Array Design for Automotive Digital Beamforming Radar ... - IEEE Xplore

2 downloads 0 Views 888KB Size Report
In this paper a Digital Beamforming Radar System for automotive Short Range Radar system is presented. It takes advantage of multiple transmitters for ...
Array Design for Automotive Digital Beamforming Radar System Karin Schuler, Marwan Younis, Rainer Lenz, Werner Wiesbeck University Karlsruhe (TH), IHE, Germany Key Words: Digital Beamforming, Array Design, Ambiguity Suppression

1. Introduction

In contrast to current approaches, the here presented Digital Beamforming SRR obtains angular resolution out of the phase differences of reflected signals received at the same time by different antennas. The resolution is the better, the more combinations of received signals can be analyzed. Using switched transmitters, all illuminating sequentially the same area, the number of combination even increases and with that the angular resolution. Due to the advantage of independent

To cover the complete area around a vehicle with DBFSRR, the area is divided into several angular segments as shown in fig. 1. Each segment is illuminated by one DBFtransceiver module, which is described in detail below. Regarding the integration, only a certain number of modules are needed, e.g. two modules in each bumper and in each side. Since all of them are working independently, there is no RF-synchronization needed and a link to the common data bus is sufficient. Depending on the applications in the different angular segments, the desired performance can be adjusted in each DBF-transceiver module separately.

DBF-SRRModules Bli n

dS pot

nd Bli

Side Impact

ot Sp

Side Impact

Stop-and-Go Pre-Crash Warning e Lan ol tr Con

The importance of sensors in vehicles has grown enormously over the last years. After the rollout of safety relevant systems like Automatic Cruise Control (ACC), also driver-assisting systems like parking aid were brought onto the market. By now, a number of different sensor systems exist, mostly varying in techniques and applications. Despite the different applications, some of them can also be based on one common hardware platform. For this, we propose a Short Range Radar (SRR) system covering the complete surrounding of a vehicle and therewith delivering all the information needed for parking aid, blind spot surveillance, pre-cash warning, etc.. Regarding the current constraints due to frequency regulation and transmit power issues on one side and the required performance aspects on the other, the introduced SRR-System, based on Digital Beamforming (DBF), appears highly promising. Up to now the existing systems are mainly based on scanned or switched antenna beams. Beam scanning is achieved mechanically by rotating the antenna. To achieve good accuracy this is a complex procedure, especially in vehicles with all the specific problems as for example vibration. With the other current approach of switched antenna beams, data is collected from one covered area after the other. This is for example implemented in ACC sensors. To achieve good angular resolution, sharp radiation patterns are required. Covering the whole area around a vehicle with switched antenna beams, this would however lead to an enormous number of antennas.

2. Integration of DBF-SRR-System in Vehicles

Lan Con e trol

In this paper a Digital Beamforming Radar System for automotive Short Range Radar system is presented. It takes advantage of multiple transmitters for increasing the angular resolution. The main focus of this paper is on the optimization of the array configuration to obtain a system for integration into vehicles. A Digital Beamforming simulator shows the comparison of different configurations.

range and angular resolution, this approach claims only a bandwidth of about 100 MHz to 500 MHz. As will be shown, compared to any other concept based on switched or scanned antenna beams, the DBF-concept with multiple transmitters offers the best angular resolution.

Parking Aid

Summary

Covered Angular Segments

Fig. 1: Applications and integration example for a DBFSRR. 3. DBF-Transceiver Concept Each DBF-transceiver module consists of multiple transmitters and multiple receivers. The use of multiple switched transmitters offers the advantage of highly increased resolution with insignificantly bigger array extension [1]. A block diagram of one DBF-SRR-transceiver module is shown in fig. 2.

0-7803-8882-8/05/$20.00 (C) 2005 IEEE

Transmitter

24GHz

RFout,1

24GHz

PRF

Switch

Mod.

PROCESSOR

A/D - Conversion

IQ - Downconversion

IQ - Downconversion

RFout,NT

Receiver RFin,1

RFin,NR

Fig. 2: Block diagram of a DBF-SRR-transceiver module. All antenna beams overlap. A RF-chirp signal is generated and modulated with a pulse needed for range resolution. The switch is used to select sequentially one of the multiple transmit antennas to radiate into the covered angular segment. After reflections on obstacles, the radar signal is received simultaneously from all the receiving antennas. In each receiver chain the signal is coherently down converted and separated into inphase and quadrature components. For data processing, each channel output must be A/D-converted and stored separately. After one cycle of transmitters switches, the received signals are stored separately for all transmitter-receiver combinations. They are now available for Digital Beamforming processing. 4. Digital Beamforming Processing First a common range compression is performed. In principle, this is a correlation between the transmit and the receive signal over time. The correlation gives a maximum at the time corresponding to the propagation time to the reflecting object. In this case of multiple transmitters and receivers, the distances to objects are calculated for each receiver separately. The range compressed signal src is the correlation between transmit and receive signal. For a transmit signal of antenna xTm and a reflection at object xn with the related propagation time τn and the reception by antenna xRn this gives:

rate.

has to be considered whereas the amplitude term can be neglected. The kernel represents the theoretical phase of a signal reflected by a point scatterer in the focused direction. The azimuth compressed signal sac can be expressed as

where TP is the chirp duration and ke = B/(2TP)is the chirp

After range compression the azimuth compression can be performed with the Digital Beamforming. In principle, this is a multiplication of the range-compressed signal with a summation kernel. For this, in the range compressed signal src only the phase term directly depending on the angular position

where the distances between object and transmitter and receiver are expressed by RTm and RRn respectively. The direction (ψ0,θ0) is the focusing angle of the two arrays and Kac is a focusing term. The summation results in a high absolute value if the kernel and the range compressed signal match and in a lower absolute value the more the assumed and the real direction are different. Assuming an uniform array of NT receive antennas and a different uniform array of NR switched transmit antennas, this summation can be written in general as a multiplication of the two array factors [2]: sac (N T ,N R ,d T ,d R ,ψ 0 ,θ 0 ) = sin

(

πNT λ

sin sin

(

π λ

(

sin

(

π λ

)

d T (sin(ψ )sin(θ )− sin (ψ 0 )sin(θ 0 ))

πNR λ

)⋅

d T (sin(ψ )sin (θ )− sin(ψ 0 )sin(θ 0 ))

(3)

)

d R (sin (ψ )sin (θ )− sin(ψ 0 )sin(θ 0 ))

)

d R (sin(ψ )sin(θ )− sin (ψ 0 )sin(θ 0 ))

where dR and dT are the distances between the receivers and the transmitters respectively. The resolution is therewith the resolution of the total array factor. 5. Array Design Considerations Regarding the implementation in a vehicle, not only the resolution, but also hardware costs, integration possibilities, and sensibility to environmental stress have to be considered. As can be seen in (3), the resolution of the total array configuration is dependant on the number of used antennas. The number of the antennas together with the distance in between them leads also to the dimension of the array and also to the position of grating lobes. As the existence of grating lobes cannot be tolerated, the covered angular segment and the shape of the element factor is therewith defined. In the following the relation between the array parameters is presented leading to the optimal configuration. Starting from the theory above, a configuration has to be found to realize the optimum DBF SRR system. For simplification this is reduced to azimuth, meaning θ = 90°. As already mentioned, the whole area around the car is divided into several angular segments. Each of them is covered by one DBF-transceiver including a transmit and as well as a receive array. To maximize the angular resolution, the question for the number of segments Nseg, the number of transmit antennas NT and receive antennas NR and the antenna distances dT and dR respectively arises with the constraints like hardware costs, integration possibilities, and sensitivity to environmental

0-7803-8882-8/05/$20.00 (C) 2005 IEEE

stress. In the following, the dependencies during array design are briefly described. They are also visualized in fig. 3. DBF - SRR Configuration

N tot

Nseg DBF-Transceiver Receive NR Array ψseg

dR

Transmit Array

NT

The security factor γ describes the factor of the angular segment between the two grating lobes which is actually used for observation. The number of segments Nseg around the vehicle is then:

dT ψ 3dB,R

N seg =

ψ3dB,T ψ 3dB,tot

Fig. 3: Parameter relation in DBF - SRR. The number of segments around the vehicle Nseg, their size ψseg and therewith also the number of DBF-transceivers is defined by the receive antenna distance dR: The usable angular segments ψseg must not have any grating lobes. In principle, the segment without grating lobes is larger, the smaller the distance between the receive antennas dR becomes. However, the smaller is dR, the smaller is the receive array extension dR(NR-1) resulting in a worse resolution. The position of the grating lobes is a function of the scan angle and the distance between the antennas d/λ. For defining the maximum covered angular segment, it is not sufficient to consider the position of the grating lobes for an unscanned receive array, but for the maximum and minimum scan angle to assure that for any scan angle no grating lobe is shifted into the covered angular segment. Fig. 4 shows the position of the grating lobes for different distances between the antennas. When scanning the array or digitally focusing - as done in Digital Beamforming processing – not only the main beam but also the grating lobe moves.

360 o

ψ seg

(5)

Now the transmit array has to be considered: Grating lobes of the transmit array may occur within the covered angular segment ψseg. As long as they are outside the HPBW of the receive array factor ψ3dB,R, they do not lead to ambiguities because they are multiplied with the receive array factor and therewith attenuated. This is why the transmit antenna distance dT must be chosen smaller, the larger the HPBW of the receive array factor ψ3dB,R. Therefore, the distance between the transmit antennas dR can be chosen to:

For dT >> λ the HPBW of the transmit array can be approximated to

The resolution of the DBF array can be approximated with the resolution of the transmit array:

An example for array factors is shown in fig. 5. The grating lobes of the transmit array are outside the HPBW of receive array. For simplicity an amplitude distribution on the antennas is not considered but naturally would decrease the side lobes.

Fig. 4: Grating lobe position over scan angle. Grey square is maximum angular segment useable with d/λ = 0.7 and γ = 0.7. Together with a security factor γ the maximum covered segment with the size ψseg is expressed with:

0-7803-8882-8/05/$20.00 (C) 2005 IEEE

Covered angular segment

Number of Receivers

Number of Transmitters

Resolution / Degree

8 segments with 45°, 6 antennas each

4

2 (1)

9.8 (15)

3

3

8.4

6

2 (1)

8.4 (12)

5

3

6.4

4

4

5.8

6 segments with 60°, 8 antennas each

Tab. 1: Resolution for different array configuration with 48 antennas in total. Fig. 5: Array factor of transmit and receive arrays with constant amplitude distribution. NR = 4, dR = 0.7 λ, NT = 4, dT = 3.2 λ according to tab. 1 last row. In Digital Beamforming processing the two array factors are virtually multiplied according to eq. 3. The product of the two array factors in fig. 5 is shown in fig. 6. It does not have grating lobes within the covered segment from –30° to 30°.

A conventional transceiver system of 24 transceivers each covering its own angular segment, offers a resolution of 360°/24 = 15°. The DBF-system offers with the same hardware effort (24 transmitters + 24 receivers = 48 antennas) around three times better resolution as can be seen in tab. 1 in the last column. As already can be assumed by looking at the table, the Digital Beamforming with a fixed total number of antennas gives a better performance for large angular segments. This short comparison already shows the evident advantage of multitransmit DBF for vehicular applications: The less SRRmodules shall be used for integration reasons, the more multitransmit DBF-SRR stands out! 6. Simulation Example

Fig. 6: Mulitplication of transmit and receive array factor according to fig. 5. The resolution ψ3dB,R is an implicit function of the total number of antennas Ntot and the number of segments Nseg. To get an idea of the performance of DBF-SRR, some example values are given in tab. 1. The configuration of the first and the last row are considered in paragraph 6.

For evaluating the performance of different array configurations, a simulation program has been established. Starting from an arbitrary scenario, the signals for all combinations of transmit and receive antennas are simulated. All antennas are assumed to be isotropic radiators. The used signal is a chirp-modulated signal with variable bandwidth. These simulated signals are then processed as described in paragraph 4. Up to now, the simulation is limited to point scatterers. Amplitude degradation is also neglected as only the phase carries the information for Digital Beamforming. For illustrating the effects of different array configurations, the simulation results for the configurations of tab. 1 in the first and last row are shown in the following. Fig. 7 shows the distribution of point scatterers along range and cross-range used for the following simulations.

0-7803-8882-8/05/$20.00 (C) 2005 IEEE

Fig. 7: Position of ideal point scatterers for simulation. In both cases the bandwidth is fixed to 100 MHz resulting in a range resolution of 1.5 m. The first simulation is performed with NR = 4, dR = 0.9 λ, NT = 2, dT = 4.1 λ according to the first row in tab. 1 and a half-power-beamwidth of around 10°. The result is shown in fig. 8. The 5 point scatterers in the right part of the scanario can be seen, however only with a poor resolutuion. The 4 point scatteres at range position 24 m are blured. Their number and position cannot be determined.

Fig. 9: Processed image for scenario shown in fig. 7. B = 100 MHz, NR = 4, dR = 0.7 λ, NT = 4, dT = 3.2 λ, according to tab. 1 last row. Comparing the two simulations it must be noted, that for the first configuration (fig. 8) only 6 antennas are used to cover 45° degree whereas the second (fig. 9) uses 8 antennas to covers 60°. If one wants to cover 360°, which is necessary for Short Range Radar Systems, the total number of antennas and the hardware effort has to be considered. For the first configuration 6 antennas in 8 segments each with a size of 45° are needed leading to a total number of 48 antennas. In the second configuration 8 antennas cover a 60°-segment of which 6 are needed so that the total number of antennas is also 48. Looking at the two simulation results (fig. 8 and fig. 9) the benefit of the second configuration in fig. 9 is evident. Conclusion In this paper an automotive Short Range Radar, based on Digital Beamforming, is presented and the advantages compared to other Radar approaches are discussed. Array considerations and comparisons show the evident advantage of Digital Beamforming for large angular segments. In contrast to conventional Digital Beamforming Systems, the considerable performance improvement by the use of multiple transmitters is proved. References

Fig. 8: Processed image for scenario shown in fig. 7. B = 100 MHz, NR = 4, dR = 0.9 λ, NT = 2, dT = 4.1 λ, according to tab. 1 first row.

[1]

The second simulation uses the distribution of the last row of tab. 1. with NR = 4, dR = 0.7 λ, NT = 4, dT = 3.2 λ giving a half-power-beam-width of around 6°. Its result is shown in fig. 9. Unlike in the previous simulation, all point scatterers can be identified.

[2]

W. Wiesbeck and M. Younis, “A Novel Radar Sensor Concept for Automotive Safety Systems,” Workshop on Environmental Sensor Systems for Automotive Applications, 33rd European Microwave Conference EuMC03, München, Germany, October 2003. M. Younis, “Digital Beam-Forming for High Resolution Wide Swath Real and Synthetic Aperture Radar,” Dissertation, Forschungsberichte aus dem Institut für Höchstfrequenztechnik und Elektronik der Universität Karlsruhe (TH), 2004.

0-7803-8882-8/05/$20.00 (C) 2005 IEEE

Biography Karin Schuler was born in St. Georgen, Germany in 1976. In 2000 she spent six months as a visiting scientist at National Oceanic and Atmospheric Administration (NOAA), Boulder, USA, where she worked on passive remote sensing. She received the DEA (M.S.E.E) in 2002 from Ecole Nationale Supérieure d’Electronique et de Radioélectricité (ENSERG), Grenoble, France, and the Dipl.-Ing. degree in 2003 from the Universität Karlsruhe (TH), Germany. Since then she is with the Institut für Höchstfrequenztechnik und Elektronik at the Universität Karlsruhe (TH) as a Reseach Scientist working towards her Dr.-Ing. (Ph.D.). Her research areas have been focused on millimeter wave antennas, Digital Beamforming and automotive Radar. Karin Schuler is an IEEE student member. She won the EADS student award 2003 for her work on millimeter wave antennas and is co-author of the paper wining the EEEfCOM Innovationspreis 2003 issued by Rohde&Schwarz, together with Gerotron GmbH. Adress: Universität Karlsruhe (TH), Institut für Höchstfrequenztechnik und Elektronik (IHE), Kaiserstrasse 12, D76128 Karlsruhe, Germany, Email: [email protected]

0-7803-8882-8/05/$20.00 (C) 2005 IEEE