Interferometric Inverse Synthetic Aperture Radar 1
2
T G Kostis C J Baker H D Griffiths 1
University of the Aegean
2
2 University College London
Target identification (classification) as friend or foe (IFF) as an automatic radar function is connected with the concepts and techniques of High Resolution Radar systems. The basic concept employed is Inverse Synthetic Aperture Radar (ISAR). ISAR can determine the radar cross section of a target in its x (slant range) and its y (cross range) directions, providing a two-dimensional aspect of the observed target. Ways to extend the ISAR techniques towards the acquisition of the target’s third (height range) direction, based on Range-Doppler processing and Interferometric concepts, are presented in this paper.
1 Introduction Inverse Synthetic Aperture Radar (ISAR) [1] is an imaging process capable of providing the outline of a given target in a two dimensional plane, namely its slant range and its cross range dimensions. In the interferometry [2] method, data from two radar antennas are used simultaneously to produce a higher quality image than either radar system alone. The challenge of this project is to associate the above two methods by combining their relevant parts leading to a system called Interferometric Inverse Synthetic Aperture Radar (InISAR) [3]. The InISAR system is targeted to the provision of three-dimensional (3D) information about a generally non-cooperative target. This paper presents the related feasibility analysis by employing an organisation model for the simulation of high resolution radar systems. Finally the need for this project is connected with the ever demanding requirement for accurate target identification as friend or foe involving great distances between radar and target. 2 Theatre of Operations : Simulation Geometry The target used in this project is the KMS Bismarck, the flagship of the German military navy (Deutsche Kriegsmarine) in 1939. The battleship is observed by a hypothetical double radar pod mounted on the WQ-Z PBY Catalina that actually spotted the Bismarck on May 26th, 1941, as depicted in Figure 1.
Figure 1 KMS Bismarck as the sea target. The double radar pod is mounted on the WQ-Z PBY Catalina. The target is simulated by the determination of one hundred and forty five (145) point scatterers. Each one of these points contains the necessary information that needs to be processed by the radar, like in the real situation. Analytically each scattering reference is composed by its slant range, cross range, height range, reflectivity amplitude and initial phase information as shown in Figure 2. (x2,y2,z2)
Height Range Axis
Center Scatterer Aejφ
R RADAR
(x1,y1,z1)
SHIP
45
Slant Range Axis (0,0,0)
Figure 2 KMS Bismarck target simulation.
Cross Range Axis
And the simulation geometry is shown in Figure 3. The ship's center of gravity and center of buoyancy is the base of the mast for simulation purposes.
A second antenna acquires the interferometric data.
z
The ship's mast takes 1 sec to rotate by 2 degrees. 2000m
45
PRF = 1000Hz Sea Surface
Radar Beam Sweep Area
4 5k
Antenna Height 2001m
m
The radar uses a 32 pulse burst. Each pulse is spaced 1msec apart from each other.
Antenna Height 2000m
Every pulse is sampling a position rotated by 0.002 degrees.Therefore the motion of the ship's mast is sampled 1000 times every second.
The radar beam captures the target's profile every one (1) millisecond.
Sea Level
Figure 3 Simulation geometry. 3 Simulation Realisation : 3D Engine & Information Propagation The 3D Engine provides translational and rotational motion of the target. Furthermore the engine is a deterministic model for echo collection, as shown in Figure 4. FB 14 Implemented Bus - September 2001
Sensor Block
3D Engine 2
Radar Coordinates
1
xrotation
2
magnitude 3D
Target Coordinates Signal Procesing Block Reflectivity Data
3
3
slmulrotpoints
4
Results
Figure 4 FB-14 Data Bus Implementation. Numbers in arrows show the sequence of data circulation. The coordinates of the targets and radars are explicitly (a priori) known to the receiver block. In other words range is computed directly from radar and target three dimensional coordinates. Then this data is made available to the Interferometric and ISAR decision making processes as shown in Figure 5. R
RC
thetav
λ 0.06
SRxxx
CRxxx
HRxxx
phirot 0.002
CLxxx
RFxxx
RC 8m
IPxxx
R 45Km
thetav 45
Radar Parameters
Movement Engine
mainshipsl
mainship
Save coordinate data into a file.
SR
CR
HR
phirot
Β
mulrotpoints xrotation magnitude3d
(32 loop)
Increase rotation angle until loop ends
R1
All values are kept in arrays so they can be called outside the function by their index
R2
Radar Antenna 1
Radar Antenna 2
α
R2
λ
α
B
λ : Radar Wavelength (0.06 = 5GHz) phirot : Rotation Angle (0.002 degrees per 1msec) RC : Reflectance Cell Size R : Distance from Radar to middle RC thetav : Radar Aspect Angle
Environment Model
CRxxx
HRxxx
phirot
CLxxx
a01 ... a32
RFxxx
Range Delay
IPxxx
Increase rotation angle until loop ends
Delay 2Ro/c
b01 ... b32
SRX(n,t)=σeφ-Range Delay
Now ALL a and b return to mainshipsl and are assigned to diffent variables before they change for the next point
Target Point P
R1 H
Target
Antenna
Make vectors SR, CR and HR in order to be able to plot all above points via MATLAB.
R2
θ
STX(n,t)=Aeφ Exciter
Save all above info (145 cells) to file c:\matlab\ship\shipcoor.txt for reference.
SRxxx
RDHH
irif1 (interferometer)
CLxxx : Cell ID RFxxx : Initial Reflectance Amplitude IPxxx : Initial Phase SRxxx : Slant Range Coordinate (x axis) CRxxx : Cross Range Coordinate (y axis) HRxxx : Height Range Coordinate (z axis)
MOVEMENT PROCESSING BLACK BOX Note : Contents are shown separately
32 sets of R1 and R2 accessed by their index.
R1
Declare :
ANTENNA 1 : a01 = PxxxAyy, where xxx is the Point ID (ie 001) and yy is the movement profile (ie 01, 02, ...,32) ANTENNA 2 : b01 = PxxxByy, where xxx is the Point ID (ie 001) and yy is the movement profile (ie 01, 02, ...,32)
Now the REFLECTANCE MAP is initiated.
Windowing (Reduces Range Side-lobes)
MIXER Pulse Compression - Matched Filter Function - Dechirp
SREF(n,t)=eφ-Delay
Group all points that have common yy (ie Points P001A01 ... P145A01). (mainshipslb groups all common B points)
SIF(n,t-nT)=σeφ-Range Delay-Dechirp
Pass them through to the radar beam function ResCellsANT1 mainshipsla
mainshipslb
ANTENNA 1 : a01 = PxxxAyy, where xxx is the Point ID (ie 001) and yy is the movement profile (ie 01, 02, ...,32)
ANTENNA 2 : b01 = PxxxByy, where xxx is the Point ID (ie 001) and yy is the movement profile (ie 01, 02, ...,32)
(32 loop) PxxxAyy [xxx = 001 to 145] [yy = 01 ...32]
Estimated Target Height
ResCellsANT1
z
PxxxByy [xxx = 001 to 145] [yy = 01 ...32]
Low Pass FIltering for A/D Help
Estimated Target Range y
REVAxxx
32 sets of ETH and ETR accessed by their index.
z
SLRnnVmm [nn=01...32] [mm=01...33]
NOTE : The prefix ANT1 is used by both antennas.
NOTE : The prefix REVA is used by both antennas.
NOTE : The processes for the two antennas are not parallel they are performed by two files (mainshipsla - mainshipslb).
REVAxxx
SLRnnVmm [nn=01...32] [mm=01...33]
Analog to Digital Conversion
Now the SLANT RANGE PROFILES are in mainshipsl. Make vectors SLRnnV that represent the 32 slant range profiles. Plot SLRnnV vs Resolution Cell ID (1...33)
32 sets of ETH and ETR also saved into a file.
Make vectors VmmANT1 that represent the 33 vertical cut of the SRP. Make associated vectors DFTAVmm that are the fft of the VmmANT1 vectors. Plot DFTAVmm vs Resolution Cell ID (1...33)
Estimated Target Height and Range Module. Produces 32 sets of interferometric and phase data.
R
Range Profile Module- Automated Processing. Produces Slant Range Profiles and ISAR Height Information per pulse requested.
N −1 K −1
S DIG (n, k ) = σ ∑ (∑ eϕ ( n ,k ) ) n=0 k =0
Recorder
Figure 5 Data flow design charts and relative theory for the Interferometry and ISAR processes.
From the above software results concerning target information are extracted. In this stage very sophisticated software blocks can be introduced in order to provide, for example, complicated roll, yaw, pitch and translation motions.
4 Results First the interferometry technique for the determination of the height of the corresponding target scatterers yields the results that appear in Figure 6. Interferometer Profiles (32) Data 50
45
45
40
40
35
35
Es tim ated Target Height
Real Target Height
Real Target Height vs Resolution Cell ID 50
30 25 20 15 10
30 25 20 15 10
5
5
0 0
5
10
15 20 Resolution Cell ID
25
30
0 0
35
5
10
15 20 Resolution Cell ID
25
30
35
Figure 6 Interferometry simulation explanation and corresponding results. And in the second part the Range-Doppler [4] [5] method is now employed for the determination of the height profile of the target, as shown in Figure 7. TARGET SLANT RANGE PROFILE - FREQUENCY DOMAIN SIGNATURE
TARGET SLANT RANGE PROFILE - TIME DOMAIN HISTORY
Η/φ amplitude/phase of I and Q echo samples from the ith step of the kth burst
h/θ amplitude/phase of the slant range cell of the synthetic range profile from the kth burst
Steps from 1 to i Bursts from 1 to k
RESULTS
nth
Steps from 1 to n
PULSE 01
PULSE 02
PULSE 33
H/ˆ(1,1)
H/ˆ(1,2)
H/ˆ(1,33)
SLANT RANGE CELL 01
SLANT RANGE CELL 02
SLANT RANGE CELL 33
h/Ë(1,1)
h/Ë(1,2)
h/Ë(1,33)
E1 MULTILAYER MODEL - SRP Ant 1 (32)
4
2.5
BURST 1
FFT-1
x 10
2
Reflectance Value
1.5
1
0.5
BURST 32
H/ˆ(32,1)
H/ˆ(32,2)
FFT-1
H/ˆ(32,33)
h/Ë(32,1)
h/Ë(32,2)
h/Ë(32,33) 0 0
FFT
FFT
FFT
RD(1,1)
RD(1,2)
RD(1,33)
5
TARGET SLANT RANGE PROFILE - PHYSICAL DOMAIN
15 20 Resolution Cell ID
25
30
35
E1 MULTILAYER MODEL - ISAR Ant 1 (32)
5
7
10
x 10
6
Steps from 1 to j
5
4 Doppler
Catalina distance is 45 Km from target.
3 2
1
Aspect Angle is 45 degrees.
RD(32,1)
RD(32,2)
0
RD(32,33)
-1 0
Radar Sweep
FFT
FFT
FFT
RESOLUTION CELL 01
RESOLUTION CELL 02
RESOLUTION CELL 33
ISAR IMAGE - DOPPLER (HEIGHT) vs RESOLUTION CELLS
3 MORE MODEL C - SRP Ant 1 (32)
4
2.5
x 10
10
15 20 Resolution Cell ID
x 10
6 2
4 Doppler
Reflectanc e Value
5
1.5
1
3
2
1 0.5 0
0
0
5
10
15 20 Resolution Cell ID
25
30
35
-1
0
5
10
25
30
35
RD magnitude of the nth slant range cell and the jth doppler cell of the Range-Doppler image.
3 MORE MODEL C - ISAR Ant 1 (32)
5
7
5
15 20 Resolution Cell ID
25
30
Figure 7 Range-Doppler simulation methodology, geometry and results for height determination.
35
The comparison of Figure 6 with Figure 7 shows the general relevancy of results between the Interferometric and the Range-Doppler methods. And a careful examination of the two figures shows the differences and similarities between the two methods. This comparison is very useful to the radar operator trying to estimate the type of target that exists forty-five kilometres away from the radar. Analytically the Interferometric method will show the actual dimensions of the target, thus coming closer to an optical type of observation. While the Range-Doppler method will exhibit higher reflectance on parts of the target that may be gun placements or antennas, leading to a microwave type of observation. 5 Conclusions The objective of this work has been to establish the methodology of constructing an organisation system [6] and the corresponding mathematical model for the simulation of high resolution imagery radar systems. The main concepts are sound theoretical understanding, correct simulation geometry, structured data flow design and a fast three dimensional pace control engine. Coupled with the aid of mathematical modelling the radar engineer can extract realistic feasibility analysis results for a prospective complicated and expensive project. Furthermore the overall design was proven to be engineered in a way so to easily allow the tasks of reconfiguration, modular upgrade, debugging and generally readily provide for any future maintenance. Analytically target frequency response designation was illustrated for the KMS Bismarck. The way to receive this frequency response and transform its equivalent time domain planar return response was illustrated. The potential to process the above slant range profiles of the target into height, as indicated by its Doppler and Interferometric signatures, was shown. Finally the similarities and differences of the results from both techniques present more information about a target than with each method alone. Reflecting all above statements the simulation of an Interferometric Inverse Synthetic Aperture Radar (InISAR) imaging system was presented. Acknowledgements This work was supported by the Engineering & Physical Sciences Research Council (EPSRC). The authors thank Mrs M Diakaki for helping with the data processing requirements of the different Bismarck models used in this project. References [1]. BAKER, C.J., “High Resolution 3D Radar Imaging”, page 190, Malvern : DRA, Worcs WR14 3PS, UK, 1995. [2]. GRIFFITHS, H.D., “Interferometric Synthetic Aperture Radar (InSAR)”, 1995. London : IEE Electronics & Communication Engineering Journal Volume 7 Number 6. [3]. KOSTIS, T.G., “Interferometric Inverse Synthetic Aperture Radar”, University http://rootca.aegean.gr/tkostis/ (Range-Doppler Processing & Interferometry for ISAR), 2001.
College
London,
Master’s
Thesis,
[4]. WEHNER, D.R., “High Resolution Radar”, page 368 (Range-Doppler Processing), Artech House, 1994. [5]. SON, J.S., THOMAS, G. and FLORES, B.C., “Range Doppler Radar Imaging and Motion Compensation”, page 14 (Range-Doppler Processing), Artech House, 2001. [6]. LILLY, A.S., BRENNAN, D.V., GRIFFITHS, H.D., and MONEY, D.G., 1993, “A Generic Approach to Radar Performance Modelling”, IEE Radar 97, Conference Pub., No. 449, Oct. 1997.
Postal Addresses Theodoros G. Kostis CEng MIEE Research Unit, UNIVERSITY OF THE AEGEAN, 30 Voulgaroktonou Street, Athens 114 72, GREECE. Tel.: 0030 210 649 2059, E-mail :
[email protected] – Internet : http://rootca.aegean.gr Prof. Chris J. Baker Department of Electronic & Electrical Engineering, UNIVERSITY COLLEGE LONDON, Torrington Place, LONDON WC1E 7JE, U.K. Tel. : 0207 679 3966 - E-mail :
[email protected] - Internet : http://www.ee.ucl.ac.uk Prof. Hugh D. Griffiths Department of Electronic & Electrical Engineering, UNIVERSITY COLLEGE LONDON, Torrington Place, LONDON WC1E 7JE, U.K. Tel. : 0207 679 7310 - E-mail :
[email protected] - Internet : http://www.ee.ucl.ac.uk
