Improvement of Detection and Localization Performance Using the Receiving Array Response Difference Between Ocean Noise and Signal in Shallow Water

It was observed that when the short vertical line array (SVLA) is located in the deeper part of the water column, where sound velocity is lower, a groove always exists in the receiving array response in the horizontal direction for distant sound sources in the shallower part of the water column, where the sound velocity is higher. Normal mode modeling is used to explain this result. According to the normal mode theory, the receiving array response of the SVLA to a distant sound source can be expressed in terms of modal beams weighted in accordance with the normal mode amplitude. This modal representation offers a physical interpretation of the receiving array response to a distant sound source in terms of normal modes. The environmental effects of the shape of the sound velocity profile and geo-acoustic properties of the seabed on the receiving array response are analyzed. Based on the results, three conditions for the existence of the groove in the receiving array response are obtained: 1) a gradient in the sound velocity profile, 2) an SVLA in a water column in which the sound velocity is lower and low-order normal modes are trapped, and 3) a distant sound source in a shallow water column in which the sound velocity is higher, and acoustic source couples weakly with low-order normal modes and strongly with high-order normal modes. Finally, the receiving array response of the SVLA to ocean noise and distant sound source are analyzed and discussed using the Mediterranean Sea data. It is shown that the receiving array response to ocean noise differs from that to a distant sound source. Utilizing this difference, the array can be steered carefully to improve the output signal-to-noise ratio and increase the passive detection range against a submerged target in shallow water.

ABSTRACT It was observed that when the short vertical line array (SVLA) is located in the deeper part of the water column, where sound velocity is lower, a groove always exists in the receiving array response in the horizontal direction for distant sound sources in the shallower part of the water column, where the sound velocity is higher. Normal mode modeling is used to explain this result. According to the normal mode theory, the receiving array response of the SVLA to a distant sound source can be expressed in terms of modal beams weighted in accordance with the normal mode amplitude. This modal representation offers a physical interpretation of the receiving array response to a distant sound source in terms of normal modes. The environmental effects of the shape of the sound velocity profile and geo-acoustic properties of the seabed on the receiving array response are analyzed. Based on the results, three conditions for the existence of the groove in the receiving array response are obtained: 1) a gradient in the sound velocity profile, 2) an SVLA in a water column in which the sound velocity is lower and low-order normal modes are trapped, and 3) a distant sound source in a shallow water column in which the sound velocity is higher, and acoustic source couples weakly with low-order normal modes and strongly with high-order normal modes. Finally, the receiving array response of the SVLA to ocean noise and distant sound source are analyzed and discussed using the Mediterranean Sea data. It is shown that the receiving array response to ocean noise differs from that to a distant sound source. Utilizing this difference, the array can be steered carefully to improve the output signal-to-noise ratio and increase the passive detection range against a submerged target in shallow water.
INDEX TERMS Localization, receiving array response, groove, signal-to-noise ratio, sensitivity.

I. INTRODUCTION
It is very difficult to model and predict the ocean noise in shallow water [1]- [3]. The reason is that ocean noise in shallow water is strongly dependent on time and space [4], [5]. If there The associate editor coordinating the review of this manuscript and approving it for publication was Honghao Gao. exists an isothermal layer in the shallower part of the water column, the receiving array response of the short vertical line array (SVLA) to ocean noise divides into two arms enclosing a region in which the ocean noise strength is very small. Hereafter, this structure is referred to as a groove. The main reason for this phenomenon is that the downward refraction of acoustic rays causes the ocean noise to propagate forward at a steeper angle [6]. However, changes to the ocean environment may redistribute the ocean noise into a shallower angle, which in turn fills the groove. For example, when subjected to ocean internal wave, strong mode coupling may occur during sound propagation, which may cause the groove of the ocean noise to be weakened or even disappear [7].
For a given sea area, the receiving array response to ocean noise over a relatively short period of time is often relatively stable. When the receiving array response of the SVLA to ocean noise forms a groove in the horizontal (0 • ) direction, this provides a very powerful window for the sonar to detect the signal reaching the transducer array from the 0 • direction. However, due to the multi-path effect in the shallow water, the signal radiated by the distant sound source (the term distant in this study means that the distance between the sound source and the receiver is larger than five time the water depth) tends to split into beams with different elevation angles, so that the signal does not always reach the transducer array from the 0 • direction.
In fact, the receiving array response to distant sound sources is mainly affected by the physical mechanism of sound propagation in shallow water [8]. The sound pressure field can be expressed as an expansion of normal modes by the normal mode theory. The receiving array response to distant sound sources in shallow water can be factored into modal beams. A normal mode is the sum of down-going and up-going plane waves. Both these normal modes have the grazing angles θ m = arctan(k zm /k rm ), where k zm and k rm are the vertical and horizontal wavenumbers, respectively. Loworder modes correspond to small grazing angles, while highorder modes correspond to large grazing angles.
Underwater networks provide a means for data collection and positioning [9]- [11]. Methods used in cyber-physical systems and hybrid networks, such as particle swarm optimization [12], [13], can also be used in underwater acoustic sensor networks to improve positioning performance. Recently, a stratification-based data collection scheme in underwater acoustic sensor networks was found to be able to combine the advantages of a multi-hop transmission scheme and autonomous underwater vehicle (AUV)-aided data collection scheme to reduce network consumption and improve network lifetime [14].
Specifically, the following are the contributions of this work to the literature.
(1) It is observed that when the SVLA is located in the lower slow sound speed water column in shallow water, a deep notch of the spatial response always exists in the near horizontal direction for sources at the upper high sound speed water column.
(2) The normal mode modeling is used to explain this result. According to the normal mode theory, the receiving array response of the SVLA to a distant sound source can be expressed in terms of the modal beams weighted in accordance to the normal mode amplitude. This modal representation offers a physical interpretation of the receiving array response to a distant sound source in terms of normal modes.
(3) The environmental effects of the shape of the sound velocity profile and geo-acoustic properties of the seabed on the receiving array response are analyzed. Three conditions for the existence of the groove in the receiving array response are obtained.
(4) Experimental data from the Mediterranean Sea show that the receiving array response to ocean noise differs from that to a distant sound source. An approach based on this difference is presented to improve the output signal-to-noise ratio (OSNR) in shallow water by controlling the SVLA receiving direction.
The remainder of this paper is organized as follows. In section II, the normal mode method is used to theoretically derive the receiving array response of an SVLA to a distant sound source, and represent it as a modal beam weighted summation whose weight value depends on the amplitude of the normal mode. In section III, the receiving array response of an SVLA to a distant sound source in the typical summer environment of shallow water is analyzed by computer simulation. Section IV discusses environmental effects on receiving array response of an SVLA to a distant sound source. In section V, a shallow water experiment on Elba Island in the Mediterranean is briefly introduced, and the experimental data is used to analyze the receiving array response of an SVLA to ocean noise and a distant sound source. A summary and discussion of the findings of the study is presented in Section VI.

A. PLANE WAVE BEAMFORMING
Let the signal incident from the θ 0 direction, the output on the i-th hydrophone is x i (θ 0 ). Under the narrow band approximation, the data received by all hydrophones are weighted and summed to obtain the output signal of the line array. Thus, and w (θ) is the complex weight vector. For an equally spaced line array, the complex weight vector can be expressed as where θ is the steering angle, k is the wavenumber, and d is the array spacing. For a given direction, the conventional beam power of the signal or noise field is where R x (θ 0 ) = E[x(θ 0 ) * x(θ 0 ) T ] is the covariance matrix of the data vector x(θ 0 ). When the incident direction of the signal is the same as the direction of observation, the maximum value of Equation (4) is reached. The sound pressure field can be expressed as an expansion of normal modes. The eigenvalue and eigenfunction can be obtained by solving the Helmholtz equation which satisfy certain boundary conditions. In cylindrical coordinates, the sound pressure field is the sum of each normal mode. Thus, where ρ (z s ) is the medium density at the sound source, k rm and ψ m are the horizontal wavenumber and eigenfunction of the m-th mode, respectively, and M is the whole normal mode number. For a given ocean environment, the KRAKEN [15] normal mode calculation software can be used to calculate the mode function and the corresponding horizontal wavenumber very efficiently and accurately. Therefore, given the amplitude of the normal mode and the horizontal wavenumber, the sound pressure field at any position of the sound source at the receiving array can be computed quickly.

C. RECEIVING ARRAY RESPONSE
Considering the harmonic process, omitting the time factor exp(−iωt), the received data vector of the vertical line array at the distance r is x (θ 0 ) = [p(r, z 1 ; r s , z s ), p(r, z 2 ; r s , z s ), · · · , p(r, z N ; r s , z s )] T (6) For the sake of convenience, we analyze the array beam y (θ). Substituting Equations (5) and (6) into Equation (1), Introducing vector symbols, Equation (7) can be written as The normal mode amplitude is The modal beam B m is Thus, Equation (9) becomes The receiving array response of the SVLA can be expressed as the weighted summation of the modal beams. The contribution of each modal beam is determined by the amplitude of the normal mode.

III. COMPUTER SIMULATION A. SHALLOW WATER WAVEGUIDE
Simulation data were used for experimental studies. A typical range independent shallow water environment was taken as the simulation environment, as shown in Figure 1. The water depth is 129 m, and the sound velocity profile is shown by the curve. The acoustic parameters of the sedimentary layer and the sub-bottom are also given in the figure. It can be seen from the figure that the sound velocity profile is a typical summer sound velocity profile. At a depth of 60 m from the sea surface, the sound velocity is approximately equal to that at the surface, and there is a strong transition layer at 60-80 m. It is assumed that the sea surface is a pressure release surface. Ocean noise is not considered, and it is assumed that the entire ocean environment does not change over time during processing.

B. RECEIVING ARRAY RESPONSE
Twenty-one hydrophones form a uniform SVLA. The depth of the SVLA is equal to the depth of the SVLA center. The spacing is equal to half the wavelength. The signal radiated by the source is a rectangular waveform, with a frequency of 1500 Hz and 50 cycles. Figure 2 shows the signal waveform of the source radiation and its spectral structure. This frequency was selected for the simulation primarily to obtain a small-aperture SVLA of approximately 10 m. The horizontal distance between the sound source and the SVLA is 5.0 km. Narrow band time domain (NBTD) beamforming is used here to analyze receiving array responses. The signal covariance matrix can be estimated by the NBTD snapshot model where x(n) is the data vector. The beam power can be obtained by Equation (4). The beam scanning space of 180 • is discretized into 1801 angles in increments of 0.1 • , and then the beam output of each discrete angle is calculated separately. For convenience, let the horizontal direction be 0 • . Figure 3 shows the receiving array response of the SVLA as a function of source depth. Positive steering angles indicate beam steering towards seafloor to capture signals that propagate upwards by seafloor reflection; negative steering angles correspond to beam steering towards the sea surface to capture signals propagating downwards by sea surface reflection. Figure 3(a) corresponds to the case of an SVLA of 50 m depth. As can be seen from the figure, the main vertical lobes of the receiving array response appear in the 0 • direction. For some source depths, the receiving array response exhibits small grooves in the 0 • direction, but the existence of these small grooves is irregular. Figure 3(b) corresponds to the case of an SVLA of 100 m depth. It can be seen from the figure that for the sound source in the shallower water, the receiving array response exhibits a deep groove in the 0 • direction, and they are clearly visible. This means that the maximum beam power of the SVLA has deviated significantly from the 0 • direction. As the depth of the sound source increases, the groove of the receiving array response gradually weakens until it disappears. At this point, the main lobe of the receiving array response returns to the 0 • direction.

C. PHYSICAL EXPLANATION
Each order of normal mode can be easily calculated by KRAKEN software. For a sound source with a frequency of 1500 Hz, the shallow sea environment shown in Figure 1 can support approximately 81 normal modes. One can obtain the angular distribution of the modal beam by using Equation (11), as shown in Figure 4. The modal beams corresponding to each normal mode are symmetric about the 0 • direction. The reason is that each normal mode can be regarded as a pair of plane waves incident on the upper and lower boundaries at a certain angle. The grazing angle of each normal mode is where f is the signal frequency, and k rm is the horizontal wavenumber, which is independent of depth. Given that the sound velocity c(z) depends on depth, the vertical wavenumber k zm also depends on depth as the grazing angle θ m (z) is a function of depth. When the vertical wavenumber k zm at some depth tends to zero, the grazing angle will also become zero. Figure 4(a) corresponds to an SVLA at 50 m depth. As the low-order normal modes are restricted to the deeper part of the water column, they are negligible compared to the high-order normal modes. Only when the normal mode order is greater than 19, does the high-order normal mode contribute to the received sound field of the SVLA. It is noted that as the order of the normal mode increases, the grooves in the 0 • direction gradually become deeper and wider. Figure 4(b) corresponds to an SVLA at 100 m depth. When the SVLA corresponds to the deeper part of the water column, and the low-order normal mode contributes to the received sound field of the SVLA. Similarly, as the order of the normal mode increases, the grooves in the 0 • direction gradually become deeper and wider. Figure 5 shows the normalized normal mode amplitudes for different sound source depths. The source frequency is 1500 Hz, the distance between the source and SVLA is 5 km, and the sound source depth is shown for 6 m and 100 m. For quantitative analysis of the amplitude of the normal mode, see Equation (10). When the sound source depth is 6 m, the amplitude of the first 19 normal modes are 100 dB smaller than the strongest normal mode, as shown in Figure 5(a). Thus, low-order normal modes do not contribute to the sound pressure field, and do not contribute to the receiving array response. In contrast, when the sound source is located in the deeper part of the water column, it can excite low-order normal modes, as shown in Figure 5(b).
The low-order normal modes are mainly restricted to the deeper part of the water column, while the high-order normal modes can be distributed throughout the whole water column. Therefore, sound sources in the shallower part of the water column, where the sound velocity is higher, can be strongly coupled with the high-order normal modes. Sound sources in the deeper part of the water column, where the sound velocity is lower, can excite all normal modes efficiently. The receiving array response can be expressed as the summation of modal beams B m . Based on the above analysis, we can conclude that the weak groove in the 0 • direction is irregular, only occurring when the SVLA is located in the shallower part of the water column, where the sound velocity is higher. When the SVLA is located in the deeper part of the water column, where the sound velocity is lower, the deep groove in the 0 • direction is visible over all the sound sources in the shallower part of the water column. However, as the sound source is moved towards the deeper part of the water column, where the sound velocity is lower, the groove disappears.

IV. ENVIRONMENTAL EFFECTS ON RECEIVING ARRAY RESPONSE
The simulations show that there indeed exists a groove at 0 • when a distant sound source is located in a shallower part of the water column, where the sound velocity is higher and the array is in the deeper part of the water column, where the sound velocity is lower, for a typical downward refracting summer environment. However, many factors affect the propagation of sound in shallow water [16]. The most important of these are the shape of the sound-velocity profile (SVP) and the bottom type. Other parameters, including the roughness of the seabed, surface disturbances, random inhomogeneities in the water layer, and sea currents, are less important at lower frequencies.
This section presents the results of numerical simulations of the receiving array response of the SVLA under various conditions, such as for different frequencies, different seabed types, and different SVPs. In what follows, the deviation angle θ is defined as the difference between the steering angle of maximum beam power and the 0 • direction, and P is defined as the difference between the maximum beam power and the beam power at 0 • , i.e., the horizontal sidelobe level.

A. EFFECT OF FREQUENCY
The receiving array response of the SVLA to a distant sound source was investigated for three different frequencies (300, 1500, and 3750 Hz). The total length of the SVLA for the lowest and highest frequency signals were 50 and 4 m, respectively. Figure 6 shows the characteristics of receiving array response of the SVLA at different depths with the distant sound source depth covering the whole water column. The deviation angle θ and P are shown in the upper and lower plots, respectively. The left plots correspond to the low frequency case. The right plots correspond to the higher frequency case.
The deviation angle θ is large and P is small when the SVLA lies in the lower water column and the distant sound source is located at the thermocline and in the upper column. This pattern is very clear for the high frequency cases. For the low frequency case, that pattern is much less clear, with the groove at 0 • becoming very irregular. The reason for this is that the number (18) of normal modes excited by the lower frequency source is small, and the beam pattern becomes more sensitive to the source depth and array depth.  Figure 7 shows the characteristics of the beam pattern of the SVLA at different depths, with the distant sound source depth spanning the whole water column. The deviation angle θ and P are shown in the upper plots and lower plots, respectively. The plots on the left correspond to the soft bottom case. The plots on the right correspond to the hard-bottom case.
Inspection of Figure 7 shows that θ is large and P is small when the SVLA lies at the thermocline and in the deeper part of the water column and the distant sound source is located at the thermocline and in the shallower part of the water column. Thus, the beam pattern has a wide groove at 0 • . When the array and the sound source are located at the same depth, θ and P are approximately 0. This means that there is only a main lobe around the 0 • direction. Furthermore, when there is a wide groove, there is a deviation angle θ that is larger for the hard-bottom case than for the soft one. This is because the soft bottom strips away some of the high-order normal modes, while the acoustic energy consists of more normal modes in the hard-bottom case.

C. EFFECT OF SOUND VELOCITY PROFILE
Three SVPs, corresponding to an iso-velocity condition (SVP1: 1500 m/s), a linear downward-refracting summer condition (SVP2: 1520-1500 m/s), and a linear upwardrefracting winter condition (SVP3: 1500-1510 m/s), were examined in this study. Figure 8 shows the characteristics of the beam pattern of the SVLA at different depths for a source depth that spans the whole water column. The deviation angle θ and P are shown in the upper plots and lower plots, respectively. The plots on the left, in the middle, and on the right correspond to SVP1, SVP2, and SVP3, respectively.
The iso-velocity results of the left plots show that the existence of the groove is irregular. In the SVP2 case, the deviation angle θ is large and P is small when the SVLA lies in the deeper part of the water column and the distant sound source is in the shallower part of the water column. These results are similar to those obtained for the typical summer profile in the Mediterranean, as shown in Figure 1. Conversely, the deviation angle θ is large and P is small for the SVP3 case, when the SVLA lies in the shallower part of the water column and the distant sound source is in the deeper part of the water column. For this case, the area of the groove at 0 • is smaller than for the SVP2 case because the linear upward-refractive index is smaller.
Using Equation (1), one can interpret and predict the directionality of a distant sound source in terms of environmental acoustic parameters. The conditions that result in a groove at high frequencies include: (1) a gradient in the sound velocity profile, (2) an SVLA in a water column in which the sound velocity is lower and low-order normal modes are trapped, and (3) a distant sound source in a shallow water column in which the sound velocity is higher and the acoustic source couples weakly with low-order normal modes and strongly with high-order normal modes.
The coherence and receiving array response of the SVLA to ocean noise depend on the ocean noise distribution and environment in shallow water. The receiving array response of the SVLA to ocean noise is often stable for short time periods, whether the ocean noise groove is present or not. However, the receiving array response of the SVLA to source signal changes with the source depth. When there is an ocean noise groove, it creates a window through which one can observe the signal arriving in the 0 • direction. If the ocean noise field is centered in the 0 • direction but the receiving array response of the SVLA to acoustic source is relatively strong in other directions, one can steer the SVLA to the strongest signal power direction to improve the OSNR.

V. EXPERIMENTAL DATA ANALYSIS A. EXPERIMENT DESCRIPTION
In October 1993, the SACLANT Center organized a shallow sea experiment near Elba Island in the Mediterranean [17]- [19]. The experiment used a vertical receiving array to collect shallow water data. Although the main purpose of the experiment was to test the proposed matching field inversion algorithm, we can still use the data collected by its vertical line array to verify the fact that the receiving array response of the SVLA to distant sound sources has a groove in the 0 • direction. The experiment was conducted for two days, from October 26 to 27. The location of the experimental sea area is shown in Figure 9. During the experiment, the wave heights in the experimental sea area ranged from 0.2 to 1.0 m. On the first day, the wind speed on the sea surface was 12 to 20 knots, and the wind speed on the second day was 7 knots. The entire experimental sea area can be approximated as a flat seabed covered with a layer of clay and sediment. The sound velocity profile and the ground acoustic parameters were measured during the experiment, and the obtained baseline model of the environmental parameters was the same as that of Figure 1. The sound velocity in water is a typical negative gradient summer sound velocity profile. The depth of the experimental sea area varies between 120 and 140 m.
In Figure 9, point A represents the position of the vertical receiving array, and point B represents the position of the sound source buoy, which are approximately 5.6 km apart. The receiving array is surrounded by three monitors arranged in equilateral triangles for monitoring changes in the formation. During the first day of the experiment, the acoustic source was fixed, anchored at point B with a buoy, and in the next day's experiment, a ship was used to drag the acoustic source from point A to point B. The depth of the vertical receiving array was approximately 127 m, as shown in Figure 10. The vertical receiving array of this experiment was a 48-element array. The depth of the first hydrophone was 18.7 m, the depth of the last one was 112.7 m, the spacing of array elements was 2 m, and the aperture of the array was 94 m. Note that the above array parameters are ideal depths assuming that the vertical receiving array does not shift. In fact, due to the movement of sea water, etc., the array formation is constantly changing. Three monitors were used in the experiment to monitor changes in the array formation. The monitoring results show that the maximum deviation of hydrophone 48 (closest to the sea are surface) was no more than 1 m. The data acquisition system samples the output of the 48-element vertical receiving array at a sampling rate of 6 kHz and low-pass filters the signal with a 255-order finite-impulse response (FIR) low-pass filter. The lower limit of the filter is 420 Hz and the transition bandwidth is 60 Hz. The response of the filter at 408 Hz is 60 dB lower than the passband response. To reduce the memory space, the output of the filter is down-sampled so that the output sampling rate is 1 kHz.
The fixed sound source was approximately 5.6 km away from the receiving array, and the used buoy is anchored on VOLUME 7, 2019 the seabed, as shown in Figure 11. Due to the influence of wind speed, the surface buoy moves, causing an error of approximately 200 m in the sum of the transmission and reception distances.  The emission source is named as HX-90G, and its frequency response curve is shown in Figure 12. At the sound source location, the sea depth is approximately 130 m, and the sound source is approximately 79 m below the sea surface. The transmitting sound source mainly emits two signals. The first signal, called RM5, is a pseudo-random sequence generated by a 6-bit shift register with a length of 52.9 ms. This sequence is modulated onto a carrier frequency with a center frequency of 170 Hz. The repeat length of this sequence is 3.15 s, the −3 dB bandwidth is approximately 12 Hz, and the sound source level is approximately 163 dB (relative to 1µPa/ √ Hz). The second signal, called RM2, is also generated by the 6-bit shift register, but each bit is 20 ms, and the carrier frequency is 335 Hz. The sequence has a repeat length of 1.3 s, a −3 dB bandwidth of approximately 30 Hz, and a sound source level of approximately 164 dB (relative to 1µPa/ √ Hz).

B. OCEAN NOISE
Moving sound source experiments were conducted on the afternoon of October 27. The HX-90G sound source was dragged by the ship from point A to point B. Unlike the fixed sound source experiment, the pseudo-random signal is transmitted intermittently at an interval of 30 s. Although the sound source only transmits signals for 30 s per minute, the data acquisition system samples the output of the 48-element hydrophones continuously. Therefore, when the sound source does not emit a signal, pure ocean noise data near 335 Hz can be obtained. The ocean noise frequency was chosen to be 335 Hz for comparison with the signal data center frequency, which was analyzed later. This ocean noise was used to analyze the receiving array response of the SVLA to ocean noise. The ocean noise data is first filtered by a 333-337 Hz bandpass filter, making it narrowband ocean noise. The narrowband time domain snapshot model is used to sample the stable 25 s ocean noise data to estimate its covariance matrix R x (θ 0 ). Figure 13 shows the receiving array response of the SVLA to this ocean noise. The SVLA consisted of 21 receivers. The depth of the SVLA is represented by the mid-point of each sub-array. Weak ocean noise grooves existed at 0 • for subarrays in the deeper part of the water column where the sound velocity was lower. The ocean noise in the groove was approximately 3-4 dB lower than the local maxima. When the SVLA corresponded to a shallower part of the water column where the sound velocity was higher, the weak ocean noise groove disappeared, and the ocean noise level peaked in the 0 • direction.

C. SIGNAL
This section uses the receiving array response of the SVLA to the RM2 source signal of the distant sound source. The center frequency of this sound source is 335 Hz, the data sampling rate is 1 kHz, and the −3 dB bandwidth is approximately 30 Hz. A total of approximately 10 minutes of data is obtained. It is worth noting that the experimental data includes both distant sound source signals and ocean noise. The hydrophone input signal-to-noise ratio (SNR) is approximately 10 dB. Given that the main concern is the narrowband signal, narrowband time domain beamforming method is utilized to analyze the receiving array response to distant sound source signals. The received experimental data is filtered by an FIR bandpass filter with a passband of 333-337 Hz to make it into a narrowband signal. The signal covariance matrix is estimated using the average of 60 s of experimental data.   Figure 14(a) is the result of numerical simulation using the inversion parameter model (environmental parameters obtained by inversion of fixedpoint emission signals in the same sea area in [17]- [19]). Figure 14(b) corresponds to the experimental data. A rectangular pulse signal is used for simulation, with a center frequency of 335 Hz, and length of 20 cycles. The groove in the 0 • direction is manifest.
It is noted that the groove of the receiving array response of the experimental data is deeper than that obtained by numerical simulation through the inversion model. The main reason is that the ocean environmental parameters are not the same for the simulation and experiment. In addition, the experimental data contains ocean noise components.

D. IMPROVEMENT OF OSNR
The coherence and directionality of the ocean noise depend on the noise-source distribution and environmental parameters in shallow water. The receiving array response of the SVLA to the ocean noise is stable over short time periods, regardless of the ocean noise groove. However, the receiving array response of the SVLA to the source signal changes with the source depth. Figure 15 shows the power spectra of the SVLA output obtained using the experimental data. The SVLA is comprised of the top 21 receivers from 18.7 m to 58.7 m. Figure 15(a) shows that the spectral components of the source signal were apparent when the SVLA was steered to the 0 • direction, and have an average OSNR of 13.6 dB. When the SVLA was steered to the highest power direction, the average OSNR was 23.7 dB, as shown in Figure 15(b).  Figure 15 illustrates the significance of the vertical directionality of the signal. The OSNR improved by approximately 10 dB when the array was steered from the 0 • direction (groove) to the direction of the highest signal power. However, steering the array to the direction of the highest signal power will not always yield the largest OSNR, as the ocean noise field may also be strong in that direction. However, the receiving array response to ocean noise and distant sound sources are not the same, and the array could be steered carefully using this difference to improve the OSNR.

VI. SUMMARY AND DISCUSSION
In this paper, the receiving array response of an SVLA in shallow water to a distance sound source is studied. By using the normal mode method, the receiving array response of the SVLA in shallow water to a distance sound source is analyzed by theoretical derivation. The receiving array response of the SVLA can be expressed as the weighted summation of the modal beams. The contribution of each modal beam is determined by the amplitude of the normal mode.
The low-order normal modes in the summer environment investigated were restricted to the deeper part of the water column, where the sound velocity is lower. When the SVLA was located in the deeper part of a water column, where the sound velocity is lower, a groove always occurred in the receiving array response in the near horizontal direction for sources in shallower parts of the water column, where the sound velocity is higher. This phenomenon was simulated for different seabed types and SVPs.
The receiving array response of the SVLA to ocean noise and distant sound sources were analyzed using experimental data. Experimental data from the Mediterranean Sea show that the receiving array response of the SVLA to ocean noise and distant sound sources are different. An approach to improve the OSNR (by approximately 10 dB) in shallow water by controlling the SVLA receiving direction is presented. The results can be used to improve the detection and localization of submerged targets.
Both the theoretical analysis and computer simulation use a typical range-independent shallow water waveguide, which is different from the actual ocean environment. To better apply the research results of this paper to an actual sonar system, it is necessary to consider the more complex shallow water environment in future theoretical and applied studies, such as the variation of the seabed with distance, fluctuation of the sea surface, and internal waves in the seawater. In addition, it is necessary to model ocean noise and evaluate its groove effect, which would require a large number of maritime observation data.
The prototype of the underwater Internet of Things (IoT) is an underwater wireless sensor networks (UWSNs), and there are still many problems of acoustic detection, localization, communication, network routing, and optimization [20]- [24]. In this paper, we mainly discuss acoustic detection and localization problems in shallow water. At the same time, we also want to call on more IoT experts to solve the problems of the underwater IoT and UWSNs, and jointly promote the development of the underwater IoT and UWSNs. He has authored more than 30 papers published in related international conference proceedings and journals, including IEEE ACCESS, the Journal of Navigation, Review of Scientific Instruments, and the International Journal of Advanced Robotic Systems. He is the holder of ten patents. His research interests include ocean acoustic modeling, signal processing, sonar engineering, underwater positioning, and navigation.