sensors
Article
A High-Resolution, Wide-Swath SAR Imaging System Based on
Tandem SAR Satellites
Liwei Sun * and Chunsheng Li
Citation: Sun, L.; Li, C. A
High-Resolution, Wide-Swath SAR
Imaging System Based on Tandem
SAR Satellites. Sensors 2022, 22, 7747.
https://doi.org/10.3390/s22207747
Academic Editor: Renato Machado
Received: 24 August 2022
Accepted: 4 October 2022
Published: 12 October 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Sensors 2022, 22, 7747. https://doi.org/10.3390/s22207747
School of Electronics and Information Engineering, Beihang University, Beijing 100083, China
* Correspondence: sunliwei@buaa.edu.cn; Tel.: +86-180-1009-3451
Abstract: For the spaceborne synthetic aperture radar (SAR), it is difficult to obtain high resolution
and wide width at the same time. This paper proposes a novel imaging system based on tandem SAR
satellites, where one obtains coarse resolution and wide swath by the scanning mode, and the other
obtains the undersampled echo from the same swath. The high resolution is achieved by associating
the tandem SARs’ echo and using the minimum-energy-based algorithm. Finally, a high-resolution
wide-swath SAR system is designed, and its imaging performance is verified by simulated data and
real airborne SAR data.
Keywords: synthetic aperture radar (SAR); high-resolution wide-swath; minimum-energy criterion
1. Introduction
The synthetic aperture radar (SAR) is a powerful sensor that can observe the Earth’s
surface day and night, regardless of the weather conditions [1–3]. This sensor has been
widely adopted in the military and civilian remote sensing areas, such as oceanographic
observations, sea-ice monitoring, and vegetation mapping [4–6]. For future SAR develop-
ment, high-resolution wide-swath (HRWS) technology plays an important role in these
applications. However, the high resolution and the wide swath contradict each other due
to the minimum antenna area constraint [7]. On the one hand, the high resolution requires
a high-pulse repetition frequency (PRF) to suppress the Doppler ambiguity. At the same
time, the wide swath needs a low PRF to avoid range ambiguity. Therefore, HRWS cannot
be obtained in a monostatic SAR system simultaneously [8].
For achieving monostatic HRWS-SAR imaging, researchers proposed many methods,
which can be divided into two categories: 1. The first approach focuses on the range
ambiguity issue, where a sufficiently high PRF is selected to avoid the Doppler ambiguity.
The frequency diversity array (FDA) technique is proposed in [9]. It introduces a small
increment across the all array elements, then the echoes received from different ambiguous
regions are separated. Moreover, the multiple-elevation-beam technology is adopted in [10].
It employs real-time beamforming on receipt to form multiple narrow elevation beams,
each following the radar echo of a different transmitted pulse; thus, the range ambiguity is
suppressed. 2. The second approach is to cope with the Doppler ambiguity issue, where
a low PRF is chosen to avoid range ambiguity. For example, the azimuth multichannel
technology compensates for the lack of temporal sampling by the spatial sampling of points
of multiple channels [11]. Thus, the effective sampling rate is increased, and the Doppler
ambiguity is suppressed [12]. Those technologies mentioned above can effectively improve
the width of the swath while ensuring the azimuth resolution for monostatic SAR. However,
they all have a complex structure, leading to greater challenges in designing a large satellite.
To reduce hardware complexity [13–15], separates the transmitter and receiver of SAR to
a multi-satellite network, but the increased launch volume and mass for the deployment
of multiple satellites and additional hardware for accurate phase synchronization are
indispensable [16]. Furthermore, [17] develops a monostatic SAR−HRWS imaging system
https://www.mdpi.com/journal/sensors
Sensors 2022, 22, x FOR PEER REVIEW 2 of 16

Sensors 2022, 22, 7747 2 of 16

accurate phase synchronization are indispensable [16]. Furthermore, [17] develops a mon-
ostatic SAR−HRWS imaging system based on compressed sensing (CS) theory, but its im-
aging performance depends on the sparsity of the observed scene.
based on compressed sensing (CS) theory, but its imaging performance depends on the
In order to achieve HRWS-SAR imaging, this paper proposes a novel imaging system
sparsity of the observed scene.
based on tandem SAR satellites. The first one works in the scanning mode to obtain the
In order to achieve HRWS-SAR imaging, this paper proposes a novel imaging system
coarse observations of a wide swath. The second one obtains the downsampled echo from
based on tandem SAR satellites. The first one works in the scanning mode to obtain the
the same swath. Then, the high-resolution imaging is achieved by associating the tandem
coarse observations of a wide swath. The second one obtains the downsampled echo from
SARs’ echoswath.
the same and using
Then,the
theminimum-energy-based
high-resolution imagingalgorithm. Since
is achieved the precise phase
by associating syn-
the tandem
chronization between two satellites is unnecessary, the system cost can
SARs’ echo and using the minimum-energy-based algorithm. Since the precise phase be reduced.
This paper isbetween
synchronization organized
twoassatellites
follows:isSection 2 introduces
unnecessary, the working
the system modes
cost can be of the
reduced.
two satellites. Section 3 constructs the signal model and describes the minimum-energy-
This paper is organized as follows: Section 2 introduces the working modes of the two
based algorithm.
satellites. Section Section 4 designs
3 constructs an example
the signal system
model and and verifies
describes its imaging perfor-
the minimum-energy-based
mance. Finally, Section 5 concludes this paper.
algorithm. Section 4 designs an example system and verifies its imaging performance.
Finally, Section 5 concludes this paper.
2. Imaging System Based on Tandem SAR Satellites
2. Imaging Systemsystem
The proposed Based consists
on Tandem SAR
of two Satellites
satellites flying successively in the same orbit.
The distance between
The proposed two satellites
system consistsisoffar
twoenough to ensure
satellites flying that they do not
successively interfere
in the same with
orbit.
each other. Both
The distance satellites
between twoadopt a scanning
satellites mode, but
is far enough the scanning
to ensure methods
that they are different.
do not interfere with
The
eachintroduction is presented
other. Both satellites adoptin detail as follows.
a scanning mode, but the scanning methods are different.
The introduction is presented in detail as follows.
2.1. Working Mode of the First Satellite
2.1. Working Mode of the First Satellite
The first satellite, namely, Sat-1, operates in the traditional scanning mode [18], which
adopts The
thefirst satellite,
burst mode, namely, Sat-1, operates
and assigns in the traditional
the aperture scanning
time to multiple mode [18],Itwhich
subswaths. can
adopts the burst mode, and assigns the aperture time to multiple subswaths.
broaden the range swath at the cost of coarsening the azimuth resolution. The observation It can broaden
the range of
geometry swath
Sat-1atisthe cost of
shown in coarsening
Figure 1. To the azimuth
achieve anresolution. The observation
imaging system with azimuth geometry
reso-
of Sat-1 is shown in Figure 1. To achieve an imaging system with azimuth
lution δ az , the necessary burst time Tn of the nth subswath should satisfy the following: resolution δaz ,
the necessary burst time Tn of the nth subswath should satisfy the following:
0.886 ⋅ λ R0 ,n
Tn =0.886 · λR (1)
Tn = 2V δ az 0,n (1)
2Vδaz
where λ is the wavelength, R0 ,n is the mean slant range in the zero Doppler plane of
where λ is the wavelength, R0,n is the mean slant range in the zero Doppler plane of the
the nth subswath, and V is the speed of the SAR platform.
nth subswath, and V is the speed of the SAR platform.

Figure
Figure1.1.Pulse
Pulseassignment
assignmentdiagram
diagramof ofthe
thetraditional
traditionalscanning
scanningmethod,
method,where
wherethe
theantenna
antennabeam
beam
jumps periodically among different subswaths. N indicates the number of subswaths.
jumps periodically among different subswaths. N indicates the number of subswaths.

According
Accordingto
tothe
theprinciple
principleof
oftraditional
traditionalScanSAR,
ScanSAR, we
we have
have

BDB= ( ( N++ 11))⋅·BBn n

D =N (2)
(2)
where B n and B D denote the Doppler bandwidth of the nth subswath and the 3 dB
where Bn and BD denote the Doppler bandwidth of the nth subswath and the 3 dB band-
bandwidth
width of the of the azimuth
azimuth antenna, respectively.
antenna, respectively. According toAccording to the
Equation (2), Equation (2),ofthe
coarsening the
azimuth resolution is in the order of N + 1, while the range swath compared to a strip-map
system is increased by N.
Sensors 2022, 22, x FOR PEER REVIEW 3 of 16
coarsening of the azimuth resolution is in the order of N + 1, while the range swath com
pared to a strip-map system is increased by N.

coarsening of the
2.2. Working Modeazimuth resolution
of the Second is in the order of N + 1, while the range swath com-
Satellite
Sensors 2022, 22, 7747 pared The
to a strip-map systemwhich
is increased 3 of 16
byasN.Sat-2, adopts a new scanning method.
second satellite, is noted It ran
domly assigns each pulse to one of the subswaths on the ground and receives the reflecte
2.2. Working
echo, ModeinofFigure
as shown the Second
2. Satellite
2.2. Working Mode of the Second Satellite
The second satellite, which is noted as Sat-2, adopts a new scanning method. It ran-
The secondeach
domly assigns satellite,
pulsewhich
to oneisofnoted as Sat-2, adopts
the subswaths on theaground
new scanning method.
and receives the Itreflected
randomly assigns each pulse to one of the subswaths on the ground and receives the
echo, as shown in Figure 2.
reflected echo, as shown in Figure 2.

Figure 2. Pulse assignment diagram of the new scanning method, where each pulse is randoml
assigned to one of the subswaths on the ground and different subswaths have different squint an
gles.
Figure 2.
Figure 2. Pulse
Pulseassignment
assignment diagram
diagram of the newnew
of the scanning method,
scanning wherewhere
method, each pulse
eachispulse
randomly
is randomly
assignedSince
to onespaceborne
of the subswaths SAR on platforms
the ground usually
and differentwork at
subswathsan
assigned to one of the subswaths on the ground and different subswaths have different altitude
have differentof more
squint than
angles. 100an-
squint km
[19], the reflected echoes of transmitted pulses return to the receiver after several puls
gles.
Since spaceborne SAR platforms usually work at an altitude of more than 100 km [19],
intervals (PIs). Because all subswaths share one receiver, the waiting PI number for diffe
the reflected echoes of transmitted pulses return to the receiver after several pulse intervals
ent Since
(PIs). Becausespaceborne
subswaths all must SAR
be
subswaths the platforms
sharesame usually
onetoreceiver,
avoid thework
overlapping
waitingat PI
an altitude
ofnumber ofdifferent
the received
for more than
echoes.
sub- 100 km
Assumin
[19], the
that the
swaths reflected
mustwaiting echoes
PI number
be the same of
to avoid transmitted
isoverlapping pulses
NPI , as shown return
of theinreceivedto the
Figureechoes.receiver
3, the reflectedafterthat theof pulse
Assumingechoes several all th
intervals
waiting PI(PIs).
number Because
subswaths should return is N PIall
, assubswaths
shown in share
Figure one
3, the receiver,
reflected the waiting
echoes of all PI
thenumber
subswaths
to the receiver during the N PI +1th PI after transmission. To tha for differ-
ent subswaths
should return tomust be the during
the receiver same to theavoid
NPI +overlapping of the receivedthat
1th PI after transmission. echoes. Assuming
end, we set the farthest subswath work in broadside mode and To the otherend, we
subswaths i
set
that the farthest
the mode.
waitingsubswath work
PI squint
number in broadside
is Nincreases mode and the other subswaths in squint mode.
PI , as shown in Figure 3, the reflected echoes of all the
squint The angle as the distance of subswath to nadir decreases, a
The squint angle increases as the distance of subswath to nadir decreases, as shown in
subswaths
shown should return to the receiver during the N PI +1th PI after transmission. To that
Figure 2. in Figure 2.
end, we set the farthest subswath work in broadside mode and the other subswaths in
squint mode. The squint angle increases as the distance of subswath to nadir decreases, as
shown in Figure 2.

Figure
Figure Timing
3. 3. Timingdiagram of theof
diagram subswath’s pulse transmitting
the subswath’s and receiving,
pulse transmitting andwhere the vertical
receiving, where line
the vertic
indicates
line indicates the pulse emission time, and the long red line indicates that the currenttopulse is a
the pulse emission time, and the long red line indicates that the current pulse is assigned
this subswath.
signed The
to this short gray The
subswath. line suggests that line
short gray the current
suggestspulse is assigned
that to other
the current subswaths.
pulse S1 , to othe
is assigned
S2subswaths.
, and S5 represent the reflected echoes of pulses 1, 2, and 5, respectively.
S1, S2, and S5 represent the reflected echoes of pulses 1, 2, and 5, respectively.
Figure 3. Timing diagram of the subswath’s pulse transmitting and receiving, where the vertical
line indicates the pulse emission
In the trans-receive view, thetime,
timeand the long
interval redtwo
of the linetransmitted
indicates that the current
pulses is fixedpulse
to is as-
signed In
to the
this trans-receive
subswath. The view,
short the
graytime
line interval
suggests of the
that thetwo transmitted
current pulse is
1/PRF, while the time interval of the two received pulses of any subswath is not certain, pulses
assigned istofixed
othert
the ,average
subswaths.
1 PRF
and while
S1, S2,the
andtime interval
S5 represent
sampling ofreflected
rate isthe thethan
lower twothereceived
echoes pulses
rate.1, of
of pulses
Nyquist any5,subswath
2, and is not certain
respectively.
Therefore, the received
and the
echoes areaverage sampling rate is lower than the Nyquist rate. Therefore, the received ech
downsampled.
oesIn
arethe trans-receive view, the time interval of the two transmitted pulses is fixed to
downsampled.
12.3. Comparison
PRF , while the of the Twointerval
time Scanningof Methods
the two received pulses of any subswath is not certain,
and The
2.3.the scanning method
average sampling
Comparison adopted
of the Tworate in
is lower
Scanning Sat-2 canthe
than
Methods keep the entire
Nyquist rate.aperture time the
Therefore, of targets,
received ech-
which is different from the traditional method. As shown in Figures 4 and 5, each point
oes areThe downsampled.
scanning
represents an azimuthmethod
sample inadopted
the echoinofSat-2 can keep
a ground targetthe
andentire aperture
corresponds to atime of target
trans-
which is different from the traditional method. As shown in Figures
mitted pulse. The traditional scanning method assigns a continuous pulse train to one 4 and 5, each poin
2.3. Comparison
represents
subswath, thus
of the
an truncatingTwo
azimuth sampleScanning
the target’s
Methods
in the echo oftime.
aperture a ground target and
The proposed corresponds
scanning methodto a tran
The pulse.
mitted
randomly scanning
choosesThe method
the pulses adopted
traditional in
scanning
assigned Sat-2
to one can keep
method it the
can entire
assigns
subswath; aretain aperture
continuous
the time train
entirepulse
apertureof targets,
to on
time,
which although
is with
different downsampling.
from the traditional method. As shown in Figures
subswath, thus truncating the target’s aperture time. The proposed scanning metho 4 and 5, each point
represents an azimuth sample in the echo of a ground target and corresponds to a trans-
mitted pulse. The traditional scanning method assigns a continuous pulse train to one
subswath, thus truncating the target’s aperture time. The proposed scanning method
 Sensors
Sensors 2022,
2022, 22,
22, xx FOR
FOR PEER
PEER REVIEW
REVIEW 44 of
of 16
16

Sensors 2022, 22, 7747 4 of 16

randomly chooses the pulses assigned to one subswath; it can retain the entire aperture
time, although with downsampling.

Figure4.4.Traditional
Figure Traditionalscanning
scanningmethod.
method.

Figure5.5.Proposed
Figure Proposedscanning
scanningmethod.
method.

An
Anexample
examplesystem
systemisisdesigned
designedin inSection
Section4.4.After
Afterthe
theechoes
echoesare
arereceived,
received,the
thecoarse
coarse
observations obtained by Sat-1 are used as prior knowledge to achieve high-resolution
observations obtained by Sat-1 are used as prior knowledge to achieve high-resolution
imaging
imagingfrom
fromthe
thereceived
receivedSat-2
Sat-2data.
data.
3. Signal Model and Imaging Algorithm
3. Signal Model and Imaging Algorithm
According to the system introduced above, Sat-1 works in traditional scanning mode;
According to the system introduced above, Sat-1 works in traditional scanning mode;
its received data can be processed by existing algorithms [1–3]. This section mainly focuses
its received data and
on the modeling can be processed
imaging by existing algorithms [1–3]. This section mainly focuses
for Sat-2.
on the modeling and imaging for Sat-2.
3.1. Signal Model of the Sat-2
3.1. Signal Model of the Sat-2
According to the SAR principle described in [20], the original SAR echo of Sat-2 can
According
be expressed as to the SAR principle described in [20], the original SAR echo of Sat-2 can
be expressed as (  )
2R(η ) 2
    
2R(η ) 4π
exp −2jR (η R  2 R (η )  
S0 (η, τ ) = A0 Wa (η − ηc )Wr τ − ) (η )  · exp
4π jπkr τ −
22
  (3)
S00 (η ,τ ) = A00Waa (ηc − η cc ) Wrr τ − λ  exp  − j R (η )  ⋅ exp  jπ krr c τ −   (3)
 c  λ    c
     
where τ and η denote the range and azimuth time, respectively, and ηc is the beam center
where τtime.
crossing andR(ηη )denote the range
is the slant distanceandbetween
azimuththe time, respectively,
observed ηcc platform
target and the is the beam
at
azimuth time η, ctime.
center crossing is the RR (η ) isofthe
speed light, kr distance
slant is the pulse modulation
between frequency,
the observed A0 is
target a complex
and the plat-
constant, and Wa (·)
form at azimuth timeandηW , r c(·) is
are
thethe azimuth
speed and range
of light, antenna mode, respectively. M is
krr is the pulse modulation frequency,
the number of targets on the observed scene.
A00 is a complex constant, and W Waa ( ⋅⋅) and Wrr ( ⋅) are the azimuth and range antenna
After range compression, Equation (3) can be expressed as
mode, respectively. M is the number of targets   on the observed  scene. 
After range compression, Equation (3) can be2R (η )
expressed as 4π
S1 (η, τ ) = A0 Wa (η − ηc )sin c Br τ − exp − j R(η ) (4)
c λ
  2 R (η )    4π 
S11 (η ,τ ) = A00Waa (η − ηcc ) sinc  Brr τ −   exp  − j R (η )  (4)
where Br is the bandwidth of the transmitted 
 pulse. S
c (   τ ) can
η,  λ
also be expressed in the
 1 
matrix form as
Si = A ·pulse.
where Brr is the bandwidth of the transmitted σi S11 (η ,τ ) can also be expressed in (5)
the
matrixSform
where as
i and σi are the observation vector and scene vector of the ith range gate, respec-
tively. σi = [σi,1 , σi,2 , · · · , σi,K ]T and σi,k is the Sii =equivalent
A ⋅ σ ii backscatter coefficient of the kth (5)
pixel, K is the number of pixels in the ith range gate. A = [A1 , A2 , · · · , Ak , · · · , AK ] is the
where matrix
sensing S ii andwith
σii are the observation vector and scene vector of the ith range gate, re-
 
TT 4π
spectively. σii = σ ii,1,1,σ ii,2,2,A  a (and
k,σ=ii,,KKW η − ησc,k
ii,,kk) ·is
exp − j R(η, kbackscatter
the equivalent ) coefficient of the
λ
kth pixel, K is the number of pixels in the ith range gate. A A == A A22 ,,
A11 ,,A Akk ,,
,,A AKK  is the
,,A
where η = [η1 , η2 , · · · , η Na ]T is the azimuth time vector, Na is the total number of sampling
sensing matrix with
points in the azimuth direction, ηc,k is the beam center crossing time of the kth pixel, and
R(η, k ) is the slant distance vector between the kth pixel and the platform.
According to Equation (5), the traditional imaging algorithms [21,22] based on matched
filtering can be expressed as
σ̂i = AH · Si (6)
 where η = η1 ,η 2 , ,η N  is the azimuth time vector, Na is the total number of sampling
a

points in the azimuth direction, ηc , k is the beam center crossing time of the kth pixel, and
R ( η, k ) is the slant distance vector between the kth pixel and the platform.
According to Equation (5), the traditional imaging algorithms [21,22] based on
Sensors 2022, 22, 7747 matched filtering can be expressed as 5 of 16

σˆ i = A H ⋅ Si (6)
where AHH is
where is the
theconjugate
conjugatetranspose
transposematrix
matrixofofA.ASince thethe
. Since received echo
received of Sat-2
echo is seri-
of Sat-2 is
ously downsampled,
seriously downsampled, severe ambiguity
severe energy
ambiguity occurs
energy in theinimaging
occurs resultresult
the imaging of Equation (6).
of Equa-
The(6).
tion CS-based SAR system
The CS-based introduces
SAR system the sparsity
introduces prior to
the sparsity thetoscene
prior reconstruction
the scene reconstruc-to
compensate for the information lost by downsampling, and then selects the
tion to compensate for the information lost by downsampling, and then selects the solu-solution with
the best sparsity as the estimation of the observed scene, as shown in (7).
tion with the best sparsity as the estimation of the observed scene, as shown in (7).

σ̂iσˆ = = argmin
arg minkσ
σ̂ = Aσ
ˆ i i k0 s.t. SSi i= ˆ ii
Aσ̂ (7)
i 0 (7)
σ̂
ˆi
σ
i

However,
However,the theestimation
estimationaccuracy
accuracyofof (7)(7)
depends
depends ononwhether
whether thethe
scene is sparse
scene or
is sparse
whether it can
or whether be sparsely
it can represented
be sparsely accurately.
represented accurately. In order to achieve
In order betterbetter
to achieve imaging per-
imaging
formance, this this
performance, paper modifies
paper modifiesthethe
traditional
traditional minimum-energy
minimum-energy criterion
criterion[23]
[23]introducing
introducing
the
thecoarse
coarseobservation
observationas asprior
priorinformation
informationand andconstructs
constructsthe
theoptimization
optimizationproblem
problemas as
−1
σˆ = arg min σˆ i , Q− ⋅ σˆ i s.t. Si = Aσ ˆi
D E
σ̂i =i argmin σ̂i , Q 1 · σ̂ i s.t. Si = Aσ̂i
(8)
(8)
σˆ
σ̂
where Q = diag ( σd ,i
⋅ σ *d , i ) is a K × K matrix and σ d ,i is the amplitude envelope of the im-
where = diag · ∗ K ×the
is a on K matrix
Q
aging result of the ithσ d,irange σ d,i gate ground,and is the
σd,i can
which be amplitude
obtained byenvelope of the
interpolating
imaging
the result
imaging of the
result of ithSat-1. range gate on
Finally, thethe ground,
solution which
with thecan be obtained
minimum by interpolating
weighted energy is
the imaging result of
chosen as the scene estimation. Sat-1. Finally, the solution with the minimum weighted energy is
chosen as the scene estimation.
3.2. Minimum-Energy-Based Imaging Algorithm
3.2. Minimum-Energy-Based Imaging Algorithm
Based on Equation (8), this part proposes a minimum-energy-based algorithm that
Based on Equation (8), this part proposes a minimum-energy-based algorithm that
mainly includes five parts: range compression, observation vector extraction, sensing ma-
mainly includes five parts: range compression, observation vector extraction, sensing
trix construction, interpolation for coarse observation, and minimum-energy-based esti-
matrix construction, interpolation for coarse observation, and minimum-energy-based
mation. The procedure is shown in Figure 6, and each step is introduced as follows.
estimation. The procedure is shown in Figure 6, and each step is introduced as follows.

com ( τ )
Procedureofofthe
Figure6.6.Procedure
Figure theproposed
proposedalgorithm,
algorithm,where
whereH Hcom f( f τ )isisthe
thematched
matchedfiltering
filteringfunction,
function,
σ̂sum is the reconstructed scene, σ̂sum ( : , i ) = σ̂i indicates that σ̂i is assigned to the ith column of
ˆ sum ( : , i ) =σ
σ̂sum is the reconstructed scene, σ ˆ indicates that σˆ i is assigned to the ith column of σ̂sum
σ̂sum , and Ng is the number of range gates,i the numbers 1–5 mark the main steps of the algorithm.
, and N g is the number of range gates, the numbers 1–5 mark the main steps of the algorithm.
3.2.1. Range Compression
The range compression filter Hcom ( f τ ), as shown in Figure 6, can be constructed by

f2
 
Hcom ( f τ ) = exp − jπ τ (9)
kr

where f τ is the range frequency. After range compression, S1 (η, τ ) is obtained.

  r 

where fτ is the range frequency. After range compression, S1 (η ,τ ) is obtained.

Sensors 2022, 22, 7747

3.2.2. Observation Vector Extraction 6 of 16
In S1 (η ,τ ) , the target echoes appear as curves with range cell migration (RCM
the RCM must be corrected to extract the observation vector Si , which can be exp
3.2.2. Observation Vector Extraction
as
In S1 (η, τ ), the target echoes appear as curves with range cell migration (RCM). And
the RCM must be corrected toKextract the observation vector Si , which can be expressed
T as

( )  S (η ,τ )  S (η )
K K
Si =   S1 η1 ,τ η1 , k 1 2 η2 , k  1 Na
,#τTηN ,k
  k =1  K 
" a
K k =1 K k =1
∑ S1 ∑ S1 ∑ S1


Si = η1 , τη1 ,k η2 , τη2 ,k ··· η Na , τη Na ,k (10)
where τη n ,k
= 2R (ηn , k ) c is the main lobe’s peak position of the kth pixel at η n . The
k =1 k =1 k =1

where of
tration = extraction
τηn ,kthe 2R(ηn , k )/c of
is the
S i main lobe’s in
is shown peak 7. of the kth pixel at ηn . The
position
Figure
illustration of the extraction of Si is shown in Figure 7.

(a) (b)
Figure
Figure 7.
7. Illustration
Illustration ofof
thethe extraction
extraction of Si . of
(a) S . (a) Diagram
Diagram
i of the ith of thegate,
range ith range gate,point
where each where each
represents aapixel.
represents pixel.(b)(b)
Illustration of the RCM
Illustration correction
of the of the compressed
RCM correction of theecho of the ith range
compressed gate.
echo of the ith
gate.
3.2.3. Observation Vector Extraction
According to the definition in Equation (5), the sensing matrix can be constructed by
3.2.3. Observation Vector Extraction
A = [A1in
According to the definition 2 , · · · , Ak , · (5),
, AEquation · · , Athe (11)
K ] sensing matrix can be construc
with A =  A 1 ,A 2 , , A k , ,A K 
4π
Ak = Wa (η − ηc,m ) · exp − j R(η, k )
λ
with
3.2.4. Interpolation for Coarse Observation
 4π 
Since the coarse observations from A k =Sat-1,
Wa ηwhich (
− ηc , m are
⋅ exp − j
processed )( ηthe
Rby , k ) existing algo-
rithms [1–3], are used as the weighting coefficients on pixels  inλeach range  gate of the
final imaging result, the nearest neighbor interpolation algorithm [24] is adopted to ensure
that the coarse observations are consistent with the final image’s sampling space. The
3.2.4. Interpolation
interpolation principlefor
canCoarse Observation
be expressed as
Since the coarse observations from Sat-1,  & which '! are processed by the existing
0 x ye
rithms [1–3], are usedσas ethe weighting coefficients on pixels in each range(12) gate of th
e
d (x , ye) = σd ,
∆d x /∆d0x ∆dy /∆d0y
imaging result, the nearest neighbor interpolation algorithm [24] is adopted to ensu
the coarse
where ( xe, yeobservations are consistent
) is the 2D coordinates of pixels with
in the the final image’s
interpolated result, sampling
and σ0d and space.
σd are The in
0
the interpolation, respectively. ∆d x and ∆d x are the
lation principle
the image can be
amplitudes expressed
before and after as
sampling space before and after interpolation along the X direction, respectively, and ∆dy
and ∆d0y are the sampling space before and after interpolation along Y direction, respectively.
The diagram of interpolation is shown in Figure 8.
 where ( x , y ) is the 2D coordinates of pixels in the interpolated result, and σ′d and σ d
are the image amplitudes before and after the interpolation, respectively. Δdx and Δd′x
are the sampling space before and after interpolation along the X direction, respectively
and Δdy and Δd′y are the sampling space before and after interpolation along Y direction
Sensors 2022, 22, 7747 7 of 16
respectively. The diagram of interpolation is shown in Figure 8.

Figure8.8.Illustration
Figure of the
Illustration of nearest neighbor
the nearest interpolation
neighbor algorithm,
interpolation where ( x, ywhere
algorithm, ( y ) is the 2D coor
) is the 2Dx ,coordinate
of pixels in the original image.
dinate of pixels in the original image.
3.2.5. Minimum-Energy-Based Estimation
3.2.5.ToMinimum-Energy-Based Estimation
obtain the minimum weighted energy solution required in Equation (8), we define
a newTo obtain
norm the
in the minimum
space weighted
consisted energy sequences
of band-limited solution requiredas in Equation (8), we define
a new norm in the space consisted of band-limited
D Esequences as
kσ̂i knew = σ̂i , Q−1 · σ̂i (13)
σˆ i = σˆ i , Q−1 ⋅ σˆ i .
new
(13
The validation of the new norm can be found in Appendix A. Then, the solution with
The validation
minimum of theisnew
weighted energy norm can
the solution be the
with found in Appendix
smallest new norm.A.According
Then, the tosolution
the with
minimumtheorem
projection weighted energy
in [25], is the
σ̂i must besolution with
orthogonal in the
h·, ·ismallest new
new to ℵ(A norm.
), where ℵ(According
A) is the to the
null space of A and h· ,
projection theorem in new ·i [25], σˆ i must be orthogonal in ⋅, ⋅ new to ℵ ( Anew
is the new inner product corresponding to k · k ) , .where
So, we ℵ ( A ) i
obtain
the null space of A and ⋅, ⋅ new is σ̂ the new ∗inner product corresponding to (14)
i ∈ <(A )
⋅ new . So, we
obtain<(A∗ ) is the range space of A∗ , and A∗ is the adjoint operator of A, which can be
where
calculated by
σˆ i ∈
A∗ = QA Hℜ A
*
( ) (15) (14

The ℜ
where ( )
specific
A * is derivation
the rangeof Equation
space of(15)
A*can be found
, and A* isinthe
theadjoint
appendix of [26]. According
operator of A , which can
to Equations (14) and (15), σ̂i can be expressed as
be calculated by
σ̂i = QAH* · v H (16)
A = QA . (15
where v is a Na × 1 vector. According to Si = Aσ̂i in Equation (8), v can be calculated by
The specific derivation of Equation (15) can be found in the appendix of [26]. Accord
ing to Equations (14) and (15), σˆ i can
 be expressed
H
 −1 as
v = AQA Si (17)
σˆ i = QA ⋅ v .
H
(16
Then, the σ̂i can be obtained by
where v is a Na ×1 vector. According to Si =−
Aσ ˆ
1 i in Equation (8), v can be calculated
by σ̂i = QAH AQAH Si (18)

( )
−1
It can be divided into the following two
v=steps:
AQA H Si . (17
  −1
Then, the σˆ i can be obtained by AQAH
σ̂tem = Si (19)

σ̂i = Q · AH σ̂tem (20)

Step one normalizes the amplitude of the targets in Si as shown in Equation (19), and
step two is to image the normalized result of step one as shown in Equation (20). Due to the
Sensors 2022, 22, 7747 8 of 16

inevitable model error and noise in the real data, we modify Equations (19)–(21) to achieve
more robust imaging performance.

σ̂tem = Qs −1 Si (21)
 ρ
where Qs is a diagonal matrix, its diagonal elements are |Si,1 |ρ , |Si,2 |ρ , · · · , |Si,Na | , and
|Si,n | is the absolute value of the nth element in Si , ρ is the normalization factor and this
paper sets ρ = 0.5. At the same time, Equation (20) can be modified to an iterative form, as
shown in Equation (22).
σ̂i,j = Q j−1 · AH σ̂tem (22)
where σ̂i,j is the jth reconstruction result and σ̂i,0 = σd,i , σd,i is the ith column of the coarse
 
∗
observation of Sat-1, Q j−1 = diag σ̂i,j−1 · σ̂i,j −1 . This paper sets the iteration order to 2,
and the detailed steps are listed in Table 1.

Table 1. Processing steps of the real airborne SAR data.

Processing Steps Content

Step one Obtains S1 (η, τ ) by range compression
 of the original
echo.
Step two Extracts the observation vectors Si |i = 1, · · · , Ng from S1 (η, τ )
For i = 1: Ng
σ̂tem = Qs−1 S
i 
Q0 = diag σd,i · σ∗d,i
Step three σ̂i,1 = Q0 ·AH σ̂tem 
∗
Q1 = diag σ̂i,1 · σ̂i,1
σ̂i,2 = Q1 · AH σ̂tem
End
Output:
Step four  
σ̂sum = σ̂1,2 σ̂2,2 ··· σ̂ Ng ,2

4. Verification and Analysis

This section aims to verify the discussion in the previous sections. It can be divided
into three parts. Based on the SAR imaging system introduced above, Section 4.1 designs
an example system and verifies the imaging performance. Section 4.2 verifies the effec-
tiveness of the proposed algorithm with real airborne SAR data. Section 4.3 analyzes the
performance variation of the proposed system with the number of subswaths. The details
are as follows.

4.1. System Design Example and Imaging Performance

Table 2 shows the satellite parameters used in this part.

Table 2. System parameters for Sat-1 and Sat-2.

Parameters Value
Carrier frequency (GHz) 9.6
Range bandwidth (MHz) 100
Sampling frequency (MHz) 120
Pulse width (µs) 20
Orbital height (km) 630
Platform velocity (m/s) 7545
Resolution (m) 3

Based on the parameters in Table 2, this part designs an example HRWS-SAR system.
The azimuth and range resolutions of the system are set to 3 m. The range swath is set to
200 km and divided into six subswaths. Each subswath has the same PRF and different
 Sensors 2022, 22, 7747 9 of 16

squint angles. The beam position parameters of the two satellites are the same as in Table 3.
The scanning parameters of Sat-1 are listed in Table 4.

Table 3. Beam position parameters of six subswaths.

Look Angle in the

PRF (Hz) Squint Angle (◦ ) Subswath Width (km)
Zero Doppler Plane (◦ )
Subswath 1 32.01~34.41◦ 2673 0◦ 39.90
Subswath 2 29.79~32.19◦ 2673 8.70◦ 37.58
Subswath 3 27.51~29.91◦ 2673 15.30◦ 35.55
Subswath 4 25.23~27.63◦ 2673 19.40◦ 33.82
Subswath 5 22.91~25.31◦ 2673 22.50◦ 32.31
Subswath 6 20.55~22.95◦ 2673 26.80◦ 31.00

Table 4. Scanning parameters of Sat-1.

Burst Duration(s) 3 dB Doppler Bandwidth (MHz)

Subswath 1 0.077 2446
Subswath 2 0.075 2428
Subswath 3 0.074 2323
Subswath 4 0.072 2231
Subswath 5 0.070 2148
Subswath 6 0.069 2032

According to the designed SAR imaging system, the received echo of subswath 1 is
simulated based on the scene as shown in Figure 9. Then, the BP algorithm, the L1 algorithm,
and the proposed algorithm are carried out to image the simulated echo. According to the
ors 2022, 22, x FOR PEER REVIEW
imaging results in Figure 10, there is obvious ambiguity energy in the imaging result of the
BP algorithm, while the L1 algorithm and the proposed algorithm achieve good imaging
quality. Figure 11 shows the azimuth profiles of the range gate marked by a red arrow in
Figure 10.

Figure 9. Illustration of the observed scene including 25-point targets.

Figure 9. Illustration of the observed scene including 25-point targets.
Sensors 2022, 22, 7747 10 of 16

Figure
Figure9.9.Illustration
Illustrationof
of the
the observed scene including
observed scene including25-point
25-pointtargets.
targets.

(a) (b) (c)

(a) (b) (c)
Figure10.
Figure 10. Illustration
Illustration of the
ofthe imaging
imagingresults,
theimaging results,where
where(a–c)
(a–c)are the
are results
the of of
results thethe
BPBP
algorithm, thethe
algorithm, L1 L1
Figure 10. Illustration of results, where (a–c) are the results of the BP algorithm, the L1
algorithm, and the proposed algorithm, respectively.
algorithm,and
algorithm, andthe
theproposed
proposed algorithm,
algorithm, respectively.
respectively.

Figure 11. Illustration of the azimuth profile.

Figure Illustrationof
11.Illustration
Figure 11. ofthe
theazimuth
azimuthprofile.
profile.
Then, the azimuth resolution, the peak side lobe ratio, the integrated side lobe ratio,
Then,
andThen, the azimuth
the structure resolution,
similarity are used the peak side
to peak
measure lobe ratio, the integrated side lobe ratio,
the azimuth resolution, the sidethe performance
lobe of the threeside
ratio, the integrated algorithms.
lobe ratio,
and
Thethe structure
evaluated similarity
results are used
are shown to measure
in Table 5. the performance of the three algorithms.
and the structure similarity are used to measure the performance of the three algorithms.
The evaluated results are shown in Table 5.
The evaluated results are shown in Table 5.
Table 5. Evaluation results.

Proposed
BP Algorithm L1 Algorithm
Algorithm
Azimuth resolution (m) 2.94 2.73 2.73
PSLR (dB) −16.96 −20.55 −22.62
ISLR (dB) −9.64 −15.16 −15.74
SSIM 0.61 0.80 0.81

Consistent with the description above, the evaluation results in Table 5 show that the
proposed algorithm has the highest SSIM, PSLR, and ISLR; the imaging quality of the L1
algorithm is slightly worse; and the performance of the BP algorithm is the worst. Since
the simulation scene in Figure 9 is sparse, and cannot verify the imaging performance
sufficiently, the next section uses real airborne SAR data to further measure the performance
of the three algorithms.
 proposed algorithm has the highest SSIM, PSLR, and ISLR; the imaging quality of the L1
algorithm is slightly worse; and the performance of the BP algorithm is the worst. Since
the simulation scene in Figure 9 is sparse, and cannot verify the imaging performance
sufficiently, the next section uses real airborne SAR data to further measure the perfor-
Sensors 2022, 22, 7747 mance of the three algorithms. 11 of 16

4.2. Verification of Real Airborne SAR Data

This section
4.2. Verification usesAirborne
of Real real airborne SAR data obtained in January 2017 in Crown Head
SAR Data
Ridge, Guangxi, China, to verify the
This section uses real airborne SAR performance of the
data obtained inproposed imaging
January 2017 system;
in Crown the
Head
parameters are listed in Table 6.
Ridge, Guangxi, China, to verify the performance of the proposed imaging system; the
parameters are listed in Table 6.
Table 6. System parameters for airborne SAR.
Table 6. System parameters for airborne SAR.
Parameters Value
Carrier frequency (GHz)
Parameters 35 Value
Range bandwidth (MHz)
Carrier frequency (GHz) 480 35
SamplingRange
frequency (MHz)
bandwidth (MHz) 500 480
Sampling frequency
Pulse width (µs) (MHz) 25 500
Pulse width (µs) 25
Slant range (km) 27.21
Slant range (km) 27.21
VelocityVelocity
(m/s) (m/s) 90.86 90.86
Doppler Doppler
bandwidth (MHz)(MHz)
bandwidth 600 600
DurationDuration
time (s)time (s) 15 15

The
The optical
opticalimage
imageofofthethe
observed scene
observed is shown
scene is shownin Figure 12. The
in Figure 12.SAR
Theimaging result
SAR imaging
processed by the by
result processed traditional BP algorithm
the traditional [27] is presented
BP algorithm in Figure
[27] is presented in 13a, where
Figure 13a, awhere
stronga
target
strongistarget
marked. Since the
is marked. airborne
Since SAR system
the airborne SAR works
systeminworks
strip-map mode, the
in strip-map original
mode, the
SAR dataSAR
original are data
randomly downsampled
are randomly by a factor
downsampled by of six to of
a factor construct the received
six to construct echo of
the received
one
echosubswath in the proposed
of one subswath system. The
in the proposed imaging
system. Theresult of the
imaging traditional
result BP algorithm
of the traditional BP
with downsampled SAR data is presented in Figure 13b, which is seriously
algorithm with downsampled SAR data is presented in Figure 13b, which is seriously pol- polluted by the
ambiguity
luted by theenergy.
ambiguity energy.

Figure 12. Panorama of the observed scene.

The result of the proposed algorithm, σ̂sum , is shown in Figure 14c, and the coarse
observation of Sat-1 and the result of the L1 algorithm are shown in Figure 14a,b for
comparison.
According to Figure 14, the coarse observation in Figure 14a has low azimuth resolu-
tion, making the targets stack together and reducing imaging interpretation performance.
The result of the L1 algorithm in Figure 14b improves the imaging resolution, but some
weak target information is lost, and there are a lot of reconstruction errors in the imag-
ing result. The proposed algorithm provides the best imaging performance, as shown
in Figure 14c. It not only achieves a resolution similar to the original image as shown in
Figure 13a, but also suppresses the targets’ side lobes. Then, the azimuth resolution, PSLR
(peak side lobe ratio), ISLR (integral side lobe ratio), and SSIM (structure similarity index
 Sensors 2022, 22, 7747 12 of 16

Sensors 2022, 22, x FOR PEER REVIEWmeasure) are used to quantitatively evaluate the imaging quality. The calculation method 12 of 16
of SSIM is represented in Equation (23), and all of the evaluation results are listed in Table 7.




2µ x µy + c1 2s xy + c2
Figure 12. Panorama SSIM
of the Im
observed
x , Im y scene.
=    (23)
µ 2 + µ2 + c s 2 + s2 + c
Sensors 2022, 22, x FOR PEER REVIEW x y 1 x y 2 12 of 16

where Imx and Imy are the evaluated image and the reference image, respectively. µ x and
µy denote the average value of Imx and Imy , respectively. s2y and s2x denote the variance of
Figure 12.Im
Imx and Panorama of the observed
y , respectively. scene.
s xy is the covariance of Imx and Imy .

(a)

(a)

(b)
Figure 13. SAR images of traditional BP algorithm. (a) Result of original SAR data. (b) Result of
downsampled SAR data.

The result of the proposed algorithm,(b) σ̂ sum , is shown in Figure 14c, and the coarse
observation
Figure ofimages
Figure 13. SAR
SAR Sat-1 and
images the result
ofoftraditional
traditional of
BPBP the L1 algorithm
algorithm.
algorithm. ofare
(a) Result
(a) Result shown
of SARindata.
original
original Figure
SAR 14a,b
data.
(b) for
(b) of
Result com-
Result of
parison.
downsampled SARdata.
downsampled SAR data.

The result of the proposed algorithm, σ̂ sum , is shown in Figure 14c, and the coarse
observation of Sat-1 and the result of the L1 algorithm are shown in Figure 14a,b for com-
parison.

(a)
Figure 14. Cont.

(a)
Sensors 20222022,
Sensors , 22, 22,
x FOR
7747PEER REVIEW 13 of 16 13 of 16

(b)

(c)
Figure14.
Figure Imaging results
14. Imaging resultsofofreal
realairborne SAR
airborne data.
SAR (a) Coarse
data. observation
(a) Coarse of Sat-1.
observation of (b) Reconstruc-
Sat-1. (b) Reconstruc-
tion result of the L1 algorithm. (c) Reconstruction result of the proposed
tion result the L1 algorithm. (c) Reconstruction result of the proposed algorithm.algorithm.

Table 7. Evaluation results.

According to Figure 14, the coarse observation in Figure 14a has low azimuth reso-
lution, making the targets stack together Coarseand reducing imaging interpretation
Proposed perfor-
Reference L1 Algorithm
mance. The result of the L1 algorithm Observation
in Figure 14b improves the imaging Algorithm
resolution, but
some Azimuth
weak target information
0.6
is lost, and
3.6
there are a lot0.6
of reconstruction
0.6
errors in the
resolution
imaging (m) The proposed algorithm provides the best imaging performance, as shown
result.
PSLR (dB) −12.27 −11.77 −10.47 −17.84
in Figure 14c. It not only
ISLR (dB) achieves a resolution
−7.37 −9.35 similar to the original image
−7.97 −17.13 as shown in
Figure SSIM
13a, but also suppresses
1 the targets’
0.89 side lobes. 0.87Then, the azimuth0.93 resolution,
PSLR (peak side lobe ratio), ISLR (integral side lobe ratio), and SSIM (structure similarity
indexThemeasure) are results
evaluation used toin quantitatively evaluate
Table 6 are consistent withthe
the imaging quality.
imaging results The calculation
in Figure 13.
method of SSIM isachieves
The L1 algorithm representedhigh in Equation
azimuth (23), and
resolution, allthe
but of reconstruction
the evaluationerrors
results are listed
lead
to the deterioration
in Table 7. of SSIM. The proposed algorithm achieves high azimuth resolution
and higher sidelube suppression performance. Compared with the reference result, the
proposed algorithm improves the PSLR and ISLR2by (
μ x μ5.58
+ cdB )(
sxy +9.76
2and )
c2 dB, respectively.
(
SSIM Im x ,Im y = 2 ) y 1
(23)
4.3. Analysis of the Subswath Number ( )(
μ x + μ y2 + c1 sx2 + sy2 + c2 )
whereAccording to the introduction above, Sat-2 randomly assigns the transmitted pulse
Im x and Im y are the evaluated image and the reference image, respectively. μx
to multiple subswaths. Therefore, the received echo of each subswath is undersampled,
and μ ygreater
and a denote the average
number value ofcorresponds
of subswaths Im x and to y , respectively.
Imfewer sy2 and
sampling points andsx2worse
denote the
reconstruction performance. A detailed analysis is listed in this part. The reconstruction
variance of Im x and Im y , respectively. sxy is the covariance of Im x and Im y .

Table 7. Evaluation results.

Coarse Proposed
Reference L1 Algorithm
Observation Algorithm
Azimuth resolution (m) 0.6 3.6 0.6 0.6
PSLR (dB) −12.27 −11.77 −10.47 −17.84
 posed algorithm improves the PSLR and ISLR by 5.58 dB and 9.76 dB, respectively.

4.3. Analysis of the Subswath Number

According to the introduction above, Sat-2 randomly assigns the transmitted pulse
Sensors 2022, 22, 7747 to multiple subswaths. Therefore, the received echo of each subswath is undersampled, 14 of 16
and a greater number of subswaths corresponds to fewer sampling points and worse re-
construction performance. A detailed analysis is listed in this part. The reconstruction er-
ror is is
error regarded asasnoise.
regarded noise.The
Thesignal-to-reconstruction-noise
signal-to-reconstruction-noiseratio
ratioisisdefined
definedto
to analyze
analyze the
the
reconstruction performance, that
reconstruction performance, that is is
2

kσσ iknew
2
SNRSNR = = σ −i σˆnew
ME ME 2
2
(24)
(24)
kσi i− σ̂ii knew
new

where σσi and

where i
and σˆ i the
σ̂i are areideal
the ideal
resultresultand the and the estimated
estimated result of result
the ithofrange
the ith range
gate, gate, re-
respectively.
2 2
k · knew is the⋅ newisnorm
spectively. new
the new defined norm in Section
defined3.2.5. The kσ3.2.5.
in Section i − σ̂i k new in
The σ iEquation
− σˆ i new in(24) can be
Equation
expanded as
(24) can be expanded as
−1H −1
( )
D E  
kσi − σ̂i k2newσ= ( −1
) ( ) 22 HH H H
2
i
− σˆ ( σ i
i new
−= σ̂i σ
) ,i
−
Q σˆ i
(,σQi
−1
− σσ̂i
i
−
) σˆ ≤
i
≤
k σσ
i k
i new
new
−−σ σ A
i i
A AQAAQAH
A σ i Aσi (25)
(25)

Then,
Then, we can obtain
2
kσ k2 σ i new 1 1
= ME =i new2 2 ≥ ≥
SNR ME SNR (26)
(26)
( )
−1
σˆ i new 1 − σ iHA H AQA H H A−σ1i σ i
2
kσi − σ
σ̂ii k−new H H

1 − σ A AQA i k σ k2
Aσ /new i i new

Figure 15 shows the variation curves of SNRME on the subswath number, where the
Figure 15 shows the variation curves of SNR ME on the subswath number, where the
lowest SNRME and
lowest SNR the highest lower bound, calculated by Equation (26), among all range
ME and the highest lower bound, calculated by Equation (26), among all range
gates
gates are
are selected.
selected.

Figure 15.
Figure 15. Illustration SNR MEcurves
Illustrationofof SNR curvesfor
fordifferent
differentsubswath
subswath numbers,
numbers, where NN
where is the
is the number
number of
ME
of subswaths.
subswaths.
According to Figure 15, when the subswath number approaches 20, SNR ME is close to
According to Figure 15, when the subswath number approaches 20, SNR is close
the theoretical lower bound, and the imaging result may be completely submerged
ME
by the
to the theoretical
reconstruction lower bound, and the imaging result may be completely submerged by
error.
the reconstruction error.
5. Conclusions
In SAR observations, high resolution and wide swath contradict each other. This
paper proposes a novel SAR imaging system based on tandem SAR satellites, where Sat-1
works in the traditional scanning mode to obtain coarse observations of a wide swath,
and Sat-2 adopts a new scanning method to obtain the downsampled echo from the same
swath. Then, a minimum-energy-based algorithm is proposed to associate the two SARs’
data and achieve high resolution. Since the accurate phase synchronization between two
satellites is unnecessary, the system cost can be significantly reduced. The simulations and
processed results of real airborne SAR data have proven that the proposed imaging system
can expand the range swath of the strip-map mode by up to six times and achieve the
high-resolution imaging of observed scenes.
Sensors 2022, 22, 7747 15 of 16

Author Contributions: Conceptualization, L.S. and C.L.; methodology, L.S. and C.L.; software,
L.S.; validation, L.S.; formal analysis, L.S.; investigation, L.S.; resources, L.S.; data curation, L.S.;
writing—original draft preparation, L.S.; writing—review and editing, L.S. and C.L.; visualization,
L.S.; supervision, C.L.; project administration, C.L. All authors have read and agreed to the published
version of the manuscript.
Funding: This research received no external funding.
Conflicts of Interest: The authors declare no conflict of interest.

Appendix A
Given a vector space X over a subfield F of the complex numbers C, a norm on X is
a real-valued function k · k: X → R with the following properties, where | a| denotes the
usual absolute value of a scalar a:
(1) Triangle inequality: kx + yk ≤ kxk + kyk for all x, y ∈ X.
(2) Absolute homogeneity: k axk = | a| · kxk for all x ∈ X and all scalar a.
(3) Positive definiteness: for all x ∈ X, kxk ≥ 0 and kxk = 0 ⇔ x = 0 .
Then, the validation of the new norm k · knew defined in Equation (13) is derived as
follows. According to the definition, kxknew can be calculated by
rD E r
kxknew = x, Q−1 x = ∑ q− 1
nn · x n xn (A1)
n

where xn is the nth element of x, and qnn is the nth row and nth column element in Q. Since
x n xn > 0 and q− 1
nn > 0, the positive definiteness of kxknew is easily proved. Then, according
to Equation (A1), we have
r
kx + yknew = ∑ q− 1
nn · ( x n + yn )( xn + yn )
n
r
= ∑ q− 1 −1 −1 −1
nn · x n xn + ∑ qnn · x n yn + ∑ qnn · yn xn + ∑ qnn · yn yn
n n n n (A2)
s r r
≤ ∑ q−
nn
1
· x n xn + ∑ q−
nn
1
· yn yn + 2 ∑ q−
nn
1
· x n xn ∑ q−
nn
1
· yn yn
n n n n
= kxknew + kyknew

and r r
k axknew = ∑ q−nn1 · axn axn = |a| · ∑ q−nn1 · xn xn = |a| · kxknew (A3)
n n

So k · knew satisfies property 1 and 2. Then, the validation of k · knew is proven.

References
1. An, D.; Huang, X.; Jin, T. Extended Nonlinear Chirp Scaling Algorithm for High-Resolution Highly Squint SAR Data Focusing.
IEEE Trans. Geosci. Remote Sens. 2012, 50, 3595–3609. [CrossRef]
2. Xu, W.; Deng, Y.; Huang, P. Full-Aperture SAR Data Focusing in the Spaceborne Squinted Sliding-Spotlight Mode. IEEE Trans.
Geosci. Remote Sens. 2014, 52, 4596–4607. [CrossRef]
3. Sun, G.; Wu, Y.; Yang, J. Full-Aperture Focusing of Very High Resolution Spaceborne-Squinted Sliding Spotlight SAR Data. IEEE
Trans. Geosci. Remote Sens. 2017, 55, 3309–3321. [CrossRef]
4. Krieger, G.; Gebert, N.; Moreira, A. Multidimensional waveform encoding: A new digital beamforming technique for synthetic
aperture radar remote sensing. IEEE Trans. Geosci. Remote Sens. 2008, 46, 31–46. [CrossRef]
5. Hu, C.; Li, Y.; Dong, X. Performance Analysis of L-Band Geosynchronous SAR Imaging in the Presence of Ionospheric Scintillation.
IEEE Trans. Geosci. Remote Sens. 2017, 55, 159–172. [CrossRef]
6. Dong, X.; Quegan, S.; Yumiko, U.; Hu, C.; Zeng, T. Feasibility Study of C- and L-band SAR Time Series Data in Tracking
Indonesian Plantation and Natural Forest Cover Changes. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2015, 8, 3692–3699.
[CrossRef]
Sensors 2022, 22, 7747 16 of 16

7. Freeman, A.; Johnson, W.T.K.; Huneycutt, B. The “Myth” of the minimum SAR antenna area constraint. IEEE Trans. Geosci. Remote
Sens. 2000, 38, 320–324. [CrossRef]
8. Zhang, Y.; Wang, W.; Deng, Y.; Wang, R. Signal Reconstruction Algorithm for Azimuth Multichannel SAR System Based on a
Multiobjective Optimization Model. IEEE Trans. Geosci. Remote Sens. 2020, 58, 3881–3893. [CrossRef]
9. Chen, Z.; Zhang, Z.; Zhou, Y. Elevated Frequency Diversity Array: A Novel Approach to High Resolution and Wide Swath
Imaging for Synthetic Aperture Radar. IEEE Geosci. Remote Sens. Lett. 2022, 19, 1–5. [CrossRef]
10. Krieger, G.; Huber, S.; Villano, M. SIMO and MIMO System Architectures and Modes for High-Resolution Ultra-Wide-Swath SAR
Imaging. In Proceedings of the EUSAR: European Conference on Synthetic Aperture Radar, Hamburg, Germany, 6–9 June 2016.
11. Liu, B.; He, Y. Improved DBF Algorithm for Multichannel High-Resolution Wide-Swath SAR. IEEE Trans. Geosci. Remote Sens.
2016, 54, 1209–1225. [CrossRef]
12. Sun, L.; Yu, Z.; Li, C. A Minimum-Energy-Based Algorithm for Multichannel Reconstruction. IEEE Geosci. Remote Sens. Lett. 2022,
19, 1–5. [CrossRef]
13. Xia, Z.; Zhao, Z.; Zhang, T. Wide-Swath SAR Based on Networking of Multiple Small Satellites for Maritime Applications.
In Proceedings of the IET International Radar Conference, Chongqing, China, 4–6 November 2020.
14. Krieger, G.; Moreira, A. Spaceborne bi- and multistatic SAR: Potential and challenges. IEE Proc. Radar Sonar Navig. 2006, 153,
184–198. [CrossRef]
15. Zhang, Y.; Zhang, H.; Hou, S. An Innovative Superpolyhedron (SP) Formation for Multistatic SAR (M-SAR) Interferometry. IEEE
Trans. Geosci. Remote Sens. 2021, 59, 10136–10150. [CrossRef]
16. Krieger, G.; Zonno, M.; Mittermayer, J. MirrorSAR: A Fractionated Space Transponder Concept for the Implementation of
Low-Cost Multistatic SAR Missions. In Proceedings of the Eusar: European Conference on Synthetic Aperture Radar, Aachen,
Germany, 4–7 June 2018.
17. Chen, Y.; Zhao, Y.; Li, G. High-Resolution and Wide-Swath Monostatic SAR Imaging via Random Beam Scanning. In Proceedings
of the 2020 IEEE 11th Sensor Array and Multichannel Signal Processing Workshop, Hangzhou, China, 8–11 June 2020.
18. Gebert, N.; Krieger, G.; Moreira, A. Multichannel Azimuth Processing in ScanSAR and TOPS Mode Operation. IEEE Trans. Geosci.
Remote Sens. 2010, 48, 2994–3008. [CrossRef]
19. Huang, L.; Qiu, X.; Hu, D. Medium-Earth-Orbit SAR Focusing Using Range Doppler Algorithm With Integrated Two-Step
Azimuth Perturbation. IEEE Geosci. Remote Sens. Lett. 2015, 12, 626–630. [CrossRef]
20. Cumming, I.G.; Wong, F.H. Digital Processing of Synthetic Aperture Radar Data: Algorithms and Implementation; Artech House:
Boston, MA, USA, 2005.
21. Lao, D.; Zhu, B.; Yu, S. An Improved SAR Imaging Algorithm Based on a Two-Dimension-Separated Algorithm. In Proceedings
of the 2018 China International SAR Symposium (CISS), Shanghai, China, 10–12 October 2018.
22. Li, D.; Lin, H.; Liu, H. Focus Improvement for High-Resolution Highly Squinted SAR Imaging Based on 2-D Spatial-Variant
Linear and Quadratic RCMs Correction and Azimuth-Dependent Doppler Equalization. IEEE J. Sel. Top. Appl. Earth Obs. Remote
Sens. 2017, 10, 168–183. [CrossRef]
23. Papoulis, A. A new algorithm in spectral analysis and band-limited extrapolation. IEEE Trans. Circuits Syst. 1975, 22, 735–742.
[CrossRef]
24. Parker, J.A.; Kenyon, R.V.; Troxel, D.E. Comparison of Interpolating Methods for Image Resampling. IEEE Trans. Med. Imaging
1983, 2, 31–39. [CrossRef] [PubMed]
25. Luenberger, D.G. Optimization by Vector Space Methods; Wiley: New York, NY, USA, 1969.
26. Potter, L.C.; Arun, K.S. Energy concentration in band-limited extrapolation. IEEE Trans. Acoust. Speech Signal Process. 1989, 37,
1027–1041. [CrossRef]
27. Sahiner, B.; Yagle, A.E. A fast algorithm for backprojection with linear interpolation. IEEE Trans. Image Process. 1993, 2, 547–550.
[CrossRef]