bubble latest after tpds review

8/11/2019 Bubble Latest After TPDS Review

1/12

Effects of Wind-induced Near-surface Bubble

Plumes on the Performance of Distributed

Underwater Acoustic Sensor Networks

AbstractThis paper analyzes the effect of the presence ofnear-surface oceanic bubble plumes on the overall performanceof distributed Underwater Wireless Acoustic Sensor Networks(UWASNs). The existence of bubble plumes in surface and sub-surface ocean water columns is inevitable in most windy oceanicenvironments. There exists studies reporting the degradationof acoustic signal propagating through oceanic bubble plumes.However, most of the existing network protocols designed foruse in underwater sensor networks are ignorant of these effects.In this paper, we study these effects and observe how bubbleplumes have impact on the overall performance of UWASNs,

with respect to different parameters such as the bit error rate(BER), delay, and signal-to-interference-plus-noise-ratio (SINR).Simulation studies show that in the presence of bubble plumes,SINR decreases by approximately 29 %, BER increases by 71 %,and delay increases by 56 %, approximately.

Keywords-Shallow-water, acoustics signal, multipath-

reflection, acoustic sensor network, bubble plumes

I. INTRODUCTION

This paper presents an analysis of the effect of oceanic

bubble plumes on the performance of distributed UWASNs.

UWASNs are envisioned to enable applications such as

oceanographic data collection, pollution monitoring, offshore

exploration, disaster prevention, and assisted navigation [1][2].

Large volume of research literature (e.g., [3], [4], [5], [6],

[7]) exists on UWASNs. However, most studies ignored the

effect of bubble plumes on the overall network performance,

despite the fact that the existence of near-surface bubble

plumes is inevitable in most windy oceanic environments. This

work is an attempt towards addressing this important research

lacuna.

A UWASN consists of variable number of sensor nodes,

which are capable of sensing, processing, and communicating,

and are deployed over a given area of interest to perform

application-centric collaborative monitoring tasks [8]. Under

the surface of water, electromagnetic signals (RF) cannot beused efficiently, primarily due to the absorption by the conduc-

tive sea water. Similarly, optical signals cannot be used in the

underwater column, as they suffer from increased scattering

effects. Additionally, the precision for sending optical signal

should be high. On the contrary, in an underwater column,

acoustic signal can traverse significant distance without atten-

uation. So, for reliable communication in such an environment,

acoustic signal is normally used.

Unlike terrestrial wireless sensor networks, UWASNs suffer

from multiple challenges. Some of the key inherent character-

istics [9] in designing a UWASN are listed below.

The available acoustic bandwidth under the surface of wa-

ter depends on the transmission range [10] and frequency.

Due to high environmental noise at low frequency [11],

communication quality also degrades.

UWASNs are highly dynamic in nature. Except for the

nodes anchored with surface buoys, the other nodes have

mobilities ranging from minimum to a finite value. The

relative mobility between the source and receiver nodesleads to Doppler spread. Additionally, nodes mobility

leads to deformation in architecture of a sensor network,

which, in turn, affects its communication performance.

In UWASNs, internode communication is affected for the

acoustic signal fluctuation caused by many factors such

as path loss, noise, and multi-path fading [12].

End-to-end delay in underwater environments is very

high [13]. The acoustic signal propagation speed in such

environments is five orders of magnitude higher than the

propagation speed in terrestrial environment. Propagation

speed of wireless acoustic signal in underwater environ-

ment is about 1500 m/s, which is five orders of magnitude

lower than propagation speed of signal used in terrestrial

sensor network.

Due to the adverse channel condition, there is a probabil-

ity of having high Bit Error Rate (BER) and temporary

losses of connectivity, thereby affecting the overall net-

work performance.

Battery power of sensor nodes is limited and remote

replenishment of power source is nearly impossible. Also,

solar power cannot be exploited as an energy source under

water. Therefore, energy consumption per node in the

network is crucial aspect of study to increase the overall

network lifetime.

Power attenuation of the signal depends primarily onfrequency and distance.

There is a probability of hardware failure due to foul-

ing and corrosion. Failure rate of nodes in underwater

environments is more than that in the terrestrial sensor

network.

An ocean column is typically segmented into three lay-

ers mixed, thermocline, and deep isothermal. Among all

these layers, the mixed layer is normally the most turbu-

lent. The upper surface of this layer is directly subjected


2/12

to tangential stress due to surface winds. In other words,

transfer of momentum takes place from wind to the water

surface, which often results in splashing/breaking of the waves.

This phenomenon creates concentrated layers of micro-bubbles

underlying the ocean surface, typically up to a depth of 10

meters [14]. The layers of bubble plumes have been observed

to vary not only in depth, but also in range and time [15].

Their evolution in density, radius, and spatial distribution has

profound influence on the quality of acoustic communication

[16].

A. Contribution

As mentioned above, the communication performance of

distributed UWASN systems is very sensitive to the presence

of near-surface bubble plumes. Again, wind has an important

role on the nature and intensity of plumes at subsurface region.

So, it is also important to study the effect of wind velocity on

the overall performance of distributed UWASN.

In sum, the contributionsof this work is cataloged as below.

We have analyzed the variation of plume depth with the

variation of wind speeds. We have also studied how the spatial distribution of

bubble plumes varies as a function of wind velocity.

We have analyzed the effect of individual bubble, as

well as the combined effect of bubble plumes, on the

propagation characteristics of the acoustic signal.

Finally, we have analyzed the performance of the network

in terms of three metrics: SINR, BER, and delay.

I I . RELATED WORK

The unique characteristics of acoustic signal propagation in

underwater environment makes UWASNs distinct from those

of the terrestrial wireless sensor networks. There exist studies

on the performance analysis of UWASNs in the physicallayer. Most of these studies are focused on the physical

characteristics of acoustic signal. There is deficiency of work

analyzing the performance of UWASNs by considering the

interaction of wireless acoustic signals in realistic physical

phenomena occurring in the ocean, especially in the surface

and subsurface regions.

Ismail et al. [17] analyzed the performance of UWASNs in

terms of propagation loss by considering the acoustic signal to

be propagated through underwater ocean channel. The authors

considered only the inherent characteristics of the underwater

channel from the communication point of view. Stefanov

and Stojanovic [18] analyzed the performance of underwater

acoustic ad-hoc networks in the presence of interference.They have assumed the uniform distribution of nodes in the

entire channel. They have modeled path loss in inter-node

communication to be frequency dependent, and have assumed

Rician distribution of acoustic fading.

Fox et al. [19] proposed a methodology for predicting

the underwater acoustic communication performance. They

showed how performance varies with respect to the variation

in relative locations of source and receiver nodes in the per-

spective of acoustic speed profile in the channel. The authors

have taken parameters such as variable sound speed profiles

as a function of environment variables between the source

and the receiver nodes. These parameters help in assessing

the overall communication performance of UWASNs. Llor

and Malumbes [20] studied how physical layer modeling

affects the performance of higher layer protocols. Zhang et

al. [21] considered a linear multihop communication architec-

ture to model acoustic propagation. They took into account

frequency-dependent signal attenuation, interhop interference,

and propagation delay.

A synthesis from the review of existing works reveals that

the effect of near-surface bubble plumes on overall network

performance needs investigation.

Fig. 1: Architecture profile of UWASN

Fig. 2: Link establishment between any two nodes in the

network

III. COMMUNICATION ARCHITECTURE

We have considered a 3-D architecture where large number

of nodes are randomly distributed in space in the ocean

column, as shown in Figure 1. Due to the several advantages

of single-hop communication over multi-hop [22], we have

considered the multi-hop communication architecture between

the source and the surface sink node. We have neglected the

vertical movement of nodes, as their deviation with respect to

their equilibrium position is very small.


3/12

Typically, nodes in UWASN are equipped with acoustic

transceivers, whereas the surface sink nodes are equipped

with both acoustic and electromagnetic transceivers. The nodes

placed in between the source and surface sink nodes act as

the relay nodes. The relay nodes help in routing informa-

tion from the source node to the surface sink. The nodes

communicate with one another in the wireless medium using

acoustic signal. In this paper, we have not considered inter-

sink communication. Nodes are assumed to be homogeneous

in nature with respect to their transmission capability. One of

the challenging aspects of network modeling is establishing

connectivity amongst the sensor nodes. Connectivity between

any two nodes is established when they come in the transmis-

sion range of each other. In this paper, the network model is

presented in terms of a graph

G= (N, Cl) (1)

In Equation (1), N=N1, N2,....,Nn is the number of nodesdeployed over the particular region of interest, and Cl is thelink established between two nodes. The term, Cl, can be

expressed as:(ni, nj) Cl (2)

The scenario is depicted in Figure 2.

IV. CHARACTERIZATION OF NEAR-SURFACE WIND

INDUCED BUBBLE PLUMES

A. Categorization of bubble plumes

The physical unit formed by the combination of many

discrete bubbles is termed as bubble plume or bubble clouds

[23]. Based on Thorps measurements and observations [23],

Monohan [24] classified bubble plumes into three categories:

, , and . These plumes bear their names on the basis oftheir individual lifetime during active bubble generation pro-

cess induced by wind action on the ocean surface. Whitecaps

[15] are associated with two stages, Stage-A and Stage-B, due

to wave breaking. Stage-A is involved with the crest of spilling

breaker and Stage-B with the foam patch. Theplume is thesubsurface extension of Stage-A whitecap [15], and it contains

the highest void fraction ranging from 101 to 102. They

have very short lifetime, typically in the order of less than

1 second. Due to the dissipation of momentum of downward

moving jets that originate due to wave-breaking, plumesdecay into plumes. The plumes have less void fraction(ranging from 104 to 103 ) than plumes, and they

remain attached with the Stage-B foam patch. They, typically,persist only upto 4 seconds. In course of time, theplumesevolve into plumes, and gradually become detached fromthe originating whitecap. Among the three plume categories,

plumes have the lowest void fraction ranging from 106 to107 with a time span 10-100 times greater than plumes.Due to their high longevity and size, the plumes areaffected by the local circulation process, and finally decay to

a weak-stratified background layer. Even if,m plume hashigh void fraction, due to their very short lifetime with respect

to andplume, we have omitted the analysis of the theireffect on the performance of distributed UWASNs.

B. Physical characterization of bubble plumes

This section discusses the spatial distribution of bubble

plumes and the quantification of their population. Monahan

[24] extensively studied the characteristics of oceanic bubbles,

and their importance in the gas exchange between air-sea

interface. As proposed by Hall [25], the Bubble Population

Spectral Density (BPSD), , for any level of bubble plume ismathematically expressed in the functional form as:

(r,z,w10) = N0(r, z)Z(z, w10)W(w10) (3)

In Equation (3), r is the bubble radius, z is the depth fromocean surface, w10 is the wind speed 10 m above the seasurface, W is a wind-dependent factor, and N0 is a constanthaving value corresponding to a particular value of, which isobtained for particular values of radius (r), depth (z), and wind

velocity (w10). The term (r, z) denotes the spectral shapefunction,Z(z) is the depth dependent function signifying the

e-folding depth [26] for each level of bubble plumes. The e-folding depth is normalized with respect to the reference depth

z = 0. Thus, the general form of Equation (3) is written as:

(r,z,w10) = N0(r)Z(z)W(w10) (4)

C. Wind dependent factor

As observed by Monahan [24], the rate of generation of

bubble plumes is equal to the rate of formation of whitecaps.

The study of the relation between the fraction of whitecaps

on the sea surface and the wind velocity was undertaken by

Monahan and Muircheartaigh [27]. After a couple of decades,

Andreas and Monahan [28] presented a premier work, which

states that the fraction of whitecaps present on the sea surface

can be approximated to be proportional to the third power

of wind speed. Hall [25] undertook an integrated study on

several research works done on the dependence of BPSD on

wind velocity (w10). Finally, the author proposed a functionalrelation, which is given below [15]:

W(w10) = (w1013

)3 (5)

In Equation (5), w10 is the wind velocity 10 m above thesurface. As it is assumed that wind is the main factor to

induce the formation of all bubbles, the function W(w10)is regarded the same for all stages of plume development,

including the last stage, i.e., the background layer at any

instant of time.

D. Nature of spectral shape function

Based on the analysis of several experimental data using

multiple measuring techniques, Hall [25] postulated that the

spectral shape function follows power law with an exponent

close to -4. It has been observed that this power law is valid

for bubbles having radii in the range 20-30 m upto 50-80m.Again, on the basis of the report presented by Hall [25], it can


4/12

be stated that the power law for spectral shape function holds

good for small and intermediate shaped bubbles. Therefore,

power law of spectral shape function is invariant for small

and intermediate shaped bubbles, except for the changes in

population size. Therefore, the generalized form of the spectral

shape function forplumes,plumes, and the backgroundlayer is expressed as [25]:

(r) =

0, forrref =rmin

(rrefr1

)3, forrmin rref< r1

1.0, forr1 rref r2

( r2rref

)4, forr2 > rref r3

(6)

In Equation (6), rmin = 10 m, r1 = 15 m, r2 = 20 m, r3= 54.4 + (1.984)z m, and z is in meter. In the subsequentsections, BPSD for different types of bubble plumes is

discussed.

1) Form of BPSD for Plume: From the previoussection, it has been observed that the spectral slope of

the power law is a function of bubble size. Higher slopes

correspond to bubbles of larger radii. As the lifetime of a

plume is comparatively shorter than that of plume,the spectral slope for a plume is set as -4, whereas forplume and background layer a slightly higher value of -2.6is taken. Again, in our study it has been assumed that bubbles

of larger radii exist near to the surface, and with increasing

depth, the population density decreases. Considering all

the aspects stated above, the functional form of BPSD for

plumes is expressed as follows [15].

(r,z,w10) = N0(r)Z(z)W(w10) (7)

In Equation (7), the value of the constant, N0 , is taken tobe 2.0107 m3m1 [26]. The functional form for spectralshape function, (r), for plume can be expressed as:

(r) =

0, forrref > rmin

(rrefr1

)3, forrmin rref < r1

1.0, forr1 rref r2

( r2rref

)4, forr2 < rref r3

( r2r3

)4( r3rref

)2.6, forrref > r3

(8)

Following the work by Monahan and Lu [29], which considers

uniform distribution of plume as a function of e-folding

depth, the general form of depth dependent function is math-ematically expressed as:

Z(z) =

1.0, forzref zmax

0, forz > zmax(9)

In Equation (9), the parameter,Zmax, refers to the maximumpenetration depth of a plume, and it is related to orbitalmotion of breaking wave. Therefore, it controls the vertical

axis of plume. Normally, the penetration depth ofplume extends to one half of the significant wave height

[26]. The expression of maximum penetration depth for

plume,Zmax, is presented as [15]:

Zmax = (1.23 102)w210 (10)

2) BPSD for plume: Mathematically, bubble densityspectrum for plume can be expressed in a manner similarto that in Equation (7) for plume. Therefore, consideringvarious parameters associated with the plume, the densityspectrum is expressed as:

(r,z,w10) = N0(r)Z(z)W(w10) (11)

Asplumes are the transformed products ofplumes, thespectral shape function ofplumes,(r), for small and in-termediate shaped bubbles remains unchanged. However, their

concentration gets weaker. During the transformation phase

from plume to plume, a significant change is observedin bubble population size. The spectral shape function for

plume is expressed as:

(r) =

0, forrref< rmin

(rrefr1

), forrmin rref < r1

1.0, forr1 rref r2

( r2rref

)4, forr2 < rref r3

( r2r3

)4( r3rref

)p(z), forrref> r3

(12)

In Equation (12),p(z)is the spectral slope, which has a steepervalue for the bubbles having radius larger than 60m, and itis expressed as:

p(z) = 4.37 + ( z

2.55)2 (13)

The representative values forr1,r2, andr3 are taken the sameas in Equation (7). Majority of the experimental observations

show that the depth dependent function Z(z) in Equation(12) is expressed as:

Z(z) = exp(z/d) (14)

In Equation (14), d is the the e-folding depth for plume.The e-folding depth as a function of wind speed, w10, isexpressed as:

d(w10) = davg(w10)

ln(Mv(atz=0)Mn

)(15)

In Equation (15), davg is the average penetration depth, Mvis the back-scattering cross-section per unit volume, and Mnis the average noise level. The parameter, davg , is expressedas:

davg(w10) = 0.6(w10 5) + 3.5 (16)

Equation (16) is valid for any value ofw10 greater than 5 m/s.The parameter, Mv, used in Equation (15) is expressed as:

0

n(r)bsdr (17)


5/12

(a) BPSD for small and intermediate bubbles (b) BPSD for near-large bubbles

(c) BPSD for large bubbles

Fig. 3: Distribution of BPSD for andplumes

Fig. 4: Variation of penetration depths of plumes with the variation in wind speed


6/12

0

50

100

150

200

250

10 15 20 25 30 35 40

(Ceff)[m/s]

Frequency [kHz]

w10= 11m/sw10= 12m/sw10= 13m/s

(a) Effective speed, (Ceff), of acoustic signal throughGamma plume

0

20

40

60

80

100

120

10 15 20 25 30 35 40

((C

eff)+(Ceff))avg

[m/s]

Frequency [kHz]

w10= 11m/sw10= 12m/sw10= 13m/s

(b) Effective speed of acoustic signal through Beta as well asGamma-plume

Fig. 5: Acoustic signal propagation through Beta and Gamma plume induced region

0

0.05

0.1

0.15

0.2

0.25

0.3

10 15 20 25 30 35 40

(s

)[dB/m]

Frequency [kHz]

w10= 11m/sw10= 12m/sw10= 13m/s

(a) Attenuation, (s) of acoustic signal through Gammaplume

0

0.1

0.2

0.3

0.4

0.5

0.6

10 15 20 25 30 35 40

(s

)+

[dB/m]

Frequency [kHz]

w10= 11m/sw10= 12m/sw10= 13m/s

(b) Attenuation of acoustic signal due to combined effect ofBeta and Gamma plume

Fig. 6: Attenuation of acoustics signal through Beta and Gamma plume

In the Equation (17), bs is the back scattering cross-sectionfor a single bubble.

Figures 3 (a), (b), (c) depict the distribution of BPSD

for both the plumes. From the figures, we observe that the

distribution curves get shifted upward for plume withrespect to the plume.

E. Nature of penetration of andPlumes

Wind velocity is an important factor in deciding the pen-

etration depth of a particular type of plume. It is usuallyseen that the penetration depth ofplume is lower than thatof the plume for a particular wind velocity. Variation ofpenetration depth for both types of aforementioned plumes

is shown in Figure 4. From Figure 4, we infer that in the

near-surface bubble-induced region, wireless acoustic signal

propagates partly through the plume induced region andpartly through the andplume induced regions, together.Figure 7 shows the bubble induced regions under different

wind speeds.

0

2

4

6

8

10

11 12 13

PenetartionDepth(m)

Wind speed w10(m/s)

- and -plume induced region

-plume induced region

Fig. 7: Bubble induced regions under different wind speed

V. ANOMALOUS BEHAVIOR OF ACOUSTIC SIGNAL IN THE

PRESENCE OF BUBBLE PLUMES

During propagation through oscillating bubble clouds,

acoustic signal gets attenuated. To calculate the attenuation

and perturbation caused by the bubble plumes, we followed


7/12

the approach used by Norton and Novarini [15] and Hall [25].

The assumptions made in this approach are as follows:

The compressional speed of acoustic signal is assumed

to be approximately constant throughout the medium.

There is no interaction between any pair of bubbles.

Due to the oscillation mechanism, the medium becomes

dispersive.

Multiple scattering among the bubbles is negligible.The steps to be followed to calculate the sound perturbation

are:

Compressibility is first calculated for a single bubble.

Total change in compressibility (K) is calculated byintegrating the result obtained for a single bubble, for

sizes of bubbles ranging in radius from maximum (rmax)to minimum (rmin).

Then, the effective speed of acoustic signal (Ceff) isexpressed in terms ofK.

Under the condition when typical square distance between

bubbles is larger than the scattering strength, the effective

compressibility, Keff, of the bubble mixture is expressed asfollows:

Keff = (1 Vfrac)K0+ K (18)

In Equation (18), K0 is the compressibility of ocean waterin the absence of bubbles, and Vfrac is the fractional volumeoccupied by the bubbles. Assuming kr < 1, where k is thepropagation vector of acoustic signal, Kis expressed as:

K(f, z) = N

0f2r

(frf 1 +i)

(19)

For a distribution of plumes, the expression is written as:

K(f, z) = 1

0f2

rmaxrmin

[ r(r)

((frf

)2 1 +i) ] dr (20)

In Equation (19), the physical parameter, fr, is the resonancefrequency. Under adiabatic oscillation and no viscous effect,

for a spherical air-bubble in water, the resonance frequency is

expressed as:

fr = (1 + z

10)0.5(

3.25 106

r ) (21)

where, fr is expressed in Hz, and the radius r is in m.The parameter, , in Equation (19), is the damping constant.Damping is observed in three formsdue to radiation,r, dueto shear viscosity, , and due to thermal conductivity, t.Damping constant, , in terms of these three components isexpressed as:

= r++ t (22)

Mathematical calculation shows that out of these three, damp-

ing due to viscosity and temperature can be neglected as

compared to damping due to radiation. So, the effective

damping is expressed as:

= r =2f r

C0(23)

where,C0 is the velocity of acoustic signal in the absence ofbubble plumes.

For low void fraction, the effective sound speed, Ceff, isexpressed as [15]:

1

C2eff=

1

C20+

1

f2

rmaxrmin

[ r(r)

((frf

)2 1 +i)] dr (24)

Equating the numerator and the denominator of Equation (24),

we have:

1

C2eff=

1

C20+

1

f2

rmaxrmin

r(r)[((frf

)2 1) ir]

[((frf

)2 1)2 +2r ]dr (25)

By putting the value of r from Equation (23) in Equation(25), and writing the real and imaginary part separately, we

have:

1

C2eff=

1

C20+

1

f2

rmaxrmin

r(r)[((frf

)2 1)]

[((frf

)2 1)2 +2r ]dr (26)

1

C2eff=

1

C20+

1

f2

rmax

rmin

r(r)(r)

[((frf

)2 1)2 +2r ]dr (27)

Equations (26) and (27), respectively, represent the real and

imaginary parts. Next, we analyze the sound speed modified

during propagation through and plumes.

A. Sound speed perturbation by andplume

Sound perturbation due to bubble plume depends on the

wind speed and the type of plume. In this section, we have

computed the effective sound speed caused due to only the

and

plumes.1) Perturbation byplume: On the basis of their sizes,

we have classified the plumes into three subtypes: small-and-intermediate, near-large, and large. Contribution to the

effective sound speed, Ceff, through plume comes fromthese three subtypes. The parameter, Ceff, for plume isexpressed as:

( 1

Ceff) =

1

C20 I (28)

In the Equation (28), the quantity I is known as the integra-tion term and it can be expressed as:

I = 1

f2

rmaxrmin

r(r)

[((fr

f )2 +2r 1)

2 +2r ]dr (29)

In the Equation (29), fr is known as the resonance frequencyofplume. It is a function of bubble radius and the depthof influence of plume. It has also a implicit dependency on

the wind speed. For three values of wind speed, namely, 11

m/s, 12 m/s, 13 m/s, values of resonance frequency are given

in the set of Equations (30) as:


8/12

fr =3.47 106

r (30a)

fr =3.53 106

r (30b)

fr =3.58 106

r (30c)

Now, putting the values of BPDS, , of plume, fordifferent bubble sub-categories, we have calculated the inte-

gral, I for these plume subtypes. Depending on the bubblesubtypes, integration limits have been varied. For small-and-

intermediate, rmin = rmin and rmax = r1, for near-large,rmin = r1 and rmax = r2, for large, rmin = r2 andrmax= r3.As an example, we have evaluated the integral, I, for small-and-intermediate bubbles which we have denoted as Is fora particular wind speed, w10, as parameter. Here, for windspeed, w10 = 11m/s, we have calculated I for small-and-intermediate bubbles. We have denoted it as Is, the calculation

is given below as:

Small and intermediate bubbles

Wind speed w10 = 11 m/s

Is= (1.51 1016)f3

r1

rmin

r9

[((134.39 106)2 f2C0r2 106)2+ (2f3r3 109)2]

dr (31)

Next we have calculated Is for other two values of w10,namely, 12 m/s and 13 m/s.

In the same way as above, we have calculated I for near-large and large bubbles for three wind speeds.

2) Perturbation by Plume: Like the plume,the plume is further classified into three categories.Contribution to the effective sound speed under variable

wind speed for each of these categories has been taken into

consideration.

The effective sound speed, Ceff, through plume isexpressed as:

( 1

Ceff)=

1

C20 I (32)

The integration term, I, for plume for three kinds ofplumes is expressed in the Equation (33) as:

fr =4.25 106

r (33a)

fr =4.32 106

r (33b)

fr =4.40 106

r (33c)

Now, for different values of BPDS, , of plume, wehave carried out the integration, I for different bubblesubtypes. Depending on the bubble subtypes, integration

limits have been varied.

For instance, we have evaluated the integral, Is, for small-and-intermediate bubbles under the wind speed, w10. Here,for wind speed, w10 = 11m/s, we have calculated I forsmall-and-intermediate bubbles. We have represented the

integration, Is, as:Small-and-intermediate bubbles

Wind speed w10 = 11 m/s

The integration term, Is, is expressed as:

Is = (5.61 1017)f3

r1rmin

r7

[((164.60 106)2 f2C0r2 106)2

+ (2f3r3 109)2]

dr (34)

Again, we have calculated I for near-large and largebubbles for other values of wind speed.

B. Acoustic signal attenuation by near-surface bubble plumes

In Section A, we have the effective sound speed, Ceff, in

the presence of both types of plumes: as well as . Theterm, Ceff, is expressed in terms of the real and imaginarycomponents as:

Ceff=Cs+is (35)

In Equation (35), the real component, Cs, represents theacoustic phase speed, and the imaginary component, s,represents the attenuation of the acoustic signal due to the

presence of bubble plumes. Equation (24) is derived under

the condition where multiple scattering among the bubbles is

neglected. The two terms on the right side of Equation (35),

can be, respectively, written as:

Cs = Re(Ceff) (36)

s(dB/m) = 20

In(10)Im(

1

Ceff) (37)

In Figure 5, we have shown the variation of signal propagation

through the and plume induced underwater region.From the two figures, we can infer that with the same

increment in frequency, the acoustic signal is subjected to less

resistance from theplume induced region, compared to theregion induced by both the andplumes. Alternatively,we can say that the acoustic signal propagates faster through


9/12

TABLE I: Simulation Parameters

Parameters Values

Deployment Region 100m 100m 100mBubbles Penetration Depth plume: 1.4m - 2.1m; plume: 7.1m - 8.3m

Signal frequencies 10 kHz, 15 kHz, 20 kHz, 25 kHz, 30 kHz, 35 kHz, 40 kHz

Data rate 500 bps

Packet interval 10 sec

Modulation scheme 64-QAM

Number of nodes (N) 150Spreading co-efficient 1.5

Wind Speed (w10) 11 m/s, 12 m/s, 13 m/s

the plume induced region. In Figure 6, we show theacoustic signal attenuation during propagation through andplume induced region. From the two figures, we infer thatthe attenuation rate of acoustic signal is higher in the and plume induced region, than in the region with onlyplume.

VI . PERFORMANCE EVALUATION

In this Section, we present the result of simulation-based

UWASN in terms of different metrics, which are explainedbelow.

A. Simulation settings

Settings of the simulation parameters are shown in Table

I. We have used the NS-3 simulator for our work. We have

considered the network to be consisting of 150 nodes. We have

considered shallow region for the deployment of network. The

nodes are deployed in a region of dimension 100m100m100m. The nodes are deployed randomly at different depths inbetween 0 to 100 m. The sink node is placed at the sea-surface.

We have used the Meandering-current mobility model [30].

Packets are transmitted from the source node to the sink node

via the intermediate relay nodes. We have considered seven

operating frequencies, namely, 10, 15, 20, 25, 30, 35, and 40

kHz.

B. Performance metrics

The performance of UWASN is assessed using the

following metrics.

Signal-to-Interference-Plus-Noise-Ratio (SINR): In any

communication system, signal gets affected due to the

presence of noise generated by system itself or injected from

outside. Like the Signal-to-Noise-Ratio (SNR), SINR is an

important metric for estimating the link quality in wireless

communication among nodes. Interference occurs when

signals from transmitters, other than the desired source, arrive

at the receiver coincidentally with the desired signal [31].

Mathematically, SINR is expressed as:

SINR= SiI+N

(38)

In the Equation (38), Si indicates the signal from the ith

source,I indicates the interference between the signals at thereceiving end, and N is noise in the channel.

Bit Error Rate (BER): In digital transmission, BER is

defined as:

BE R=Percentage of bits with errors

Total no of bits sent (39)

BER and SNR are interrelated with each other, but they

are distinct. In our experiment, we have used the 64-QAM

modulation scheme for evaluating BER. We have:

BE R= 7

12 (0.5 erfc(z)) (40)

Different topologies may exhibit different BER in the same

situation.

Delay (Latency):

Delay is defined as the time taken by a packet to reach the

destination node from the source node [32]. In our work, we

have defined delay in terms of the propagation time, i.e., the

time for a message to reach the sink node from the source

node. Mathematically, it is expressed as:

Delay= Distance covered by the signal

Speed of acoustic signal through channel(41)

In propagating through the underwater channel, the acoustic

signal traverses through bubble-free and bubble-induced en-

vironments. Regarding speed, there will be combination of

ideal channel propagation speed, which is C0 (1500 m/s), andanother, the effective speed, Ceff, through the bubble plume.

C. Benchmark

We have compared the simulated results with the existing

Thorp propagation [33] model in underwater environment.

In Thorp model, internode communication is affected due toattenuation of acoustic signal by conductive and saline ocean

water. So, in this case, it is observed that the performance

of the network is solely dependent on the underwater oceanic

channel characteristics. In our study, part of the channel is

induced by bubble plume. So, in such a scenario, the perfor-

mance of the network differs from that using the Thorp model.

As the Thorp model includes the ideal channel scenario, we

have compared our simulation results with the results obtained

using the Thorp model.


10/12

0

10

20

30

40

50

60

70

80

10 15 20 25 30 35 40

SINR[

dB]

Frquency [kHz]

Bubble Plume, N = 30Bubble Plume, N = 90

Bubble Plume, N = 150

Thorpe, N = 30Thorpe, N = 90

Thorpe, N = 150

(a) SINR for wind speed 11 m/s

0

10

20

30

40

50

60

70

80

10 15 20 25 30 35 40

SINR[

dB]

Frquency [kHz]




Thorpe, N = 150

(b) SINR for wind speed 12 m/s

Fig. 8: SINR for bubble induced and bubble free channel

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

10 15 20 25 30 35 40

BER

Frquency [kHz]




Thorpe, N = 150

(a) BER for wind speed 11 m/s

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

10 15 20 25 30 35 40

BER

Frquency [kHz]




Thorpe, N = 150

(b) BER for wind speed 12 m/s

Fig. 9: BER for bubble induced and bubble free channel

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

10 15 20 25 30 35 40

DELAY

[sec]

Frquency [kHz]




Thorpe, N = 150

(a) Delay for wind speed 11 m/s

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

10 15 20 25 30 35 40

DELAY

[sec]

Frquency [kHz]




Thorpe, N = 150

(b) Delay for wind speed 12 m/s

Fig. 10: Delay for bubble induced and bubble free channel

D. Result and discussion

The results obtained with respect to the three metrics defined

in Section VI-B are discussed below.

1) SINR: Figure 8 shows how SINR varies with frequencyfor different number of nodes under variable wind speed. We

see from the figure that for the same number of nodes at the

same frequency, SINR is higher using the Thorp model than in

the case where bubble plume is considered. Again, we observe

from the figure that towards the higher frequency region, SINR

increases with the increase in the number of nodes. In case of

Thorp, we observe that the distribution of SINR is frequency-

independent. This is due to the fact that even if signal and noise

values are different, the ratio is the same for all the cases. Our

study reveals that the overall SINR for bubble plume decreases

by 28.80% for wind speed 11 m/s and 29.80 %for wind speed12 m/s.

2) BER: Variation of BER with the variation in frequencyis shown in Figure 9. A close observation reveals that,

compared to Thorp, BER is higher in the bubble-induced

channel. Again, we see that BER gradually decreases with

the increase in frequency. Additionally, a comparative analysis

directs that as wind velocity shifts from 11 m/s to 12 m/s,

there is significant increase in the BER. This is due to the

fact that as the wind velocity increases, the bubble plumes

penetrate more deeper region. So, the signal traverses more

distance to reach the surface sink. As expected, the bubble-


11/12

induced channel invokes more errors in bits than the bubble-

free channel. In sum, we infer that compared to Thorp, BER

increases by 69.27% for wind speed 11 m/s, and 71.72 % forthe wind speed 12 m/s.

3) Delay: Figure 10 shows the average time taken by a

packet to reach the destination from the source node. As seen

from the figure, in the bubble-induced channel, it takes longer

time for a packet to reach the destination from the source

node than in the bubble-free channel. Again, we observe

that with the increase in the number of nodes N, the delay

decreases. As the number of nodes increases, the number of

hops also increases. As a consequence, a packet takes lesser

propagation time. Another noticeable observation is that, with

the increase in wind speed, the delay also increases. As wind

speed increases, bubbles penetration depth also increases.

Therefore, the depth of near-surface bubble-induced region

increases, and the duration of interaction with bubbles also

increases. Simulation results show that delay for bubble plume

induced region increases by 47.21 %for wind speed of 11 m/s,and 64.34 % for wind speed of 12 m/s.

VII. CONCLUSION

Increased wind speed near the ocean surface induces the

generation of bubble plumes. Bubble plumes are important

physical entities that persist near to the ocean surface. In

UWASNs, we have studied the effect of near-surface bubble-

induced region on network performance. Simulation-based

network performance was undertaken in terms of three per-

formance metricsSINR, BER, and delay. The simulation

results were compared with the results obtained using the

Thorp model. Performance comparison shows that there is an

increase in SINR by 29.3 %, whereas, for BER and delay, itis increased by 70.5 % and 55.78 % respectively.

In this work, bubble-plumes were considered to be station-

ary. However, situations may arise where there is collective

motion of bubble plumes. In future, we want to observe the

effect of near-surface moving bubble plumes on the perfor-

mance of distributed UWASNs.

REFERENCES

[1] I. F. Akyildiz, D. Pompili, T. Melodia, Underwater acoustic sensornetwork: Research challenges, Ad Hoc Networks, no. 3, pp. 257279,February 2005.

[2] J. Xu, K. Li, Reliable and energy-efficient multipath communicationsin underwater sensor networks, IEEE Trasactions on Parallel and

Distributed Systems, vol. 23, no. 7, pp. 13261335, July 2012.

[3] J. Heidemann, M. Stojanovic, M. Zorzi, Underwater sensor networks:

applications, advances and challenges, Philosophical Transactions ofThe Royal Society, vol. 370, pp. 158175, August 2012.

[4] J. Heidemann, U. Mitra, J. Preisig, M. Stojanobic, M. Zorzi,, Underwa-ter wireless communication networks, IEEE Journal of Selected Areasin Communication, vol. 26, no. 9, pp. 16171619, December 2008.

[5] M. Erol-Kantarci, H. T. Mouftah, S. Oktug, A survey of architecturesand localization techniques for underwater acoustic sensor networks,

IEEE Communications Surveys & Tutorials, vol. 13, no. 3, pp. 487502, 24th March 2011.

[6] P. Nicopolitidis, K. Christidis, G.. Papadimitriou, P. G. Sarigiannidis,A. S. Pomportsis, Performance evaluation of acoustic underwater databroadcasting exploiting the bandwidth-distance relationship, Jounal of

Mobile Information Systems, vol. 7, no. 4, pp. 285298, 7th 2011.

[7] S. Sendra, J. V. Lamparero, J. Lloret, M. Ardid, Underwater com-munications in wireless sensor networks using wlan at 2.4 ghz, inProceedings of IEEE 8th International Conference on Mobile Ad-Hocand Sensor Systems, Valencia, 17-22 October 2011, pp. 892 897.

[8] H. Lu, J. Li, M. Guizani, Secure and efficient data transmission forcluster-based wireless sensor networks IEEE Transactions on Paralleland Distributed Systems, pp. 112, 25th February 2013.

[9] P. Xie, J. H. Cui, L. Lao,VBF: Vector-Based Forwarding Protocol forUnderwater Sensor Networks. Sringer, 2006, vol. 3976, pp. 12161221.

[10] D. Pompili, T. Melodia, I. F. Akyildiz, Distributed routing algorithms

for underwater acoustic sensor networks IEEE Transactions on Wire-less Communications, vol. 9, no. 9, pp. 29342944, September 2010.

[11] D. Pompili, Rutgers, I. F. Akyildiz, Overview of networking protocolsfor underwater wireless communications, IEEE Communication Mag-azine, p. 5, January 2009.

[12] S. Prabhavat, H. Nishiyama, N. Ansari, N. Kato, Effective delay-controlled load distribution over multipath networks.IEEE Transactionson Parallel and Distributed Systems, vol. 22, no. 10, pp. 17301741,October 2011.

[13] T. Abdelkader, K. Naik, A. Nayak, N. Goel, V. Srivastava, SGBR: ARouting protocol for delay tolerant networks using social grouping,

IEEE Transactions on Parallel and Distributed Systems, vol. 24, no. 12,pp. 2472 2481, December 2013.

[14] X. Lurton,An introduction to underwater acoustics, 1st ed., ser. SpringerPraxis Books / Geophysical Sciences. Springer, December 2002.

[15] J. C. Novarini, R. S. Keiffer, G. V. Norton, A model for variationsin the range and depth dependence of the sound speed and attenuationinduced by bubble clouds under wind-driven sea surfaces,IEEE Journalof Oceanic Engineering, vol. 23, no. 4, pp. 423438, October 1998.

[16] D. M. Farmer, G. B. Deane, S. Vagle, The influence of bubble cloudson acoustic propagation in the surf zone, IEEE Journal of Oceanic

Engineering, vol. 26, no. 1, pp. 113124, January 2001.

[17] N. S. N. Ismail, L. A. Hussein, S. H. S. Ariffin, Analyzing theperformance of acoustic channel in underwater wireless sensor, in Pro-ceedings of IEEE International Conference on Mathematical/Analytical

Modelling and Computer Simulation, Kota Kinabalu, Malaysia, 26-28May 2010, pp. 550555.

[18] A. Stefanov, M. Stojanovic, Design and performance analysis ofunderwater acoustic networks, IEEE Journal of Oceanic Engineering,vol. 29, no. 10, pp. 20122021, December 2011.

[19] W. L. J. Fox, P. Arabshahi, S. Roy, N. Parrish, Underwater acousticcommunications performance modeling in support of ad hoc networkdesign, in Proceedings of IEEE Oceans, Vancouver, BC, Sep. 29-Oct.

4 2007, pp. 15.[20] J. Llor, M. P. Malumbres, Underwater wireless sensor networks: How

do acoustic propagation models impact the performance of higher-levelprotocols? Journal of Sensors, vol. 12, no. 2, pp. 13121335, 31st

January 2012.

[21] W. Zhang, M. Stojanovic, U. Mitra, Analysis of a linear mulihopunderwate acoustic network, IEEE Journal of Oceanic Engineering,pp. 110, 7thJune 2010.

[22] M. Vajapeyam, S. Vedantam, U. Mitra, J. C. Preisig and M. Stojanovic,Time cooperative schemes for underwater acoustic communications,

IEEE Journal of Oceanic Engineering, vol. 33, no. 4, pp. 489501,October 2008.

[23] S. A. Thorpe, On the clouds of bubbles formed by breaking waves indeep water and their role in the air-sea gas transfer, Philos. Trans. R.Soc. London Ser. A, vol. 304, pp. 155210, 1982.

[24] E. C. Monahan, Natural Physical Sources of Underwater Sound, B. R.Kerman, Ed. Boston, MA: Kluwer, 1993.

[25] M. V. Hall, A comprehensive model of wind-generated bubbles in theocean and predictions of the effects on sound propagation at frequenciesup to 40 khz, Journal of Acoustical Scociety of America, vol. 86, pp.11031116, 1989.

[26] U. Saxena, P. K. Bhaskaran, A bubble population density spectrum forcentral arabian sea, Journal of Ship Technology, vol. 6, no. 2, 2010.

[27] E. C. Monahan, O. Muircheartaigh, Optimal power-law descriptionof oceanic whitecap coverage dependence on wind speed, Journal ofPhysical Oceanography, vol. 10, pp. 20942099, 1980.

[28] E. L. Andreas, E.C Monahan, The role of whitecap bubbles in air-sea heat and moisture exchange, Journal of Physical Oceanography,vol. 30, pp. 433442, 2000.

[29] E. C. Monahan, M. Lu, Acoustically relevant bubble assemblages


12/12

and their dependence on meteorological parameters, IEEE Journal ofOceanic Engineering, vol. 15, no. 4, pp. 340349, 1990.

[30] A. Caruso, F. Paparella, L. Filipe, M. Vieira, M. Erol, M. Geria, Themeandering current mobility model and its impact on underwater mobilesensor networks, in Proceedings of IEEE INFOCOM, Phoenix, AZ,April 2008.

[31] C. R. Benson, M. J. Ryan, M. R. Frater, Implications of simplifyingsinr in underwater acoustic networks, inProceedings of IEEE OCEANS2009, Biloxi, MS, 26-29th October 2009, pp. 15.

[32] B. A. Forouzan, Data Communication and Networking, 4t h ed.

McGraw-Hill, February 2006.[33] W. H. Thorp, Analytic description of the low frequency attenuation

coefficient, Journal of Acoustical Scociety of America, p. 270, 8th

March 1967.

bubble latest after tpds review

Documents