1
00:00:00,079 --> 00:00:03,210
hi my name is Rick Gentile I'm the

2
00:00:03,210 --> 00:00:04,770
phased array system toolbox product

3
00:00:04,770 --> 00:00:06,839
manager I'd like to show you how easy it

4
00:00:06,839 --> 00:00:08,460
is to process a radar data cube with

5
00:00:08,460 --> 00:00:09,719
MATLAB in the phased array system

6
00:00:09,719 --> 00:00:12,480
toolbox as we saw in part 1 of the

7
00:00:12,480 --> 00:00:14,099
recording the radar data cube is a

8
00:00:14,099 --> 00:00:16,230
collection of radar returns received by

9
00:00:16,230 --> 00:00:18,539
a phased array radar it's organized into

10
00:00:18,539 --> 00:00:20,820
a three-dimensional matrix the vertical

11
00:00:20,820 --> 00:00:22,410
axis of the queue is composed of the

12
00:00:22,410 --> 00:00:24,449
received time samples representing each

13
00:00:24,449 --> 00:00:26,970
individual radar pulse the horizontal

14
00:00:26,970 --> 00:00:28,740
axis contains the time samples collected

15
00:00:28,740 --> 00:00:31,590
in each the array elements for the depth

16
00:00:31,590 --> 00:00:33,329
axis there's a collection of consecutive

17
00:00:33,329 --> 00:00:35,280
pulses focus on a particular direction

18
00:00:35,280 --> 00:00:37,500
or object this is sometimes called the

19
00:00:37,500 --> 00:00:39,840
dwell time in this presentation will

20
00:00:39,840 --> 00:00:41,940
talk about the three array processing

21
00:00:41,940 --> 00:00:43,710
algorithm starting with beam forming a

22
00:00:43,710 --> 00:00:46,559
beamforming is a process of combining

23
00:00:46,559 --> 00:00:48,420
the individual receive channels into a

24
00:00:48,420 --> 00:00:51,000
single receive signal you can think of

25
00:00:51,000 --> 00:00:52,590
beam forming as a spatial filter that

26
00:00:52,590 --> 00:00:54,390
applies different complex weights to the

27
00:00:54,390 --> 00:00:56,760
channels and that sums the results to

28
00:00:56,760 --> 00:00:58,350
effectively steer a beam in a specific

29
00:00:58,350 --> 00:01:00,840
direction a beam forming works across

30
00:01:00,840 --> 00:01:02,699
the horizontal axis of the radar data

31
00:01:02,699 --> 00:01:04,350
cube and it gives you information about

32
00:01:04,350 --> 00:01:06,090
the bearing elevation of the object

33
00:01:06,090 --> 00:01:08,400
you're trying to track the second

34
00:01:08,400 --> 00:01:10,680
algorithm is matched filtering or pulse

35
00:01:10,680 --> 00:01:12,750
compression which enhances the radar

36
00:01:12,750 --> 00:01:14,610
return by correlating the receive time

37
00:01:14,610 --> 00:01:16,320
samples with a sampled version of the

38
00:01:16,320 --> 00:01:18,780
source waveform match filtering is

39
00:01:18,780 --> 00:01:20,340
performed on the data that makes up the

40
00:01:20,340 --> 00:01:23,189
vertical axis of the radar data cube the

41
00:01:23,189 --> 00:01:24,210
range bin where the peak amplitude

42
00:01:24,210 --> 00:01:26,520
occurs can be used to determine the

43
00:01:26,520 --> 00:01:27,960
distance of an object you are trying to

44
00:01:27,960 --> 00:01:30,990
track the radar returns from moving

45
00:01:30,990 --> 00:01:32,909
objects will have a Doppler shift that's

46
00:01:32,909 --> 00:01:34,680
proportional to the radial velocity of

47
00:01:34,680 --> 00:01:36,990
the object so if we take the FFT of the

48
00:01:36,990 --> 00:01:38,520
radar returns along the depth dimension

49
00:01:38,520 --> 00:01:40,829
of the radar data cube we can extract

50
00:01:40,829 --> 00:01:42,270
information about the speed of the

51
00:01:42,270 --> 00:01:45,509
moving object now let's take a look at

52
00:01:45,509 --> 00:01:47,340
some matlab code that can be used to

53
00:01:47,340 --> 00:01:51,960
process the radar data cube will use

54
00:01:51,960 --> 00:01:53,579
some code from a previous recording

55
00:01:53,579 --> 00:01:55,619
called building the radar data cube to

56
00:01:55,619 --> 00:01:57,659
quickly generate our radar data cube for

57
00:01:57,659 --> 00:01:59,670
a target located at 20 kilometers away

58
00:01:59,670 --> 00:02:02,250
from the radar with a radial velocity of

59
00:02:02,250 --> 00:02:05,189
150 meters per second so you can see the

60
00:02:05,189 --> 00:02:06,420
data cube variable in my matlab

61
00:02:06,420 --> 00:02:09,149
workspace it's built up as a thousand

62
00:02:09,149 --> 00:02:11,670
time sample by eight receive element x

63
00:02:11,670 --> 00:02:13,500
32 pulse variable

64
00:02:13,500 --> 00:02:15,450
it's complex data type because there are

65
00:02:15,450 --> 00:02:17,730
complex inq samples that have been

66
00:02:17,730 --> 00:02:20,970
received here after the array now we'll

67
00:02:20,970 --> 00:02:22,470
start with beamforming which can give us

68
00:02:22,470 --> 00:02:24,060
information about the bearing of the

69
00:02:24,060 --> 00:02:26,700
moving object we're tracking you can use

70
00:02:26,700 --> 00:02:28,380
the phased array system toolbox to

71
00:02:28,380 --> 00:02:30,630
create a phase shift beamformer and set

72
00:02:30,630 --> 00:02:33,500
the properties that we would like to use

73
00:02:33,500 --> 00:02:36,000
now once we've created this object we

74
00:02:36,000 --> 00:02:37,920
can use the step method as a way of

75
00:02:37,920 --> 00:02:40,980
executing the beamformer now for

76
00:02:40,980 --> 00:02:42,090
illustration purposes we're going to

77
00:02:42,090 --> 00:02:44,340
steer the beam both 20 degrees azmuth

78
00:02:44,340 --> 00:02:46,260
and to a 30-degree azmuth where the

79
00:02:46,260 --> 00:02:48,750
moving object is actually located so go

80
00:02:48,750 --> 00:02:53,459
ahead and execute this cell on the

81
00:02:53,459 --> 00:02:54,690
resulting plot we can see three

82
00:02:54,690 --> 00:02:56,880
different things the first is the plot

83
00:02:56,880 --> 00:02:59,670
response this is the response of the

84
00:02:59,670 --> 00:03:02,550
uniform linear array object we can see

85
00:03:02,550 --> 00:03:04,290
with no beam forming the peak of the

86
00:03:04,290 --> 00:03:06,020
response is at zero degrees azmuth

87
00:03:06,020 --> 00:03:08,489
there's a deep null at 30 degrees where

88
00:03:08,489 --> 00:03:11,520
our targets located if we just some the

89
00:03:11,520 --> 00:03:12,900
receive elements we can't see any

90
00:03:12,900 --> 00:03:15,300
indication that the targets present this

91
00:03:15,300 --> 00:03:17,850
is because we have the beam steered to

92
00:03:17,850 --> 00:03:20,489
the zero Degree azmuth location if we

93
00:03:20,489 --> 00:03:22,110
use some phased array weights in the

94
00:03:22,110 --> 00:03:24,120
beam former to steer the array to a 30

95
00:03:24,120 --> 00:03:26,670
degree azmuth where the target is we now

96
00:03:26,670 --> 00:03:28,560
see an enhanced target return or we can

97
00:03:28,560 --> 00:03:30,000
really see the target return for the

98
00:03:30,000 --> 00:03:33,360
noise now when you're processing the

99
00:03:33,360 --> 00:03:35,100
radar data cube you have to do

100
00:03:35,100 --> 00:03:36,570
beamforming for all the pulses along the

101
00:03:36,570 --> 00:03:38,970
depth dimension and that's the

102
00:03:38,970 --> 00:03:41,310
processing you see here the figure we

103
00:03:41,310 --> 00:03:43,500
see shows the improved target returns

104
00:03:43,500 --> 00:03:45,420
for the entire sequence of 3G pulses

105
00:03:45,420 --> 00:03:48,989
when we use beamforming we can now use

106
00:03:48,989 --> 00:03:51,150
pulse compression to determine the

107
00:03:51,150 --> 00:03:54,209
distance to the target and you'll see

108
00:03:54,209 --> 00:03:56,100
here we can use the get matched filter

109
00:03:56,100 --> 00:03:59,940
method to return the match filter for

110
00:03:59,940 --> 00:04:02,250
the transmitted wave form we use this

111
00:04:02,250 --> 00:04:04,320
filter object to process the time

112
00:04:04,320 --> 00:04:08,730
samples for the radar pulse return now

113
00:04:08,730 --> 00:04:10,350
you can see here the result is a sharp

114
00:04:10,350 --> 00:04:11,910
peak at the range bin of the moving

115
00:04:11,910 --> 00:04:14,160
object this is one of the advantages of

116
00:04:14,160 --> 00:04:15,959
using a modulated pulse like the linear

117
00:04:15,959 --> 00:04:17,989
FM waveform we see here on the Left

118
00:04:17,989 --> 00:04:20,100
there's a very strong correlation

119
00:04:20,100 --> 00:04:21,720
property and therefore produces a strong

120
00:04:21,720 --> 00:04:24,750
narrow peak at the target return once

121
00:04:24,750 --> 00:04:26,220
we've detected this peak we can

122
00:04:26,220 --> 00:04:27,389
calculate the distance the

123
00:04:27,389 --> 00:04:29,939
moving object the index of the time

124
00:04:29,939 --> 00:04:31,560
sample where the peak exists tells us

125
00:04:31,560 --> 00:04:34,499
the distance of the object in this case

126
00:04:34,499 --> 00:04:35,849
we have to use the time deranged

127
00:04:35,849 --> 00:04:37,349
function and subtract the length of

128
00:04:37,349 --> 00:04:39,029
match filter to convert the time to

129
00:04:39,029 --> 00:04:41,460
distance when we execute this it gives

130
00:04:41,460 --> 00:04:44,099
us a distance of 19.9 4 kilometers which

131
00:04:44,099 --> 00:04:45,810
is within the range resolution of the

132
00:04:45,810 --> 00:04:49,139
waveform that we selected and last of

133
00:04:49,139 --> 00:04:50,639
all we can determine the speed of the

134
00:04:50,639 --> 00:04:52,319
target using the Doppler shift observed

135
00:04:52,319 --> 00:04:55,740
in the target returns in this case the

136
00:04:55,740 --> 00:04:57,719
FFT across the range bin where the

137
00:04:57,719 --> 00:05:00,800
moving object has been detected yields a

138
00:05:00,800 --> 00:05:03,560
plot where we can see the peak here and

139
00:05:03,560 --> 00:05:06,240
it's at a frequency 150 Hertz which

140
00:05:06,240 --> 00:05:08,400
corresponds to a radial velocity 150

141
00:05:08,400 --> 00:05:11,550
meters per second now this Doppler

142
00:05:11,550 --> 00:05:13,199
resolution is a function of the number

143
00:05:13,199 --> 00:05:14,969
of pulses in the dwell time so if we

144
00:05:14,969 --> 00:05:16,590
wanted to improve the accuracy of the

145
00:05:16,590 --> 00:05:18,389
speed measurement we can send out more

146
00:05:18,389 --> 00:05:20,810
pulses at this particular object a

147
00:05:20,810 --> 00:05:23,099
phased array system toolbox also has a

148
00:05:23,099 --> 00:05:25,430
range Doppler response system object

149
00:05:25,430 --> 00:05:27,779
this object essentially computes the f

150
00:05:27,779 --> 00:05:29,580
of T across the range bins in the data

151
00:05:29,580 --> 00:05:31,740
cube so when you look at this plot you

152
00:05:31,740 --> 00:05:33,360
can see the strong return at a distance

153
00:05:33,360 --> 00:05:35,729
of twenty kilometers and a speed of 150

154
00:05:35,729 --> 00:05:38,639
meters per second now this example is

155
00:05:38,639 --> 00:05:40,259
meant to show how to process a radar

156
00:05:40,259 --> 00:05:42,689
data cube to determine the range speed

157
00:05:42,689 --> 00:05:45,569
and bearing of a moving object the

158
00:05:45,569 --> 00:05:47,819
phased array system toolbox has other

159
00:05:47,819 --> 00:05:50,219
very processing techniques like do a

160
00:05:50,219 --> 00:05:52,379
algorithms and stab space-time adaptive

161
00:05:52,379 --> 00:05:54,689
processing or you can extract even more

162
00:05:54,689 --> 00:05:56,189
information about the target returns

163
00:05:56,189 --> 00:05:58,979
contained in the rdc you can find more

164
00:05:58,979 --> 00:06:01,830
information at the product page and if

165
00:06:01,830 --> 00:06:04,949
you're interested in this code you can

166
00:06:04,949 --> 00:06:07,289
find it in matlab central with the title

167
00:06:07,289 --> 00:06:11,310
processing the rdc thank you again for

168
00:06:11,310 --> 00:06:13,430
your time