Automatic radar signal analysis speeds up ADAS development
According to a study by the Audi Accident Research Unit, more than 90 % of all accidents on the road can be attributed to human error. Furthermore, the accident rate could be drastically reduced by implementing automated driving (similar to the autopilot systems used in aircraft). Although it might sound like something from a science-fiction novel, automated driving has already become reality in many cars in the luxury class and now increasingly in the medium-price class too. Besides the now classic parking-assistance system, there are other functions to help with everyday driving such as a lane-change assistant, blind-spot detection and adaptive cruise control. While a parking-assistance system is based on a clear yes/no procedure where information is paramount, adaptive cruise control can involve modifying the driving speed in response to the vehicle in front, for example.
Maintaining the flow of traffic
Another reason why automated driving has become such an important topic is related to the rapid development of megacities. The International Energy Agency has noted that as cities like Moscow, Shanghai, Tokyo and Mexico City attain populations of 20 or even 30 million, they are experiencing a dramatic increase in vehicle usage. Today, there are already more than one billion vehicles in the world. In 2025, there will be 1.5 billion vehicles, including 400 million in China alone, where they will be concentrated in metropolitan areas. In such a context, automated driving will no longer be merely a question of road safety and convenience. Instead, it will offer the only way to keep traffic flowing in cities, where the average speed is already less than 20 km/h due to extreme road usage.
Radar technology for automotive applications
Radar technology for use in cars differs in several ways from the military applications for which radar was originally developed. Firstly, the automotive industry is subject to enormous cost pressures, so components must be much more economical. In addition, due to the very limited space for radar sensors behind plastic bumpers, the sensors must be extremely compact.
Compared with camera and ultrasonic applications, radar has a major advantage in that no visual contact is required between the radar sensor and the object to be detected. This saves costs in the production of bumpers and can also be exploited in the vehicle design. However, it is still a challenge to compensate for the attenuation of the transmit and receive signals as they pass through the different layers of material in the bumper as well as the (metallic) paint. Such compensation is implemented with postprocessing in the radar sensor.
For automotive applications, vehicle manufacturers can currently make use of four frequency bands at 24 GHz and 77 GHz with different bandwidths. While the 24 GHz ISM band has a maximum bandwidth of 250 MHz, the 24 GHz ultrawideband (UWB) already offers up to 5 GHz; however, this is allowed only until the end of 2022 due to international regulations. The band that will be available past this date with up to 4 GHz bandwidth lies between the frequencies of 77 GHz and 81 GHz. It is already used for forward-looking applications. Since the signal bandwidth determines the range resolution, it is very important in radar applications. Accordingly, the other allocated frequencies of 122 GHz and 244 GHz for this application with a bandwidth of only 1 GHz will see little use in the automotive industry and are restricted to research projects until further notice.
Speed plus distance
When using radar signals in such applications, developers generally want to simultaneously determine the speed and distance of multiple objects within a single measurement cycle. However, ordinary pulse radar cannot easily handle such a task. Based on the timing offset between the transmit and receive signals within a cycle, only the distance can be determined. If the speed must also be determined, a frequency-modulated signal is used, e.g. a linear frequency-modulated continuous-wave (LFMCW) signal.
Fig. 1 The components of an LFMCW radar signal
The frequency offset between the transmit and receive signals is also known as the beat frequency. It has a Doppler frequency component fD and a delay component fT. The Doppler component contains information about the velocity, and the delay component contains information about the range. This equation has the unknowns of range R and velocity vr, and two beat-frequency measurements are needed to determine the desired parameters. Immediately after the first signal, a second signal with a linearly modified frequency is incorporated into the measurement. If there are multiple targets, however, it is no longer possible to unambiguously determine the beat-frequency pairs for several fast frequency changes ("chirps"). "Ghost targets" are generated that do not really exist. This problem can be solved using various transmit signals with different chirp rates, but the measurement time increases accordingly.
Fig. 2 Speed and distance can be determined within a single measurement cycle using FM chirp sequences.
Determination of both parameters within a single measurement cycle is possible with FM chirp sequences. Since a single chirp is very short compared with the total measurement cycle, each beat frequency is determined primarily by the delay component fT. In this manner, the range can be ascertained directly after each chirp. The Doppler frequency is neglected initially. However, if you determine the phase shift between several successive chirps within a sequence, the Doppler frequency can be determined using a Fourier transformation, making it possible to calculate the speed of the vehicles in front. The speed resolution improves as the length of the measurement cycle is increased. This complex process requires radar components that use state-of-the-art circuits and advanced signal processors with high processing power.
Characterization of the radar signal
Engineers working to develop radar sensors with LFMCW signals are faced with a major challenge: Any deviations from the ideal shape of the transmit signals cause errors in the determination of the velocity and range. Especially in safety-relevant applications, this can have disastrous consequences. Important parameters such as the frequency linearity of a chirp, its length and its reproducibility within a chirp sequence must all be verified.
Signals of this type with rapidly changing frequencies and wide bandwidths can be characterized using a time-domain signal analysis technique known as transient analysis. A spectrum analyzer such as the R&S FSW from Rohde & Schwarz is suitable for this application. A transient analysis option designed for radar applications is available for this instrument. The option allows automatic detection and analysis of linear FM chirp sequences. Important chirp parameters such as the chirp rate, chirp length and chirp rate deviation are displayed in a result table, eliminating the need for manual analysis with marker functions. I/Q-based data analysis lies at the heart of this method. By recording and saving all of the I/Q data, it is possible to determine an analysis range in terms of the frequency, measurement bandwidth and recording time. The results can be displayed in graphical format, making the analysis process more efficient and providing a clearer presentation. The size of this range determines how many chirps are subsequently measured, and the chirp rate deviations relative to an ideal chirp are presented in a result table. The maximum measurement time decreases if a larger measurement bandwidth is selected. In addition, a timing window can be defined in order to neglect transients that occur during the measurement. An ideal chirp of this sort is determined by measuring the average chirp rate and the power.
Analysis of the radar signals begins during the measurement process, since the I/Q data is recorded asynchronously and evaluated. Especially when working with signals with large bandwidths or in case of long measurement times, the duration of the analysis can be significantly reduced in this manner. Various choices are available for displaying the measurement results (e.g. RF spectrum, amplitude/frequency/phase modulation vs. time), and simultaneous display can be enabled. The spectrum analyzer can display the entire data memory, a user-defined interval or individual chirps.
For characterization of FM chirps, the FM linearity is very important, since it influences the accuracy of the object parameters. This can be displayed especially well using the spectrogram mode, which depicts how the signal’s spectrum fluctuates vs. time. Along with the frequency (x‑axis) and time (y‑axis), the signal strength is presented using a color-coding scheme. This provides a good overview of the signal behavior and allows assessment of the timing even for brief signal impairments. If impairments of this sort occur, additional functions can be used for further detailed analysis.
Fig. 3 FM linearity can be well displayed with the R&S FSW using the spectrogram mode. Any deviations are clear at a glance.
The RF spectrum diagram, for example, provides a picture of the measured signal’s overall spectrum at a selected time. It shows the spectrum not only of the wanted signal but also of possible impairments. With the "Frequency Deviation Time Domain" diagram, the frequency errors determined in this manner for a complete chirp can then be demodulated and displayed separately. If you are interested in detecting very low-amplitude impairments, you can additionally use video filters or averaging over multiple chirps to minimize any noise that is present.
Accuracy of the speed measurement
Another important parameter during development of radar sensors is the deviation from the ideal chirp length, since this influences the accuracy of the speed measurement. For this purpose, the measurement results are displayed in a table with the start time and length of the chirp along with the parameters mentioned above. All of the chirps that fall into a previously defined result range are taken into account. Each individual chirp can be identified based on a timestamp. In addition, the chirps are sequentially numbered in a table so they can be better distinguished from one another.
Fig. 4 The R&S FSW displays important chirp parameters such as the chirp rate, chirp length and chirp rate deviation in a result table, eliminating the need for manual analysis with marker functions. For full resolution, click here.
The measurements described here can be performed with the spectrum analyzers up to a frequency of 67 GHz without any additional accessories. In situations requiring measurement of radar signals at frequencies over 67 GHz, harmonic mixers must be used to convert the input signal to the analyzer’s intermediate frequency (IF). It is important for the analyzer to use the highest possible IF, since this results in a wide and unambiguous frequency range. This is especially critical when analyzing broadband signals such as LFMCW signals.
The technologies described in this article can be expected to further influence ‒ if not revolutionize ‒ the world of driving. Radar technology will continue to play a key role. As technology advances, it will be feasible to generate increasingly complex signals with a wide bandwidth, allowing greater resolutions along with improvements in safety. However, the proliferation of signals of this sort, especially at intersections, will make it necessary to ensure that sensors can clearly distinguish the right signals. For example, signal coding can be used in one possible implementation.
As automated driving becomes increasingly prevalent, car-to-car communications (C2C) will supplement the described technologies. C2C is based on WLAN standard 802.11p and will allow future vehicles to communicate with one another as well as with the road infrastructure. For example, construction sites will alert drivers to their presence, and traffic lights will transfer information about coordinated traffic signals to passing vehicles, allowing the driving assistant to take this information directly into account. As in the past, new technologies will come along and influence the world of driving. One thing will stay the same, however: You can keep safe by respecting your fellow drivers!
R&S FSW spectrum analyzer’s key features at a glance
- Frequency range from 2 Hz to 8/13.6/26.5/43.5/50/67 GHz; with harmonic mixers, expandable up to 500 GHz
- Intermediate frequency (IF) from 1310 MHz to 1530 MHz (depending on bandwidth)
- Up to 500 MHz signal analysis bandwidth
- Low phase noise of –137 dBc (1 Hz) at 10 kHz offset (1 GHz carrier)
- Standard-compliant analysis of digital wireless communications standards such as LTE and IEEE 802.11p (car-to-car)
- Multiple measurement applications can be run and displayed in parallel
- High-resolution 12.1" (31 cm) touchscreen for convenient operation.
About the author:
Christoph Wagner heads business development for the automotive market segment at Rohde & Schwarz in Munich. He studied communications engineering at the Deutsche Telekom University of Applied Sciences in Berlin as well as the Copenhagen University College of Engineering.