Vol50,3,2007 411 ANNALS OF GEOPHYSICS, VOL. 50, N. 3, June 2007 Key words underwater acoustics – signal-to-noise ratio – shallow water – high frequency – Gulf of Naples 1. Introduction Over the last ten years, there has been a rap- id extension in the use of digital underwater acoustic communication from mainly military purposes to commercial sues to satisfy the grow- ing needs for underwater acoustic wireless com- munication systems in real time and oversee the submerged instrumentation without the direct in- tervention of man. Underwater acoustic wireless sensor net- works are an emerging topic of research and are envisioned to enable applications on the sur- veillance and monitoring of sea areas (Adams et al., 2000; Sozer et al., 2000; Akyildiz et al., 2005; Benson et al., 2006; Jaffe et al., 2006). There is, in fact, significant interest in monitor- ing aquatic environments for scientific, envi- ronmental, commercial, safety and military rea- sons such as detect tectonic movements, incom- ing tsunamis, water pollution, global warming, distributed tactical surveillance, mine recon- naissance, and many other facts that are bound to affect our lives (Badia et al., 2006; Pompili et al., 2006). The main problem of underwater acoustic communication is the complexity of the marine Underwater acoustic channel properties in the Gulf of Naples and their effects on digital data transmission Tony Alfredo Stabile (1), Aldo Zollo (1), Maurizio Vassallo (1) and Giovanni Iannaccone (2) (1) Dipartimento di Scienze Fisiche, Università degli Studi di Napoli Federico II, Napoli (2) Istituto Nazionale di Geofisica e Vulcanologia, Osservatorio Vesuviano, Napoli Abstract In this paper we studied the physical properties of the Gulf of Naples (Southern Italy) for its use as a commu- nication channel for the acoustic transmission of digital data acquired by seismic instruments on the seafloor to a moored buoy. The acoustic link will be assured by high frequency acoustic modems operating with a central frequency of 100 kHz and a band pass of 10 kHz. The main operational requirements of data transmission con- cern the near horizontal acoustic link, the maximum depth of the sea being about 300 m and the planned hori- zontal distance between seismic instruments and buoy 2 km. This study constructs the signal-to-noise ratio maps to understand the limits beyond which the clarity of the transmission is no longer considered reliable. Using ray- theory, we compute the amplitudes of a transmitted signal at a grid of 21×12 receivers to calculate the transmis- sion loss at each receiver. The signal-to-noise ratio is finally computed for each receiver knowing also the trans- mitter source level and the acoustic noise level in the Gulf of Naples. The results show that the multipath effects predominate over the effects produced by the sound velocity gradient in the sea in the summer period. In the case of omnidirectional transmitters with a Source Level (SL) of 165 dB and a baud rate of 2.4 kbit/s, the results al- so show that distances of 1400-1600 m can be reached throughout the year for transmitter-receiver connections below 50 m depth in the underwater acoustic channel. Mailing address: Dr. Tony Alfredo Stabile, Diparti- mento di Scienze Fisiche, Università degli Studi di Napoli Federico II, Napoli, Italy; e-mail: stabile@na.infn.it CHECKING COPY 412 Tony Alfredo Stabile, Aldo Zollo, Maurizio Vassallo and Giovanni Iannaccone acoustic channel: background noise from the environment and from human activities; un- evenness in the water caused by the presence of solid and gaseous particles; the presence of spa- tial and temporal sound velocity gradients that can arise from salinity, pressure and tempera- ture variations; multipath effects caused by multiple reflections on the sea surface and bot- tom; reverberations caused by the agitation of the sea and the topography of the sea bottom; and marine currents. In particular, with a digital communication system, the multipath that oc- curs with the propagation of signals results in intersymbol interference (ISI), which limits the capacity of the underwater acoustic communi- cation channel. All of these phenomena are magnified in the situation of acoustic transmis- sion in shallow water with respect to transmis- sion in deep water, due to the depth of the sea and the vicinity to the coast (Cazzolato et al., 2001; Smith et al., 2003), so the shallow water acoustic communication channel is character- ized by strong signal degradation. The use of high frequencies in acoustic communication improves the velocity of digital transmission of data, but also causes a greater attenuation of the signal (Thorp, 1967; Fisher and Simmons, 1977), with the consequent de- crease in the range of distance, i.e. in the abili- ty to transmit data over distance. Acoustic communication has been progres- sively improving over the years, both in terms of the distances and the data transmission capacity, due to the development of new techniques of sig- nal modulation. In more recent years, the feasi- bility of modulation techniques for underwater communication and telemetry have been demon- strated (Baggeroer, 1984; Catipovic, 1990; Coates, 1993; Stojanovic et al., 1993, 1994, 1995; Stojanovic, 1996), minimizing the ISI to improve the capacity of the transmission chan- nel. Internationally, the systems of acoustic links that are now most advanced use frequencies above 20 kHz (ultrasound), reaching bit rates of 10 kbit/s (Stojanovic, 2004). To be able to complete any underwater acoustic communication system, it is necessary to know the physical properties of the transmis- sion channel to be able to understand what the limits on signal transmission are and how its complexity can influence the quality of the in- formation transmission. In the Gulf of Naples (Southern Italy) a fea- sibility test is ongoing which provides for the installation of seismic instruments on the seafloor to extent offshore the geophysical monitoring of the Neapolitan volcanoes, Mt. Vesuvius and Campi Flegrei caldera. The seis- mic data loggers will be acoustically linked to a moored buoy using a two-way communication modem, then a radio apparatus on the buoy will telemeter data on land. Here, we studied the physical properties of the water layer of the Gulf of Naples for its use as an acoustic communication channel for high- frequency digital transmissions (around 100 kHz), and analyzed the influence of the season- al variations in its physical parameters on un- derwater transmission and the limits that the channel forces upon this communication. For this purpose we are interested to understand the effects produced by the underwater acoustic channel on a high frequency digital transmis- sion to choose the best configuration of the communication system. We do not take into ac- count the effects of bubble scattering and ab- sorption because the layer that contains bubbles is near the surface (Preisig, 2006) and, as our models use wind speeds less than 6 m/s, bub- bles cannot significantly attenuate propagating signals in this layer (Dahl, 2004). We also con- sider the sea as a steady propagation environ- ment with smooth sea surface and seafloor dur- ing the transmission of data. The first section provides a description of the physical properties of the Gulf of Naples as a one-dimensional (1D) communication chan- nel in shallow water, considering that the aver- age depth of this water layer is less than 300 m. In the second section, the potential noise sources are described, and the Noise Levels (NL) expected at the 100 kHz frequency are calculated. In the third section, we proceed to the simulation of underwater acoustic propaga- tion, describing the calculation method carried out to determine the Signal-to-Noise Ratio (SNR). Finally, in the fourth section, the results we obtained for a standard piece of apparatus for digital acoustic transmission of information are reported, evaluating the limits of high fre- 413 Underwater acoustic channel properties in the Gulf of Naples and their effects on digital data transmission quency acoustic transmission in the Gulf of Naples, starting from the SNR calculated from the sea surface to the bottom, and up to a dis- tance of 2 km. 2. One-dimensional velocity model of the Gulf of Naples The Gulf of Naples is located along the south-western coast of Italy, between 40°35l and 40°50l latitude north, and between 14°00l and 14°30l longitude east. The depth of the sea in the area of interest is not deeper than 300 m, except for the northern part of the Dohrn canyon (fig. 1), in which the sea depth reaches 400 m. Since the Gulf of Naples is not deeper than 300 m, it can be considered a shallow wa- ter acoustic communication channel. This im- plies a greater complexity on the sound propa- gation because in shallow water the surface, volume, and bottom properties are all impor- tant. Moreover the seasonal variation in sound- speed structure is significant with winter condi- tion being nearly isospeed (Jensen et al., 2000). The essential characteristic of sea water is, in fact, the velocity at which the sound can be propagated through it. The square of the veloc- ity of the sound depends on the ratio between the density and the bulk modulus of the propa- gation medium. In particular, in the sea the den- sity is a function of the pressure, temperature and salinity of the sea, while the bulk modulus is a function of the pressure and temperature of the sea. For this reason, the propagation veloci- ty of acoustic energy in the sea can be described Fig. 1. Bathymetric and topographic image of the Gulf of Naples. A bright ellipse indicates the area of inter- est for our studies, where the researchers of the Anton Dohrn Zoological Station carried out measurements of the physical properties of the seawater in the period from February 2002 to February 2003. 414 Tony Alfredo Stabile, Aldo Zollo, Maurizio Vassallo and Giovanni Iannaccone as a function of three physical parameters: pres- sure, temperature and salinity (Jensen et al., 2000). It is not only necessary to know the mean values of these three physical parameters for the sea, but it is also important to know their spatial and temporal variabilities, as these vari- ations significantly modify the propagation ve- locity of the acoustic waves in the sea. Over the years, various formulae have been suggested that link the sound velocity in the sea to the three parameters of pressure, temperature and salinity (Kinsler et al., 2000; Mackenzie, 1981; Coppens, 1981), and in particular, the UN- ESCO algorithm (Chen and Millero, 1977; Wong and Zhu, 1995) and the equation of Del Grosso (Del Grosso, 1974) are commonly used. If the sea is considered to be a resting fluid with a constant density ρ, the pressure is solely a function of the depth. There are also some very accurate, empirical relationships between sea depth and pressure. The sea temperature and salinity, however, depend upon the ambient conditions of the location, such as the solar ir- radiation, evaporation, the atmospheric condi- tions, the water circulation, the wind, the pres- ence of rivers, and the geographical position. The characterization of the Gulf of Naples starts, therefore, from the knowledge of the val- ues and variabilities of these parameters in space and time. For this, the temperature and salinity data were analyzed from measurements carried out by the researchers of the Anton Dohrn Zoological Station (Ribera, personal communication, 2005) in the Gulf of Naples from the sea surface to a depth of 300 m, in the period from February 2002 to February 2003. The sound velocity data, calculated by the same with the equation of Del Grosso (Del Grosso, 1974) from these salinity and temperature measurements, were analyzed. Figure 2 shows the salinity profiles of the sea between the surface and the bottom at 300 m in depth, with a sampling rate of 0.1 m, across 15 different periods of the year. The salinity went through small variations through- out the year, with a value of about 38 ppm. It can be seen from fig. 2 that there was an in- crease in the salinity with depth of about 0.5 ppm, from the surface to the bottom, and also rapid oscillations in the salinity values in the first 100 m in the non-winter periods due to sur- face marine currents, where at some depths the salinity can change of about 2 ppm from the mean value. In the winter periods, the strong agitation of the sea causes a mixing of the wa- ter down to great depth, resulting in stability of the salinity values. Figure 3 shows the temperature profiles measured at the same time as the salinity at the same depths. The temperature varies with time, and on large spatial scales, in concomitance with the daily and seasonal variations of solar irradiation; fine spatio-temporal variations are determined locally by turbulent movements of the seawater. The variability in the temperature of the sea depends on the variability of the ther- mal content of the water as a consequence of the heat transfer between the surface and the adjacent atmosphere (De Dominicis Rotondi, 1995). From fig. 3 it can be seen that at the sur- face the higher temperatures are reached in the summer months, with maximum values of 27°C in August, while the lowest temperatures are reached in the winter months, with minimum values of 13°C in February; in the deepest wa- ter, the mean temperature is lower and the tem- perature variations between the summer and winter seasons are greatly reduced. In summer, water mixings hardly occur and results in a large negative temperature gradient in the first 75 m of depth, where the temperature can drop from 27 °C at the surface to 14°C at 75 m in depth. In autumn and winter, the mixing of the water is very intense, which generates isother- mic surface layers that reach down to a depth of 50 m in October to 160 m in February. For depths greater than 160 m the mean tempera- ture is 14°C and its seasonal variation rarely reaches 0.1°C; at this depth the temperature is no longer influenced by the outside meteoro- logical conditions. Figure 4 shows the sound velocity profiles calculated with the equation of Del Grosso (Del Grosso, 1974), starting from the salinity and tem- perature data. It can be inferred from this figure that the highest variations in the sound velocity are at the surface, and are between 1508 m/s and 1541 m/s, equivalent to a percent variation of ±1% of the mean velocity value. At increasing depths, the seasonal variations decrease until 415 Underwater acoustic channel properties in the Gulf of Naples and their effects on digital data transmission they become lesser than 5 m/s below 100 m and negligible below 160 m. A comparison between the sound velocity profiles with those of temper- ature reveal an overall correlation at all depths for all the periods of the year, except for a positive linear trend followed by the sound velocity pro- files for increasing depths, caused by a rise in pressure with growing depth. From 160 m to 300 m in depth, where the temperature variations are negligible, the velocity only depends upon the Fig. 2. Salinity profiles in the Gulf of Naples from the sea surface to the bottom, at 300 m in depth, measured across 15 different periods between February 2002 and February 2003. 416 Tony Alfredo Stabile, Aldo Zollo, Maurizio Vassallo and Giovanni Iannaccone pressure variations, increasing by around 2 m/s. The comparison between the salinity and sound velocity profiles shows that the rapid salinity os- cillations in the hottest months, and for depths less than 100 m, do not correspond to apprecia- ble variations in the sound velocity profiles, since the latter are less then 0.5 m/s (that is 0.03% of the mean sound velocity in the seawater). Conse- Fig. 3. Temperature profiles in the Gulf of Naples from the sea surface to the bottom, at 300 m in depth, meas- ured across 15 different periods between February 2002 and February 2003. The main temperature variations are above 50 m depth. For depths greater than 160 m the seasonal temperature variation rarely reaches 0.1°C. 417 Underwater acoustic channel properties in the Gulf of Naples and their effects on digital data transmission quently, the rapid variations in salinity have not been considered in the construction of a sound velocity model of the Gulf of Naples. Knowledge of the effects of the three phys- ical parameters of temperature, pressure and salinity on the sound velocity profiles has al- lowed the construction of a multiparameter function of depth z that simulates the sound ve- locity profiles throughout the year and for depths from 0-300 m in the Gulf of Naples, as- Fig. 4. Velocity profiles in the Gulf of Naples calculated from the salinity and temperature data obtained from the measurements carried out between February 2002 and February 2003. 418 Tony Alfredo Stabile, Aldo Zollo, Maurizio Vassallo and Giovanni Iannaccone suming that the variations seen over the sam- pling period are stationary. The parameters of the function were obtained through the inver- sion of the 15 sound velocity profiles using the non-linear Simplex global optimization algo- rithm (Nelder and Mead, 1965; Press et al., 1992) and creating a function of time for each parameter that interpolates the values obtained from the inversion. The chosen multiparameter model is described by the following function: . (2.1) The linear dependence with depth (βz) express- es the increase in the mechanical pressure. The parameter α represents the value of the sound velocity in sea water at a temperature of 15 °C and a pressure of 1 atm (depth of z = 0 m); the parameter V0 represents the variation with re- spect to α in the sound velocity at the sea sur- face for a different temperature. The parameter zc gives information on the thickness of the sur- face isothermic layer. The adimensional param- eter γ amplifies the z/zc ratio and determines the negative slope of the function V(z) for z ≈ zc. 3. Noise sources in the underwater acoustic channel in the Gulf of Naples The noise signal of an underwater acoustic sensor is composed of electronic and acoustic noise, generated by the pressure fluctuations in the sea water due to the action of ambient sources (De Dominicis Rotondi, 1996). Here, we consider only acoustic noise, and in particular we calculate the noise level around a frequency of 100 kHz in the Gulf of Naples. Analytically, the Noise Level (NL) will be ex- pressed in decibels (dB) as the ratio between the noise pressure (or intensity) revealed with omnidirectional hydrophones and a reference pressure (or intensity) of a plane acoustic wave in the same frequency bandwidth. Typically, the reference pressure level used in underwater acoustics is 1 µPa. In the sea, there are various noise sources that are both artificial and natural, which have inten- sities that vary in time, space and frequency. The natural sources have physical origins (molecular ( ) ( / )V z V z z z1 c0 α β= + + +γ6 @ thermal agitation, sea tides and turbulence), me- teorological (wind, rain, hail, storms, wave back- wash, glaciers), terrestrial (earthquakes, volca- noes, distant storms) and biological (cetaceans, crustaceans, fish); the artificial sources are due to human activities like seaside traffic, coastal in- dustrial works, ports, boreholes for research into energy sources, and underwater acoustic com- munications besides those of interest. This noise covers a vast frequency range that goes from frequencies lower than 1 Hz to frequencies higher than hundreds of kHz. For the various frequency ranges, the trend of the spectrum shows particular characteristics as a consequence of the peculiarities of the various sources: each source has its own dominant fre- quency band in which it has a predominant as- pect with respect to the other noise sources; sometimes, different sources can predominate in the same frequency band. The range going from a few kHz to 100 kHz is of interest in our study and is dominated by the noise generated by the impact on the sea surface of drops and water spray that arise from the breaking of the waves. A classical representation of the spectral distribution in the range from 100 Hz to 25 kHz comes from the historic Knudsen curves (Knudsen et al., 1948), which came from a vast measurement campaign carried out during the Second World War by a group of researchers guided by V.O. Knudsen, along the coastal wa- ters of various ocean regions. The curve, parame- terized according to the condition of the sea or equivalently as a function of the wind velocity (in knots), shows a linear trend on a logarithmic frequency scale with a slope of around −20 dB/dec. A series of subsequent investigations in the Knudsen range, carried out by Wenz (1962) and Urick (1986), established a correlation be- tween wind speed and the ambient noise spec- trum level. Using the empirical formulae given by Coates (Coates, 1990; Stojanovic, 2006), it is possible to calculate the Noise Level (NL), from 1 Hz to 1 MHz, as a function of frequency (in kHz), wind speed w (in m/s) and shipping activ- ity factor s, whose value ranges between 0 and 1 for low and high activity, respectively. From these formulae it is possible to infer that the ship- ping activity factor s is negligible at frequencies of about 100 kHz. 419 Underwater acoustic channel properties in the Gulf of Naples and their effects on digital data transmission The calculation of the noise level of the Gulf of Naples for a frequency of 100 kHz, was car- ried out using Coates formulae and the mean wind velocity in the Gulf of Naples for both the summer and winter months. The wind velocities at 10 m above the sea surface were considered with the READY program starting from the FNL model (the model and program are available on- line on the website of the National Oceanic and Atmospheric Administration) they were sampled every six hours from February 2002 to February 2003, the period in which the measurements of the physical parameters of the sea were taken. The mean wind velocity calculated for the sum- mer months was 3.9 m/s, while for the winter months it was 5.6 m/s. The noise levels in the Gulf of Naples at the frequency of f = 100 kHz are then 28 dB/µPa in summer and 30 dB/µPa in winter. 4. Efficiency of the acoustic link for standard digital data transmission apparatus Knowledge of the physical parameters and the use of the function designed to simulate the sound velocity variations with depth and time in the Gulf of Naples has allowed us to carry out various simulations of high frequency acoustic propagation in the Gulf of Naples. Us- ing eq. (2.1), a mean velocity profile was con- structed for the summer and winter months, with the aim of studying the behavior of the acoustic propagation according to the variations in the sound velocity in the sea; the two veloci- ty profiles are shown in fig. 5. We also calculat- ed (fig. 5) the residuals for the summer and winter months as the difference between the values of the mean velocity profiles constructed using eq. (2.1) and those calculated using the profiles of fig. 4. The residuals are less than 0.5 m/s for both summer and winter mean velocity profiles, except for the first 50 m depth in sum- mer profile where the residuals can reach a maximum value of 1.8 m/s (only 0.1% of the sound velocity in the water). Below, we de- scribe the calculation method used for the sim- ulation of the signal propagation and for the calculation of the signal-to-noise ratio, consid- ering a standard digital data transmission appa- ratus. This considers an omnidirectional trans- mitter that delivers sinusoidal acoustic signals for the transmission of a digital binary se- quence, with a velocity of 2.4 kbit/s at a domi- nant frequency of 100 kHz, in an operative band of 10 kHz. The power delivered to the an- tenna was assumed to be 0.4 W with a electroa- coustic efficiency of 50%. 4.1. Description of the calculation method for simulation of signal propagation The propagation medium is described as- suming a sea water layer (as previously de- scribed) and two solid layers (sea bottom sedi- ments). The sediments samples taken from the first few meters of the sea bottom in the Gulf of Naples show that it is composed of clayey silt sediments in the first five meters and sand-silt- clay sediments below, because of lower per- centage of silt. From Hamilton studies on phys- ical properties of marine sediments (Hamilton, 1978, 1979, 1980) we can therefore define a geoacoustic model of the seafloor assigning, for each solid layer, the thickness, the sound veloc- ity (P-waves), the Vp/Vs ratio, and the density. The values of the physical parameters of the propagation medium used for the simulation are given in table I. We also consider a variation of the Vp/Vs ratio for both the solid layers to un- derstand if it influences the propagation. The discontinuity surfaces between the lay- ers are flat and horizontal and no rough sea sur- Table I. Physical values assigned to the propagation medium for the simulation: the medium is described according to a liquid layer (sea water) and two solid layers (clayey silt sediments above sand-silt-clay sediments). Stratum Thickness Sound Vp/Vs Density typology (m) velocity (g/cm3) Vp (m/s) Sea water 300 1510-1534 ∞ 1.03 Clayey silt 5 1550 13.0±2 1.40 Sand-silt-clay ∞ 1650 9.4±2 1.80 420 Tony Alfredo Stabile, Aldo Zollo, Maurizio Vassallo and Giovanni Iannaccone face and seafloor are considered in this paper. Within the water layer, the sound velocity is as- sumed to vary continuously with depth, follow- ing the trend given in fig. 5. The model describ- ing the characteristics of the propagation medi- um in the summer months essentially differs from the one relative to the winter months in terms of the sound velocity values in the water layer above the cut-off depth of about 50 m. Table II gives the sound velocity values for the two models at increasing depths. The simulations were carried out with the 3D dynamic ray-tracing code (Farra and Madariaga, 1987) with modifications to allow for the anelas- tic attenuation in the sea, which is considered by multiplying for each phase (direct P, multiple P reflections in the water) the amplitude of the non- attenuated wave A0 by a negative exponential (4.1) where α is the attenuation coefficient and has the dimensions of inverse length and x is the length of the ray-path. From the empirical Thorp’s formula (Thorp, 1967; Fisher et al., 1977; Jensen et al., 2000) ( )expA A x0 α= − Table II. Sound velocities of the summer and winter models for increasing depths. The main differences be- tween the two models are above 50 m depth. Depth (m) 0 10 20 30 40 50 60 80 100 160 300 Summer model Vp (m/s) 1534 1533 1527 1523 1517 1514 1512 1511 1510 1511 1513 Winter model Vp (m/s) 1514 1514 1514 1514 1514 1514 1513 1512 1511 1511 1513 Fig. 5. Mean velocity profiles (left) for the summer and winter seasons calculated from eq. (2.1) and residuals (right) evaluated respect to the mean velocity profiles computed using the 15 velocity profiles obtained from da- ta. Below 50 m depth the residuals are lesser than 0.5 m/s for both summer and winter mean velocity profiles. Barbara Evidenziato Barbara Casella di testo oppure Fisher and Simmons, 1977 (vedi Bibliografia)? Prego verificare. Grazie 421 Underwater acoustic channel properties in the Gulf of Naples and their effects on digital data transmission (4.2) the attenuation is expressed as a loss in decibels per unit of distance as a function of the frequen- cy f (in kHz) of the wave that propagates in the sea and it is indicated by αl (Jensen et al., 2000), instead of α. From the eq. (4.2) and considering a frequency of 100 kHz we find that αl is equal to 34.3 dB/km. Taking into consideration eq. (4.1), we can obtain α from αl as follow: (4.3) To study the reliability of the acoustic commu- nication, we simulated the propagation from the transmitter to the receivers of a signal consist- ing of a sinusoidal transient of a 1 s duration and with a frequency of 100 kHz. The transmitter, i.e. the source of the signal, is omnidirectional and was positioned at the three different depths of 1 m, 190 m and 299 m from the sea surface; for each transmitter posi- tion, the reception of the signal was simulated on a grid of 21×12 receivers, arranged in a ver- tical plane inside the area of interest showed in fig. 1 by a bright ellipse. The horizontal dis- tance between the receivers is ∆x = 100 m from the vertical axis through the source for up to 2 km in distance, while the vertical distance be- tween the receivers is ∆z = 25 m, from 5 m in depth to 280 m in depth. For each receiver, the particle displacement velocity v=(vx, vy, vz) was simulated, from which it is possible to deduce the intensity of the sig- nal, information that is useful for the determina- tion of transmission losses. Indeed, the instanta- neous acoustic intensity I of field produced by force F that acts normally on a unitary surface A is I = Fv/A= pv, where p is the pressure and v is the particle displacement velocity along the nor- mal to the surface A. Considering that the time- scale of oceanographic changes is an order of magnitude greater than the time-scale of acoustic propagation, it can be assumed that the . km3 95& .α = . . logx A A x20 8 686 8 686 0 1 &.α α α = − − l l c m 44 ( ) 3.3 10 . . dB/km f f f f f 1 0 11 4100 3 3 10 3 4 $ $ .α + + + + + + − − 2 2 2 2 2 l density ρ and the sound velocity c are independ- ent of time. With this assumption, both the pres- sure p and the velocity potential φ(v=∇φ) satis- fy the wave equation, and using the linearized Euler equation (Jensen et al., 2000), this gives (4.4) where v is the particle displacement velocity along the direction of acoustic wave propagation. The simulated signal at each receiver is per- turbed by the multipath effects, arising from the multiple reflections from the surface and the bottom of the sea. Because of this effect, with the transmission of digital data, the coding sig- nals applied to the preceding bit are superim- posed on the coding signals of the succeeding bit upon reception, causing an increase in the intersymbol interference. 4.2. Calculation of the Signal-to-Noise Ratio (SNR) An object that radiates with a Source Level (SL) will be registered by a receiver with a low- er Signal level (S) due to the Transmission Loss (TL) that it undergoes in its passage through the propagation medium. The Signal-to-Noise Ra- tio (SNR) for this receiver is obtained by sub- tracting the Noise Level (NL) from the Signal level (S), such that: (4.5) where all of the quantities under consideration are expressed in dB. To determine the limits beyond which the transmission is no longer reliable, using eq. (4.5) we can construct the SNR maps for a determined Source Level (SL) starting from the knowledge of the Transmission Loss (TL) at each receiver and the Noise Level (NL) for the Gulf of Naples. The Noise Level (NL) was discussed above in the sec- ond section, so we now consider the determina- tion of the Source Level (SL) and calculate the Transmission Loss (TL) at the frequency of 100 kHz on the whole grid of 21×12 receivers. The source level is determined according to the rela- tionship (De Dominicis Rotondi, 1990): SNR (db S NL SL TL NL) = − = − − p cv I cv2&ρ ρ= = 422 Tony Alfredo Stabile, Aldo Zollo, Maurizio Vassallo and Giovanni Iannaccone (4.6) where We represents the electrical power sup- plied to the antenna, η the electroacoustic effi- ciency, and DI the directivity index. Consider- ing a realistic example where the transmitter has an efficiency of 50% and the current sup- plied to the antenna is 185 mA with a tension of 3.24 V, thus giving it an electrical power of 0.6 W, from eq. (4.6) the source level will be SL= =165 dB. The transmission loss undergone by the signal at any specific receiver are obtained in- stead from the ratio between the mean intensity I of the signal simulated at each receiver and the mean intensity I0 of the signal simulated at one meter distance from the transmitter, giving the value in dB according to the relationship (4.7) which, taking into account eq. (4.4), becomes: (4.8) Using eq. (4.8), it is therefore possible to calcu- late for each receiver the transmission loss know- ing the mean value of the particle displacement velocity v = (vx, vy, vz). The mean value is calcu- lated across a period of time equal to the duration of the signal sent by the transmitter. 5. Results obtained Starting from the knowledge of the trans- mission loss in the acoustic communication channel at the frequency of 100 kHz, and from the knowledge of the noise level in the Gulf of Naples, and considering a source with a 165 dB level, using eq. (4.5) we have constructed the maps that give the Signal-to-Noise Ratio (SNR) for the summer and winter months, for three different transmitter positions, considering both 10 10 TL (dB) log log x x y z y z0 2 0 2 0 2 2 2 2 0 2 2 ν ν ν ν ν ν ν ν = − − + + + + =d f n p 10TL (dB) log I I 0 = − d n DI+10 ( )log η+10 ( )log We+170.5=SNR (db) the multiple reflections and the direct phase on- ly. If it is assumed that the band pass of the acoustic communication is 10 kHz around the frequency of 100 kHz, and that the transmission velocity of the data from the emitter is 2.4 kbit/s, from Shannon’s Capacity Theorem for a communication channel (Shannon, 1948) the signal-to-noise ratio must be greater than 22 dB for reliable information transmission. Figures 6, 7, and 8 indicate in black the re- gions in the section of water in which the com- munication is not reliable, as in those zones of the channel in which the signal-to-noise ratio is less than 22 dB. Where the transmission should be reliable (SNR >22 dB) a different color has been assigned to each SNR value calculated, in decreasing order from red to violet. Figure 6 shows the signal-to-noise ratio for a transmitter positioned at 1 m in depth. A com- parison between the summer and winter maps relative to the direct phase only shows that if the source is positioned at a low depth, the ef- fect of the negative velocity gradient that is seen in summer does not favor surface commu- nication for increasing distances. If, however, a comparison is made between the summer and winter maps that takes into account multiple re- flections between the surface and the bottom, we see that the multipath effect predominates over the velocity gradient effect. Figures 7 and 8 show the signal-to-noise ratio for transmitters positioned at 190 m and 299 m in depth, respec- tively. From a comparison of the summer and winter maps, relative to the direct phase only, it can be seen that for these source depths the neg- ative velocity gradient of the summer months no longer influences communication. Comparing figs. 6, 7, and 8 for the cases in which multiple reflections are taken into ac- count, it can be seen that if the source is a long way from the bottom and from the surface of the sea, the communication is less disturbed for up to 1.0 to 1.2 km, because of the lesser influence of the surfaces that delimit the communication channel. Above this distance, the effects pro- duced by the bottom and the free surfaces are similar, independently of the position of the source. The Vp/Vs ratio does not influence the communication. In fact changing the Vp/Vs ratio in the range illustrated in table I we obtain the 423 Underwater acoustic channel properties in the Gulf of Naples and their effects on digital data transmission F ig . 6. T he s ig na l- to -n oi se r at io f or t he s um m er a nd w in te r m on th s in t he G ul f of N ap le s w it h th e tr an sm it te r po si ti on ed a t a de pt h of 1 m . T he r e- ce iv er g ri d is p lo tt ed w it h w hi te d ot s. A c om pa ri so n be tw ee n th e su m m er a nd w in te r m ap s re la ti ve t o th e di re ct p ha se o nl y sh ow s th at i n su m m er t he ef fe ct o f th e ne ga ti ve v el oc it y gr ad ie nt d oe s no t fa vo r su rf ac e co m m un ic at io n fo r in cr ea si ng d is ta nc es . C on si de ri ng a ls o m ul ti pl e re fl ec ti on s, th e m ul - ti pa th e ff ec t pr ed om in at es o ve r th e ve lo ci ty g ra di en t ef fe ct . F ig . 7. T he s ig na l- to -n oi se r at io f or t he s um m er a nd w in te r m on th s in t he G ul f of N ap le s w it h th e tr an sm it te r po si ti on ed a t a de pt h of 1 90 m . T he re ce iv er g ri d is p lo tt ed w it h w hi te d ot s. F or t hi s so ur ce d ep th t he n eg at iv e ve lo ci ty g ra di en t of t he s um m er m on th s no l on ge r in fl ue nc es c om m un ic a- ti on ; m or eo ve r th e co m m un ic at io n is l es s di st ur be d fo r up t o 1. 0 to 1 .2 k m b ec au se o f th e le ss er i nf lu en ce o f bo th t he s ea a nd b ot to m s ur fa ce s, si nc e th ey a re f ar f ro m t he s ou rc e. 6 7 424 Tony Alfredo Stabile, Aldo Zollo, Maurizio Vassallo and Giovanni Iannaccone same results as reported in figs. 6, 7, and 8. Moreover, from these three figures (6, 7, and 8), it can be seen that the transmission is slightly fa- vored in the summer period due to a lower noise level from agitation of the sea. It is nevertheless important to bear in mind that on average the dif- ference in the noise levels between the summer and winter months due to the conditions of the sea is only 4 dB in the Gulf of Naples. Finally considering the information of fig. 5 we can see that below 50 m depth the mean sound velocity profiles for summer and winter season are similar, so also the SNR is stable during the year for transmitter-receiver connec- tions below this depth. 6. Conclusions Here, we have simulated the propagation of sinusoidal acoustic signals sent by an omnidi- rectional transmitter at the frequency of 100 Fig. 8. The signal-to-noise ratio for the summer and winter months in the Gulf of Naples with the transmitter positioned at a depth of 299 m. The receiver grid is plotted with white dots. For this source depth the negative velocity gradient of the summer months no longer influences communication, but the influence of the seafloor is magnified. 425 Underwater acoustic channel properties in the Gulf of Naples and their effects on digital data transmission kHz for the binary sequential digital transmis- sion of information, in a propagation medium with the physical characteristics of the Gulf of Naples. Three different transmitter positions have been considered, at 1 m, 190 m and 299 m (1 m above the bottom) in depth. The reception of the transmitted signal was simulated, for all three cases, by a 21×12 receivers grid in the sea of 300 m in depth, for a range of 2 km; from the simulation, the SNR maps have been construct- ed for all of the cases under consideration. Considering a transmitter with a source lev- el of 165 dB and a velocity of 2.4 kbit/s, for a band pass of 10 kHz, it is seen that the transmis- sion of information is reliable up to distances of 1400-1600 m between source and receiver be- low 50 m depth and in all seasons of the year, because the mean sound velocity profiles for summer and winter months are similar below this depth. The summer seasonal effects on the sound velocity cannot be overlooked when the transmitter is positioned at the surface or at a low depth. The presence of the sea surface and bottom produce, moreover, important multipath effects that result in errors in the decoding process of the bit sequence of the transmitted information, particularly if the transmitter is positioned near the surfaces limiting the acoustic channel. Although the sea surface and bottom produce multipath effects, the Vp/Vs ra- tio of the sedimentary layers does not influence the communication. If the transmitter is supplied with direction- ality, the source level can then be increased up to 40 dB, and in this case it is possible to reach a distance of 2000 m while minimizing the mul- tipath effects due to the multiple reflections. Figures 6, 7, and 8 that are relative to the direct phase only give indications as to the SNR that would be present in the case in which the trans- mitter is directional; these values must then be increased by the factor of directionality (DI) of the source, which in the case examined in this article is equal to zero, because it involves a non-directional source. This study represents a feasibility test of us- ing the acoustic communication to transmit da- ta acquired on the seafloor of the Gulf of Naples. In this region a complex geophysical monitoring system operates for the surveillance of the Neapolitan volcanoes (Mt.Vesuvius and Campi Flegrei caldera). All the instruments are installed inland, while on the seafloor only tem- porary experiments have been carried out. This study offers a new perspective on the future de- velopments of the monitoring system by the in- stallation of geophysical stations on the seafloor with data transmission by acoustic link to a moored buoy that provides connectivity back to the land by a radio link. Acknowledgements The authors would like to thank the engi- neers of the Whitehead Alenia Sistemi Subac- quei SpA of Monteruscello (NA) for providing the technical specifications of the underwater acoustic transmitters and of the modalities of the underwater acoustic transmission of digital data. We also thank Dr. M. Ribera for providing us with the data on the physical characteristics of the water channel in the Gulf of Naples after the measurements carried out by A. Dohrn Zo- ological Station staff in the period February 2002-February 2003. This work was funded by MIUR (Ministry of Education, University and Research) PON- SisMa project. REFERENCES ADAMS, A., O. HINTON, B. SHARIF, G. SALLES, N. ORR and C. TSIMINEDES (2000): An experiment in sub-sea net- works - The LOTUS sea trials, in Proceedings of the ECUA’00, Lyon, France. AKYILDIZ, I.F., D. POMPILI and T. MELODIA (2005): Under- water acoustic sensor networks: research challenges, Ad Hoc Networks, 3 (3), 257-279. BADIA, L., M. MASTROGIOVANNI, C. PETRIOLI, S. STE- FANAKOS and M. ZORZI (2006): An optimization frame- work for joint sensor deployment, link scheduling and routing in underwater sensor networks, in Proceedings of the First ACM International Workshop on Underwa- ter Networks (WUWNet’06), Los Angeles, California, U.S.A., 56-63. BAGGEROER, A. (1984): Acoustic telemetry – An overview, IEEE J. Ocean. Eng., 9, 229-235. BENSON, B., G. CHANG, D. MANOV, B. GRAHAM and R. KASTNER (2006): Design of a low-cost acoustic modem for moored oceanographic applications, in Proceedings of the First ACM International Workshop on Underwa- ter Networks (WUWNet’06), Los Angeles, California, U.S.A., 71-78. 426 Tony Alfredo Stabile, Aldo Zollo, Maurizio Vassallo and Giovanni Iannaccone CATIPOVIC, J.A. (1990): Performance limitations in under- water acoustic telemetry, IEEE J. Ocean. Eng., 15, 205-216. CAZZOLATO, B.S., P. NELSON, P. JOSEPH and R.J. BRIND (2001): Numerical simulation of optimal deconvolu- tion in a shallow-water environment, J. Acoust. Soc. Am., 110, 170-185. CHEN, C.T. and F.J. MILLERO (1977): Speed of sound in sea- water at high pressure, J. Acoust. Soc. Am., 62, 1129- 1135. COATES, R.F.W. (1993): Underwater acoustic communica- tions, in Proceeding of OCEANS ‘93, Victoria, Canada, III.420-III.425. COATES, R.F.W. (1990): Underwater Acoustic Systems (John Wiley & Sons Ltd., New York), pp. 188. COPPENS, A.B. (1981): Simple equations for the speed of sound in Neptunian waters, J. Acoust. Soc. Am., 69, 862-863. DAHL, P.H. (2004): The sea surface bounce channel. Bub- ble-medianted energy loss and time/angle spreading, in High Frequency Ocean Acoustics, edited by M. PORTER, M. SIDERIUS and W. KUPERMAN (American In- stitute of Physics), 194-203. DE DOMINICIS ROTONDI, A. (1990): Principi di Elettroacus- tica Subacquea (Ed. ELSAG, Genova), Vol. I, pp. 422. DE DOMINICIS ROTONDI, A. (1995): Principi di Elettroacus- tica Subacquea. Le Caratterizzazioni dell’Smbiente Operativo Marino (Edizioni Scriba), vol. II, pp. 256. DE DOMINICIS ROTONDI, A. (1996): rincipi di Elettroacusti- ca Subacquea. Il Canale Acustico Marino (Edizioni Scriba), vol. III, pp. 512. DEL GROSSO, V.A. (1974): New equation for the speed of sound in natural waters (with comparisons to other equations), J. Acoust. Soc. Am., 56, 1084-1091. FARRA, V. and R. MADARIAGA (1987): Seismic waveform modeling in heterogeneous media by ray perturbation theory, J. Geophys. Res., 92, 2697-2712. FISHER, F.H. and V.P. SIMMONS (1977): Sound absorption in sea water, J. Acoust. Soc. Am., 62, 558-564. HAMILTON, E.L. (1978): Sound velocity gradients in marine sediments, J. Acoust. Soc. Am., 65, 909-922. HAMILTON, E. L. (1979): Vp/Vs and Poisson’s ratios in ma- rine sediments and rocks, J. Acoust. Soc. Am., 66, 1093-1101. HAMILTON, E.L. (1980): Geoacoustic modeling of the sea floor, J. Acoust. Soc. Am., 68, 1313-1340. JAFFE, J. and C. SCHURGERS (2006): Sensor networks of freely drifting autonomous underwater explorers, in Proceedings of the First ACM International Workshop on Underwater Networks (WUWNet’06), Los Angeles, California, U.S.A., 93-96. JENSEN, F.B., W.A. KUPERMAN, M.B. PORTER and H. SCHMIDT (2000): Computational Ocean Acoustics (Springer-Verlag), pp. 600. KINSLER, L.E., A.R. FREY, A.B. COPPENS and J.V. SANDERS (2000): Fundamentals of Acoustics (John Wiley & Sons Lodt., New York), 4th edition, pp. 560. KNUDSEN, V.O., R.S. ALFORD and J.W. EMLING (1948): Un- derwater ambient noise, J. Mar. Res., 22, 410-429. LEROY, C.C. and F. PARTHIOT (1998): Depht-pressure rela- tionships in the oceans and seas, J. Acoust. Soc. Am., 103, 1346-1352. MACKENZIE, K.V. (1981): Nine-term equation for the sound speed in the oceans, J. Acoust. Soc. Am., 70, 807-812. NELDER, J.A. and R. MEAD (1965): A simplex method for function minimization, Comput. J., 7, 308-313. POMPILI, D., T. MELODIA and I.F. AKYILDIZ (2006): Deploy- ment analysis in underwater acoustic wireless sensor networks, in Proceedings of the First ACM Internation- al Workshop on Underwater Networks (WUWNet’06), Los Angeles, California, U.S.A., 48-55. PREISIG, J. (2006): Acoustic propagation considerations for underwater acoustic communications network develop- ment, in Proceedings of the First ACM International Workshop on Underwater Networks (WUWNet’06), Los Angeles, California, U.S.A., 1-5. PRESS, W.H., S.A. TEUKOLSKY, W.T. VETTERLING and B.P. FLANNERY (2003): Numerical Recipes in Fortran 77: the Art of Scientific Computing (Cambridge University Press), 2nd edition, pp. 992 (available online at: http://www.nr.com/). SHANNON, C.E. (1948): A mathematical theory of commu- nication, Bell System Technical J., 27, 379-423, 623- 656. SMITH, K.B., A.A.M. ABRANTES and A. LARRAZA (2003): Examination of time-reversal acoustics in shallow wa- ter and applications to noncoherent underwater com- munications, J. Acoust. Soc. Am., 113, 3095-3110. SOZER, EM., M. STOJANOVIC and J. G. PROAKIS (2000): Un- derwater acoustic networks, IEEE J. Ocean. Eng., 25, 72-83. STOJANOVIC, M. (1996): Recent advances in high-speed un- derwater acoustic communications, IEEE J. Ocean. Eng., 21, 125-136. STOJANOVIC, M. (2004): Refocusing techniques for high rate acoustic communications, J. Acoust. Soc. Am., 117, 1173-1185. STOJANOVIC, M. (2006): On the relationship between capacity and distance in an underwater acoustic communication channel, in Proceedings of the First ACM International Workshop on Underwater Networks (WUWNet’06), Los Angeles, California, U.S.A., 41-47. STOJANOVIC, M., J.A. CATIPOVIC and J.G. PROAKIS (1993). Adaptive multichannel combining and equalization for underwater acoustic communications, J. Acoust. Soc. Am., 94, 1621-1631. STOJANOVIC, M., J.A. CATIPOVIC and J.G. PROAKIS (1994): Phase coherent digital communications for underwater acoustic channels, IEEE J. Ocean. Eng., 19, 100-111. STOJANOVIC, M., J.A. CATIPOVIC and J.G. PROAKIS (1995): Reduced complexity spatial and temporal processing of underwater acoustic communication signals, J. Acoust. Soc. Am., 98, 961-672. THORP, W.H. (1967): Analytic description of the low-frequen- cy attenuation coefficient, J. Acoust. Soc. Am., 42, p. 270. URICK, R.J. (1986): Ambient Noise in the Sea (Peninsula Publishing), pp. 205. WENZ, G.M. (1962): Acoustic ambient noise in the ocean: spectra and sources, J. Acoust. Soc. Am., 34, 1936- 1956. WONG, G.S. K. and S. ZHU (1995): Speed of sound in sea- water as a function of salinity, temperature and pres- sure, J. Acoust. Soc. Am., 97, 1732-1736. (received .............; accepted ................) Barbara Evidenziato Barbara Casella di testo Questo riferimento sembra non essere citato nel testo. Prego verificare. Grazie