Depth Discrimination of an Acoustic Source Based on Modal Energy
Distribution - Performance Analysis
by
Jakov Kostjukovsky
M.E., Systems Engineering (2005)
Technion,
B.Sc., Applied Mathematics and Software Engineering (1998)
Jerusalem College of Technology
Submitted to the Department of Mechanical Engineering
in partial fulfillment of the requirements for the degree of
Master of Science in Ocean Engineering
at the
Massachusetts Institute of Technology
ARCHIVES
OF
ECNOOGY
MAY 0 5 2010
February 2010
LIBRARIES
copyright, 2010 Massachusetts Institute of Technology
All rights reserved
Signature of author ...................................
..
.
D
Q
. . .. .. .
mn of Mechanical Engineering
January 29, 2010
C ertified by ..................................
. ........... ... .. ...
. ........ ..............
Professor Henrik Schmidt
Professor of Mechanical and Ocean Engineering
A
A ccepted by ....................................
.
.00111
Thesis Supervisor
. .....................................................
Professor David Hardt
Chairman, Department Committee on Graduate Students
Depth discrimination of an acoustic source based on modal energy
distribution - performance analysis
by
Jakov Kostjukovsky
Submitted to the Department of Mechanical Engineering
in partial fulfillment of the requirement for the degree of
Master of Science in Ocean Engineering
Abstract
A recently proposed method [Apelfeld, Y., Depth Discriminationof an Acoustic Source Based
on Modal Energy Distribution.MIT, 2007] for rough, but fast and robust depth classification of
acoustic sources was investigated in realistic towing scenarios. The method uses modal energy
distribution as a categorical indicator for the source depth.
In this study we analyzed the method performance in two major towing patterns: the "yo-yo"
pattern and the "one-time" depth sampling pattern. The position of the array elements was
obtained using the MOOS-IvP simulation package. Among the main towing parameters of the
performance evaluation was the pitch angle of the array.
The main finding of this study is the discovery of a relationship between vertical orientation of
the array and the relative source bearing. Thus, downward array orientation leads to relatively
high separation capability for sources located at 00 relative bearing. In the contrary, upward array
orientation leads to high separation capability for sources located at the opposite horizontal
relative bearing (1800). This relationship implies direct tactical considerations on the AUV
deployment that are discussed and summarized in this study.
Thesis Supervisor: Professor Henrik Schmidt
Title: Professor of Mechanical and Ocean Engineering
Acknowledgments
First and foremost, I wish to thank to my advisor, Prof. Henrik Schmidt. Without your guidance,
patience and support this work would never be done. For the past 18 months you were not only a
lecturer and academic advisor, but most importantly, a friend.
I want to thank to Prof. Nicholas C. Makris for a very interesting and fascinating way to teach
acoustics. It was a big intellectual challenge and my great pleasure to be your student. Also, I
wish to thank to Prof. Gilbert Strang from Department of Mathematics for making my passion
for applied mathematics even stronger.
I would like to express my regards to the Department of Mechanical Engineering, to Graduate
Office staff and in particular to Ms. Leslie Regan, to administrative assistant Mr. Geoff Fox and
to my fellow students and teaching assistants, Kevin Cockrell and loannis Bertsatos. Thank you
for making the learning process so pleasant and productive.
My special appreciation goes to the Israeli Navy for giving me the unique opportunity to study at
MIT and providing me a full scholarship. In particular, I am thankful to Rear Admiral Nitzan
Shaked for your trust and recommendation. I am sincerely thankful to my commanding officers:
Captains Itzik Maya and Eitan Tzuker, Commanders Slomo Azar and Yaron Abutbul for
supporting me and making my dream a reality. My fellows, MIT alumni, Commanders Sela
Meyouhas (1997) and Eran Naftali (2000), Lieutenant Commanders Gilhad Bar-Yehoshua
(2003) and Yan Apelfeld (2007), thank you for helping and guiding me during the preparation to
this incredible journey.
I would like to thank to Prof. Yossi Ben-Asher and Dr. Miri Doron for your recommendation
letters that made my studies at MIT possible.
I wish to thank to my whole family, who are always beside me, even when I am on the other side
of the Globe. Last, but certainly not least, I am deeply thankful to my beloved wife, Inessa,
whose love and support have the most important role in my success. Thank you from all my
heart.
Contents
1
Introduction.............................................................................................................................
2
N orm al m odes for a point source in shallow w ater ..............................................................
3
4
M athem atical derivation.............................................................................................
12
2.2
Soft surface, hard bottom w aiveguide............................................................................
15
The SI algorithm ...................................................................................................................
18
3.1
General description ......................................................................................................
18
3.2
Initial perform ance evaluation ....................................................................................
20
The SI algorithm evaluation in realistic AUV scenarios ......................................................
25
Methodology ..................................................................................................................
25
4.1.1
Simulation tools ....................................................................................................
25
4.1.2
Towing patterns ...................................................................................................
26
4.1.3
Range dependence analysis..................................................................................
31
4.2
6
12
2.1
4.1
5
9
Results............................................................................................................................
32
4.2.1
The yo-yo towing pattern....................................................................................
32
4.2.2
One-w ay depth sam pling ......................................................................................
43
4.2.3
Range dependence ...............................................................................................
51
D iscussion.............................................................................................................................
64
5.1
V ertical orientation.......................................................................................................
64
5.2
Tow ing pattern param eters........................................................................................
68
Sum mary and Conclusion..................................................................................................
69
Appendix A: Main simulation tool user's guide and source code ............................................
72
A .1. M ain simulation module.............................................................................................
72
Secondary functions ....................................................................................................
78
A .2.
List of Figures
Figure 2.1: Plane wave schematic interpretation of the modal field for a homogeneous layer.... 16
Figure 3.1: Schematic description of the SI averaging algorithm (adapted from [5]).............. 19
21
Figure 3.2: Sound speed profile used in simulations .................................................................
Figure 3.3: SI results for 300Hz source, 32-element array, Cb = 1575 m/s, bearing 00, taken from
21
[5 ]).................................................................................................................................................
from
taken
Figure 3.4: SI results for 300Hz source, 32-element array, Cb = 1625 m/s, bearing 00,
22
[5 ]).................................................................................................................................................
Figure 3.5: SI results for 300Hz source, 32-element array, cb = 1612 m/s, bearing 50, taken from
22
[5 ]).................................................................................................................................................
Figure 3.6: SI results for 300Hz source, 32-element array, Cb = 1612 m/s, bearing 150, taken from
23
[5 ]).................................................................................................................................................
Figure 3.7: SI results for 300Hz source, 32-element array, cb = 1612 m/s, bearing 350, taken from
23
[5 ]).................................................................................................................................................
Figure 3.8: SI results for 300Hz source, 32-element array, cb = 1612 m/s, bearing 550, taken from
24
[5 ]).................................................................................................................................................
26
Figure 4.1: A typical yo-yo towing pattern in x-z plane..........................................................
Figure 4.2: A schematic description of the towed array during the one-time depth sampling
29
tow ing pattern ...............................................................................................................................
Figure 4.3: Typical array position for different pitch angles during free diving maneuver. The
green dot represents the first element of the array. The x axis represents the distance from the
30
tow ing A U V ..................................................................................................................................
Figure 4.4: The yo-yo towing pattern, diving vs. climbing SI results for bearing 00 ................... 33
Figure 4.5: The yo-yo towing pattern, diving vs. climbing SI results for bearing 50.............. 34
Figure 4.6: The yo-yo towing pattern, diving vs. climbing SI results for bearing 10 ................. 35
Figure 4.7: The yo-yo towing pattern, diving vs. climbing SI results for 300Hz source, 32elem ent array,
Cb =
36
1612 m /s, bearing 150 .................................................................................
Figure 4.8: The yo-yo towing pattern, diving vs. climbing SI results for bearing 20 .............. 37
Figure 4.9: The yo-yo towing pattern, diving vs. climbing SI results for bearing 1800............... 38
Figure 4.10: The yo-yo towing pattern, diving vs. climbing SI results for bearing 1750............. 39
Figure 4.11: The yo-yo towing pattern, diving vs. climbing SI results for bearing 1700 ............. 40
Figure 4.12: The yo-yo towing pattern, diving vs. climbing SI results for bearing 1650 ............. 41
42
Figure 4.13: The yo-yo towing pattern, diving vs. climbing SI results for bearing 1600 ......
Figure 4.14 The one-way towing pattern, diving vs. climbing averaged SI results for pitch ±20.43
Figure 4.15 The one-way towing pattern, diving vs. climbing averaged SI results for pitch
+40.44
Figure 4.16 The one-way towing pattern, diving vs. climbing averaged SI results for pitch +6.44
Figure 4.17 The one-way towing pattern, diving vs. climbing averaged SI results for pitch ±80.45
Figure 4.18 The one-way towing pattern, diving vs. climbing averaged SI results for pitch +100.
45
.......................................................................................................................................................
Figure 4.19 The one-way towing pattern, diving vs. climbing averaged SI results for pitch +12*.
.... 4 6
.................................................................................................................................................
Figure 4.20 The one-way towing pattern, diving vs. climbing averaged SI results for pitch ±140.
.... 4 6
.................................................................................................................................................
Figure 4.21 The one-way towing pattern, diving vs. climbing averaged SI results for pitch ±16*.
....................................................................................................................
.... 4 7
............................
Figure 4.22 The one-way towing pattern, diving vs. climbing averaged SI results for pitch ±180.
.... 4 7
................................................................
................................................................................
Figure 4.23 The one-way towing pattern, diving vs. climbing averaged SI results for pitch ±200.
48
.......................................................................................................................................................
Figure 4.24 The one-way towing pattern, diving vs. climbing averaged SI results for pitch ±250.
...........................................................................................
................................................
....... 4 8
Figure 4.25 The one-way towing pattern, diving vs. climbing averaged SI results for pitch ±300.
........................................................................................................................
..............................
Figure 4.26: The one-way towing pattern, diving vs. climbing averaged SI results for pitch
.......
................................
........
.
............
.
.
.................................................
49
±35*.
.......................
49
Figure 4.27: Averaged SI results for MOOS-IvP simulated diving towing pattern, characteristic
pitch 19 .......................................
...... ...................................................................................
50
Figure 4.28: Averaged SI results for MOOS-IvP simulated diving towing pattern, characteristic
p itch 32 .............................................
...............
.......... .......................................................
50
Figure 4.29: Averaged SI results for diving towing pattern for different initial source ranges,
p itch 2 .......................................................................................................................................
51
Figure 4.30: Averaged SI results for diving towing pattern for different initial source ranges,
p itch
4
. .........................................................................................................................................
5 2
Figure 4.31: Averaged SI results for diving towing pattern for different initial source ranges,
pitch 6 ...............
....................................................................
53
Figure 4.32: Averaged SI results for diving towing pattern for different initial source ranges,
p itch 8 ...........................................................................................................................................
54
Figure 4.33: Averaged SI results for diving towing pattern for different initial source ranges,
pitch 10 ..................................................................................................................................
55
Figure 4.34: Averaged SI results for diving towing pattern for different initial source ranges,
p itch 12 .................................................................................................................
56
Figure 4.35: Averaged SI results for diving towing pattern for different initial source ranges,
pitch 14 .....................................................................................................................................
57
Figure 4.36: Averaged SI results for diving towing pattern for different initial source ranges,
p itch 16 ........................................................................................................................................
Figure 4.37: Averaged SI results for diving towing pattern for different initial source ranges,
p itch 180......................................................................................................................................59
58
Figure 4.38: Averaged SI results for diving towing pattern for different initial source ranges,
p itch 2 0 ........................................................................................................................................
60
Figure 4.39: Averaged SI results for diving towing pattern for different initial source ranges,
p itch 2 5 ........................................................................................................................................
Figure 4.40: Averaged SI results for diving towing pattern for different initial source ranges,
61
p itch 30 .........................................................................................................................................
62
Figure 4.41: Averaged SI results for diving towing pattern for different initial source ranges,
p itch
3 5
........................................................................................................................................
Figure 5.1: A comparison of averaged SI performance for downward and upward array
orientation .....................................................................................................................................
Figure 5.2: A comparison of averaged SI performance for near 00 array pitch angles (regular
environm ental settings).................................................................................................................
Figure 5.3: The receiver beam-former output for a typical submerged source (60m depth) and
pitch angles of the receiving array: 00 (horizontal), 100 (upward), and -10* (downward). The
source is located at (a) 00 and (b) 1800 relative bearing to the receiver...................................
6 3
65
65
3
66
List of Tables
Table 4.1: The array geometry at each step of the yo-yo towing pattern. The green dot represents
the first element of the array. The x-axis shows the distance from the towing AUV............... 27
1
Introduction
Over the past decade, autonomous underwater vehicles (AUVs) have been excessively used in
wide spectrum of oceanographic research and marine operations. In many AUV missions there is
a need in rapid and robust source localization. However, in some missions only a rough estimate
of the source position is needed.
Localization of an acoustic source is one of most studied problems in underwater acoustics
research. Various methods have been suggested in the literature [1, 2, 3, 4]. Generally, these
methods refer to either of two main categories: matched field processing and modal field
decomposition. Both categories provide relatively exact source location under several
assumptions that we will describe shortly.
The matched field processing (MFP) method [3, 4] can be seen a generalization of a plane-wave
beam-forming. In the beam-forming process the matching is done for plane-wave replicas of
different angles. In the MFP method the matching is done for all possible source locations. The
receiving field is calculated using a propagation model for each test point and the resulting
replica is matched in term of best correlation to the actual receiving field. The validity of the
MFP method depends mostly on the validity of the propagation model that is used in the
calculation of the field replicas. In many practical applications the exact environmental
parameters, such as sound speed profile or bottom properties are not known. Therefore, the
applicability of the MFP method to real-time source localization in an unknown environment is
highly limited.
The modal field decomposition [1, 2] method is based on the modal decomposition of the
receiving field. The normal modes functions depend on the source position and environmental
conditions. Thus, knowing the modes functions and using their orthogonality property one can
retrieve the source location. As for the MFP method, the applicability of the modal field
decomposition method is also limited in complex unknown environments.
As mentioned above, in some AUV missions the exact source location is not needed. Instead, a
fast and most importantly, robust method that provides only a rough source classification is
Chapter 1. Introduction.
required. Classifying an acoustic source as surfaced or submerged in a fast and robust way is an
example of such a method. Apelfeld [5] has shown that this task can be performed using the
Submergence Index (SI) algorithm.
The SI algorithm is a method for depth classification of acoustic sources based on the modal
energy distribution. The basic idea of the SI algorithm is based on the fact that near-the-surface
sources in shallow water environment will excite more energy in higher modes than submerged
sources. Therefore, the ratio of the energy concentrated in lower modes to the energy
concentrated in higher modes can serve as an indicator for source depth. This ratio is called the
Submergence Index.
Compared to the modal field decomposition techniques, the SI algorithm does not require the
ability to calculate the full modal field decomposition. Instead, the method uses the fact that each
incoming mode can be viewed as a plane wave, coming with different propagation angle to a
receiving array. Moreover, the propagation angle is directly proportional to the mode number [6].
Thus, placing the sources at the end-fire direction of the array will narrow any angle deviation to
vertical (grazing) angles only. Comparing the energy concentrated in shallow vs. steep grazing
angles leads to the SI estimation.
In his pioneering work Apelfeld [5] has shown that in order to achieve a robust and practical
separation performance, two additional averaging processes are required. The first is averaging
the SI results over the receiver depth. The second is averaging over range. Both averaging
processes can be performed during the same depth-changing maneuver.
The main purpose of the current research is to analyze the SI performance in various realistic
towing patterns and determine the most effective (in terms of higher SI capability) towing
pattern. We have investigated two major towing patterns: the "yo-yo" path, where the towing
AUV is gradually changing its depth in a sinusoidal pattern and the "one-time" depth sampling,
where the AUV is performing either a diving or climbing maneuver with a constant pitch angle.
Chapter 2 of this thesis presents the mathematical foundations that lie in the basis of the
presented approach to depth classification of acoustic sources. Chapter 3 describes the SI
algorithm in details presents some typical results from the previous study [5]. Chapter 4 is the
core part of this thesis. It presents the methodology and results of the current study.
10
Chapter1. Introduction.
In Chapter 5 we discuss the main findings of the current research and in Chapter 6 we summarize
and draw conclusions.
Appendix A contains the source code of the main simulation tool used in the performance
evaluation and a short user's guide to it.
2
2.1
Normal modes for a point source in shallow water
Mathematical derivation
A derivation of the normal modes solution for a point source in a horizontally stratified medium
is considered a classical problem in ocean acoustics and it is widely studied in literature. Here,
let us follow the derivation presented by Jensen et al [7] and Frisk [8]. For a point source located
at ro = (0,0, zo) in a horizontally stratified medium, where both density p and sound speed c
depend only on depth, it is natural to choose cylindrical coordinates. In these coordinates the
problem becomes two-dimensional and the corresponding Helmholtz equation can be written as
r c'r
- 1~r + p(z_)
Br
~apzp
8Z)
P
8p1
+k
(
p(z)
(z
(r )1(z - zo
2nr
2.1
where k(z) = w)/c(z) is the total wavenumber.
Using the method of separation of variables, we assume the solution to be of a form p(r, z) =
<b(r)Y(z). Substituting p(r,z) to a homogeneous form of ( 2.1) leads to
1 [1 d
--
(D
1
dCD
1
(rr )] +-[p(z)(+
r dr
dr
T _
d=
z p(z) dz
k2()t]0
2.2
Notice that all partial derivatives in (2.1) become regular derivatives in because CD(r) and YP(z)
are functions of a single variable. Moreover, the two parts of (2.2) are functions of different
variables (r and z); therefore, in order to equation (2.2) to be valid for all possible values of r and
z both parts should be equal to a constant. Denoting this constant by k2, multiplying both sides
by I(z) and dividing by p(z) leads the z-dependent part of (2.2) to the following ordinary
differential equation:
t+
dz p (z) dz
Let us define
p(z)
k2(z)
-
k2 T=0.
p(z) '"
2.3
Chapter2. Normal modes for a point source in shallow water.
13
k, =k 2 (z)-k2
2.4
and call krn the horizontal wavenumber and kz, the vertical wavenumber. We will use these
notations in the following chapters.
Under the assumption that both p(z) and c(z) are real, the equation (1.3) has the well-known
Strum-Liouville form Lu(x) - Aw(x)u(x) = 0, where L is a second-order differential operator
Arfken et al. [9]. In our case,
d 1 d
L=-i-i+-k2(z),
dz (p(z) dz
12
p(z)
S=k2.5
1
According to the properties of Strum-Liouville problems, there is an infinite number of solutions
(eigenfunctions, modes) T, (z) of equation (2.3), each of them corresponds to some eigenvalue
kr
The eigenfunctions are orthonormal (with the weighting factor w(z)) , i.e.,
D
D1
f w(z)T, (z)Wn(z)dz=
1 'm , (z),{ (z)dz
0 p(z)
n
0,
m#n
2.6
.
Another important property of the Strum-Liouville problems is that the modes form a complete
set; therefore, the acoustic pressure field p(r, z) can be written as an infinite sum
, (r), (z).
p(r, z) =
2.7
n=1
Substituting (2.7) into the original Helmholtz equation (2.1) leads to
n=
1 dJ
i~ r doJ (r)4 (z)+
dr )
r dr
(()
ldz
d
1 d,
(z) +k2 (z)n(z)
p(z) dz )
This equation can be simplified using (2.3) to become
= -
g(r)(z - zo). 2.8
2nr
Chapter2. Normal modesfor a point source in shallow water.
1 d (r do(r
n(z)+k2,(),r?()
nr dr
dr )f
8(r)(z-zo)
2.9
2nr
Due to the orthogonality property of the modal functions Wn(z), multiplying by
-and
p (Z)
integrating both sides with respect to z leads to the following:
Irrd(
r dr
+ (
dr
r) =
)
S(r)m (zo)
2nrp(zo)
2.10
Equation (1.10) has a standard solution, which is given in terms of the Hankel function
(D
) ' v (zo)HO()(k,.r).
(
2.11
Therefore, the pressure field p (r, z) is given by
p(r,z)=
J
'n(zo )TnW(z)
2.12
(kr).
4p(z)n=1
Using the asymptotic form of the Hankel function for large argument (x >> 1),
H 1
(x)
=
2
2.13
i(x-/4)
we can write the pressure field as
p (r , z )
ie -i
p(zo W
4
((
n=1
ik
,,m
r
2.14
km
Chapter2. Normal modes for a point source in shallow water.
2.2
Soft surface, hard bottom waiveguide
In this chapter we consider a simple model of an isovelocity acoustic waiveguide with pressure
release surface and perfectly reflecting hard bottom. These assumptions can be formulated as the
following boundary conditions:
=
= 0,
T(0)
dz
0.
2.15
z=D
Under the assumption of a constant density p, equation (2.3) becomes
+Tk 2T = 0,
2.16
dz 2
which has a solution of form
2.17
Tn (z)= A sin k,,z.
The vertical wavenumber kzn can be determined using the boundary conditions (2.15). Thus,
(n-1/2)r
kzn =
D
2.18
, n =1,2,3,...
The A constant in equation (2.17) can be determined using the orthonormality property (2.6) of
eigenfunctions -P
1 (z). To satisfy this property, the equality
D
,2(z) =1,
2.19
0h
should hold for every n. Substituting (2.17) into (2.19) leads to
A2
-
D
sin2 k2
p0
Z
A=,2
[
1 kzD
pkzn _2
sin2kz,,D
2.20
2
Because equation (2.20) should hold for every n, we can choose the easiest value of n - one.
Substituting kz1 from (2.18) into (2.20) leads to
~
A2
1r
-kznD-si
pk,
2
Z
~snkD~2D21x
D) =k D2[
2 2
p
_
2
sing
2
-
DA 2
2p
=1;
Chapter2. Normal modesfor a point source in shallow water.
therefore,
2.21
A =L
Using equations (2.14), (2.17) and (2.21), the acoustic pressure p (r, z) for kr
>> 1 can now be
written as
2ie
"I
p(r,z)~-sink
zD8
o*ir4
,zosinkznz
n=1
e i~
2.22
k/
or, using Euler's formula as
p(r z) ~e'
[e i(k'r+k'''r) -~i(k'r+kr)l
sin kzzo
D n=1 85 J.
rI
2.23
Equation (2.23) in conjunction with the definition of vertical and horizontal wavenumbers, which
are formulated in equation (2.4) opens a way for representation of the total field as a
superposition of up- and down-going plane waves with angle of inclination On, such that (see
Figure 2.1)
kzn = k COs O,
kn= k sin On.
Figure 2.1: Plane wave schematic interpretation of the modal field for a homogeneous layer
2.24
Chapter2. Normal modes for a point source in shallow water.
As it was mentioned in the Introduction section, this observation is one of the main parts that lie
in the basis of the current approach for depth classification of acoustic sources.
Let us rewrite the definition in (2.4) as
k,
=n
2.25
or, using the definition of k and equation (2.18), as
k, =
(c
j[
2
c
-(n -1/2),,r
_
2.26
D
It is clear from (2.26) that for some values of n the square root argument in (2.26) may be
negative. When this happens the horizontal wavenumber becomes imaginary and the
corresponding mode does not propagate. Such, exponentially decaying modes are called
evanescent modes. The nth mode, which frequency is less than the cutofffrequency will not
propagate and become evanescent. The cutoff frequency, is given by
'n =
(n -1/2)=c
D
or f, =
(n -l/2)c
2D
2.27
Now, let us look back on the total field expression given in (2.23). The sin kzn zo term represents
the excitation of the mode at the source depth. Therefore, different modes will have different
initial excitation, depending on the source depth and the mode number. For example, a shallow
source will excite higher modes with more energy than a deep source. As it was mentioned in the
Introduction section, this conclusion is one of the main parts that lie in the basis of the current
approach for depth classification of acoustic sources.
3
3.1
The SI algorithm
General description
As we have seen in Section 2.2, near-the-surface sources excite more energy in higher modes
than submerged sources. In addition, each mode can be represented as a plane wave traveling
with some grazing angle. If the source is placed in either end-fire direction of the receiving line
array (zero or 1800 horizontally), the only angular deviation of the received field will be due to
the vertical source position. Therefore, a comparison of the relative amount of received energy
concentrated in shallow versus steep grazing angles can serve as an indicator for the energy
prevalence in higher modes.
Based on the above observations, the SI algorithm divides the receiving field in two zones of
some empirical relative ratio. The shallow angles zone correspond to the lower modes, while the
deep angles correspond to the higher modes. The ratio of the two zones was chosen empirically
in the previous research [5] to be 1/3. Thus, two Hanning [10] (other window functions were
examined in [5]) windows are applied to the regular beamformer output. The total energy
accumulated in the windows' output is compared and the Submergence Index is defined as
SJ-l
l ower zone
SI=1o1g
Eupperzone
3.1
As it was broadly examined in the previous research [5], the SI algorithm is highly sensitive to
many factors. In particular, its performance depends on the relative range-depth configuration
between the source and the receiver. Therefore, averaged over depth and range SI results are of
more practical interest.
Such averaging can be performed using depth-changing moving pattern of the carrying AUV.
The range averaging will take place simultaneously with the depth averaging due to the
horizontal AUV movement. The following figure (Figure 3.1, adapted from [5]) schematically
illustrates the SI averaging algorithm.
Chapter3. The SI algorithm.
Beamform the receiving
pressure field
Divide the total field into
two angle subspaces using
Hanning window function
Calculate the ratio of the energy
concentrated in the subspaces
(local SI)
Calculate the average SI
over all receiver positions
Move the receiver to a new
position, until the whole
watercolumn sampling is
finished
4
Figure 3.1: Schematic description of the SI averaging algorithm (adapted from [5])
Various depth-changing patterns were suggested, among them, the yo-yo pattern, in which the
towing AUV is gradually changing its depth in a sinusoidal shape, and a step-wise discrete depth
sampling. The main purpose of the current research is to analyze the proposed algorithm in
various realistic towing scenarios, determine the optimum deployment technique and evaluate its
performance.
Chapter3. The SI algorithm.
3.2
Initial performance evaluation
The initial performance evaluation of the SI algorithm was performed by Apelfeld [5]. We
provide here a brief summary of his main findings and conclusions. The evaluation was
performed using computer simulations that simulate the receiving field. The receiving field was
simulated by OASES package [11], which produces plain wave replicas in the MultipleConstrains beamforming Mode (MCM) [12]. The OASES output The SI is calculated directly
from the OASES output, making the main simulation design easy and flexible.
The initial performance evaluation was performed for 300 Hz, 500 Hz and 1 kHz sources in
range-independent and range-dependent environments. Two different arrays were used as the
receiving array: a 128-element equally spaced array with d = 0.75m and a 32-element equally
spaced array with d = 2.5m. The spacing interval was chosen to satisfy the space sampling
criteria d <
for frequencies under IkHz.
The simulations were performed using an approximate Monterey Bay sound speed profile (see
Figure 3.2) in a 100m depth water-channel. The sea surface and the sea bottom were assumed to
be completely plain with no roughness. The ambient noise level was assumed to be equal 50dB
and various values of bottom sound speed (cb
=
1575, 1600, 1612 and 1625 m/s) were analyzed.
The source was located at near the end-fire direction of the receiver (0 to 55 degrees), at 10 km
separation distance. The source Source Level (SL) was set to 120dB.
In the following figures we present some typical results from [5]. Figure 3.3, for example,
demonstrates the SI values for 300 Hz source and 32-element receiver, located at the end-fire
direction of the receiver in an environment with bottom sound speed of 1575m/s. Figure 3.3(a)
shows the complete set of SI values for all possible source/receiver depth combinations, while
Figure 3.3(b) shows the averaged over the receiver depth SI values. As it can be seen from
Figure 3.3, for this source/receiver configuration, there is a difference of roughly 1.7dB between
averaged SI for a surfaced (2 m depth) and submerged (10 m and more) source.
Chapter3. The SI algorithm.
sound speed [mis]
Figure 3.2: Sound speed profile used in simulations
(a) SI for all source/receiver depth combinations
(b) Averaged over the receiver depth SI
Figure 3.3: SI results for 300Hz source, 32-element array, cb = 1575 mn/s, bearing 0*, taken from
[5]).
Chapter3. The SI algorithm.
Figure 3.4: SI results for 300Hz source, 32-element array,
[5]).
Submw~m
kubx-
Cb =
1625 m/s, bearing 0*, taken from
ftamp da
c
VA"MDp
--
10k
/:10d
ri50d
T"W
D91hi
(a) SI for all source/receiver depth combinations
Figure 3.5: SI results for 300Hz source, 32-element array,
[5]).
TwW~~ht/
(b) Averaged over the receiver depth SI
Cb =
1612 m/s, bearing 50, taken from
..........
.............
Chapter3. The SI algorithm.
sammn
n.
tr
ma-
(a) SI for all source/receiver depth combinations
Figure 3.6: SI results for 300Hz source, 32-element array,
[5]).
DepO4-Avge
Submrgence
kKndx
- sme
-f
-z
30.
(b) Averaged over the receiver depth SI
cb =
1612 m/s, bearing 15*, taken from
N-0d
T.W
DP*'
(a) SI for all source/receiver depth combinations
Figure 3.7: SI results for 300Hz source, 32-element array,
[5]).
T~l Dp, (.
(b) Averaged over the receiver depth SI
cb =
1612 m/s, bearing 350, taken from
Chapter3. The SI algorithm.
120
d8
WdI
(a) SI for all source/receiver depth combinations
Figure 3.8: SI results for 300Hz source, 32-element array,
[5]).
(b) Averaged over the receiver depth SI
Cb =
1612 m/s, bearing 550, taken from
Let us summaries the main findings of [4] for a range-independent environment (Monterey Bay
bathytermic conditions):
1. The SI algorithm is sufficiently accurate for bearing angles up to 200.
2. The separation between surfaced and submerged source increases as the bottom sound
speed increases; however, small oscillations in the bottom sound speed will degrade the
SI separation ability.
3. The SI algorithm performance has strong dependence on the range between the source
and the receiver. An addition of the range averaging process to the SI averaging
algorithm can significantly improve its robustness.
4. A diving angle of 150 was found to be an optimal range-depth averaging technique for a
horizontally oriented array.
4
The SI algorithm evaluation in realistic AUV scenarios
4.1
Methodology
As described in the Introduction section above, the main purpose of the current research was to
evaluate the SI performance in realistic AUV scenarios. This section describes the methods and
tools that were used in this evaluation.
4.1.1
Simulation tools
In general, the evaluation was performed according the scheme shown on Figure 3.1. The main
difference between the current evaluation process and the one described in Section 3.2 is in
setting the receiver array position. In the original evaluation process, the array was considered to
be horizontally oriented (pitch angle 00) and its coordinates were synthetically generated. In
contrary, here, we used realistic towed array geometry.
The towing path and the array elements positions were simulated using the MOOS-IvP
simulation package [13]. The MOOS-IvP package was originally developed as an integrated
command and control software for multiple AUVs operations. In addition, the software can be
used as a simulation tool to simulate an AUV behavior. Here, we used the MOOS-IvP package to
simulate an AUV and its towed array movement. During the run, the simulation stores the x, y
and z coordinates of the AUV and each element of its towed array in an external file. The SI
calculations were performed using this stored data. Thus, after simulating the towed path and
elements positions for a particular towing pattern, the SI performance was evaluated for a
particular mutual source/receiver geometrical configuration in the same way it was described
earlier in Section 3.2.
Chapter 4. The SI algorithm evaluation in realisticA UV scenarios
4.1.2
Towing patterns
The main consideration of the current research was given to the towing pattern and its
parameters. Two major towing patterns were considered - a yo-yo path (see also Section 3.1)
and a one-time, either downward or upward depth sampling.
Because of a limitation of currently available MOOS-IvP tools, the yo-yo path was simulated
using step-wise depth changing maneuver, where the AUV is commanded to change its depth in
small increments, creating a relatively smooth sinusoidal pattern, see Figure 4.1. More detailed
description of the array geometry at each step is presented on Table 4.1.
AUV track (x-z plane)
0
-10
-20
-30
'
-40
-50
-60
-70
-800
2800
3000
3200
3400
3600
3800
x [m]
Figure 4.1: A typical yo-yo towing pattern in x-z plane.
4000
4200
4400
4600
Chapter4. The SI algorithm evaluation in realisticA UV scenarios
Table 4.1: The array geometry at each step of the yo-yo towing pattern. The green dot represents
the first element of the array. The x-axis shows the distance from the towing AUV.
Characteristic
i
dhe
depth
Upward direction
Downward direction
10 -20
9
22
020
24
20 m
29
21
20
25
30
40
35
5
45
20
25
30
35
40
45
50
22
300m
31
--
3 -
34
40 m
3
40
44"
32
15
x41
50
15
2
54
4435
Chapter4. The SI algorithm evaluation in realisticA UV scenarios
Characteristic
dhe
i
depth
Upward direction
Downward direction
504m
52
54
-45
54
60m
60 m
45
S2
I5
2
25
30
35
40
45
5s
20
25
30
35
40
45
55
55
Chapter 4. The SI algorithm evaluation in realisticA UV scenarios
The one-time depth sampling pattern was simulated using an assumption of a linear distribution
of elements in space during the towing with constant speed, pitch and heading. Figure 4.2 shows
a schematic description of the array and its elements during the towing maneuver. The position
of the array elements is calculated using the following formulas:
xi,=x, -l cosa
zi+1
4.1
,+l
1, sina
where li is the distance between element i and element i+ 1.
x
(xi, zi)
a, the pitch angle
(X2, Z2)
(X32, Z32)
zI
Figure 4.2: A schematic description of the towed array during the one-time depth sampling
towing pattern
In order to validate this towed array model two typical runs were performed using the MOOSIvP simulation package. In the first run the resulted pitch angle of the array was 19 degrees and
in the second - 32 degrees. In both runs the AUV performed a diving maneuver, but in opposite
bearings. Figure 4.3 presents the position of the array elements for both runs.
I.....................
..............................
Chapter 4. The SI algorithm evaluation in realisticA UV scenarios
30
Typica
vrW co4lgwblo *AV' dying mfwuv, pitch I deg
82
84
-S
-. 5
-Q
.3
*
88)
30
Pic*nl
Tyia
.2
-0
1
x
wco*
1ere
rto
-n
yn
ic
2e
30*
31*
32
-
33
-
34*
35836
-
37
70
-
8
1012
1
16
s
20
2
22
xn
(b)
Figre
gren
towin
.3
araypoitin
Tyica
fr
dt thrpreentfrsteleentof
he
AUV
dgree
Ptch ngle32
iffret
ptchanlesdurngfre
rry.
he
dvin
axs
rpreens
te
maeuer.Th
dstace
romth
Chapter 4. The SI algorithm evaluation in realisticA UV scenarios
4.1.3
Range dependence analysis
As it was shown in the previous research [5], the SI algorithm has relatively high range
dependence, which can be minimized using range averaging in addition to depth averaging. The
purpose of this section is to describe the method that was used to evaluate the robustness of the
SI algorithm when used in a free-towing AUV scenario.
For this purpose, the SI performance was calculated at the same environmental and towing
settings for different source ranges: 5 km, 7 km, 10 km and 12 km. In all runs a diving maneuver
with pitch angle 2o-35* (see Section 4.1.2) was used.
31
Chapter4. The SI algorithm evaluation in realisticA UV scenarios
4.2
4.2.1
Results
The yo-yo towing pattern
As described above in Section 4.1.2, the yo-yo towing pattern consists of two major parts - the
diving and the climbing maneuvers. The following figures present the SI results for both
downward and upward directions of movement for different bearing angles. All runs were
performed using the same environmental data as presented in Section 3.2.
Each of the figures consists of four parts: part (a) presents the SI results for all source/receiver
depth combinations during the diving maneuver; part (b) presents the SI results for all
source/receiver depth combinations during the climbing maneuver; part (c) presents the averaged
over receiver depth SI results for the diving maneuver; and finally, part (d) presents the averaged
over receiver depth SI results for the climbing maneuver. Figure 4.4 up to Figure 4.8 present the
SI results for bearing angles 00 up to 200, while Figure 4.9 up to Figure 4.13 present the SI
results for bearing angles 1800 down to 1600.
The opposite nature of the diving and climbing parts of the yo-yo towing pattern can be easily
seen on the figures. The diving part and the climbing part have similar performance when the
source is located at the opposite end-fire direction. For example, examine Figure 4.4 and Figure
4.9. Notice the 5-6dB separation in the SI between surfaced and submerged sources in
Figure 4.4(c) and compare it to 4-6dB separation in Figure 4.9(d). Although the SI performance
for sources located at 180* relative to the receiving array is weaker than for sources located at 00,
there is a reasonable similarity between these two scenarios. More detail analysis of this
phenomenon is presented in Section 5.1 below.
Chapter 4. The SI algorithm evaluationin realisticA UV scenarios
Submergence index - summerMBHF3
Submergence Index - summerMBHF3
~so
045
U
W 40
20
10
40
30
50
60
70
80
20
10
90
Depth-Range Averaged Submergence Index - summerMBHF3
10
20
30
40
50
50
60
70
80
90
Target Depth (m)
(a) Diving SI for all source/receiver depth
combinations
0
40
30
Target Depth (m)
s0
70
80
90
Target Depth (m)
(c) Diving, averaged over the receiver depth SI
(b) Climbing SI for all source/receiver depth
combinations
Depth-Range Averaged Submergence Index - summerMBHF3
0
10
20
30
40
50
60
70
80
90
Target Depth (m)
(d) Climbing, averaged over the receiver depth SI
Figure 4.4: The yo-yo towing pattern, diving vs. climbing SI results for bearing 00.
Chapter 4. The SI algorithm evaluation in realisticA UV scenarios
3
Submergence Index - summer BHF
0
10
20
30
40
70
60
50
Submergence Index - summerBHF3
80
0
10
20
Depth-Range Averaged Submergence Index - summermBHF3
10
20
30
40
50
so
40
70
s0
0
(b) Climbing SI for all source/receiver depth
combinations
(a) Diving SI for all source/receiver depth
combinations
0
30
Target Depth (m)
Target Depth (m)
80
70
80
90
Target Depth (m)
(c) Diving, averaged over the receiver depth SI
Depth-Range Averaged Submergence Index - summerMBHF3
0
10
20
30
40
50
0
70
80
90
Target Depth (m)
(d) Climbing, averaged over the receiver depth SI
Figure 4.5: The yo-yo towing pattern, diving vs. climbing SI results for bearing 5*.
Chapter4. The SI algorithm evaluation in realisticA UV scenarios
Submergence Index - summerBHF3
Submergence Index - summerBHF3
0
10
20
30
50
40
70
60
0
80
10
20
(a) Diving SI for all source/receiver depth
combinations
Depth-Range Averaged Submergence Index - summermBHF3
0
10
20
30
40
so
30
40
60
so
70
80
Target Depth (m)
Target Depth (m)
so
70
80
90
Target Depth (m)
(c) Diving, averaged over the receiver depth SI
(b) Climbing SI for all source/receiver depth
combinations
Depth-Range Averaged Submergence Index - summerMBHF3
0
10
20
30
40
50
so
70
80
90
Target Depth (m)
(d) Climbing, averaged over the receiver depth SI
Figure 4.6: The yo-yo towing pattern, diving vs. climbing SI results for bearing 100.
own
Chapter 4. The SI algorithmevaluation in realisticA UV scenarios
Submergence Index - summerMBHF3
Submergence Index - summermBHF3
707
707
Go
5
5
E4
o
4)4
2
0 40
202
E-
cc
30
0
-1
-1
20
0
10
20
30
so
40
60
70
-2
60
0
10
20
30
40
so
so
70
-2
so
Target Depth (m)
Target Depth (m)
(b) Climbing SI for all source/receiver depth
combinations
(a) Diving SI for all source/receiver depth
combinations
Depth-Range Averaged Submergence Index - summermBHF3
Depth-Range Averaged Submergence Index - summerMBHF3
0.3
f - 300 HAz
sl -120 dB
nI - 50 dB
- J0 km
M
S2 -init
Hz
r5 km
1-15
SI 120 B
nl 50 d
it- 10
02 -f
- 9,,95 km
2Sr
0.1
-
~r
0
-45
C
4)4
C
U -0.1
C
E
4-02
-0
0.5 -
0
-0A4
to
20
30
40
so
so
70
80
90
Target Depth (m)
(c) Diving, averaged over the receiver depth SI
-0.5
0
10
20
30
40
so
60
70
90
(d)Climbing, averaged over the receiver depth SI
Figure 4.7: The yo-yo towing pattern, diving vs. climbing SI results for 300Hz source, 32element array, Cb = 1612 m/s, bearing 15*.
80
Target Depth (m)
Chapter 4. The SI algorithm evaluation in realisticA UV scenarios
Submergence Index - summermBHFS
Submergence Index - summerMBHF3
7
7
0
4
o
70
2
0
60
0
20
0
10
20
30
40
50
70
s0
0
80
10
20
Target Depth (m)
30
50
40
80
-2
80
70
Target Depth (m)
(b) Climbing SI for all source/receiver depth
combinations
(a) Diving SI for all source/receiver depth
combinations
Depth-Range Averaged Submergence Index - summerMBHF3
Depth-Range Averaged Submergence index - summerMBHF3
n
f - 300 Hz
r - 9.895 km
8- 20
sI - 120 dB
ni - 50 dB
r init -
1
1
-0
a
10 km
f - 300 Hz
r - 9.85 km
8 = 20
s - 120 dB
ni - So dB
r init - 10 km
-
-02
45
545
3
-0
CM
4
E?
0-1
5
-0
(00
-0..7
0
10
20
30
40
50
s0
70
80
90
Target Depth (m)
(c) Diving, averaged over the receiver depth SI
10
20
30
40
50
60
70
80
90
Target Depth (m)
d) Climbing, averaged over the receiver depth SI
Figure 4.8: The yo-yo towing pattern, diving vs. climbing SI results for bearing 200.
.................
Chapter4. The SI algorithm evaluation in realisticA UV scenarios
Submergence Index - summerMBHF3
Submergence Index - summerMBHF3
8
0
0
6
05
05
4
0
3
2
2
cc
1
0
0
-1
E-2
-2
0
10
20
30
40
60
50
70
0
80
10
20
30
40
50
60
70
80
Target Depth (m)
Target Depth (m)
(b) Climbing SI for all source/receiver depth
combinations
(a) Diving SI for all source/receiver depth
combinations
Depth-Range Averaged Submergence Index - summerMBHF3
Depth-Range Averaged Submergence Index - summermBHF3
0.3
02
M
-o
S0
C
02
-01
.
f -02
-0.3
0
10
20
30
40
50
60
70
80
90
Target Depth (m)
(c) Diving, averaged over the receiver depth SI
0
10
20
30
40
50
60
70
80
90
Target Depth (m)
(d) Climbing, averaged over the receiver depth SI
Figure 4.9: The yo-yo towing pattern, diving vs. climbing SI results for bearing 1800.
Chapter 4. The SI algorithm evaluation in realisticA UV scenarios
Submergence Index - summermBHF3
Submergence Index - summermBHF3
so
50
cc
U
10
zu
:Wu
4U
u
W
u
0
Ou
10
20
30
40
60
50
70
0
Target Depth (m)
Target Depth (m)
(b) Climbing SI for all source/receiver depth
combinations
(a) Diving SI for all source/receiver depth
combinations
Depth-Range Averaged Submergence index - summerMBHF3
Depth-Range Averaged Submergence Index - summermBHF3
f - 300 Hz
r - 9.895 km
0.7 .
8 - 175
si - 120 dB
0-
ni - 50 dB
r init - 10 km
63.5
3
S2.5
0.2
C
cm
E
.0
U)
02-
0.1
0
0
0.5
10
20
30
40
so
60
70
80
90
Target Depth (m)
(c) Diving, averaged over the receiver depth SI
0/
0
10
20
30
40
so
60
70
0
90
Target Depth (m)
(d) Climbing, averaged over the receiver depth SI
Figure 4.10: The yo-yo towing pattern, diving vs. climbing SI results for bearing 175*.
Chapter 4. The SI algorithm evaluation in realisticA UV scenarios
Submergence Index - summerBHF3
Submergence Index - summermBHF3
0
10
20
30
40
50
70
60
0
80
10
20
30
40
60
50
70
80
Target Depth (m)
Target Depth (m)
(b) Climbing SI for all source/receiver depth
combinations
(a) Diving SI for all source/receiver depth
combinations
Depth-Range Averaged Submergence Index - summermBHF3
0.45
Depth-Range Averaged Submergence Index - summerMBHF3
r - .95 km
-
0.4
si - 1 dB3
:ar
8
-50
n0.3
Init-
}km
70
80
C
8
025 0.2
E
S0.150.1 0.05
0
0
10
20
30
40
50
60
90
Target Depth (m)
(c) Diving, averaged over the receiver depth SI
0
10
20
30
40
50
80
70
80
90
Target Depth (m)
(d) Climbing, averaged over the receiver depth SI
Figure 4.11: The yo-yo towing pattern, diving vs. climbing SI results for bearing 1700.
Chapter4. The SI algorithm evaluation in realisticA UV scenarios
Submergence Index - summermBHF3
Submergence Index - summerBHF3
8
8
7
7
6
so
S
5
4
E
40
s
63
2
2
-1
20
0
0
10
20
30
40
so
so
70
20
-2
80
0
Target Depth (m)
E
to
20
63
30
40
50
60
70
-2
so
Target Depth (m)
(a) Diving SI for all source/receiver depth
combinations
(b) Climbing SI for all source/receiver depth
combinations
Depth-Range Averaged Submergence Index - summermBHF3
Depth-Range Averaged Submergence Index - summerMBHF3
toc
fHz
r 9. 91 km
3
0
-1
si- 1 dB
2.5
ni - 5 dB
r init - 0 km
a-45
E3
LM 2
-
W
C
E
-o
E
-0.15 1-
-02'
0
10
20
30
40
so
so
70
so
90
Target Depth (m)
(c) Diving, averaged over the receiver depth SI
0
10
20
30
40
so
so
70
so
Mo
Target Depth (m)
(d) Climbing, averaged over the receiver depth SI
Figure 4.12: The yo-yo towing pattern, diving vs. climbing SI results for bearing 1650.
..
....
. ..........
ONO
Chapter4. The SI algorithm evaluation in realisticA UV scenarios
Submergence Index - summerMBHF3
a
1U
20
30
4
3U
WU
70
Submergence Index - summerMBHF3
U
U
10
2u
Target Depth (m)
U
7U
OJ
(b) Climbing SI for all source/receiver depth
combinations
(a) Diving SI for all source/receiver depth
combinations
Depth-Range Averaged Submergence Index - summerMBHF3
0
4U
3u
Target Depth (m)
3U
Depth-Range Averaged Submergence Index - summerMBHF3
2S.
f - 300 Hz
r - 8.895 km
8- 160
s - 120 dB
ni - 50dB
r 1 21t0 km
-0.1
~-02
4
Is
C -0.3
0.
1-0.4
E
-Oz
-0.7
0
L0
20
40
L
S0
-L
S0
70
80
90
Target Depth (m)
(c) Diving, averaged over the receiver depth SI
-0.5'
0
10
20
30
40
so
60
70
o
90
Target Depth (m)
(d) Climbing, averaged over the receiver depth SI
Figure 4.13: The yo-yo towing pattern, diving vs. climbing SI results for bearing 1600.
Chapter 4. The SI algorithm evaluation in realisticA UV scenarios
One-way depth sampling
4.2.2
As mentioned above (Section 4.2.2), the SI performance for one-way depth sampling pattern was
analyzed in the same matter as the yo-yo pattern, except that the position of the receiving array
elements was calculated using the linear elements distribution.
The following figures present a comparison between diving and climbing SI performance for
different pitch angles during the one-way towing pattern. For the conciseness of presentation,
only the averaged SI results are presented. In addition, due to the symmetry in SI performance
for upward and downward array orientation, which is described in Section 5.1, the following
figures correspond to the opposite end-fire orientation for diving (00) and climbing (1800)
maneuvers.
Diepth-Range Averaged Submergence Index - summerMBHF
3
3
Depth-Range Averaged Submergence Index - summer BHF
- 300 Hz
3.5
nI =0dB
/1
2.5
r init -,10km
-
EF
(U)
0E
0
10
20
30
40
50
Target Depth (m)
(a) Diving
g0
70
80
10
20
30
40
50
60
70
80
Target Depth (m)
(b) Climbing
Figure 4.14 The one-way towing pattern, diving vs. climbing averaged SI results for pitch ±2'.
90
Chapter 4. The SI algorithm evaluation in realisticA UV scenarios
Depth-Range Averaged Submergence Index - summer BHF
3
Depth-Range Averaged Submergence Index - summer BHF3
f = 300 Hz
r - 9.076 km
8 0
3.5
sl
U
120 dB/
n/- 50 d
3
r init-
fdkm
2.5
\V/
C
O'S
0
10
20
30
40
50
60
70
80
40
90
50
60
Target Depth (m)
Target Depth (m)
(b) Climbing
(a) Diving
Figure 4.15 The one-way towing pattern, diving vs. climbing averaged SI results for pitch
Depth-Range Averaged Submergence Index - summerMBHF3
0
10
20
30
40
50
Target Depth (m)
(a) Diving
60
70
80
90
±40.
Depth-Range Averaged Submergence Index - summerMBHF3
0
10
20
30
40
50
60
Target Depth (m)
70
80
(b) Climbing
Figure 4.16 The one-way towing pattern, diving vs. climbing averaged SI results for pitch ±6'.
90
Chapter4. The SI algorithm evaluation in realisticA UV scenarios
Depth-Range Averaged Submergence Index - summer BHF3
Depth-Range Averaged Submergence Index - summerMBHF3
3.s
x
a, 2
E
115
0
10
20
30
40
50
60
70
80
40
Target Depth (m)
50
60
Target Depth (m)
(a) Diving
(b) Climbing
Figure 4.17 The one-way towing pattern, diving vs. climbing averaged SI results for pitch ±8'.
Depth-Range Averaged Submergence Index - summerMBHF3
Depth-Range Averaged Submergence Index - summerMBHF3
f -10 Hz
r - 9.5!8 km
3
nI- 50 dB
m2.5
\
r init - 10 km
~-10
S2
15
a .
40
50
60
0
Target Depth (m)
0
4
/~
05
10
20
30
40
50
50
6
60
70
80
Target Depth (m)
(a) Diving
(b) Climbing
Figure 4.18 The one-way towing pattern, diving vs. climbing averaged SI results for pitch ±100.
45
90
Chapter4. The SI algorithmevaluation in realisticA UV scenarios
Depth-Range Averaged Submergence Index - summerMBHF3
Depth-Range Averaged Submergence Index - summerMBHF3
4.5
6 0
~SI
3.5
dB
nI - 0 dB
rinit-1pkm
-sI120
.1OtHz
r=
km
a- 180 \
SI- 12O0
nI - SodB \
r init = 10 km
4
r - 9.p52km
o(
m~=-12
~=12
3
CU
S2.5
/
3-
2
E
15
2.
0
10
20
30
40
so
60
70
80
90
0
10
20
30
40
50
60
70
80
90
Target Depth (m)
Target Depth (m)
(a) Diving
(b) Climbing
Figure 4.19 The one-way towing pattern, diving vs. climbing averaged SI results for pitch ±12*.
Depth-Range Averaged Submergence Index - summer BHF3
Depth-Range Averaged Submergence Index - summer BHF3
r4km
sI- 12
ni - 50
rinit - 10 m
.
4
-dB
cc- 14
2 -
0
10
20
30
40
50
Target Depth (m)
(a) Diving
60
70
s0
90
0
10
20
30
40
50
60
70
80
Target Depth (m)
(b) Climbing
Figure 4.20 The one-way towing pattern, diving vs. climbing averaged SI results for pitch ±140.
90
Chapter 4. The SI algorithm evaluation in realisticA UV scenarios
Depth-Range Averaged Submergence Index - summerMBHF3
Depth-Range Averaged Submergence Index - summer BHF3
4,5
fHz
r 9.7 m
4
-= 180 \
1A- 120 d
-
50 dB
//-
r init= 0 km
3
g-16
-c
S2.5
CU C
,1.5
-
1
0.5
0
0
10
20
30
50
40
60
70
80
90
0
10
20
30
40
50
60
70
80
Target Depth (m)
Target Depth (m)
(a) Diving
(b) Climbing
Figure 4.21 The one-way towing pattern, diving vs. climbing averaged SI results for pitch ±16*.
Depth-Range Averaged Submergence Index - summer BHF3
Depth-Range Averaged Submergence Index - summer BHF3
6
r 9.754ign
5 -
/
/
=-
8- 0
3<
ni = 50 d
180
sl = 120 dB
ni = 50 dB
r init - 10 km
M
si - 120 B
-/
4
f 0 Hz
- 924km
18
r init - 1Pk
~--18
.S
01
2-
2s
.05
0,5
0
0
10
20
30
40
50
Target Depth (m)
(a) Diving
60
70
80
90
0
10
20
30
40
50
60
70
80
Target Depth (m)
(b) Climbing
Figure 4.22 The one-way towing pattern, diving vs. climbing averaged SI results for pitch ± 180.
90
Chapter 4. The SI algorithmevaluation in realisticA UV scenarios
Depth-Range Averaged Submergence Index - summerMBHF3
Depth-Range Averaged Submergence Index - summerMBHF3
4~.
f - 0sHz
r- 9.77 km
3.5
/0- 180\
3
s - 120 d\
ni - 50 dB
r init = 10 km
2.5
a-20
C
(3
0.5
n,
0
10
20
30
Target Depth (m)
40
50
j
i
i
60
70
80
Target Depth (m)
(a) Diving
(b) Climbing
Figure 4.23 The one-way towing pattern, diving vs. climbing averaged SI results for pitch ±20'.
Depth-Range Averaged Submergence Index - summer BHF3
Depth-Range Averaged Submergence Index - summerMBHF3
03 Hz
r
9.8N km
0-180 \
3,5
"(U
3
sIO120
ni= 50 dB
rinit- 10km
c =25
2.5
U(X
CV
CM
'O 2.5
C
CU
(3
:/3
CMs
10
20
30
40
50
Target Depth (m)
(a) Diving
60
70
80
90
0
10
20
30
40
50
Target Depth (m)
60
70
80
(b) Climbing
Figure 4.24 The one-way towing pattern, diving vs. climbing averaged SI results for pitch ±254.
90
Chapter4. The SI algorithm evaluation in realisticA UV scenarios
Depth-Range Averaged Submergence Index - summerMBHF3
Depth-Range Averaged Submergence Index - summer BHF3
4
3.5
f=3 Hz
r -9.8g km
6 -180 \
si - 120
nI - 50 dB
3.5
-3
3
r init = 10 km
X
(U
= 30
S2.5
2.5
2
CU 2
cu
M 1.5
-
1
('3
0.5
0
10
20
30
40
50
60
70
80
90
-
-
-
-
10
20
30
Target Depth (m)
40
50
60
70
80
90
Target Depth (m)
(a) Diving
(b) Climbing
Figure 4.25 The one-way towing pattern, diving vs. climbing averaged SI results for pitch ±300.
Depth-Range Averaged Submergence Index - summerMBHF3
4
.5 ,
1
I
Depth-Range Averaged Submergence Index - summerMBHF3
I
I"
4.5
30 Hz
r - 9.85km
8 180 \
3.5
3.5
sI- 120 dS
nI- 50 dB \
r init - 10 km
3
x
(J
S 2.5
(5
_ 2.5
CU
CU 2
2
1.5
0
10
20
30
40
50
60
Target Depth (m)
(a) Diving
70
80
90
01
0
10
20
30
40
s0
60
70
(b) Climbing
Figure 4.26: The one-way towing pattern, diving vs. climbing averaged SI results for pitch
The following figures present the averaged SI results for the MOOS-IvP simulated diving
patterns that were presented in Section 4.1.2. Notice a modest improvement of the SI
performance in the 19*-run (
so
Target Depth (m)
+350.
90
Chapter 4. The SI algorithm evaluation in realisticA UV scenarios
Figure 4.27) relative to 180 -run presented on Figure 4.22 and almost similar performance of the
320 -run (Figure 4.28) in comparison with 30'-run presented on Figure 4.25.
Depth-Range Averaged Submergence Index - summerMBHF3
0
10
20
30
40
50
60
70
80
90
Target Depth (m)
Figure 4.27: Averaged SI results for MOOS-IvP simulated diving towing pattern, characteristic
pitch 19*.
Depth-Range Averaged Submergence Index - summerMBHF3
3.5
3
f= 0 Hz
=0
s- 120
rn2.5
zkm
50 dB
=nI
70
r init = 19 k
cc -32
2
(U
0.)
E
-D
V)
i
n
0
10
20
30
40
50
70
80
90
Target Depth (m)
Figure 4.28: Averaged SI results for MOOS-IvP simulated diving towing pattern, characteristic
pitch 320.
Chapter4. The SI algorithm evaluation in realisticA UV scenarios
4.2.3
Range dependence
The following figures present the SI performance for different source ranges: 5 km, 7 km, 10 km
and 12 km. The results for 10 km scenario are identical to results presented in Section 4.2.2 and
they presented here for reference purposes only.
Depth-Range Averaged Submergence Index - summerMBHF3
Depth-Range Averaged Submergence Index - summer BHF3
3.5
4
35
f= 300 H
r-3l 1km
3
20dB
O
n1
=50dB
m25
-rinit
2
-
3 -SI
M
5
-/
0
10
20
30
40
50
l
/
0I
aU 1.5
V
V)
-120
km
ni =50dB
r init =7km
25
km
g15-
T3Hz
r=5.2
--
En
60
70
80
90
0
t0
20
30
40
50
Target Depth (m)
Target Depth (m)
(a) R = 5 km
(b) R
Depth-Range Averaged Submergence Index - summerMBHF3
=
60
70
80
7 km
Depth-Range Averaged Submergence Index - summer BHF3
f - 300 H
r- 1 28 km
sl = 120 dB
Ni = S0 dB
r init=-12 km
25
(U 2
/
-
CD.
0.5
0
10
20
30
40
50
60
Target Depth (m)
(c) R = 10 km
70
80
90
0
10
20
30
40
50
60
70
Target Depth (m)
(d) R = 12 km
Figure 4.29: Averaged SI results for diving towing pattern for different initial source ranges,
pitch 20.
90
80
Chapter 4. The SI algorithm evaluation in realisticA UV scenarios
Depth-Range Averaged Submergence Index - summerMBHF3
f - 300 Hz
r'= 4.076 km
8 0
si -
35
3
r = 6.076
3
OdB
km
20 d8
nI= 50,B
m2.5
r init - 5 km
\|A
Depth-Range Averaged Submergence Index - summerMBHF3
r init - 7kMn
=-4
.
S
2.5
C2
@2
gis
L)
5
-
0.5
0.5
0
'
0
/1
/
10
20
30
40
50
60
70
F
'
80
90
0
10
20
30
Target Depth (m)
40
50
70
s0
80
90
Target Depth (m)
(a) R=5km
(b) R = 7 km
)epth-Range Averaged Submergence Index - summerMBHF3
Depth-Range Averaged Submergence Index - summer MBHF3
45
4
f - 300 Hz
r - 11.076 km
3.5
3.5
3
/
= 0
S 120 d8
ni 50dB
3/X2.5
2
rn
.S
2 2
1
0.5
0.5
0
0
0
10
20
30
40
50
60
Target Depth (m)
(c) R = 10 km
70
80
90
0
10
20
30
40
50
60
70
Target Depth (m)
(d) R = 12 km
Figure 4.30: Averaged SI results for diving towing pattern for different initial source ranges,
pitch 40.
80
90
Chapter4. The SI algorithm evaluation in realisticA UV scenarios
Depth-Range Averaged Submergence index - summer BHF3
Depth-Range Averaged Submergence Index - summer BHF3
- 300 Hz
6864 km
f- 4 0
r
8
0
6 4 km
S-. 120 dB
3
1 - 50 dB
rrinit =p km
/=--6
en 25
CD
C:
CU
r init - 7 km
/
.S 2.5
Cs
-/-
sI120B'
0
ni=5 rdB
1'
-
/
-I
.01
i/I
0
20
10
30
40
50
70
60
80
-j
90
0
0
10
20
30
Target Depth (m)
40
50
60
70
60
(a) R = 5 km
(b) R= 7 km
Depth-Range Averaged Submergence Index - summerMBHF3
Depth-Range Averaged Submergence Index - summerMBHF3
4
300 Hz
3.5 -f=
r -9364 km
e 10
- 120 dB
-
3
M
i
(
2
1(.5
I
~r
- 300 Hz</
r
64
km
si - 120 dB
*5 0 dJ/
init,-,T km
X
nI = 50 dB
rinit= 12km
-3
-6
2.5
2
(U
E
CU
03
0.5
05
0
0
0
10
20
30
40
50
60
Target Depth (m)
(c) R = 10 km
70
so
90
0
6
10
20
30
40
50
60
70
Target Depth (m)
(d) R = 12 km
Figure 4.31: Averaged SI results for diving towing pattern for different initial source ranges,
pitch 60.
90
Target Depth (m)
80
Chapter 4. The SI algorithm evaluation in realisticA UV scenarios
Depth-Range Averaged Submergence Index - summer BHF3
Depth-Range Averaged Submergence Index - summerMBHF3
6
f= 300 H7
5
r - 6.S0km
sI -Y20 dB
nI = 50 dB
4
r init -
7
km
3
2
0
10
20
30
40
50
60
70
80
90
0
20
10
30
40
50
60
70
80
(b) R = 7 km
Depth-Range Averaged Submergence Index - summerMBHF3
(a) R=5km
Depth-Range Averaged Submergence Index - summerMBHF3
5
f - 300 Hz
45
4 5
r- 9,508 km
4
sl- 120dB
nI = 50 dB
r init - 10 km
3.5
4-5
/
4
-\ 300 Hz
J.508 km
nI t 50 dB
rinit= 12km
3
03
C
/
-
'O 2.5
'0.5
C
C
2
1(5
r
sI - 120 dB
M3.5
X
E"
| |
/
|
|
-
S1.5
/
-
0.5
05
10
20
30
40
5;
60
Target Depth (m)
(c) R = 10 km
70
80
1
0 0
0
0
10
20
30
40
5
6
70
80
20
30
40
50
60
70
80
Target Depth (m)
(d) R = 12 km
Figure 4.32: Averaged SI results for diving towing pattern for different initial source ranges,
pitch 80.
90
Target Depth (m)
Target Depth (m)
Chapter4. The SI algorithm evaluation in realisticA UV scenarios
Depth-Range Averaged Submergence Index - summerMBHF3
Depth-Range Averaged Submergence Index - summerMBHF3
45
a r o,9
4
f = 300 -jz'
/
-6598 km
km
sl - 120 dB
8
SI- 0120 dB
- 50 dB
ni - 50 dB
r init = 7 km
!nit = 5,km
x
o--10
/
3
25
/
CM .
/1
0.5
10
20
30
40
50
60
70
80
0
10
20
30
40
so
60
70
80
90
Target Depth (m)
Target Depth (m)
(b) R= 7 km
(a) R=5km
Depth-Range Averaged Submergence Index - summerMBHF3
Depth-Range Averaged Submergence Index - summer BHF3
f '30 r -z
-: 11.598 krn
80
s - 120dB
ni - 50 dB
rinit501d km
M--10
-3
'C
S2.5
ci2
E
.
1
ZI5
0
Target Depth (m)
(c) R = 10 km
10
20
30
40
50
60
70
Target Depth (m)
(d) R = 12 km
Figure 4.33: Averaged SI results for diving towing pattern for different initial source ranges,
pitch 100.
80
90
Chapter 4. The SI algorithm evaluation in realisticA UV scenarios
Depth-Range Averaged Submergence Index - summer BHF3
Depth-Range Averaged Submergence Index - summerMBHF3
f 00Hz
r A 4552 km
X- 0
s - 120 dB
35 -3
n1= 50 d8
/
r init - 5,km
a--12
E2.5
C
(U 2
C,)
05
7
0
/
-
10
20
-
-
30
40
50
60
70
0
80
t0
20
30
40
50
60
70
80
90
Target Depth (m)
Target Depth (m)
(b) R= 7 km
(a) R=5km
Depth-Range Averaged Submergence index - summerMBHF3
Depth-Range Averaged Submergence Index - summerMBHF3
f' 300Hkiz
4
r = 11.654m
//8-
X
00
sI - 120 dB
ni = 50 dB
3.5
rinit- 12km
= 12
3
25
W 2
M
0.5
0
Target Depth (m)
(c) R = 10 km
10
20
30
40
50
60
70
Target Depth (m)
(d) R = 12 km
Figure 4.34: Averaged SI results for diving towing pattern for different initial source ranges,
pitch 120.
80
90
Chapter4. The SI algorithm evaluation in realisticA UV scenarios
Depth-Range Averaged Submergence Index - summer BHF3
4Z
I
II
20
30
Depth-Range Averaged Submergence Index - summer BHF 3
"3
w2
a/)
05
0
0
10
40
SO
60
70
0
80
10
20
30
40
50
60
70
80
90
Target Depth (m)
Target Depth (m)
(b) R = 7 km
(a) R=5km
Depth-Range Averaged Submergence Index - summerMBHF3
Depth-Range Averaged Submergence Index - summerMBHF3
9 Km
8 0
sl = 120 dB
nI SodB
r init - ip km
cxr--14
r -9.
0
10
20
30
40
50
60
Target Depth (m)
(c) R = 10 km
70
80
90
0
10
20
30
40
50
60
70
Target Depth (m)
(d) R = 12 km
Figure 4.35: Averaged SI results for diving towing pattern for different initial source ranges,
pitch 14'.
80
90
Chapter 4. The SI algorithmevaluation in realisticA UV scenarios
Depth-Range Averaged Submergence Index - summerMBHF3
Depth-Range Averaged Submergence Index - summerMBHF3
5
4.5
f - 30
3.5
6. 0
s= 120 dB
= dB
-50
\
r init = 5,km
o -- 16
3
CM2
(U
E
1.5
0 /
0.5
0
2
0
10
20
30
4
5
40
50
0
7
0
70
80
9
0
0
60
0
90
20
10
30
Target Depth (m)
40
50
60
70
80
90
Target Depth (m)
(a) R = 5 km
(b) R = 7 km
Depth-Range Averaged Submergence Index - summerMBHF3
Depth-Range Averaged Submergence Index - summerMBHF3
6
f = 300 Hz
r - 11.73 km
30 Hz
5 -r
=9.
8
m
-
a = 0*
sI - 120 dB
ni = 50 dB
0 7
s/ = 120 d
nI - 50 dB \
4
3
/
r init - 12 km
r init - 10 km
1-16
Cx 2 5
/
W 2
,1
r
a
-I
0.5
nI
0
10
20
30
40
50
s0
Target Depth (m)
(c) R = 10 km
70
80
90
0
0
10
20
30
40
50
60
70
Target Depth (m)
(d) R = 12 km
Figure 4.36: Averaged SI results for diving towing pattern for different initial source ranges,
pitch 16'.
80
90
Chapter 4. The SI algorithm evaluation in realisticA UV scenarios
Depth-Range Averaged Submergence Index - summerMBHF3
Depth-Range Averaged Submergence Index - summer BHF3
5
f -300fiW
4.5
A4 km
r
4
s- 120 dB
r init dkm
Mc--18
S3
S2
O 25
(U2
0.2
CM
E
CU'S
/
I
0.5
I
0
10
20
30
40
50
60
70
80
90
Target Depth (m)
Target Depth (m)
(a) R=5km
(b) R = 7 km
Depth-Range Averaged Submergence Index - summer BHF3
6 ,
1
1
1
1
Depth-Range Averaged Submergence Index - summerMBHF3
1
1
3.5
3
M
(O 2
E
0
10
20
30
40
so
60
Target Depth (m)
(c) R = 10 km
70
80
90
0
10
20
30
40
50
60
70
Target Depth (m)
(d) R = 12 km
Figure 4.37: Averaged SI results for diving towing pattern for different initial source ranges,
pitch 180.
80
90
Chapter 4. The SI algorithm evaluation in realisticA UV scenarios
Depth-Range Averaged Submergence Index - summerMBHF3
Depth-Range Averaged Submergence Index - summerMBHF3
m
c(X
l)
2-
1.5
0
05-
0
10
20
30
40
50
60
70
80
-0.5'
0
90
10
20
30
Target Depth (m)
40
50
60
70
80
90
Target Depth (m)
(b) R= 7 km
(a) R=5km
Depth-Range Averaged Submergence Index - summerM BHF3
Depth-Range Averaged Submergence Index - summerMBHF3
4
'If= 300
f-/30 Hz
4-
V
3.5 .-
r/- 97 2 km
a0
-1 120 B
nI -o50 d
-
R
sI - 120 dB
nl-50dB
rinit- 12km
x 25-
cc=-20
3'
rinit- 10k\
cc--20
3
1-2'
372km
r
8 -. 0
3.5 -
~
.S2.5 CLU 2
(5
2
(9
a
0
1.5
0.50
0
10
20
30
40
50
60
Target Depth (m)
(c) R = 10 km
70
80
90
0
10
20
30
40
50
60
70
Target Depth (m)
(d) R = 12 km
Figure 4.38: Averaged SI results for diving towing pattern for different initial source ranges,
pitch 20'.
80
90
Chapter4. The SI algorithm evaluation in realisticA UV scenarios
Depth-Range Averaged Submergence Index - summerMBHF3
Depth-Range Averaged Submergence Index - summer BHF3
r 4.808 km
* -0120 dB
SnI = 50 dB
M
r nit - 5,km
* - -25
CO 25
CX2.5-
V)
CU1.5-
0.50
0
10
20
30
40
50
70
60
80
0
90
10
20
30
Target Depth (m)
40
50o
60
70
80
Target Depth (m)
(a) R=5km
(b) R = 7 km
Depth-Range Averaged Submergence Index - summer BHF 3
Depth-Range Averaged Submergence Index - summerMBHF3
fHz
r=/.8 km
4.5
.0 \
j
4 -
-_120'
r init-1,kn\
03:
-
S2.5
M
2 1.5 --
0'
0
10
20
30
40
50
60
70
s0
0
90
Target Depth (m)
10
20
30
40
50
60
70
Target Depth (m)
(c) R = 10 km
(d) R = 12 km
Figure 4.39: Averaged SI results for diving towing pattern for different initial source ranges,
pitch 250.
61
80
90
Chapter4. The SI algorithm evaluation in realisticA UV scenarios
Depth-Range Averaged Submergence Index - summerMBHF3
Depth-Range Averaged Submergence Index - summerMBHF
3
- 300 Hz
/'I - 120 dB
nI- 50 dB
r init = 5 km
-30
I
-3
2.5
.S 2<
1.5 U2
05
0
10
20
30
50
40
60
70
s0
90
Target Depth (m)
Target Depth (m)
(b) R=7km
(a) R = 5 km
Depth-Range Averaged Submergence Index - summerMBHF3
Depth-Range Averaged Submergence Index - summerMBHF3
45
f -=' Hz
r 9.8km
4
/
3.5
-
=0
- 120
nl = 50 dB\
r init = 10 kr
c--30
x
25
aD 2
C5
E1.5
0
10
20
30
40
50
60
Target Depth (m)
(c) R = 10 km
70
80
90
0
10
20
30
40
50
60
70
Target Depth (m)
(d) R = 12 km
Figure 4.40: Averaged SI results for diving towing pattern for different initial source ranges,
pitch 300.
60
90
Chapter 4. The SI algorithm evaluation in realisticA UV scenarios
Depth-Range Averaged Submergence Index - summerMBHF
3
Depth-Range Averaged Submergence Index - summer BHF3
a
O5
0
0
Target Depth(m)
10
20
30
40
50
60
70
80
90
Target Depth (m)
(b) R=7km
(a) R = 5 km
)epth-Range Averaged Submergence Index - summer BHF3
Depth-Range Averaged Submergence index - summerMBHF3
4.5
4
3.5
3
x
S25
2
c5
C,)
0.5
0
0
10
20
30
40
50
60
Target Depth (m)
(c) R = 10 km
70
80
90
to
20
30
40
50
60
70
Target Depth (m)
(d) R = 12 km
Figure 4.41: Averaged SI results for diving towing pattern for different initial source ranges,
pitch 35'.
80
90
5 Discussion
Two major types of towing patterns were examined in this work: the "yo-yo" path and "onetime" depth sampling. This section analyzes their SI performance as it is presented in Section
4.2.
5.1
Vertical orientation
The yo-yo towing pattern consist of two principal stages: diving and climbing. As it shown in
Section 4.2.1, the SI performance strongly depends on the direction of vertical movement of the
towing AUV and the source relative position in the horizontal plane. For fixed relative horizontal
position of the source, the SI algorithm produces 4-5dB separation between submerged and
surfaced sources for one part of the pattern and practically no separation for the opposite part.
Because the yo-yo part can be viewed as two separate "one-time" depth sampling maneuvers,
similar behavior can be noticed in the SI results for the one-time towing pattern, see Figure 4.21
as a typical example of this phenomenon. The following Figure 5.1 emphasizes the difference in
a typical SI performance for upward and downward array orientations.
In order to understand the dependence of the SI performance on the vertical and horizontal
orientations of the array let us closely investigate the SI behavior for pitch angles close to 00. As
it can be seen on Figure 5.2, the SI performance is gradually decaying as the pitch angle changes
from negative to positive values.
Chapter 5. Discussion
Average SI- comparsion of downward and upward arry orientation, summer MBHF scenario
10
20
30
40
50
Target Depth (m)
60
70
80
90
Figure 5.1: A comparison of averaged SI performance for downward and upward array
orientation.
Average SI - comparsion of downward and upward array orientation, summer 8HF scenario
S2.5
2
1.5
0
10
20
30
40
50
Target Depth (m)
60
70
80
90
Figure 5.2: A comparison of averaged SI performance for near 00 array pitch angles (regular
environmental settings).
Chapter 5. Discussion
Now, let us examine the beam-former output for a typical case presented on Figure 5.3. The
figure presents the receiver beam-former output for a submerged source, located at 00 and 1800
relative to the receiver. Notice the angle space boundaries for propagating modes (grazing angle
is less than the critical angle). The critical angle in the presented scenario is 21*. For the
downward and horizontal array orientation on Figure 5.3(a), one can easily recognize a clear
dependence of the beam-former output on the receiving angle, which actually represents the
grazing angle of the incoming modes because the source is located at the end-fire direction of the
array. This angular dependence corresponds to the high ratio between lower and higher incoming
modes resulting in relatively high SI values for that configuration. In contrary, there is almost no
angular dependence for the upward array orientation resulting in a negligible SI performance.
The same, but opposite behavior is shown on Figure 5.3(b). Here, the source is located at the
opposite end-fire direction (1800) of the array. The relevant angles window is now 158* to 1800.
One can notice that the beam-former output is generally the same, but shifted by 1800, which
leads the output for the upward orientation to be angular dependent, resulting again in high SI
values.
Beanformer
output
forasource
at0 degbag. Source
at60mwadreceaver
at 40m
Bemformer
output
forasourceat0 degbeing. Sourcat60m wad
receiver
at40m
70
72
-
_-
downward
68 -ward
70
66
68
62
-64
-548]o
58
5
54
52
-10
S
0
10
_
20
L5
30
40
50
rece*&n
WagWdg]
(a) Bearing 00
60
70
80
90
90
100
110
120
130
140 150
receng ane [dg]
160
170
180
(b) Bearing 1800
Figure 5.3: The receiver beam-former output for a typical submerged source (60m depth) and 3
pitch angles of the receiving array: 00 (horizontal), 100 (upward), and -100 (downward). The
source is located at (a) 0* and (b) 1800 relative bearing to the receiver.
190
Chapter5. Discussion
Such a strong SI performance dependence on the vertical orientation of the receiving array
makes the implementation of the "yo-yo" towing pattern almost irrelevant form the SI
performance point of view. In fact, for either source bearing angle one can choose the
appropriate one-time maneuver (diving for 0' relative bearing and climbing for 1800 relative
bearing) and achieve a reasonable SI performance.
Nevertheless, there may be a tactical advantage in performing the "yo-yo" towing pattern due to
other parallel missions that the AUV may be carrying out during a source depth classification
process.
Chapter5. Discussion
5.2
Towing pattern parameters
Besides the vertical orientation of the receiving array, two other major towing pattern parameters
were analyzed in this work. The first is the pitch angle of the array, while the second is the initial
range to the source.
Based on the one-time towing maneuver performance presented in Section 4.2.2 we can
conclude that the towing the array at pitch angles ±12 -±18 leads to the best averaged SI
performance with separation around 3-4dB between surfaced and submerged targets.
Moreover, every pitch angle that was used in this evaluation produces a better SI performance
than a horizontally flat array, which SI values are given in [5].
The addition of the range averaging process, which was introduced to the SI algorithm by
Apelfeld [5] is also proven to be efficient in a free towing scenario, see Figure 4.29 to Figure
4.41. The relatively weak SI performance at long ranges (12 km) is mostly due to the lower SNR
because of higher transmission loss.
6 Summary and Conclusion
The main purpose of this work was to evaluate the performance of the recently suggested method
for depth classification of underwater acoustic sources in shallow waters based on the modal
energy distribution - the SI algorithm. The SI algorithm uses a plain wave interpretation of the
modal decomposition of the acoustic field and calculates the ratio of the energy concentrated in
modes with steep vs. shallow grazing angles. The algorithm has been previously shown in [5] as
a reliable tool for classifying an acoustic source as surfaced or submerged in Monterey Bay-like
environmental conditions.
However, the recent research was performed using a simplified model of the towed array, which
did not include a realistic array position while towed. In this work we put a special emphasize on
evaluating the SI performance in realistic towed patterns that may be used, and in fact, are used
in practical AUV operations. Two major towing techniques were investigated - the "yo-yo" path
and the "one-time" depth sampling (see Section 4.1.2 for details). The position of the array
elements was simulated using MOOS-IvP simulation package.
Among the main findings of this research lies the discovery of a relationship between vertical
orientation of the array and the relative source bearing. Thus, downward array configuration
results in high averaged SI values (3-4dB) for a source located at 00 relative to the receiving
array, while upward array orientation produces the same SI values for sources located at 1800
relative to the receiver. This relationship implies direct tactical considerations on the AUV
deployment. For example, for sources located at 00 end-fire, the SI algorithm applied during a
downward "one-time" depth sampling maneuver will produce a fast and robust depth
classification of the source.
The method can be used in wide range of pitch angles; however, pitch angles in the range ±12*±180 are recommended for the best averaged SI results.
This work covered range-independent scenarios only. In the previous research, the SI algorithm
was shown as stable and generally applicable in up-slope scenarios while completely unstable in
down-slope scenarios [5]. However, as it was mentioned before, this evaluation was performed
Chapter5. Discussion
with simplified model of the array dynamics. Therefore, additional research is required to further
evaluate the applicability of the SI method for source depth classification in complex, range
depended environments.
Bibliography
1. Yang, T C. A method of range and depth estimation by modal decomposition. J. Acoust. Soc.
Am. 1987, Vol. 82, 5, pp. 1736-1745.
2. Shang, E C. Source depth estimation in waveguides. J. Acoust. Soc. Am. 1985, Vol. 77, 4, pp.
1413-1418.
3. Bucker, Homer P. Use of calculated sound fields and matched field detection to locate sound
sources in shallow water. J. Acoust. Soc. Am. 1976, Vol. 59, 2, pp. 368-373.
4. Baggeroer, A.B., Kuperman, W.A. and Schmidt, Henrik. Matched field processing: Source
localization in correlated noise as an optimum parameter estimation problem. J. Acoust. Soc. Am.
1988, Vol. 83, 2, pp. 571-587.
5. Apelfeld, Van. Depth Discriminationof an Acoustic Source Based on Modal Energy
Distribution.Cambridge, MA : MIT, 2007.
6. Brekhovskikh, L.M. and Lysanov, Yu.P. Fundamentalsof Ocean Acoustics. 2nd Edition.
New York: Springler-Verlag, 1991.
7. Jensen, Finn B., et al. ComputationalOcean Acoustics. New York: Springer, 2000.
8. Frisk, George V. Ocean and SeabedAcoustics. Upper Saddle River, NJ : Prentice Hall, 1994.
9. Arfken, George B., Weber, Hans J. and Harris, Frank. MathematicalMethodsfor
Physicists.s.l. : Academic Press, 2005.
10. Oppenheim, A. V., Schafer, R. W. and Buck, J.R. Discrete-Time Signal Processing.s.l.:
Prentice Hall, 1999.
11. Schmidt, Henrik. OASES Version 3.1. User Guide and Refernce Manual. Cambridge, MA:
MIT, 2004.
12. Schmidt, Henrik, et al. Environmentally tolerant beamforming for high-resolution matched
field processing: Deterministic mismatch. J. Acoust. Soc. Am. 1990, Vol. 88, 4, pp. 1851-1862.
13. Benjamin, Michael R, et al. An Overview of MOOS-IvP and a Brief Users Guide to the IvP
Helm Autonomy Software. Cambridge : MIT, 2009. MIT-CSAIL-TR-2009-028.
Appendix A: Main simulation tool user's guide and source code
A.1.
Main simulation module
The main simulation loop initializes the scenario and environmental settings and calculates the
local SI for each receiver and source position.
The following list summarizes the most important simulation settings:
* nsd: the number of sampling points for the source depth
* nrd: the number of sampling points for the receiver depth
* zO: the most shallow source depth [m], corresponds to the "near-the-surface" source
e thd_0: bearing angle [deg] relative to the receiving array (generally 0' or 1800)
* alpha-d: diving angle [deg] , used for range averaging and may be used as a actual pitch
of the array (see elevation, below)
* base: base file name of the environmental data files used as an input to OASES
e
sr: initial range [km] to the source
e
indexes: an array of corresponding indexes for each characteristic receiver depth in the
MOOS-IvP log file. For use with MOOS-IvP simulated data only.
e xO: the x-axis position [m] of the elements of the receiving array. For use with
synthesized data only.
e elevation: array pitch angle. For use with synthesized data only. May be set to the same
value as alpha d above.
The main module is built such that it requires several adjustments each time it used for the
different environmental and / or dynamics model. Thus, if different environmental data is
used (base variable is set to values other then 'summerBMHF3'), it is necessary to ensure
that the new file is built in the same structure as the original data file (OASES format).
The main module works with a special variable, called elem-pos to hold the position of each
element of the receiving array. Setting up this variable depends on the dynamics model simulated or synthesized array dynamics.
For simulatedarray dynamics data, i.e. taken from MOOS-IvP runs, secondary function
read dynamicsdataO is used. The function is described below, in Section A.2. The function
returns the array position during the whole run. To extract the particular position that
corresponds to some characteristic depth of the array, indexes variable is used.
Appendix A: Main simulation tool user's guide and source code
For synthesized array dynamics data, i.e. calculated based on the assumptions described in
Section 4.1.2, secondary function array-elevationo is used. The function is described below
in Section A.2.
Notice that in order to use a particular dynamics model one has to unmark the wanted part of
the code and mark out the alternative part.
%% The main simulation module
nsd=10;
nrd=7;
%# of sampling point for the source
%# of sampling point for the receiver
dz=100/nsd; %depth resolution
%source initial location (near-the-surface position)
z0=2.0;
%source depth matrix
sz=[z0:dz:z0+(nsd-1)*dz];
%receiver depth matrix
rd=[22:10:82;
%bearing angle relative to the array
thd 0=0;
%angle in rads
th_0=thd_0*pi/180;
[deg]
%speed at the sediment, changed from 1625 to 1612 (slower
cb=1612;
sediment,MB06 soundspeed profile)
cw=1500;
%water speed value, changed from 1480 to fit envir. settings of
summer.dat
freq=300;
%frequency of the source
%diving angle in degrees, must be less than 90 degrees. used
alpha_d=-10;
here for range averaging
%diving angle in rad
alpha=abs(alphad/180*pi);
%setting array tail compensation factor
comp=10;
th c=acos(cw/cb);
%critical grazing angle for the source
thb=
atan(sqrt(sin(th_0)^2 + tan(th c)^2)/cos(th_0)); %% for Odeg relative
bearing...
%th b=pi+atan(sqrt(sin(th 0)A2 + tan(th c)^2)/cos(th_0)); %%for 180deg
relative bearing...
base='summer MBHF3' %base filename with environmetal data
datfil=[ base '.dat'];
mfpfil=[base ' mfp.dat'];
% OASES input
% OASES input
% for Odeg relative bearing use:
thdl=-10;
%min steering angle of the array
%max angle
thd2=90;
(OASES input)
Appendix A: Main simulation tool user's guide and source code
%%for 180deg relative bearing use:
%thd1=190;
%min steering angle of the array
%thd2=90;
%max angle
nth=120;
nth=ceil(nth/6)*6;
%# angle sampling
%makes sure that there is no fraction when its divided by
6
thd_0=180*th_O/pi;
%converting to the degree
thd-b=180*th-b/pi;
%as above
dthd=(thd2-thdl)/(nth-1);
th=[thdl:dthd:thd2];
%sampling step in degrees
%setting the matrix of the angles
ith 0=floor((thd_0-thd1)/dth d)+l;
%index (pointer) for the init. shootin g
angle in the steering angles matrix
ith_b=ceil((thdb-thd1)/dthd )+1;
%index for bottom source angle in the
steering angle matrix
nb=ith b-ith_0+1;
%# of the sampling points in the range
subi=zeros(nsd,nrd)+1.0e-100
%setting the resulting matrix of the
indeces with infinite small values
sr=10.00;
%source range in km
sr init=sr;
sl=120.0;
%source level
nl=50.0;
%noise level
sx=sr*cos(th_0);%range projection on X axis
sy=sr*sin(th_0);%range projection on Y axis
hlam=zeros(nth,nsd);%creating matrix for beaforming results
%% use this to get array data
%indexes =
[1235,
%%
20m
%indexes =
[1980,
from a MOOS run
1305, 1365, 1420, 1490];
30m
40m
50m
60m
1910, 1870, 1810, 1700];
% going down, check!
% going up, check!
%[tt, elempos, towpos] = read dynamics data(0);
%posiotion from MOOS log file
%% --
% read the array
up to here
%% use this to get synthetic array data;
xO = [(0:5)*1.5, 7.5+(1:20)*0.75, 22.5+(1:6)*1.5]; % array spacing
elevation
alphad;
% array pitch. use the previously defined
elempos = arrayelevation(diff(xO), rd, elevation*pi/180);
% calculate the
array position
%% --
up to here
%% main loop
Appendix A: Main simulation tool user's guide and source code
for i=l:nrd-1
for j=l:nsd-1
%%setting file datfil in tmpl.dat with the current values of the source for
oases calc
cmd=['sed -e "s/SD SX SY SL/' num2str(sz(j)) ' ' num2str(sx) ' '
num2str(sy)
num2str(sl) '/"
-e "s/NL/' num2str(nl) '/g" -e "s/F1 F2/' num2str(freq)
num2str(freq) '/g" ' datfil ' > tmpl.dat')
system(cmd);
%% getting the receiver position for each element
fid = fopen('coord.txt', 'wt');
%% for data from MOOS
%fprintf(fid, '%3.2f %3.2f %3.2f\n', [elem pos.z(indexes(i),:);
elem pos.x(indexes(i),:); elem pos.y(indexes(i),:)]);
%% for syntetic data
fprintf(fid, '%3.2f %3.2f %3.2f\n',
[elempos.z(i,:); elempos.x(i,:);
elem-pos.y(i,:)]);
%%up to here
fclose(fid);
%% setting the receiver position for each element in tmp.dat
cmd=['./change coord.perl tmpl.dat coord.txt tmp.dat']
system(cmd);%setting file tmpl.dat in tmp.dat with receiver depth, x and y
%% setting the beamforming params
cmd=['sed -e "s/THMIN THMAX NTH/' num2str(thdl) ' ' num2str(thd2)
num2str(nth) '/" -e "s/Fl F2/' num2str(freq) ' ' num2str(freq) '/g"
' mfpfil
' > tmpl.dat']
system(cmd); %setting file summermfp.dat in tmphla.dat with receiver depth
and steering angles range for beamformer calc by oases
%%setting the receiver position for each element in tmp hla.dat
cmd=['./change coord.perl tmpl.dat coord.txt tmp hla.dat']
system(cmd);%setting file tmpl.dat with receiver depth, x and y
%%run OASN
cmd=['oasn tmp > tmp.log 'J;
system(cmd);%storing oases noises calc in tmp.log
%% run MFP
cmd=['mfp tmp hla tmp tmp > tmp hla.log'J;
system(cmd);%storing beamformer output in tmp hla.log
cmd=['sed -e "s/PLTEND/ /" tmphla.plt > hla.plt'];
system(cmd);%storing pressure field matrix of the beamformer in hla.plt
%% load beamformer pres field file
hla = importdata('hla.plt');
hla=reshape(hla',size(hla,l)*size(hla,2),l);
hlam(:,j)=hlam(:,j)+10.^(hla/10)/nsd;
(non dB)
%creating 1 column matrix
%matrix of the abs values
Appendix A: Main simulation tool user's guide and source code
n=length(hla);
% Here we set the split of the modal space used for the index
ww=1/3;
%window width (lower / high ratio)
fcl=(l-ww)/ww;
%normalization factor
%% creating windows vectors
w=zeros(nb,1);
nw=nb*ww;
%# of windows
nw=min(nw,0.5*nb);
% at least half overlap
nwh=2*nw;
% # hanning windows
nh=2*nwh;
% # number of points in the hanning windows
wh=hanning(nh);
%hanning windowing
w(l:nwh)=wh(nwh+l:nh);
%matrix of the windows
%% calculate LOCAL SI
subi(j, i)=fcl*mean((10.0.^(hla(ith O:ith b)/10)) .* w) /
mean((10.0.^(hla(ithO:ith b)/10)) .* (1-w));%calc ratio between steep and
lower angle of the pres field through hanning for each source/receiver
location
%%LOCAL PLOTS
figure(1);
% beamformer output
subplot(2,1,1),hold off, plot(th,hla), hold on, plot([thd_0 thd_0], [min(hla)
max(hla)] ,'k-'), plot([thdb thd_b], [min(hla) max(hla)] ,'k-');%plot pres
field results
subplot(2,1,2), hold off,
plot(th(ith O:ith b),dbp(fcl*(10.0.^(hla(ith 0:ith b)/l0)).*w),'r'),
hold on,
plot(th(ith_0:ith b),dbp((1-w).*(10.0.^(hla(ith_0:ith-b)/10))),'g ');%plot
windowing results of the pres field
title(['rd,sd=' num2str(rd(i))
','
num2str(sz(j))
'
-
f=
num2str(freq) ]);
drawnow;
% if (j>l I i>l)
figure(2)
% full SI map up to now
wavei(dbp(subi'),sz,rd,-2,8);
h=ylabel('Receiver Depth
set(h,'FontSize',14);
h=xlabel('Target Depth
set(h,'FontSize',14);
(m)');
(m)');
h=title(['Submergence Index
-
'
base]);
set(h,'FontSize',16);
drawnow;
% end
end
%updating the range according the geometry of diving at alpha angle
if(alpha)
sr=(10^-3)*(sr*1000-ceil(dz/tan(alpha))-comp);
else
sr=(10^-3)*(sr*1000-ceil(dz/tan(35*pi/180))-comp);%use 35 deg diving
angle
Appendix A: Main simulation tool user's guide and source code
end
sx=sr*cos(th_0);%range projection on X axis
sy=sr*sin(th_0);%range projection on Y axis
end %% End of the main loop
%%calculate the averaged SI
subm=mean(subi(l:nsd-1,1:nrd-l)');
savfil=[base ' ' num2str(sl)
eval(['save ' savfil]);
'
' num2str(nsd)
DBP subind=dbp(subi);%added for behavior check
DBPsubm=dbp(subm);%same as above
DBP hlam=dbp(hlam');%same as above
%%%PLOTS
sub_plot_ravg
'
'
num2str(thd_0)];
Appendix A: Main simulation tool user's guide and source code
A.2.
Secondary functions
* read dynamicsdata()
The function reads and synchronizes the position vectors of the array elements from a MOOSIvP log file. Notice, that the log two different log files are used: one for the array elements and
one for the towed AUV. Both files can be generated off-line from the main MOOS log file.
These data files have the following structure:
Time
The simulation time
when the log entry was
created
Variable
ARRAY_X, ARRA_Y, or
ARRAYZ
Remark: arrayposition is
Simulated body
pArraySim (ignore
thisfield)
Values
A sequence of floating
point values for each of
the array elements
Simulated body
pEchoVar (ignore
thisfield)
Value
A single floating point
value
always relative to the
towed
Time
The simulation time
when the log entry was
created
body
Variable
TOWPOS_X
TOWPOS_Y, or
TOW POS Z
The towing AUV and the towed array are simulated as two different bodies in the MOOS-IvP
simulation. Each simulated body creates log events in its own rate. Therefore a synchronization
of the array data and the towed AUV data is needed.
%% The
function reads and synchronizes the position of the array elements
from a MOOS-IvP run.
function [time, elempos, tow pos] = read dynamics data(draw);
% read the modified MOOS log file that contains only the time and the X Y Z
% data for each element.
[tt, pos_str] = textread('array-dynamics2.alog', '%f %*s %*s %s');
N = 32;
timel
x-pos
y-pos
z-pos
=
=
=
=
% number of array elements.
tt(1:3:end);
zeros(size(pos_str,1)/3, 32);
zeros(size(posstr,1)/3, 32);
zeros(size(posstr,l)/3, 32);
% Read the elements x,y,z, data to corresponding variable
for i = 1:size(xpos,1)
x_pos(i,:) = strread(posstr{3*(i-1)+1}, '%f', 'delimiter',',');
y pos(i,:) = strread(pos str{3*(i-1)+2}, 'If', 'delimiter',',');
z pos(i,:) = strread(pos str{3*(i-1)+3},
'%f',
'delimiter',',');
end
% read the modified MOOS log file that contains only the time and the X Y Z
.................
Appendix A: Main simulation tool user's guide and source code
% data for the towing AUV
[tt, pos] = textread('towpos2.alog',
'%f %*s %*s
%f');
time2 = tt(1:3:end);
tow_pos.x = pos(1:3:end);
tow_pos.y = pos(2:3:end);
towpos.z = pos(3:3:end);
% syncronize
combined time index = size(timel);
for i = 1:length(timel)
combined time index(i) = find(time2 < timel(i),
1, 'last');
end
% x pos, y pos and z pos are relative to the towing body
x_posabs
x_pos + repmat(tow_pos.x(combined time index), 1, N);
y_posabs = ypos + repmat(towpos.y(combined time index), 1, N);
%pay attention!!
z pos positive is downwards, while z tow pos positive is
%upwards.
z_pos_abs = -z_pos + repmat(tow_pos.z(combined time index),
1, N);
% if asked, draw a animated towing pattern
if draw
figure
hold on
axis([min([x_pos_abs(:,l); towpos.x]) - 50, max([x_pos_abs(:,l);
tow pos.x]) + 50, ...
min([zpos-abs(:,1); tow-pos.z])-10, max([z_posabs(:,l);
towpos.z])+101);
jump = 5;
for i = 1:round(length(timel)/jump)
h(l) = plot(tow_pos.x(combinedtimeindex((i-1)*jump+1)),
tow_pos.z(combinedtimeindex((i-1)*jump+1)), 'b*I);
if i > 1
plot([towpos.x(combined time index((i-2)*jump+1)),
towpos.x(combined time index((i-1)*jump+1))], ...
[tow_pos.z(combinedtime index((i-2)*jump+l)),
tow_pos.z(combined time index((i-1)*jump+1))], 'b-');
end
h(2) = plot(x_posabs((i-l)*jump+l,:), z_posabs((i-1)*jump+1, :),
'g*-'I) ;
drawnow;
delete(h);
disp(['index
=
',
num2str((i-l)*jump+l),
num2str(timel((i-1)*jump+1))]);
end
end
% set up the
elempos.x =
elem_pos.y =
elempos.z =
final output
xpos;
y-pos;
-z_posabs;
towpos.x = t ow_pos.x(combined_timeindex);
towpos.y = tow_pos.y(combinedtimeindex);
tow pos.z = tow_pos.z(combined time index);
time = timel;
'
time
=
Appendix A: Main simulation tool user's guide and source code
*
readdynamicsdataO
The function synthesizes the position vectors of the array elements using the linearity assumption
described in Section 4.1.2.
%% This function synthesizes array elements position for "linear" towing
%% with constant pitch angle
function array_pos = arrayelevation(x, zO, alpha)
% positive aplha means upward pointing array!!!
arraypos.x = zeros(length(zO),
arraypos.y = zeros(length(zO),
array_pos.z = zeros(length(zO),
length(x)+l);
length(x)+1);
length(x)+l);
array_pos.x(:,1) = zeros(length(zO),1);
arraypos.z(:,1) = zO'+sum(x(1:end/2))*sin(-alpha);
for i = 2:length(x)+1
array_pos.x(:,i) = arraypos.x(:,i-1) + x(i-1)*cos(alpha);
array_pos.z(:,i) = array_pos.z(:,i-1) + x(i-1)*sin(alpha);
end