automotive mimo radar - alp sayin · [12] j. h. holland, adaptation in natural and artificial...
TRANSCRIPT
DEPARTMENT OF ELECTRONIC, ELECTRICAL
AND SYSTEMS ENGINEERING
For more enquiries, please contact:
Head
of
MIS
L:
Pro
fessor
Mik
e C
hern
iakov,
+4
4(0
)12
1 4
14 4
28
6
m.c
hern
iakov@
bha
m.a
c.u
k U
niv
ers
ity o
f B
irm
ingh
am
, D
epa
rtm
ent
of E
ES
E, E
dg
ba
sto
n,
Birm
ingh
am
B15 2
TT.
www.birmingham.ac.uk/misl
MIMO Sensor Array for Short-Range
High-Resolution Automotive Sensing
Mr A Sayin, Dr. E Hoare, Dr. M Antoniou, Prof M Cherniakov
Name, [email protected], +44(0) 121 xxx xxxx
5.Experimental Results
2.Phased vs. MIMO ArrayA phased-array-antenna radar can be used with beam-forming techniques to direct a beam to a specific
location [3]. And then the radar can scan the desired area electronically. This can be achieved since,
the correlated signals from transmit antennas would add up constructively or destructively in different
directions in space [2]. On the other hand, a MIMO radar uses the advantage of orthogonality for
various advantages [4]. These advantages can be; reduced amount of necessary array elements,
increased angular resolution, faster scan time, simultaneous search & track etc. With a MIMO array;
beamforming process does not have to happen in the space but in the digital space, so we can
“illuminate” the whole space of interest with single transmission [1].
1.IntroductionThe aim of the project is to investigate a novel Multiple-Input-Multiple-Output (MIMO) sensor system
for automotive applications. Compared to traditional phased arrays, a MIMO array can achieve the
same fine angular resolution, but with a drastically reduced amount of sensor elements. A MIMO
array of 8 elements can deliver the same resolution as a phased array of 16 elements [1]. The other
highlight of this technology is that it can operate at short ranges, which is physically impossible with a
phased array as the beam requires significant distance from the antenna aperture to form [2].
Therefore a MIMO system can potentially provide very high angular resolution at short ranges. These
properties make such a system attractive for a number of automotive applications, including parking
aids, short-range cruise control, speed-over-ground estimation, pedestrian and object detection
and collision detection.
4. Nonlinear MIMO
3. Nearfield MIMO
To further optimize the MIMO arrays, we looked into thinned arrays. Since MIMO array patterns are
obtained from their equivalent virtual arrays, optimizing MIMO array patterns is not a straightforward
job. To achieve this task, we utilized various optimization algorithms to find solutions; namely
random descent[7][8], simulated annealing[9][10] and genetic algorithm[11][12]. Our specialized
implementations of these algorithms have successfully yielded us practical solutions. Below is a
comparison of a conventional MIMO array and one of our Nonlinear MIMO arrays.
(Scanning 0
degrees)
# of
Elements
Beamwidth Sidelobe
Level
Equivalent
PA Elements
Conventional
MIMO (sim)
8 (4+4) 6.30 -13.1 dB 16
Conventional
MIMO (exp)
8 (4+4) 6.48 -12.9 dB 16
Nonlinear
MIMO (sim)
8 (4+4) 4.20 (~65%) -10.5 dB 24
Nonlinear
MIMO (exp)
8 (4+4) 4.25 (~65%) -7.3 dB 24
14.9 19.8 21
4.1 4.543.3
mm
[1] D. Bliss, K. Forsythe, and G. Fawcett, ‘MIMO Radar:
Resolution, Performance, and Waveforms’, in Proc. 14th
Annual Adaptive Sensor Array Processing Workshop, MIT,
2006, pp. 6–7.
[2] C. A. Balanis, Antenna theory: analysis and design, 3rd
ed. Hoboken, NJ: John Wiley, 2005, ISBN: 047166782X.
[3] P. Z. Peebles, Radar principles. New York: Wiley, 1998,
ISBN: 0471252050.
[4] M. Lesturgie, ‘Tutorial on MIMO Radar’, Glasgow, UK, 21-
September-2012.
[5] R. A. Kennedy, T. Abhayapala, D. B. Ward, and R. C.
Williamson, ‘Nearfield broadband frequency invariant
beamforming’, in Acoustics, Speech, and Signal Processing,
1996. ICASSP-96.
Conference Proceedings., 1996 IEEE International
Conference on, 1996, vol. 2, pp. 905–908,
[6] R. A. Kennedy, D. B. Ward, and P. T. D. Abhayapala,
‘Nearfield beamforming using
nearfield/farfield reciprocity’, in Acoustics, Speech, and
Signal Processing, 1997. ICASSP-97.,
1997 IEEE International Conference on, 1997, vol. 5, pp.
3741–3744, Available at
http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=604683.
[7] N. Mladenović and P. Hansen, “Variable neighborhood
search,” Comput. Oper. Res., vol. 24, no. 11, pp. 1097 –
1100, 1997 [Online]. Available:
http://www.sciencedirect.com/science/article/pii/S030505489
7000312
[8] P. Hansen, N. Mladenović, and J. A. Moreno Pérez,
“Variable neighbourhood search: methods and applications,”
Ann. Oper. Res., vol. 175, no. 1, pp. 367–407, Mar. 2010
[Online]. Available: http://link.springer.com/10.1007/s10479-
009-0657-6. [Accessed: 16-Apr-2015]
[9] S. Kirkpatrick, M. P. Vecchi, and others, “Optimization by
simmulated annealing,” science, vol. 220, no. 4598, pp. 671–
680, 1983 [Online]. Available:
http://leonidzhukov.net/hse/2013/stochmod/papers/Kirkpatric
kGelattVecchi83.pdf. [Accessed: 20-Apr-2015]
[10] V. Černỳ, “Thermodynamical approach to the traveling
salesman problem: An efficient simulation algorithm,” J.
Optim. Theory Appl., vol. 45, no. 1, pp. 41–51, 1985 [Online].
Available:
http://link.springer.com/article/10.1007/BF00940812.
[Accessed: 20-Apr-2015]
[11] A. M. Turing, “A quarterly review of psychology and
philosophy,” Pattern Recognit. Introd. -Dowden Hutchinson
Ross Inc, 1973.
[12] J. H. Holland, Adaptation in natural and artificial
systems: an introductory analysis with applications to biology,
control, and artificial intelligence, 1st MIT Press ed.
Cambridge, Mass: MIT Press, 1992.
[13]J. A. Högbom, “Aperture Synthesis with a Non-Regular
Distribution of Interferometer Baselines,” \aaps, vol. 15, p.
417, Jun. 1974.
8.References
Less reflective targets can be disguised as range/angular sidelobes. Or the reverse can happen;
angular/range sidelobes can constructively add up to create fake targets. To overcome these problems
CLEAN algorithm was modified for our purposes, which is mainly used in radio astronomy[13].
• MIMO concept applied to automotive context
• Obtained further performance improvements
• Produced acoustic technology demonstrator, tested and verified
• Technology is not limited to sonar, can be moved into RF domain (e.g. for higher maximum range)
6. Conclusions
• Practical MIMO Applications
• Taking an image of a practical object with conventional MIMO
• Taking an image of a practical object with Non-Linear MIMO (e.g. bicycle, baby cart)
• Replacing receivers with miniature SMD components
• To have a fully functional “plug and play” prototype
7. Future Work
Tx/Rx
Virtual Array
Tx
Rx
Tx
Rx
MIMO
• Matched filtered signals can be interpreted as the signals out of the virtual array elements.
• Virtual Array Concept: M+N MIMO elements give the same angular resolution of MxN phased array
• Drastic reduction in antenna elements
• Example: 4+4 MIMO = 4x4 Phased Array
• Cheaper and more diverse system
Matched Filter #M
Matched Filter #mRx #1
Rx #2
Rx #n
Rx #N
…
Matched Filter #2
Matched Filter #1
Matched Filter #M
Matched Filter #m
Matched Filter #2
Matched Filter #1
Matched Filter #M
Matched Filter #m
Matched Filter #2
Matched Filter #1
Matched Filter #M
Matched Filter #m
Matched Filter #2
Matched Filter #1
Virtual Array Element #1
…
Virtual Array Element #2
Virtual Array Element #3
Virtual Array Element #4
Virtual Array Element #mxn
…
Virtual Array Element #MxN
…
Nearfield MIMO techniques can be used for scanning areas without losing focus in close ranges with a
tradeoff with more computation. The solution uses an actual geometric model rather than a far-field
approximated model to derive the array factor[5][6]. Below are our experimental results from our
nearfield algorithms at close ranges with an array with the size of 40cm looking at about 1.1 metres.
-0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8-0.5
0
0.5
1
1.5
Scenario Sketch [3 x 5]
x (meters)
Slant Range: 1.14m Az-Angle: 0.51D
RCS: 0.07
y (
mete
rs)
Transmit Array
Receive Array
Targets
Single Target at 0 degrees
-0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1-0.5
0
0.5
1
1.5
Slant Range: 1.30m Az-Angle: 31.23D
RCS: 0.07
Scenario Sketch [3 x 5]
x (meters)
y (
mete
rs)
Transmit Array
Receive Array
Targets
Single Target at 30 degrees
-0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Slant Range: 1.08m Az-Angle: 24.29D
RCS: 0.07
Scenario Sketch [3 x 5]
x (meters)
y (
mete
rs)
Transmit Array
Receive Array
Targets
Single Target at 25 degrees
-100 -50 0 50 1000.5
1
1.5
2
background2
Azimuth (Degrees)
Ra
ng
e (
mete
rs)
-100 -50 0 50 1000.5
1
1.5
2
ball0far
Azimuth (Degrees)
Ra
ng
e (
me
ters
)
-100 -50 0 50 1000.5
1
1.5
2
abs(ball0far)-abs(background2)
Azimuth (Degrees)
Ra
ng
e (
me
ters
)
-80 -60 -40 -20 0 20 40 60 80-30
-20
-10
0
Azimuth (Degrees)
Re
turn
(d
B)
Azimuth Cut at 1.15 meters [ball0far-background2]
Beamwidth: 6.8312
Sidelobe Level: -11.6715 dB
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2
-40
-30
-20
-10
0
Range (meters)
Re
turn
(d
B)
Range Cut at 0.50 degrees [ball0far-background2]
-80 -60 -40 -20 0 20 40 60 80-30
-20
-10
0
Azimuth (Degrees)
Re
turn
(d
B)
Azimuth Cut at 1.15 meters [ball0far]
Beamwidth: 6.7705
Sidelobe Level: -9.5346 dB
-20
-15
-10
-5
0
-20
-15
-10
-5
0
-20
-15
-10
-5
0
-100 -50 0 50 1000.5
1
1.5
2
background2
Azimuth (Degrees)
Ra
ng
e (
me
ters
)
-100 -50 0 50 1000.5
1
1.5
2
ball25far
Azimuth (Degrees)
Ra
ng
e (
me
ters
)
-100 -50 0 50 1000.5
1
1.5
2
abs(ball25far)-abs(background2)
Azimuth (Degrees)
Ra
ng
e (
me
ters
)
-80 -60 -40 -20 0 20 40 60 80-30
-20
-10
0
Azimuth (Degrees)
Re
turn
(d
B)
Azimuth Cut at 1.15 meters [ball25far-background2]
Beamwidth: 7.9496
Sidelobe Level: -11.5429 dB
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2
-40
-30
-20
-10
0
Range (meters)
Re
turn
(d
B)
Range Cut at 23.50 degrees [ball25far-background2]
Range Resolution: 0.19548
-80 -60 -40 -20 0 20 40 60 80-30
-20
-10
0
Azimuth (Degrees)
Re
turn
(d
B)
Azimuth Cut at 1.15 meters [ball25far]
Beamwidth: 8.8729
Sidelobe Level: -9.9224 dB
-20
-15
-10
-5
0
-20
-15
-10
-5
0
-20
-15
-10
-5
0
-100 -50 0 50 1000.5
1
1.5
2
background2
Azimuth (Degrees)
Ra
ng
e (
me
ters
)
-100 -50 0 50 1000.5
1
1.5
2
ball30far
Azimuth (Degrees)
Ra
ng
e (
me
ters
)
-100 -50 0 50 1000.5
1
1.5
2
abs(ball30far)-abs(background2)
Azimuth (Degrees)
Ra
ng
e (
me
ters
)
-80 -60 -40 -20 0 20 40 60 80-30
-20
-10
0
Azimuth (Degrees)
Re
turn
(d
B)
Azimuth Cut at 1.34 meters [ball30far-background2]
Beamwidth: 7.7805
Sidelobe Level: -7.0442 dB
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2
-40
-30
-20
-10
0
Range (meters)
Re
turn
(d
B)
Range Cut at 31.00 degrees [ball30far-background2]
Range Resolution: 0.25331
-80 -60 -40 -20 0 20 40 60 80-30
-20
-10
0
Azimuth (Degrees)
Re
turn
(d
B)
Azimuth Cut at 1.34 meters [ball30far]
Beamwidth: 9.034
Sidelobe Level: -8.7073 dB
-20
-15
-10
-5
0
-20
-15
-10
-5
0
-20
-15
-10
-5
0
-100 -50 0 50 1000.5
1
1.5
2
background2
Azimuth (Degrees)
Ran
ge (
mete
rs)
-100 -50 0 50 1000.5
1
1.5
2
ball0far
Azimuth (Degrees)
Ran
ge (
mete
rs)
-100 -50 0 50 1000.5
1
1.5
2
abs(ball0far)-abs(background2)
Azimuth (Degrees)
Ran
ge (
mete
rs)
-80 -60 -40 -20 0 20 40 60 80-30
-20
-10
0
Azimuth (Degrees)
Retu
rn (
dB
)
Azimuth Cut at 1.15 meters [ball0far-background2]
Beamwidth: 6.8312
Sidelobe Level: -11.6715 dB
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2
-40
-30
-20
-10
0
Range (meters)
Retu
rn (
dB
)
Range Cut at 0.50 degrees [ball0far-background2]
-80 -60 -40 -20 0 20 40 60 80-30
-20
-10
0
Azimuth (Degrees)
Retu
rn (
dB
)
Azimuth Cut at 1.15 meters [ball0far]
Beamwidth: 6.7705
Sidelobe Level: -9.5346 dB
-20
-15
-10
-5
0
-20
-15
-10
-5
0
-20
-15
-10
-5
0
Range-Azimuth Map
-100 -50 0 50 1000.5
1
1.5
2
background2
Azimuth (Degrees)
Ran
ge (
mete
rs)
-100 -50 0 50 1000.5
1
1.5
2
ball25far
Azimuth (Degrees)
Ran
ge (
mete
rs)
-100 -50 0 50 1000.5
1
1.5
2
abs(ball25far)-abs(background2)
Azimuth (Degrees)
Ran
ge (
mete
rs)
-80 -60 -40 -20 0 20 40 60 80-30
-20
-10
0
Azimuth (Degrees)
Retu
rn (
dB
)
Azimuth Cut at 1.15 meters [ball25far-background2]
Beamwidth: 7.9496
Sidelobe Level: -11.5429 dB
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2
-40
-30
-20
-10
0
Range (meters)
Retu
rn (
dB
)
Range Cut at 23.50 degrees [ball25far-background2]
Range Resolution: 0.19548
-80 -60 -40 -20 0 20 40 60 80-30
-20
-10
0
Azimuth (Degrees)
Retu
rn (
dB
)
Azimuth Cut at 1.15 meters [ball25far]
Beamwidth: 8.8729
Sidelobe Level: -9.9224 dB
-20
-15
-10
-5
0
-20
-15
-10
-5
0
-20
-15
-10
-5
0
Range-Azimuth Map
-100 -50 0 50 1000.5
1
1.5
2
background2
Azimuth (Degrees)
Ran
ge (
mete
rs)
-100 -50 0 50 1000.5
1
1.5
2
ball30far
Azimuth (Degrees)
Ran
ge (
mete
rs)
-100 -50 0 50 1000.5
1
1.5
2
abs(ball30far)-abs(background2)
Azimuth (Degrees)
Ran
ge (
mete
rs)
-80 -60 -40 -20 0 20 40 60 80-30
-20
-10
0
Azimuth (Degrees)
Retu
rn
(d
B)
Azimuth Cut at 1.34 meters [ball30far-background2]
Beamwidth: 7.7805
Sidelobe Level: -7.0442 dB
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2
-40
-30
-20
-10
0
Range (meters)
Retu
rn
(d
B)
Range Cut at 31.00 degrees [ball30far-background2]
Range Resolution: 0.25331
-80 -60 -40 -20 0 20 40 60 80-30
-20
-10
0
Azimuth (Degrees)
Retu
rn
(d
B)
Azimuth Cut at 1.34 meters [ball30far]
Beamwidth: 9.034
Sidelobe Level: -8.7073 dB
-20
-15
-10
-5
0
-20
-15
-10
-5
0
-20
-15
-10
-5
0
Range-Azimuth Map