Adv. Radio Sci., 8, 55–60, 2010 www.adv-radio-sci.net/8/55/2010/ doi:10.5194/ars-8-55-2010 © Author(s) 2010. CC Attribution 3.0 License.

Advances in Radio Science

Automotive radar – investigation of mutual interference mechanisms M. Goppelt1 , H.-L. Bl¨ocher2 , and W. Menzel1 1 Institute 2 Group

of Micowave Techniques, University of Ulm, 89081 Ulm, Germany Research & Advanced Engineering, Daimler AG, 89081 Ulm, Germany

Abstract. In the past mutual interference between automotive radar sensors has not been regarded as a major problem. With an increasing number of such systems, however, this topic is receiving more and more attention. The investigation of mutual interference and countermeasures is therefore one topic of the joint project “Radar on Chip for Cars” (RoCC) funded by the German Federal Ministry of Education and Research (BMBF). RoCC’s goal is to pave the way for the development of high-performance, low-cost 79 GHz radar sensors based on Silicon-Germanium (SiGe) Monolithic Microwave Integrated Circuits (MMICs). This paper will present some generic interference scenarios and report on the current status of the analysis of interference mechanisms.

1

Introduction

In the automotive field, radar sensors are key components for comfort and safety functions, for example adaptive cruise control (ACC) or collision mitigation systems (CMS). With an increasing number of automotive radar sensors operated close to each other at the same time, radar sensors may receive signals from other radar sensors. The reception of foreign signals (interference) can lead to problems such as ghost targets or a reduced signal-to-noise ratio. Figure 1 shows such a simple automotive interference scenario with direct interference from an oncoming vehicle. Up to now, interference has not been considered as a major problem because the percentage of vehicles equipped with radar sensors and therefore the probability of interference was low, and the sensors were used mainly for comfort functions. In this case it may be sufficient to detect interference and turn off the function for the duration of the interference. On the contrary, safety functions of future systems require very low failure rates. So in spite of a predicted higher number of radar systems, the probability of interference-induced Correspondence to: M. Goppelt ([email protected])

problems has to be reduced considerably. Therefore effective countermeasures have to be introduced to minimize mutual interference even with high traffic density (e.g. in large cities) and a rising percentage of vehicles equipped with radar sensors. In Germany currently three frequency bands are assigned to automotive radar applications: – 24 GHz ultra-wideband (UWB) – 76–77 GHz – 77–81 GHz In addition, some systems also are operated in the 24 GHz ISM band. Detailed information about the frequency bands can be found in Table 1. New vehicles can be equipped with 24 GHz UWB sensors only until 2013. From then on, the 79 GHz frequency band with up to 4 GHz bandwidth will replace the 24 GHz band. From 2004 to 2007, the Daimler AG took part in the German Federal Ministry of Education and Research (BMBF) funded joint project KOKON. Aim of the project was to demonstrate the feasibility of 79 GHz automotive radar sensors with SiGe-based MMICs. Presently, the Daimler AG contributes to the successor project “Radar on Chip for Cars” (RoCC) running from 2008–2011. Special emphasis of this projects lies on the following items: – reducing cost and size – improving the radio-frequency (RF) packaging technology – improving sensor performance – improving sensor reliability The last item includes the investigation of mutual interference and interference minimization techniques (countermeasures). The specific tasks are

Published by Copernicus Publications on behalf of the URSI Landesausschuss in der Bundesrepublik Deutschland e.V.

56

M. Goppelt et al.: Automotive radar – investigation of mutual interference mechanisms

Table 1. Frequency bands and equivalent isotropically radiated power (EIRP) for automotive radar applications in Germany. 24 GHz UWB short-range

24 GHz ISM Fig. short-range

77 GHz

79 GHz

frequency range (GHz)

21.650–26.650

24.000–24.250

76.000–77.000

77.000–81.000

center frequency (GHz)

24.150

24.125

76.500

79.000

max. average power (EIRP) (dBm/MHz)

−41.3

1. Simple automotive interference scenario with one long-range short-range target and one interferer.

−9

max. average power (EIRP) (dBm)

50

max. peak power (EIRP) (dBm)

20

55

55

KH2009-B-016 KH2009-B-016 Fig. 2. Block diagram of a simple generic FMCW radar.

Fig. 1. Simple automotive interference scenario with one Fig. Simple interference scenarioscenario with one target target and oneautomotive interferer. Fig.1.1. Simple automotive interference withand one one interferer. target and one interferer.

Fig. 4. Diagrams illustrating a case with interference between two FMCW radars causing a uniform increase of Fig. floor. 4. Diagrams illustrating a case with interference the noise

Fig. 2. Block diagram of a simple generic FMCW radar.

Fig. 2. Block diagram of a simple generic FMCW radar.

Fig. 2. Block diagram of a simple generic FMCW radar.

– to identify interference mechanisms, – to find efficient countermeasures and – to test countermeasures in the laboratory and with radar sensors in vehicles. At the beginning of Sect. 2 of this paper, two radar architectures are defined, and the analysis approach to interference mechanisms between different types of radar sensors is outlined. Sections 2.1 to 2.4 illustrate the interference mechanisms. Target and interference power levels are analyzed in Fig. 3. Diagrams illustrating a case where an FMCW radar Sect. 2.5 for the simple interference scenario in Fig. 1. interferes with another FMCW radar generating a ghost target. The upper diagram shows the frequency course of Fig. 3. Diagrams illustrating casethe where an received FMCW radar the local oscillator signals Adv. Radio Sci., 8,signal 55–60,(black) 2010a and interferes with another FMCW radar left-hand generating a ghost by the receive (RX) antenna. The lower diagram target.the The upper diagramIFshows the frequency course shows down-converted frequencies as a function of of time, the lower right-hand transform the local oscillator signaldiagram (black)the andFourier the signals received

two FMCW radarsacausing a uniform increaserada of Fig. between 3. Diagrams illustrating case where an FMCW the noise floor. Fig.interferes 3. Diagramswith illustrating a case where anradar FMCW radar inter- a ghost another FMCW generating feres with another FMCW radar generating a ghost target. The target. The upper diagramcourse shows thelocal frequency upper diagram shows the frequency of the oscillatorcourse of the local oscillator signal (black) and the signal (black) and the signals received by the receive (RX) signals antenna. received The by lower diagram the down-converted frequen- diagram theleft-hand receive (RX)shows antenna. The lowerIFleft-hand cies shows as a function time, the lower right-hand diagram the Fourier the of down-converted IF frequencies as a function of transform of the IF signal. time, the lower right-hand diagram the Fourier transform of the IF signal. 2

Interference mechanisms Fig. 5. Block diagram of the waveguide-based Since there are many different radar the sensors onofthe automeasurement setup to demonstrate effects motive market, it was decided to start the analysis of ininterference on an FMCW radar. Fig. 5. Block diagram the waveguide-based terference mechanisms with a of simple generic frequencymeasurement demonstrate effects of modulated continuous setup wave to(FMCW) radarthe (Fig. 2) and interference on an FMCW radar. a simple generic pulsed radar with correlation receiver (Fig. 10). www.adv-radio-sci.net/8/55/2010/

between FMCWillustrating radars causing uniform incre Fig. 4.two Diagrams a casea with interfere the noise floor. between two FMCW radars causing a uniform in

Fig. 2. Block diagram of a simple generic FMCW radar.

KH2009-B-016

M. Goppelt et al.: Automotive radar – investigation of mutual interference mechanisms the noise

floor.

57

ne

Fig. 5. Block diagram of the waveguide-based Fig. 5. Block diagram of the waveguide-based measurement setup to demonstrate the effects of Fig. 5. Block diagram of the waveguide-based measurement the setupeffects of measurement setup to demonstrate interference on an FMCW radar. to demonstrate the effects of interference on an FMCW radar. interference on an FMCW radar.

illustrating case where an FMCW radar Fig. 4. Diagramsaillustrating a case with interference r.3. Diagrams Fig. 3. Diagrams illustrating a case where an FMCW radarof Fig. 4. Diagrams illustrating a casegenerating with interference between two rferes with another FMCW radar a ghost between two FMCW radars causing a uniform increase FMCW radars causingFMCW a uniform radar of the noise floor. with another generating a ghost the noise floor. et.interferes The upper diagram shows theincrease frequency course of

target. The upper diagram the signals frequency course of local oscillator signal (black)shows and the received the local oscillator signal (black) and the signals received In a first step the The interference between the he receive (RX) antenna. lower mechanisms left-hand diagram by the receive (RX) antenna. The lower left-hand diagram different types of radar were roughly identified, followed ws the down-converted IF frequencies as a function ofby simulations and measurements to verify theas approximations shows the down-converted IF frequencies a function of e, the lower right-hand diagram the Fourier transform and to develop a more precise understanding of interferthe lower right-hand diagram the Fourierthe transform hetime, IF signal. ence mechanisms. The last item, however, has not yet been of the IFcompleted. signal. Previous research on interference mechanisms

adar t of ed am of m

had been conducted by – Brooker (2007): Investigation of the interference between an FMCW radar and a pulsed radar and propositions for interference minimization techniques. Fig. 5. Block diagram of the waveguide-based

Fig. 6. Measured FMCW radar range profile with a

Fig. 6. Measured FMCW radar range profile with Fig. 6. Measured FMCW radar range profile with a ghost target target generated by interference from another FMC generated bytarget interference from another radar. generated by FMCW interference from another FM

radar. radar.

In the second case (Fig. 4) the frequency ramp of the interfering signal starts earlier in time, consequently the crossto (2005): demonstrate the of effects of ing of the ramps lead to an IF signal with fast V-shaped fre–measurement Oprisan and setup Rohling Analysis interference quency variation over time, resulting in a uniform increase interference on an FMCW radar. effects between continuous wave (CW), frequency shift of the noise floor over the whole IF-band. Thus, the target keying (FSK), frequency-modulated continuous wave signal-to-noise-ratio can be reduced considerably. (FMCW), frequency-modulated shift keying (FMSK) Both cases were demonstrated in the laboratory using a and pulsed radars. Some interference reduction techwaveguide-based measurement setup (Fig. 5). The target was niques are proposed. simulated using an approximately three meters long waveg– Tullsson (1997): Investigation of the interference senuide terminated with a short circuit. The interfering signal sitivity of an automotive FMCW radar and analysis of was added to the reflected signal using a directional coupler. interference elimination techniques. Figure 6 shows the result for the first case (parallel frequency ramps). The ghost target’s broad spectrum originates from the fact that there is no correlation between the phase noise 2.1 FMCW radar with FMCW interference of the interfering signal and of the radar’s local oscillator sigFig. 6. Measured FMCW radar range profile with a ghostnal. To illustrate the effects of interference from an FMCW radar generated another ontarget an FMCW radar, by twointerference cases were from selected. BothFMCW radars For the second case, a CW signal was coupled into the reradar. have the same triangular FMCW modulation (same ramp duceiver of the FMCW radar. In the frequency domain, the CW ration and sweep bandwidth). In the first case (Fig. 3) the signal intersects the ramps of the FMCW radar just like the 4 FMCW signal in Fig. 4. The upper diagram of 4interfering interference leads to a constant intermediate frequency (IF) generating a second line in the radar range profile. The secFig. 7 shows the resulting IF signal in the time domain. The ond line then is interpreted as an additional target (ghost tarIF signal generated by the interfering CW signal is short in get). relation to the ramp duration . This is due to the fact that the www.adv-radio-sci.net/8/55/2010/

Adv. Radio Sci., 8, 55–60, 2010

Fig. 7. MeasuredKH2009-B-016 FMCW radar IF signal and range profile M. Goppelt et al.: Automotive radar – investigation with CW interference. of mutual interference mechanisms

58

Fig. 8. Diagrams illustrating the case of pulsed si interfering with an FMCW radar. The left-hand d Fig. 8. Diagrams illustrating the case of pulsed signals showsillustrating the spectrum ofpulsed a pulsed the right-h Fig. 8. Diagrams case of signals signal, interfering interfering with an FMCWthe radar. The left-hand diagram with an FMCW radar.shows left-hand diagram shows the spectrum thesignal, FMCW signals. Some o shows thediagram spectrum ofThe a pulsed the radar right-hand of a pulsed signal, the right-hand diagram shows the FMCW radar equidistant spectral lineslines ofofthe interfering pulsed diagram thetheFMCW radar signals. the signals. shows Some of equidistant spectral the of interfering Fig. 10. Some Block diagram of a si intersect the FMCW radar equidistant spectral of theradar interfering pulsed signal pulsed signal intersectlines the FMCW ramp. ramp. intersect the FMCW radar ramp.pulsed radar with correlation

Fig. 7. Measured FMCW radar IF signal and range profile

Fig. 7. Measured FMCW radar IF signal and range profile with CW with CW interference. interference.

radar’s IF bandwidth is much smaller than the radar’s frequency deviation (e.g. by a factor of 1000). The IF noise floor of the radar without interference is below −60 dBm, the noise floor increases to about −45 dBm due to the interfering CW signal as can be seen in the lower diagram of Fig. 7. 2.2 FMCW radar with pulsed interference

A similar situation results, when a pulse modulated signal Fig. 9. Measured FMCW radar IF signal and range profile interferes with an FMCW radar. The spectrum of a pulsed with pulsed signal interference. signal consists of equidistant lines with a sin(x)/x envelope Fig. 9. Measured FMCW IF signal and rang Fig. 11.radar Reception bandwidth Fig. 8. Diagrams illustrating the case of pulsed Fig. 9.signals Measured FMCW radar IF signal and range profile with for rectangular pulses. This is depicted in the left part of with pulsed signal interference. pulsed radar without interfere pulsed signal interference. interfering withthe anFMCW FMCW radar. The left-hand diagram Fig. 8. Some lines intersect radar ramp leading to a short IF signal V-shaped of frequency course similarthe right-hand shows thewith spectrum a pulsed signal, to the case shown in Fig. 4. diagram shows the FMCW radar signals. Some of the tD (Fig. 10) is successively increased until the maximum sigAn exemplary measurement with pulsed signal interferequidistant of the theresulting interfering pulsed nal flight signal time of interest is reached. An IF signal (in prinence is shown in Fig. 9.spectral It can be lines seen that IF 5 cipal a correlation function) appears at the mixer output if intersect theisFMCW signal (upper diagram) wider thanradar in the ramp. CW case (upper a radar signal delay and tD are equal. This is done in such diagram of Fig. 7). The resulting increase in the noise floor is a way that every range cell is sampled several times, and not as uniform as in the CW case (lower diagrams of Figs. 7 the resulting output signals are integrated in a lowpass filand 9). ter where the resulting amplitude is detected. Following this, the sampling process starts again with the lowest time de2.3 Pulsed radar with FMCW interference lay. In practice, the mixer in Fig. 10 has to be substituted Pulsed radars with correlation receiver, as typically used for by a pair of mixers to get the quadrature signals. A pulsed automotive applications and also selected for this investigaradar sensor with correlation receiver is described in detail in tion, work similar to sampling oscilloscopes. The time delay a publication by Gresham et al. (2004). Adv. Radio Sci., 8, 55–60, 2010

www.adv-radio-sci.net/8/55/2010/

KH2009-B-016

Fig. 11. Reception bandwidth and range profile of the Fig. 8. Diagrams illustrating the case of pulsed signals KH2009-B-016 pulsed radar without interference. M. Goppelt et al.: Automotive radar – investigation of mutual interference mechanisms 59 interfering with an FMCW radar. The left-hand diagram shows the spectrum of a pulsed signal, the right-hand diagram shows the FMCW radar signals. Some of the equidistant spectral lines of the interfering pulsed signal intersect the FMCW radar ramp.

Fig. 10. Block diagram of a simple generic (coherent)

10. Block diagram of a simple generic Fig.Fig. 10. Block diagram a simple generic (coherent) pulsed(coherent) radar pulsed radarof with correlation receiver. radar with correlation receiver. withpulsed correlation receiver.

ange profile

file

Fig. 12. Reception bandwidth and range profile of a pulse

Fig. 12. Reception bandwidth and range profile of a pulsed radar radar with interference by a narrowband FMCW signal with interference by a narrowband FMCW signal causing a uniform causing uniform increase noiseoffloor and therefore increase of theanoise floor and thereforeofa the reduction the target a reduction of(SNR). the target signal-to-noise-ratio (SNR). signal-to-noise-ratio

KH

Fig. 9. Measured FMCW radar IF signal and range profile with pulsed signal interference.

d signals d diagram ht-hand e of the sed signal

m

al

Fig. 11. Reception bandwidth and range profile of the pulsed radar without interference.

Fig. 11. Reception bandwidth and range profile of the Fig.pulsed 11. Reception range profile of the pulsed radar radarbandwidth withoutand interference. without interference.

5 Fig. 13. Range profile of a pulsed radar with interference

Figure 11 shows the receiver bandwidth and the range profile of the pulsed radar without interference. The amplitudes in the lower diagram are the absolute values of the complex I-Q output of the mixer. The effect of an interference from a narrowband FMCW radar on this type of radar is a uniform increase in the noise floor (lower diagram of Fig. 12). The target signal-to-noise ratio decreases accordingly.

Fig. 13. Range profile of a pulsed radar with interference from an from an identical pulsed radar with the same PRF identical pulsed radar with the same PRF generating a stationary generating a stationary ghost target. ghost target.

situation for a slightly lower PRF resulting in a ghost target with increasing distance. This diagram, however, does not yet take into account the effect of integration in the correlation receiver. Depending on the actual difference in PRF, less interference pulses may be integrated, leading to a possibly 2.4 Pulsed radar with pulsed interference improved situation. Fig. 12. Reception bandwidth and range profile of a pulsed Interference from identical pulsed radar withFMCW the same radar withaninterference by a narrowband signal 2.5 Consideration of the interference power level pulse repetition generates a ghost at causingfrequency a uniform(PRF) increase of the noise floortarget and therefore reduction of the 13). target signal-to-noise-ratio (SNR). a constanta“distance” (Fig. So far, the power levels of the target echo and the interferA slightly different PRF results in a moving ghost target ing signal have not been regarded in detail. To start with an ange profile analysis levels, the simple interference scenario in since the time difference between the transmitted pulse and Fig. of 14.power Range profile of a pulsed radar with interference the interfering pulse increases or decreases from pulse repetiFig.from 1 hasan been considered. Simplified radiation diagrams identical pulsed radar with slightly lower PRF withgenerating constant antenna gain in a given sector were used as tion period to pulse repetition period. Figure 14 displays this a ghost target with increasing distance.

Fig. 12. Reception bandwidth and range profile of a pulsed radar with interference by a narrowband FMCW signal www.adv-radio-sci.net/8/55/2010/ causing a uniform increase of the noise floor and therefore a5reduction of the target signal-to-noise-ratio (SNR).

Adv. Radio Sci., 8, 55–60, 2010

Fig. 13. Range profile of a pulsed radar with interference Fig. 15. Simple automotive interference scenario with from an identical pulsed radar with the same PRF simplified radiation diagrams. The interferer radiates directly the radar sensor’s receiver. 60 generating a stationary ghost target. M. Goppelt et al.: Automotive radarinto – investigation of mutual interference mechanisms Fig. 13. Range profile of a pulsed radar with interference from an identical pulsed radar with the same PRF generating a stationary ghost target.

Fig. 16. Target and interference power level at the radar Fig. 16. Target and(RX) interference levelthe at scenario the radar in sensor’s sensor’s receive antennapower port for receive fig. 15.(RX) antenna port for the scenario in Fig. 15.

Fig. 14. Range profile of a pulsed radar with interference Fig. 14. profile a pulsed radar withwith interference from an Fig. 14. Range Range profileofof a pulsed radar interference from an identical pulsed radar with slightly lower PRF identical pulsed radar with slightly lower PRF generating a ghost are used increasingly for safety functions. On the other hand, fromgenerating an identicalapulsed radar with slightly lower PRF ghost target with increasing distance. target with increasing distance. traffic density and percentage of vehicles equipped with radar generating a ghost target with increasing distance. sensors will increase considerably. Interference mechanisms have to be precisely understood in order to design and to verify the effectiveness of countermeasures to minimize mutual interference. This paper has described some fundamental interference effects appearing in a scenario with FMCW and pulsed radar sensors, partly by a simple analysis using generic radar sensors, partly by first experiments using radar frontends built Fig. 15. Simple automotive interference scenario with Fig. 15. Simple automotive interference scenario with by the University of Ulm. Fig. 15. Simple automotive interference scenarioradiates with simplified simplified radiation diagrams. The interferer The two major effects detected so far are the appearance of radiation diagrams. The interferer radiates directly into the radar simplified radiation diagrams. The interferer radiates directly into the radar sensor’s receiver. ghost targets and an increase of the noise or interference level sensor’s receiver. directly into the radar sensor’s receiver. in the radar receiver. Concerning the interference levels, very high signal strengths can occur if two sensors directly face depicted in Fig. 15. The positions of the radar sensors on each other. the vehicles as well as widths and orientations of the antenna Acknowledgements. The authors wish to thank the German Federal beams are just one example. Power levels then are calculated Ministry of Education and Research (BMBF) for funding this work using Friis’ formula. within the RoCC project. Figure 16 shows the simulated power levels at the antenna port of the receiving radar. The interference lasts as long as the radar sensors “see” each other. The target level is conReferences stant16. since it is and assumed that thepower targetlevel vehicle is at a conFig. Target interference at the radar stant distance. the interfering radar radiates directly sensor’s receiveAs (RX) antenna port for the scenario in into Brooker, G.: Mutual Interference of Millimeter-Wave Radar Systhe receiver, the interference level can be much higher than tems, IEEE Transactions on Electromagnetic Compatibility, 49, fig. 15. and 60 interference level This at the radar 170–181, 2007. the Fig. target16. levelTarget (more than dB above thepower target level). Oprisan, D. and Rohling, H.: Analysis of Mutual Interference behighsensor’s interference level may have differentport effects. one receive (RX) antenna for On thethe scenario in tween Automotive Radar Systems, International Radar Sympohand, it may fig. 15. drive receiver components into saturation or into sium, Berlin, Germany, Session: Automotive Radar I, 2005. their nonlinear regime, resulting in unwanted harmonics. On Tullsson, B.: Topics in FMCW Radar Disturbance Suppression, the other hand, the interference signals will be further proRadar 97, Edinburgh, United Kingdom, 1–5, 1997. cessed and can appear as ghost targets or as an increased Gresham, I., Jenkins, A., Egri, R., et al.: Ultra-Wideband Radar noise level, reducing the desired SNR of the receiving radar. Sensors for Short-Range Vehicular Applications, IEEE TransSome further investigations are necessary, however, includactions on Microwave Theory and Techniques, 52, 2105–2122, ing the respective analogue and digital signal processing; this 2004. may reduce the influence of the interference to some extent.

3 Conclusion Mutual interference between automotive radar sensors is expected to become an increasingly important topic. This is due the fact that, on the one hand, automotive radar sensors Adv. Radio Sci., 8, 55–60, 2010

6

6

www.adv-radio-sci.net/8/55/2010/

6