the FMCW chirp bandwidth for automotive radar, which ranges from a few tens of MHz to a few tens of GHz. To address this, it is necessary to measure multiple frequency ranges across the chirp bandwidth and ‘stitch’ the waveforms together. This method can capture the entire chirp bandwidth but the time taken to switch frequency ranges can give rise to a blind spot.

For the most complete solution, a combination of an oscilloscope and a spectrum analyser are often used for chirp measurement. The oscilloscope is fast enough to acquire the full-time and frequency response characteristics of the chirp as well as capture sinusoidal signals. The spectrum analyser is also used to analyse the waveforms acquired by the oscilloscope to evaluate frequency characteristics. 

EVALUATING BASIC FMCW PERFORMANCE

The compact and easy-to-handle ultra-wideband spectrum analyser MS2760A from Anritsu can measure the basic characteristics of automotive millimetre-wave radar signals such as FMCW chirp signal start/stop frequency, bandwidth, amplitude, frame time/period, and number of chirps per frame. Figure 4 shows the results of a 1GHz-band FMCW chirp measurement between 76 and 77GHz using the MS2760A. The instrument captures all the FMCW chirps in a single sweep. After the measurement, the data is processed using a PC.

A key feature of the MS2760A is its ability to cover a continuous frequency range from 9kHz to 170GHz for ultra-wideband applications. Furthermore, the pocket-sized instrument is easy to carry, install and perform measurements in production, test chamber and field test scenarios. These features are made possible by Anritsu’s patented non-linear transmission line (NLTL) technology which eliminates the need for a large mixer for down conversion. The Anritsu NLTL “Shockline” receiver can generate harmonics at very high frequencies and sample up to 170GHz. Due to its compact size, the MS2760A allows installing many spectrum analysers to improve testing and development efficiency and reduce the risk of project delays and expensive capital investment.

As millimetre-wave automotive radar becomes more advanced, the use of the dominant FMCW method creates test challenges due to rapid chirp frequency changes, ultra-wide bandwidth and higher frequencies. NLTL technology addresses these challenges with a single ultra-wideband instrument, the MS2760A, that is not only small but is ideal for basic FMWC radar testing to support ADAS and AD applications.

Tomohide Yamazaki is assistant manager at Anritsu Corporation.