sensors

Article
Observability Study on Passive Target Localization by
Conic–Angle Measurements
Taeil Suh and Woochan Kim *

Naval Combat Systems PMO, Maritime Technology Research Institute, Agency Defense Development,
Changwon-si 51698, Korea; [email protected]
* Correspondence: [email protected]

Abstract: Information from a passive linear array sensor is related to the conic angle formed by a
target and the sensor in three-dimensional (3D) space so that the target localization system using
the sensor should be also designed in 3D space. This paper presents an observability study of a
passive target localization system created using conic angle information. The study includes the
analysis of the sensor maneuver requirement needed to achieve system observability and simulations
to demonstrate the results of the analytic scheme. The proposed sensor maneuver requirements
satisfy the system observability conditions by using the local linearization approach of the Fisher
information matrix. It is also shown that this requirement can be mitigated for special cases in which
the depth difference between the sensor and the target is given. Using the simulation, it is shown
that sensors following the proposed scheme are able to obtain meaningful information that can be
used to estimate 3D target states.

Keywords: observability analysis; target tracking; passive target localization; passive linear array sensor







Citation: Suh, T.; Kim, W. 1. Introduction

Observability Study on Passive Target As passive sensors can detect a target without transmitting signals, they are widely
Localization by Conic–Angle
used in surveillance areas where the sensor needs to be concealed. In general, observa-
Measurements. Sensors 2021, 21, 6439.
tions from these sensors demonstrate no distance information (measurement), and the
https://doi.org/10.3390/s21196439
relative target distance from the sensor needs to be estimated using the observed data
for localization.
Academic Editor: Joanna Kolodziej
The target state estimation technique using these sensors is commonly referred to
as target motion analysis (TMA). For two-dimensional (2D) TMA, bearings-only TMA
Received: 28 July 2021
Accepted: 22 September 2021
(BoTMA) is a representative type of the 2D TMA, which assumes that the target and the
Published: 27 September 2021
sensor exist on the same plane used for estimating target states using in-plane azimuth
information. For conventional three-dimensional (3D) TMA systems, bearing and elevation
Publisher’s Note: MDPI stays neutral
information are used to estimate the 3D target position as well as the velocity.
with regard to jurisdictional claims in
In this paper, a passive linear array sensor is considered. From now on, “passive
published maps and institutional affil- linear array sensor” will be referred to as “sensor” for short. Information from the sensor is
iations. related to a conic angle, which is formed by the relative states of the target and the sensor in
3D space due to the beam characteristics of the sensor. The information is neither azimuth
nor elevation, so it is not applicable to conventional 2D/3D TMA algorithms. In addition,
the target localization system using the conic angle information should be designed in 3D
Copyright: © 2021 by the authors.
space. We define this system as conic angles-only TMA (CoTMA) and analyze its system
Licensee MDPI, Basel, Switzerland.
observability in this paper.
This article is an open access article
Most previous TMA-related studies have focused on proposing estimation algorithms
distributed under the terms and to overcome system nonlinearity or on analyzing system observability to prove the possi-
conditions of the Creative Commons bility of estimating target states using the given information.
Attribution (CC BY) license (https:// The following estimation algorithms for conventional TMA were previously proposed:
creativecommons.org/licenses/by/ Maximum likelihood (ML) approaches estimate the nonrandom parameters of the
4.0/). target states that maximize the measurement likelihood with respect to a set of measure-

Sensors 2021, 21, 6439. https://doi.org/10.3390/s21196439 https://www.mdpi.com/journal/sensors

Sensors 2021, 21, 6439 2 of 18

ments [1]. The ML for nonlinear measurements was proposed in [2], and the ML for
pseudo-linearized measurements was proposed in [3].
For random parameters, Bayesian approach estimators are used. Minimum mean
square error (MMSE) is one of the Bayesian estimation methods, and the Kalman Filter
is a representative MMSE algorithm. The extended Kalman filter (EKF) uses a Taylor
extension and performs forced linearization to maintain the Gaussian probability density
function (pdf) of the target estimates [1]. The unscented Kalman filter and the cubature
Kalman filter use set of sigma points that can represent the higher order terms of the Taylor
extension to reduce the linearization error [4,5]. Pseudo-measurement filter and modified
gain EKF utilize pseudo-linearization using modified measurement equations to linearize
the system [6,7]. The range-parameterized EKF improve the problem of the pdf estimates
converging to the local MMSE by constructing a bank of EKFs with different distances [8].
An approximation method of the nonlinear measurement likelihood using a mixture of
linear Gaussian components was proposed in [9].
Particle filters are sub-optimal estimators for nonlinear systems [10]. They consist
of a set of particles to represent the nonlinear and nonGaussian pdf without approxi-
mated linearization. The auxiliary particle filter, the regularized particle filter, and the
local linearized particle filter methods are variants of the particle filters and were given
in [11–13], respectively.
Data association algorithms for bearings-only tracking in cluttered environments were
proposed in [14–17].
System observability analysis, which determines whether the parameters being esti-
mated can be inferred from a given observation, should precede the adoption of a TMA
algorithm. Until system observability is achieved, no TMA algorithm can guarantee the
convergence of its estimates.
In [18], the BoTMA observability for nonmaneuvering targets was analyzed. Addition-
ally, its extended study of 3D TMA was derived from [19]. The observability conditions for
target maneuvers with dynamic models of the Nth-order were given in [20]. A linearized
structure of the pseudo-measurements was used to analyze the observability of passive
homing guidance applications [21]. In [22], an observability study of a range-only sensor
was produced. A comparison of the observability of the range-only and bearings-only
approaches was shown in [23,24]. Trajectory optimization methods associated with ob-
servability were proposed in [25–27], and work regarding matrix optimization for sensor
networks can be seen in [28–30]. At present, no observability analysis has been conducted
for the CoTMA system.
In this paper, our CoTMA system observability study includes:
• Designing the system model;
• Deriving the observability requirements needed to achieve an observable system;
• Proposing the sensor maneuver strategy to satisfy those requirements;
• Analyzing a special CoTMA case that can mitigate the requirements;
• Demonstrating the results of the analytic scheme using simulations.
As previously mentioned, a CoTMA system is designed to estimate the position and
velocity of a target in 3D space. System modeling includes the design of a target dynamic
equation and the design of a conic angle measurement equation for the sensor.
In this paper, the system observability of the CoTMA is derived using the approach
from [22]. This approach uses a Fisher information matrix (FIM) as an observability index
and analyzes the nonsigularity condition of the FIM. This condition relates to the relative
state of the sensor and the target, and a sensor maneuver strategy is proposed to satisfy the
system observability requirements.
Furthermore, a special CoTMA case is also considered. In the anti-ship submarine
warfare, the relative depth of the submarine (operator’s own ship) from the surface com-
batant (target) is equal to the diving depth of the submarine, which is determined by the
navigation instrumentation of the ship. This information provides a significant advantage
 Furthermore, a special CoTMA case is also considered. In the anti-ship submarine
warfare, the relative depth of the submarine (operator’s own ship) from the surface com-
batant (target) is equal to the diving depth of the submarine, which is determined by the
Sensors 2021, 21, 6439 navigation instrumentation of the ship. This information provides a significant advantage 3 of 18

in establishing system observability, and this special CoTMA case is defined as conic angle
with depth-supported TMA (CDTMA) in this paper.
The observability
in establishing studies for theand
system observability, CoTMA and the
this special CDTMA
CoTMA areisdemonstrated
case through
defined as conic angle
numerical experiments.
with depth-supported TMA (CDTMA) in this paper.
The
The rest of this paper
observability is organized
studies for the CoTMA as follows: and the Section
CDTMA 2 presents the CoTMAthrough
are demonstrated system
modeling. As the motivation
numerical experiments. of this paper, the influence of the vertical incidence angle for
the passive
The restlinear
of thisarray sensor
paper is shown
is organized as in Section
follows: 3. The2 observability
Section analysissystem
presents the CoTMA of the
CoTMA system is presented in Section 4. Section 5 presents
modeling. As the motivation of this paper, the influence of the vertical incidence anglethe system observability anal-
ysis
for the passive linear array sensor is shown in Section 3. The observability analysisre-
for the CDTMA. The numerical experiments are shown in Section 6. Concluding of
marks are given
the CoTMA in Section
system 7.
is presented in Section 4. Section 5 presents the system observability
analysis for the CDTMA. The numerical experiments are shown in Section 6. Concluding
2. Systemare
remarks Modeling
given in Section 7.
Figure 1 shows the geometric relationship between the target and the sensor. The
2. System
state Modeling
vectors of the target and the sensor consist of position and velocity in 3D space, and
can beFigure 1 shows
calculated as the geometric relationship between the target and the sensor. The state

[ ]
vectors of the target and the sensor consist of position and velocity in 3D space, and can be
x τk , i ≡ (p τk , i ) (v τk , i ) , τ ∈ t,s ,
T T T
calculated as (1)
  T  T T
xτ,i
k ≡ pτ,i
k vτ,i
T k
, τT∈ t, s, (1)
 x τk ,i   x τk ,i 
τ ,i =  yτ ,Ti 
p
 τ. ,i  
, v τk ,i =  y x τ,i , T
k
xkτ,i τk ,i  k 
k (2)
   z k. τ,i 
τ , i
pk =  yk  zk ,
τ,i
vτ,i  
 τ,i
k =  yk  , (2)
τ,i . τ,i
zk zk
where p, v, and k denote the position vector, the velocity vector, and the time index,
respectively. τ iskthe
where p, v, and denote the
object positionthat
indicator τ =t the
vector, velocity
is the targetvector, τ =stheis time
of, andand index,
the sensor.
respectively.
The τ isofthe
superscript theobject
vectors i means
indicator that
that = t vectors
τ the is the target of, and τ =
are represented insan is inertial
the sensor.
co-
The superscript
ordinate system. of the vectors i means that the vectors are represented in an inertial
coordinate system.

Figure 1.
Figure Relative geometry
1. Relative geometry of
of the
the target
target sensor
sensor in
in 3D
3D space.
space.

2.1. Target
2.1. Target Dynamics
Dynamics
Given the target states at k − 1 (xt,i
kt,i−1 ), a target state at k (x
t,i
t,ki ) is propagated by a linear
Given the target states at k − 1 ( x k−1 ), a target state at k ( xk ) is propagated by a linear
dynamics as
dynamics as xt,i t,i
k = F( tk −1,k )xk −1 , (3)
x k = Ffrom
where t`,m denotes a transition time interval
t ,i
tk −1,k` xtok −m, ( ) t ,i
(3)
1 , and F is the state transition matrix.
Here, if we assume that the target follows constant velocity motion, then the F is
 
I3 tk−1,k I3
F(tk−1,k ) = , (4)
O3 I3

where In and On denote the n × n identity matrix and the n × n zero matrix, respectively.
Sensors 2021, 21, 6439 4 of 18

2.2. Sensor
As shown in Figure 1, the conic angle is determined by the relative position vector
of the target sensor, which is represented in the (sensor-centered) sensor body coordinate
system. Coordinate transformation is required to formulate the conic angle in the body
coordinates with the relative position in the inertial coordinates. One can define a rotation
matrix from the inertial coordinates to the body coordinates at k by

Cik→b = Crx Cry Crz (5)

where  
1 0 0
Crx =  0 cos φk sin φk , (6)
0 − sin φk cos φk
 
cos ϕk 0 − sin ϕk
Cry =  0 1 0 , (7)
sin ϕk 0 cos ϕk
 
cos ψk sin ψk 0
Crz =  − sin ψk cos ψk 0 , (8)
0 0 1
and where φk , ϕk , and ψk are the Euler angles, which indicate the direction of the sensor
with respect to the inertial coordinates. Using Equation (5), the relative target position pk
of the body coordinates is calculated as
   T
pk = Cik→b H xt,i
k − xk
s,i
≡ xk yk zk , (9)

where xs,i
k is the sensor state at time k and is assumed to be known. H is a linear projection
matrix, which can be represented as
 
H = I 3 O3 . (10)

Note that the origin of the body coordinates is located at the sensor position ps,i s,i
k = Hxk
in the inertial coordinates.
The measurement equation consists of a nonlinear function of the relative target
position pk and the measurement error νk . In real-ocean environments, the measurement
error is difficult to model because it is influenced by complex environmental factors as well
as the relative target position. The general nonlinear function of the measurement is

θk = g(pk , νk ), (11)

In this paper, the following assumptions were added to the sensor model to focus on
the analytic study of the observability requirements for the CoTMA system according to
the target–sensor geometric relationship.
• The target can be detected in all directions by the sensor (i.e., the censor has no
blind spots);
• The measurement noise νk is an additive white gaussian noise (WGN) sequence with
a mean of zero and covariance of Rk = σ2 in all directions. The unit σ is a radian;
• The received target signals propagate directly (i.e., the acoustic ray is not bent);
• The measurement accuracy cannot be affected by the maneuvering of the sensor (or
its towing vehicle).
Although these assumptions are not suitable for calculating sensor trajectory opti-
mization in the real world, that is outside the scope of this paper and will be considered in
future work.
Sensors 2021, 21, 6439 5 of 18

Finally, Equation (11) becomes

θk = h(pk ) + νk , (12)


where h pk is the nonlinear function of pk as
xk
h(pk ) = tan−1 q , (13)
y2k + z2k

and the measurement history up until time k is defined as

θk = {θi }ik=0 = {θ0 , θ1 , . . . , θk }. (14)

3. Conventional BoTMA Using the Conic Angle Measurements

In most previous studies of the BoTMA, the state x was represented in the 2D plane
rather than in the 3D space. If the zk of Equation (13) is set to 0 for all k, then the CoTMA
system from Section 2 can be represented in the 2D plane without approximation and is
identical to the 2D BoTMA system. However, if zk is not equal to 0 or is not consistent, it is
clear that the CoTMA is different from the 2D BoTMA. This means that a modelling error
may occur when the BoTMA system is utilized using the conic angle as measurements.
In this section, the effect of the mismatch between the azimuth and the conic angle is
analyzed. An approximation error is defined, and the numerical results of the error are
analyzed in terms of a single scan and multiple scans.

3.1. Error Analysis for Single Scan

In conventional 2D BoTMA systems, the azimuth in the inertial coordinate system is
defined as
t,i s,i
−1 x k − x k
 
hi pt,i
k , p s,i
k = tan . (15)
yt,i
k − y s,i
k
Additionally, its representation of the body coordinates can be obtained as
 
h2D (pk ) = hi pt,i
k , p s,i
k − ψk (16)

with consideration that φk = ϕk = 0.

If the conic angle is assumed to be the azimuth to be applied to the BoTMA system in
the 2D plane, the approximation error denoted by ε at time k is Equation (16) subtracted
from Equation (13) as
ε(pk ) ≡ h2D (pk ) − h(pk ). (17)
The approximation error is a function of the relative position in 3D space. In order to
visualize the error tendency with respect to the relative position, the parameters α = yk /xk
and β = zk /xk are defined and can be substituted into Equation (17) as

1 1
ε(α, β) = tan−1 − tan−1 p . (18)
α α2 + β2

Then, the tendency of the approximation error in the αβ domain is represented

in Figure 2.
When α is constant, the |ε| is proportional to | β|, which is related to the vertical
incidence angle. This means that the error increases as the xk decreases or as the zk
increases for the same azimuth.
The |ε| is inversely proportional to the |α| if β is a nonzero constant. The error decreases
if the azimuth of the target becomes closer to the starboard/port direction of the sensor,
which is one of the sensor characteristics.
 1 1
ε (α , β ) = tan −1 − tan −1 . (18)
α α +β2
2

Then, the tendency of the approximation error in the αβ domain is represented in

Sensors 2021, 21, 6439 Figure 2. 6 of 18

Figure
Figure2.2.Approximation errorε ε(α(α,
Approximationerror , β )β)(Log-scale).
(Log-scale).

3.2. Error Analysis for Multiple Scans

When α is constant, the ε is proportional to β , which is related to the vertical
The BoTMA, which uses the conic angle as the azimuth, has two types of the errors.
incidence angle. This means that the error increases as the xk decreases or as the zk in-
One is the
measurement noise νk from Equation (12), and the other is the approximation
creases for the same azimuth.
error ε pk from Equation (17) for every k.
The ε is inversely
In Section proportionalnoise
2, the measurement to the α if β to
is assumed is be
a nonzero
the WGN constant.
sequences.The Unlike
error
decreases if the azimuth
the measurement noiseofsequence,
the target however,
becomes closer to the starboard/port
the approximation error isdirection
not freeofofthe
the
sensor, which because
coloredness is one of the sensor
currentcharacteristics.
target/sensor states depend on their previous states.
ors 2021, 21, x FOR PEER REVIEW Furthermore, the average of the approximation error sequence may not 7beofzero 20 in general.
3.2. Error Analysis for Multiple Scans
Since most of the estimators assume that the measurement noise is the additive WGN
sequences,
The BoTMA, such approximated
which uses themeasurements
conic angle as that do not match
the azimuth, up with
has two typestheof assumption
the errors.
can degrade
the the estimation
measurement noise accuracy
from (or tracking
Equation performance).
The effect of the approximation errorν ksequence on tracking performance difficult
One is (12), and the other is the approximation
ε (p k ) from
to interpreterror The effect
analytically. of the approximation
Therefore,
Equation it(17)
wasfor
replaced
every error
k . sequence
by numericalon tracking
analysis byperformance
means of a is difficult
simulation. to interpret analytically. Therefore, it was replaced by numerical analysis by means of
In Section 2, the measurement noise is assumed to be the WGN sequences. Unlike the
a simulation.
In themeasurement
simulation scenario, the targethowever,
and the sensor exist in 3D space,
noise sequence, the approximation error isand
nottheir tra-
free of the colored-
jectories asness In thethe
seenbecause
from simulation
horizontal scenario,
plane are the target
shown in and the3a.
Figure sensor
The exist in
initial 3D space,
distance and their
of Further-
the current target/sensor states depend on their previous states.
trajectories
the target from as seen from the horizontal
planeplane are The
shown in Figure 3a. Thetoinitial
the distance of
more,the thesensor
average onofthethehorizontal
approximation is 4 km.
error sequence target
may isnotsubjected
be zero in general.
the
constant velocity target
motion from the sensor on the horizontal plane is 4 km. The target is subjected to the
Since mostwithout the depth assume
of the estimators changing. that the measurement noise is the additive WGN
constant velocity motion without the depth changing.
sequences, such approximated measurements that do not match up with the assumption
can degrade the estimation accuracy (or tracking performance).

FigureFigure
3. Target and sensor
3. Target geometry
and sensor (a) in horizontal
geometry plane plane
(a) in horizontal and (b) in (b)
and 3D in
space.
3D space.

The sensor also maintains its speed in horizontal space, but for the present scenario,
the trajectory consists of two legs. In the first leg, the target is located on the port section
of the sensor. However, the target is located at the head section of the sensor after the
sensor maneuver. It moves to the tail section of the sensor at the end of the scenario.
 Sensors 2021, 21, 6439 7 of 18

The sensor also maintains its speed in horizontal space, but for the present scenario,
the trajectory consists of two legs. In the first leg, the target is located on the port section of
the sensor. However, the target is located at the head section of the sensor after the sensor
maneuver. It moves to the tail section of the sensor at the end of the scenario.
For the horizontal trajectory scenario, various 3D trajectory scenarios were generated
by increasing the value of zk , the vertical difference between the target and the sensor, by
10 m from 0 m to 1400 m. In each scenario, zk is a constant for every k. The trajectory
Sensors 2021, 21, x FOR PEER REVIEW scenarios for zk = 0 m and zk = 1400 m are depicted on Figure 3b as examples. 8 of 20
The differences between the conic angle and the azimuth histories in zk = 400 m are
shown in Figure 4. The target is located in the head/tail direction of the sensor at the
beginning and end of the second leg, respectively, so the difference between the conic angle
and the azimuth is conspicuous in the second leg.

Figure 4. Difference between the conic angle and the azimuth (bearing).
Figure 4. Difference between the conic angle and the azimuth (bearing).
The sensor sampling time 20 s, and the measurement noise deviation of the conic
angle
Themeasurements
root mean square equals
(RMS)
π/180 (=1for
error deg).
the z k = d scenario can be obtained as
The ML estimator designed for 2D BoTMA is used, and it estimates the initial position
and velocity of the target in the horizontal plane. 1 100 The2 estimator considers the measurement
to be the azimuth, but the measurement
ed =
is
100
noise-corrupted
the
ed , j ,
conic angle. (20)
j =1
As the approximation errors increases, the estimation error will also increase. To obtain
where A estimation
statistical denotes theerrors, a Monte
Euclidean norm of A
Carlo consisting
. of 100 trials is used for each scenario.
j–th the z k = d
Then, the performance degradation index for the z k = d error
For the trial in scenario, the estimation scenarioof the
canhorizontal
be definedplane
as
can be defined as
the ratio of the RMS error compared to the RMS "
(d,j) error for
# the z = 0 scenario, i.e.,
x̂0 − x0t,i k
ed,j = (d,j) , (19)
ŷ e − yt,i
g d =0 d , 0 (21)
(d,j) (d,j) e0
where x̂0 and ŷ0 are the estimated elements of the initial target position.
If there
The rootis no approximation
mean square (RMS) error forfor
error thethescenario zk = 0 ), then
zk = d(scenario can beg dobtained
is equalas to 1.
The tendency of g d is depicted in Figure v 5a. For the z k < 500 interval, g d increases
u 100
up to 5.2 in proportion to z k . In e500
d = tz k 1< 800
≤u
∑ k, ed,j g dk2decreases
, as z k increases. Alt- (20)
hough, they are not included in g , errors arej= 100 still
1 increasing in the bearing and course
d

for this interval. As the influence of the approximation error increases, g d increases ex-
where kAk denotes the Euclidean norm of A.
ponentially from z k ≥ 800 .
Using Equation (17), the approximation error statistics for each scenario is shown in
Figure 5b. It shows that the approximation error is a nonzero mean, and its standard de-
viation increases as z k increases. Since the statistics for the approximation errors are af-
fected by the relative geometry between the target and the sensor, it is difficult to com-
pensate for the approximation error in advance to utilize the 2D BoTMA algorithms. It is
clear that the CoTMA system should be designed in 3D space in order to avoid such ap-
Sensors 2021, 21, 6439 8 of 18

Then, the performance degradation index for the zk = d scenario can be defined as
the ratio of the RMS error compared to the RMS error for the zk = 0 scenario, i.e.,
ed
gd = , (21)
e0

If there is no approximation error for the scenario (zk = 0), then gd is equal to 1.
The tendency of gd is depicted in Figure 5a. For the zk < 500 interval, gd increases
up to 5.2 in proportion to zk . In 500 ≤ zk < 800, gd decreases as zk increases. Although,
they are not included in gd , errors are still increasing in the bearing and course for9 this
Sensors 2021, 21, x FOR PEER REVIEW of 20
interval. As the influence of the approximation error increases, gd increases exponentially
from zk ≥ 800.

Figure5.
Figure 5. (a)
(a) Performance
Performance degradation
degradation ratio
ratioand
and(b)
(b)the
theapproximation
approximationerror
errorstatistics.
statistics.

Using Equation
4. Observability (17), the
Analysis for approximation
the CoTMA Systems error statistics for each scenario is shown
in Figure 5b. It shows that the approximation
The CoTMA system has not only the distance errorambiguity
is a nonzero of amean, andsensor
passive its standard
but the
deviation increases
conic angle ambiguity ofas z increases. Since the statistics for the approximation
k a linear array sensor. The conic angle ambiguity is also errors are
called
affected byambiguity”
“left/right the relativeingeometry between
the horizontal the target and the sensor, it is difficult to com-
plane.
pensate for the approximation error in advance
The system observability includes the possibility to utilize the 2D BoTMA
of resolving algorithms.
these ambiguities It
from
is clear that the CoTMA system should be designed in 3D space in order
the given measurements. Therefore, in order to gather meaningful information, an analy- to avoid such
approximation errors.
sis of the system observability should be performed.
This section Analysis
4. Observability analyzes the
for observability of the CoTMA system. The necessary and suf-
the CoTMA Systems
ficient conditions to achieve the system observability are derived, and a sensor maneuver
The CoTMA system has not only the distance ambiguity of a passive sensor but the
strategy is proposed to satisfy the observability requirement.
conic angle ambiguity of a linear array sensor. The conic angle ambiguity is also called
“left/right ambiguity” in the horizontal plane.
4.1. Fisher Information Matrix (FIM)
The system observability includes the possibility of resolving these ambiguities from
The fisher
the given information
measurements. matrix (FIM)
Therefore, indicates
in order themeaningful
to gather quantity ofinformation,
informationan that an ob-
analysis
servable random variable carries about
of the system observability should be performed. an unknown parameter. Here, the “observable
random variable” is the set of conic angle measurements k , and the “unknown parame-
ter” is the relative target trajectory in the body coordinates, which denotes x k ≡ {x i }ik= 0 .
The relative target trajectory can be obtained by using the initial relative state vector
in the body coordinates and the given sensor trajectory in the inertial coordinates. There-
Sensors 2021, 21, 6439 9 of 18

This section analyzes the observability of the CoTMA system. The necessary and
sufficient conditions to achieve the system observability are derived, and a sensor maneuver
strategy is proposed to satisfy the observability requirement.

4.1. Fisher Information Matrix (FIM)

The fisher information matrix (FIM) indicates the quantity of information that an
observable random variable carries about an unknown parameter. Here, the “observable
random variable” is the set of conic angle measurements k, and the “unknown parameter”
 k
is the relative target trajectory in the body coordinates, which denotes xk ≡ xi i=0 .
The relative target trajectory can be obtained by using the initial relative state vector in
the body coordinates and the given sensor trajectory in the inertial coordinates. Therefore,
the “unknown parameter” xk can be summarized as x0 .
Then, the FIM J(tk , t0 ) can be defined as
" #
∂2 
k

J ( t k , t0 ) = − E ln f θ x0 x0 , (22)
∂x20
 
where E[ (·) |x0 ] denotes the conditional expectation of (·) for the given x0 , and f θk x0
is the conditional pdf of θk for the given x0 .  
Due to statistical independence of the WGN, the conditional pdf f θk x0 can be
expressed as
  k
f θ k x0 = ∏ f θ j x0 ,


(23)
j =0
   
= N h p j , σ2 , and N m, σ2 denotes the Gaussian pdf with the mean m



where f θ j x0
and covariance σ2 .  
In Equation (22), a partial differentiation of the log likelihood pdf of f θk x0 exists
with respect to x0 . It is difficult to analytically calculate the partial differentiation using the
nonlinear pdf f θk x0 .
There are various approaches that can be used for the observability analysis of nonlin-
ear systems. Examples of such approaches are the pseudo-linearization method [19,21] and
the local-linearization method [22]. Here, the authors use the Jacobian local linearization
method, which was utilized in [22].
As such, Equation (22) becomes

k ~T~
J ( t k , t 0 ) = σ −2 ∑ H j H j , (24)
j =0

where ~  
∂
Hj = h x
∂x0  j            (25)
∂h x j ∂h x j ∂h x j ∂h x j ∂h x j ∂h x j
= . . . ,
∂x0 ∂y0 ∂z0 ∂ x0 ∂ y0 ∂ z0
   
and where h x j = h p j due to p j ⊂ x j .

4.2. Observability Analysis with Local Linearization

For the system observability at t = t0 from the given information θk , the FIM J(tk , t0 )
should be nonsingular for t0 ∼ tk [21]. The nonsingularity of J(tk , t0 ) is the same as the
~
columns H j having some independence for some t j in [t0 , tk ].
Sensors 2021, 21, 6439 10 of 18

Then, the observability requirement is

~
H j µ 6= 0 for some t0 ≤ t j ≤ tk , (26)

where µ is a 6 × 1 the relative target state vector, which is represented on the body coordi-
nates at t = t0 as
 T
µ = µ px µ py µ pz µvx µvy µvz , (27)
where the elements of µ are nonzero arbitrary constants.
~
The Jacobian matrix H j can be divided into two partial differential matrices by the
chain rule as  
~ ∂h x j ∂x j
Hj = . (28)
∂x j ∂x0
 
∂h x j
The elements of ∂x j are calculated from Equation (29) through Equation (32)

  q
∂h x j y2j + z2j
= (29)
∂x j x2j + y2j + z2j
 
∂h x j −xj yj
= q   (30)
∂y j y2j + z2j x2j + y2j + z2j
 
∂h x j −xj zj
= q   (31)
∂z j y2j + z2j x2j + y2j + z2j
     
∂h x j ∂h x j ∂h x j
. = . = . = 0. (32)
∂x j ∂y j ∂z j
 
∂h x j
The ∂x j from Equation (28) is the relative state transition matrix from t0 to t j . When
∂x j
the sensor moves at a constant velocity, the ∂x0 becomes

∂x j  

I3 t0,j I3
= F t0,j = . (33)
∂x0 O3 I3
~
Using Equations (29)–(33), the H j µ can be expanded as follows.

t0,j µvx y2j − t0,j µvy x j y j − t0,j µvz x j z j + t0,j µvx z2j
 

+µ px y2j − µ py x j y j − µ pz x j z j + µ px z2j
 
~
Hj µ =  3
2
y2j +z2j
(34)
      


 y2j + z2j µ px + t0,j µvx − x j y j µ py + t0,j µvy 


 


 − x j z j µ pz + t0,j µvz 


=  3
2
y2j +z2j

Substituting Equation (13) into Equation (35),

n    o
~ µ px + t0,j µvx tan h pk y j µ py + t0,j µvy + z j µ pz + t0,j µvz
Hj µ = q − (35)
y2j + z2j y2j + z2j
 Sensors 2021, 21, 6439 11 of 18

By substituting Equation (35) into Equation (26) and rearranging it, the observability
condition for the CoTMA system is obtained as
 q
µ px + t0,j µvx y2j + z2j
tan h(pk ) 6=     (36)
y j µ py + t0,j µvy + z j µ pz + t0,j µvz

for some t0 ≤ t j ≤ tk .
Equation
 (36)
 does not hold if the relative target state maintains the constant velocity
motion in t0 , t j . Therefore, in order to satisfy the observability requirement with the
uncooperative target, the sensor maneuver is essential. This result is an extension of the
Sensors 2021, 21, x FOR PEER REVIEW 12 of 20
BoTMA observability studies in the 2D plane [18,21].
If Equation (26) is satisfied through the sensor maneuver, it indicates that the estimate
of the initial relative target state x0 in the body coordinates can be converged using the
related
conictoangle
the magnitude
measurement of the FIM of
history θkEquation (24).of
. The degree However, the established
the convergence system is
of the estimate
observability
related to does not mean solving
the magnitude the conic
of the FIM angle ambiguity.
of Equation (24). However, the established system
observability
Then, the kinddoes
ofnot mean solving
maneuver that is the conic angle
necessary ambiguity.
for the sensor to solve the conic angle
ambiguity Then, thesatisfying
while kind of maneuver
Equation that is necessary
(26) must for the sensor
be determined. In theto solve
case of athe conic
single angle
sen-
sor,ambiguity while satisfying
course maneuvers shouldEquation (26) must
be performed be determined.
on two In the case
or more nonparallel of a single
planes. sensor,
Figure 6
course
shows maneuvers should be performed on two or more nonparallel planes. Figure 6 shows
an example.
an example.

Figure
Figure 6. Target
6. Target andand
twotwo
falsefalse targets.
targets.

The initial position of the sensor is located at the origin, and its Euler angles are zero.
The initial position of the sensor is located at the origin, and its Euler angles are zero.
The sensor obtains the conic angle from the target, which moves in constant velocity motion.
The sensor obtains the conic angle from the target, which moves in constant velocity mo-
In the example, two false targets exist, and their trajectories are symmetrical with the target
tion. In the example, two false targets exist, and their trajectories are symmetrical with the
trajectories in the y = 0 plane and z = 0 plane.
target trajectories in the y = 0 plane and z = 0 plane.
The sensor makes two types of course maneuvers: the first one is a yaw change for
The sensor
100~190 s, andmakes two types
the second of course
is a pitch changemaneuvers:
for 300~345 the
s. first one isangle
The conic a yaw change for
trajectories with
100~190 s, and the second is a pitch change for 300~345 s. The
respect to the target and the false targets are shown in Figure 7. conic angle trajectories with
respect toThethetarget
targetand
andthe
thefalse
falsetargets
targetshave
are the
shown
sameinconic
Figure 7. trajectory at 0~100 s before
angle
the first maneuver of the sensor is performed.
After 100 s, when the change in the yaw angle begins, False Target #1 is separated
from the target. This change is the first maneuver on the z = 0 plane, but False Target #2,
which is symmetrical with the target in the plane, still has the same conic angle trajectory
as the target.
In order to distinguish False Target #2 from the target, the sensor needs to maneuver
on the plane, which is not parallel to z = 0. As the sensor changes the pitch angle after
300 s, the target has a unique conic angle trajectory.
 The initial position of the sensor is located at the origin, and its Euler angles are zero.
The sensor obtains the conic angle from the target, which moves in constant velocity mo-
tion. In the example, two false targets exist, and their trajectories are symmetrical with the
target trajectories in the y = 0 plane and z = 0 plane.
The sensor makes two types of course maneuvers: the first one is a yaw change for
Sensors 2021, 21, 6439 12 of 18
100~190 s, and the second is a pitch change for 300~345 s. The conic angle trajectories with
respect to the target and the false targets are shown in Figure 7.

Figure7.7.The
Figure Theconic
conicangle
angletrajectories.
trajectories.

Further numerical
The target and thesimulations
false targetsare given
have the in Section
same conic6.angle trajectory at 0~100 s before
the first maneuver of the sensor is performed.
4.3. Practical Issues
After 100 s, when the change in the yaw angle begins, False Target #1 is separated
fromAlthough the
the target. sensor
This maneuver
change is needed
is the first maneuverto satisfy
on thethez observability
= 0 plane, butrequirements
False Target of
#2,
the CoTMA system, in some applications, the movement of the sensor is either impossible
or limited.
For example, although a towed array sonar is able to be moved by a towing boat,
in sections where the course is changed, the arrangement shape of the sonar becomes
nonlinear, which degrades the measurement accuracy. Additionally, the linear array sensor
installed on the seabed for the underwater surveillance system of the harbor is not movable.
In such environments, the system observability requirements can be alleviated by
using multiple sensors or by using supported features.
Multiple sensors with different locations can use the intersection between their angles
to resolve distance/conic angle ambiguities without requiring the sensor maneuver. This is
beyond the scope of this paper.
In special cases, supported features can be obtained, which mitigate the sensor ma-
neuver requirements needed to satisfy the system observability.
If a single sensor is given and a target can be limited to a surface combatant, the diving
depth of the sensor is same as the relative depth between the target and the sensor. In this
condition, the sensor diving depth can be used as the supporting information needed for
target tracking so that the CoTMA system can be expressed in the 2D horizontal plane
without the approximation error analyzed in Section 3.
Such an environment is a special but not uncommon CoTMA case, and it is defined as
the CDTMA system in this paper.

5. Observability Analysis for the CDTMA Systems

Given zk = d, where d is a nonzero arbitrary constant, then the conic angle can be
calculated by
x
h(xk , d) = tan−1 q k , (37)
y2k + d2

where xk is a relative target state vector from the sensor that consists of the position and
the velocity in the 2D plane as
 . . T
xk = xk yk xk yk . (38)
Sensors 2021, 21, 6439 13 of 18

For some t j in [t0 , tk ], the Jacobian matrix from Equation (37) with respect to Equa-
tion (38) for k = 0 is  
~ ∂h x j , d ∂x j
Hj = , (39)
∂x j ∂x0

where  
∂h x j , d  
∂h x j ,d
 
∂h x j ,d
 
= 0 0 , (40)
∂x j ∂x j ∂y j
  q
∂h x j , d y2j + d2
= , (41)
∂x j x2j + y2j + d2
 
∂h x j −xj yj
= q  , (42)
∂y j y2j + z2j x2j + y2j + d2

∂x j  

I2 t0,j I2
= F t0,j = . (43)
∂x0 O2 I2
~
Calculating the H j µ of Equation (26) yields
    
~ y2j + d2 µ px + t0,j µvx − x j y j µ py + t0,j µvy
Hj µ = q   (44)
y2j + d2 x2j + y2j + d2

Then, the observability requirement for the CDTMA system is

xj y2j + d2
6=   (45)
µ px + t0,j µvx y j µ py + t0,j µvy

for some t0 ≤ t j ≤ tk .
For d 6= 0, the Equation (45) is satisfied, even if the sensor performs constant velocity
motion. Therefore, the sensor maneuver is not required under the CDTMA system. There
is a significant difference between the CDTMA and the BoTMA in the 2D plane.
As mentioned in Section 4.2, however, even if the condition of Equation (26) is satisfied,
it cannot be guaranteed to solve the left/right ambiguity. In order to solve the ambiguity
without the sensor maneuver, a pair of linear array sensors or information fusion with
heterogeneous sensors is still required.

6. Numerical Experiments
Simulation I and Simulation II are the numerical observability tests for the CoTMA
system determined in Section 4 and the CDTMA system determined in Section 5, respec-
tively. They are focused on the observability analysis tested by the geometric relationship
of trajectories between the sensor and the target. The sensors trajectory in this scenario is
limited to four typical cases, and the target trajectory is set to the constant velocity motion.
For each sensor trajectory, the determinant of the FIM is calculated as the performance
index over time.

6.1. Simulation I: CoTMA

 
The initial target position in the inertial coordinates is set to 2000 m, 4000 m 200 m ,
and the target moves at a constant motion velocity of 5 m/s during the scenario. The Euler
angles of the target are equal to φ0 = ϕ0 = 0 and ψ0 = −π/2.
The four sensor trajectories for the sensor are considered, and their maneuvering
scenarios are list in Table 1. The rest of the sensor parameters with the exception of the
 Sensors 2021, 21, 6439 14 of 18
Sensors 2021, 21, x FOR PEER REVIEW 15 of 20

maneuvering
Table scenario are
1. Sensor maneuvering the identical. The sensor trajectories begin at the origin of the
scenario.
inertial coordinates with the initial Euler angles of φ0 = ϕ0 = ψ0 = 0. The target and the
Sensor
sensor trajectories for 600 sSensor Sensor8.
are depicted in Figure Sensor
Time [s]
#1 #2 #3 #4
Table 1.
0~100 Sensor maneuvering scenario. CV with 5 [m/s] CV with 5 [m/s]
π Sensor π
100~220
Time [s] Sensor #1 CT#2
Sensor
2
ψ :0 → CT ψ : 0 Sensor
→
2 #3 #42
220~320
0~100 CV with 5 [m/s]CV with
CV 5with
[m/s]
5 [m/s]
CV
100~220 CV 1 with CT 2 ψ : 0 → π2CT ϕ :CT
0→ ψ :−0π→ π
350~365 with 2
CV with 54[m/s]
220~320
0 [m/s]
5 [m/s] CV
365~375 1
CV with 0 [m/s] CV
with with 5 [m/s] CV with 5 [m/s]
350~365 CT ϕ : 0 → − π4
5 [m/s] CV with 5 [m/s]
CT ϕ CV
π
375~420
365~375 : − with
→50[m/s]
4
375~420 CT ϕ : − π4 → 0
420~600 CV with 5 [m/s]
1 420~600
Constant velocity model, 2 Coordinated turn model. CV with 5 [m/s]
1 2
Constant velocity model, Coordinated turn model.

Figure
Figure 8. Simulation
8. Simulation I: target
I: target andand sensor
sensor trajectories.
trajectories.

All of the sensors measure the conic angle of the target every 1 s. The standard
All of the sensors measure the conic angle of the target every 1 s. The standard devi-
deviation of the measurement noise σ equals π/180 (=1 deg) for all of the sensors. Since the
ation of the measurement noise σ equals π / 180 (=1 deg) for all of the sensors. Since the
value of σ only affects the scale of the FIM of Equation (24), it does not have a meaningful
value of σ only affects the scale of the FIM of Equation (24), it does not have a meaning-
effect on the analysis of the system observability. The conic angle histories for all of the
ful effect on the analysis of the system observability. The conic angle histories for all of the
sensors are shown in Figure 9.
sensorsThe are shown in Figure 9.
performance index of the system observability is the determinant of the FIM
The performance index of theindex
(| J (tk , t0 )|). If the performance system observability
is equal to zero, is
thethe determinant
system of the FIMdue
is not observable (
J (to 0)
). If
t k ,tsingularitythe performance index is equal to zero, the system is not observable due
of the FIM. Otherwise, the inversion of the FIM provides the uncertainty to
singularity
variance offorthe
allFIM. Otherwise,
of the elements the inversion
to be of the
estimated. TheFIM provides
larger the uncertainty
the performance vari-the
index,
ance for allthe
smaller of the elements
estimated to be
error estimated. The larger the performance index, the smaller
variance.
Figure error
the estimated 10 shows the performance indices for all of the sensors over time. The left
variance.
subplot
Figurecontains
10 shows thethe
first maneuver interval
performance of “Sensor
indices for all of the#3” and “Sensor
sensors #4”, The
over time. while
leftthe
right subplot
subplot containsincludes
the firstthe other maneuver
maneuver intervalintervals
of “Sensorof “Sensor
#3” and#4”. The performance
“Sensor index
#4”, while the
value
right indicates
subplot the amount
includes the otherof maneuver
informationintervals
that is needed to estimate
of “Sensor #4”. Thethe target state.in-
performance
dex valueIn the left subplot,
indicates the performance
the amount index
of information values
that of “Sensor
is needed #3” and
to estimate the“Sensor #4” are
target state.
nonzero after their maneuvers. At this point, the values are not large yet, but they increase
over time. On the other hand, the performance index values of “Sensor #1” and “Sensor
#2” are equal to 0. This is because “Sensor #1” and the “Sensor #2” move with a constant
velocity (CV) model, and their relative target state vector xk is also changed in the CV
Sensors 2021,2021,
Sensors 21, x21,
FOR PEER REVIEW
6439 16 of1520of 18

Sensors 2021, 21, x FOR PEER REVIEW 16 of 20

model, which cannot satisfy the observability condition of Equation (26). This shows that
the sensor maneuver is required to achieve the observability of the CoTMA system.

Figure 9. Simulation I: conic angle trajectories.

SimulationI: I:conic
Figure9.9.Simulation
Figure conicangle
angletrajectories.
trajectories.

Figure
Figure 10. Simulation
10. Simulation I: performance
I: performance index.
index.

Figure In
10. the
Simulation I: performance
right subplot, index.
the performance index of
difference between “Sensor #3”
In the left subplot, the performance index values “Sensor #3” and “Sensor #4” areand
“Sensor
nonzero after#4” increases
their after At
maneuvers. thethis
additional
point, themaneuvers
values areof not“Sensor #4”.
large yet, butAt theincrease
they end of the
In the left
simulation, the subplot, the performance
performance index of index #4”
“Sensor values
is of “Sensor
more than #3” times
three and “Sensor
larger #4” are
than that
over time. On the other hand, the performance index values of “Sensor #1” and “Sensor
nonzero
of “Sensor after#3”.
their maneuvers.
This result At this
means thatpoint,
the the valuesof
trajectory are not large
“Sensor #4”yet,
canbut they relatively
obtain increase
#2” are equal to 0. This is because “Sensor #1” and the “Sensor #2” move with a constant
over
more time. On tracking
i target the otherinformation
hand, the performance index values
than that of “Sensor #3” can.of “Sensor #1” and “Sensor
velocity (CV)
#2” areFurthermore,
equal model, and is
to 0. This their
“Sensor
relative
because target#1”
“Sensor
#4” overcomes
state
andvector
the conic
x k is also
the “Sensor
angle ambiguity
changed
#2”mentioned
move within the CV
inaSection
constant4.2
model, which cannot satisfy the observability condition of Equation
velocity (CV) model, and their relative target state vector x k is also changed in the CV
because it maneuvers in two nonparallel planes. However, for (26).
“SensorThis shows
#3”, its that
scenario
the satisfies
sensor maneuver
the system is required to achieve
but doesthenotobservability of angle
the CoTMA system. implies
model, which cannotobservability
satisfy the observability solve the cone
condition of Equation ambiguity.
(26). This This
shows that
that the estimate for “Sensor #3” can converge on the mirror target trajectory.and “Sen-
In the right subplot, the performance index difference between “Sensor #3”
the sensor maneuver is required to achieve the observability of the CoTMA system.
sor #4” increases after the additional maneuvers of “Sensor #4”. At the end of the simula-
In the right II: subplot, the performance index difference between “Sensor #3” and “Sen-
tion,6.2.
theSimulation
performanceCDTMA index of “Sensor #4” is more than three times larger than that of
sor #4” increases after the additional maneuvers of “Sensor #4”. At the end of the simula-
“Sensor #3”.The This result scenario
trajectory means that of the
the trajectory of “Sensor
target is the same as#4” in can obtain relatively
Simulation I, but the more
sensor
tion, the performance index of “Sensor #4” is more than three times larger than that of
i target trackingare
trajectories information
changed. Two thancases
that are
of “Sensor
considered:#3” can.
one is a stationary sensor and the other
“Sensor #3”. This result means that the trajectory of “Sensor #4” can obtain relatively more
isFurthermore,
a sensor that “Sensormoves in #4”atovercomes
a CV motion theofconic angle
5 m/s. ambiguity
The sensors are mentioned
referred in
as Section
the “static
i target tracking information than that of “Sensor #3” can.
4.2 because it maneuvers in two nonparallel planes. However, for “Sensor
sensor” and the “constant moving sensor”, respectively, and they originate at the Euler #3”, its scenario
Furthermore, “Sensor #4” overcomes the conic angle ambiguity mentioned in Section
satisfies
angles theofsystem
φ = ϕ0observability
= 0 and ψ0 =but does not solve the cone angle ambiguity. This im-
π/3.
4.2 because it0 maneuvers in two nonparallel planes. However, for “Sensor #3”, its scenario
plies that the estimate for “Sensor #3” can converge on the mirror target trajectory.
satisfies the system observability but does not solve the cone angle ambiguity. This im-
plies that the estimate for “Sensor #3” can converge on the mirror target trajectory.
 The trajectory
6.2. Simulation II: CDTMAscenario of the target is the same as in Simulation I, but the sensor
trajectories are changed. Two cases are considered: one is a stationary sensor and the other
The trajectory scenario of the target is the same as in Simulation I, but the sensor
is a sensor that moves in at a CV motion of 5 m/s. The sensors are referred as the “static
trajectories are changed. Two cases are considered: one is a stationary sensor and the other
sensor” and the “constant moving sensor”, respectively, and they originate at the Euler
is a sensor that moves in at a CV motion of 5 m/s. The sensors are referred as the “static
Sensors 2021, 21, 6439 angles of φ 0 = ϕ 0 = 0 and ψ 0 = π / 3 . 16 of 18
sensor” and the “constant moving sensor”, respectively, and they originate at the Euler
The difference in the vertical
angles of φ 0 = ϕ 0 = 0 and ψ 0 = π / 3 . depth ( zk ) between the target and the sensors is 200 m,
which
Theisdifference
given as prior
in theinformation
vertical depthto the
( z kCDTMA
) between system. The and
the target simulation takesis600
the sensors 200s;m,
the
conic angledifference
which isThe
of Equationthe
given as priorin
(13)vertical
is measured
information todepth
every 1 s, and the
(zk ) between
the CDTMA
σ target
system. The
is equal
and π takes
tothe
simulation
/sensors
180 . is 200 m,
600 s; the
The target and the sensor trajectories are represented in Figure 11, and the conic
600angle
which
conic is given
angle as prior(13)
of Equation information
is measured to the
everyCDTMA1 s, and σ
system. is The
equal π /
simulation
to 180takes
. s; the
histories
conic with
angle ofrespect
Equation to the
(13)sensor
is scenarios
measured everyare 1shown
s, and in Figure
is equal 12.
to
The target and the sensor trajectories are represented in Figure 11, and the conic angle
σ π/180.
historiesThe target
with and to
respect thethesensor
sensortrajectories
scenariosare arerepresented
shown in Figurein Figure
12. 11, and the conic angle
histories with respect to the sensor scenarios are shown in Figure 12.

Figure 11. Simulation II: target and sensor trajectories.

Figure
Figure 11.11. Simulation
Simulation II: target
II: target andand sensor
sensor trajectories.
trajectories.

Figure12.
Figure SimulationII:II:conic
12.Simulation conicangle
angletrajectories.
trajectories.

Figure 12.
ForSimulation II: conic
each sensor, angle trajectories.
the performance index of the CoTMA system and that of the CDTMA
For each sensor, the performance index of the CoTMA system and that of the
system are
CDTMA compared.
system Figure 13 shows
are compared. Figureperformance index tendency
13 shows performance for the
index “static for
tendency sensor”.
the
As For each sensor,
mentioned in the performance
Simulation I, the index
sensor of the cannot
scenario CoTMA systemthe
establish and that of the in
observability
“static sensor”. As mentioned in Simulation I, the sensor scenario cannot establish the ob-
CDTMA
the CoTMAsystem are compared.
system. However, Figure
the 13 shows system,
performance index tendency
indexfor the of
servability in the CoTMA system.inHowever,
CDTMA in the CDTMA thesystem,
performance value
the performance in-
“static sensor”.
the value
trajectory As mentioned in Simulation I, the sensor scenario cannot establish the ob-
dex of theistrajectory
nonzero, is
indicating
nonzero, that it satisfies
indicating that the observability
it satisfies conditionscondi-
the observability of the
servability
CDTMA in the CoTMA system. However, in the CDTMA system, the performance in-
tions of thesystem.
CDTMA system.
dex value of the trajectory is nonzero, indicating that it satisfies the observability condi-
tions of the CDTMA system.
 Sensors 2021, 21, x FOR PEER REVIEW 18 of 20

Sensors 2021, 21, x FOR PEER REVIEW 18 of 20

Sensors 2021, 21, 6439 17 of 18

Figure 13. Simulation II: performance index (static sensor scenario).

The
Figure
Figure 13. 13.tendency
Simulation II:of
Simulation the
II: performance
performance
performance indices
index
index (static
(static for the
sensor
sensor “constant
scenario). moving sensor” are shown
scenario).
in Figure 14. The results show that the CDTMA system can achieve system observability
Thesensor
without tendency of the performance indices for the “constant moving sensor” are shown
maneuvers.
The tendency of the performance indices for the “constant moving sensor” are shown
in Figure 14. The results show that the CDTMA system can achieve system observability
in Figure 14. The results show that the CDTMA system can achieve system observability
without sensor maneuvers.
without sensor maneuvers.

Figure 14. Simulation II: performance index (constant moving sensor scenario).
Figure 14. Simulation II: performance index (constant moving sensor scenario).
7. Conclusions
7. Conclusions
Figure 14.
InSimulation II: performance
this paper, index (constant
target localization moving
using conic sensor
angle scenario). was defined as the
information
CoTMA In this paper,
system, andtarget localization
the system using conic
was modeled in 3Dangle
space.information
The CoTMAwas defined
system as the
is different
7. Conclusions
CoTMA system, and the system was modeled in 3D space. The CoTMA
from the conventional 2D/3D TMA in the previous studies and is applicable to linear array system is different
from
sonar the
In this conventional
withpaper, 2D/3D
targetincidence
a vertical TMA
localization in the previous
using
angle. conic anglestudies and is applicable
information was definedto linear array
as the
sonarThe
CoTMA withobservability
system,a vertical
and theincidence
system angle.
was
requirements modeled
for the in CoTMA
3D space.system
The CoTMA system
are derived byisthe
different
local lin-
from theThe
earized observability
conventional
Fisher 2D/3D
information requirements
TMA
matrix, thefor
inand the CoTMA
previous
a sensor studies
maneuversystem
and isare
strategy derived
applicable bylinear
to
was proposed the local
toarrayline-
achieve
arized
sonar Fisher information
with observability.
system a vertical incidenceWe also matrix,
angle. and a sensor maneuver strategy
proposed the CDTMA system, which can mitigate the sen-was proposed to
achieve
The
sor system
observability
maneuver observability.
requirements
limitation. We also proposed
for the
The results CoTMA
of the the
analytic CDTMA
system
scheme system,
are were
derived which can mitigate
by the localthrough
demonstrated line-
the sensor
Fisher maneuver
simulation
arized studies.
information limitation. The results
matrix, and a sensor of maneuver
the analyticstrategy
schemewas wereproposed
demonstratedto
through
achieve Basedsimulation
system on studies.
the results
observability. ofWethis paper,
also the nonlinear
proposed the CDTMA tracking
system,algorithm,
which can trajectory
mitigate opti-
Based
the mization,
sensor on the
data
maneuver results
association, ofand
limitation. this paper, theofnonlinear
information
The results fusion tracking
will
the analytic become
scheme algorithm,
future
weretopicstrajectory
for CoTMA
demonstrated opti-
mization,
system
through data association,
research.
simulation studies. and information fusion will become future topics for CoTMA
system research.
Based on the results of this paper, the nonlinear tracking algorithm, trajectory opti-
Author data
mization, Contributions:
association,Conceptualization,
and information T.S.;fusion
methodology, T.S. andfuture
will become W.K.; software,
topics for W.K.;
CoTMAwriting—
original draft
system research. preparation, W.K.; writing—review and editing, T.S. and W.K.; visualization, W.K.;
supervision, T.S. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Sensors 2021, 21, 6439 18 of 18

Data Availability Statement: Data sharing not applicable.

Conflicts of Interest: The authors declare no conflict of interest.

References
1. Bar-Shalom, Y.; Li, X.-R.; Kirubarajan, T. Estimation with Applications to Tracking and Navigation; Wiley: Hoboken, NJ, USA, 2001.
2. Nardone, S.; Lindgren, A.; Gong, K. Fundamental properties and performance of conventional bearings only target motion
analysis. IEEE Trans. Autom. Control 1984, 29, 775–787. [CrossRef]
3. Lindgren, A.G.; Gong, K.F. Properties of a nonlinear estimator for determining position and velocity from angle of arrival
measurements. In Proceedings of the 14th Asilomar Conference Circuits, System, and Computer, Pacific Grove, CA, USA, 17–19
November 1980.
4. Julier, S.; Uhlmann, J. Unscented Filtering and Nonlinear Estimation. Proc. IEEE 2004, 92, 401–422. [CrossRef]
5. Arasaratnam, I.; Haykin, S. Cubature kalman filters. IEEE Trans. Autom. Control 2009, 54, 1254–1269. [CrossRef]
6. Aidala, V.J.; Nardone, S.C. Biased Estimation Properties of the Pseudolinear Tracking Filter. IEEE Trans. Aerosp. Electron. Syst.
1982, AES-18, 432–441. [CrossRef]
7. Song, T.; Speyer, J. A stochastic analysis of a modified gain extended Kalman filter with applications to estimation with bearings
only measurements. IEEE Trans. Autom. Control 1983, AC-30, 940–949. [CrossRef]
8. Peach, N. Bearings-only tracking using a set of range-parameterised extended Kalman filters. In IEE Proceedings Control Theory
and Applications; Institution of Engineering and Technology (IET): London, UK, 1995; Volume 142, pp. 73–80.
9. Mušicki, D. Bearings only single-sensor target tracking using Gaussian mixtures. Automatica 2009, 45, 2088–2092. [CrossRef]
10. Risfic, B.; Arulampalam, S.; Gordon, N. Beyond the kalman filter—Book Review. IEEE Aerosp. Electron. Syst. Mag. 2004, 19, 37–38.
[CrossRef]
11. Pitt, M.; Shephard, N. Filtering via simulation: Auxiliary particle filters. J. Am. Stat. Assoc. 1999, 49, 590–599. [CrossRef]
12. Musso, C.; Oudjane, N.; LeGland, F. Improving regularized particle filters. In Sequential Monte Carlo Methods in Practice; Springer:
Berlin/Heidelberg, Germany, 2001.
13. Doucet, A.; Godsill, S.; Andrieu, C. On sequential Monte Carlo sampling methods for Bayesian filtering. Stat. Comput. 2000,
10, 197–208. [CrossRef]
14. Kirubarajan, T.; Bar-Shalom, Y.; Lerro, D. Bearings-only tracking of maneuvering targets using a batch-recursive estimator. IEEE
Trans. Aerosp. Electron. Syst. 2001, 37, 770–780. [CrossRef]
15. Mušicki, D.; Song, T.; Kim, W.; Nešić, D. Non-linear automatic target tracking in clutter using dynamic Gaussian mixture. IET
Radar Sonar Navig. 2012, 6, 937–944. [CrossRef]
16. Xie, Y.; Song, T.L. Bearings-only multi-target tracking using an improved labeled multi-Bernoulli filter. Signal Process. 2018,
151, 32–44. [CrossRef]
17. Li, X.; Zhao, C.; Lu, X.; Wei, W. Underwater Bearings-Only Multitarget Tracking Based on Modified PMHT in Dense-Cluttered
Environment. IEEE Access 2019, 7, 93678–93689. [CrossRef]
18. Nardone, S.C.; Aidala, V.J. Observability Criteria for Bearings-Only Target Motion Analysis. IEEE Trans. Aerosp. Electron. Syst.
1981, AES-17, 162–166. [CrossRef]
19. Hammel, S.; Aidala, V. Observability Requirements for Three-Dimensional Tracking via Angle Measurements. IEEE Trans. Aerosp.
Electron. Syst. 1985, AES-21, 200–207. [CrossRef]
20. Fogel, E.; Gavish, M. Nth-order dynamics target observability from angle measurements. IEEE Trans. Aerosp. Electron. Syst. 1988,
24, 305–308. [CrossRef]
21. Song, T.L. Observability of target tracking with bearings-only measurements. IEEE Trans. Aerosp. Electron. Syst. 1996, 32,
1468–1472. [CrossRef]
22. Song, T.L. Observability of target tracking with range-only measurements. IEEE J. Ocean. Eng. 1999, 24, 383–387. [CrossRef]
23. Pillon, D.; Perez-Pignol, A.-C.; Jauffret, C. Observability: range-only vs. bearings-only target motion analysis for a leg-by-leg
observer’s trajectory. IEEE Trans. Aerosp. Electron. Syst. 2016, 52, 1667–1678. [CrossRef]
24. Jauffret, C.; Perez, A.-C.; Pillon, D. Observability: Range-Only Versus Bearings-Only Target Motion Analysis When the Observer
Maneuvers Smoothly. IEEE Trans. Aerosp. Electron. Syst. 2017, 53, 2814–2832. [CrossRef]
25. Oshman, Y.; Davidson, P. Optimization of observer trajectories for bearings-only target localization. IEEE Trans. Aerosp. Electron.
Syst. 1999, 35, 892–902. [CrossRef]
26. Rao, S.K. Bearings-Only Tracking: Observer Maneuver Recommendation. IETE J. Res. 2021, 67, 193–204. [CrossRef]
27. He, S.; Shin, H.-S.; Tsourdos, A. Trajectory Optimization for Target Localization With Bearing-Only Measurement. IEEE Trans.
Robot. 2019, 35, 653–668. [CrossRef]
28. Lin, Z.; Lin, M.; de Cola, T.; Wang, J.; Zhu, W.; Cheng, J. Supporting IoT with rate-splitting multiple access in satellite and
aeri-al-integrated networks. IEEE Internet Things 2021, 8, 11123–11134. [CrossRef]
29. Lin, Z.; Lin, M.; Champagne, B.; Zhu, W.-P.; Al-Dhahir, N. Secure and Energy Efficient Transmission for RSMA-Based Cognitive
Satellite-Terrestrial Networks. IEEE Wirel. Commun. Lett. 2021, 10, 251–255. [CrossRef]
30. Lin, Z.; Lin, M.; Champagne, B.; Zhu, W.-P.; Al-Dhahir, N. Secure Beamforming for Cognitive Satellite Terrestrial Networks With
Unknown Eavesdroppers. IEEE Syst. J. 2021, 15, 2186–2189. [CrossRef]