Underwater Channel Characterization to Design Wireless Sensor Network by Bellhop

Acoustic underwater link due to low attenuation is employed for underwater sensor network (UWSN). Due to water changing physical properties and different environmental conditions, the sound propagation for every underwater channel are different. So, the most important parameters to design reliable UWSN are: the shadow zone determination, optimum nodes placement, high signal to noise ratio and more power efficiency of nodes. This paper simulates underwater channel for 7 KHz carrier frequency, with measured physical water properties and remote sensing data. Based on defined scenario, the research determines optimum nodes placement and link budget analysis based on bottom to surface acoustic link. The paper utilizes Bellhop acoustic toolbox as a simulator, GeoMapApp program to collects bathymetry data and Aviso+ database to determine wind speed on sea surface. As a result the paper based on simulation of sound propagation in channel and transmission loss determination in depth and range, finds the optimum nodes positions and link budget calculation to prove the results.

 cosβ cosβ c c (1) Accuracy condition: Refraction is cause of different speed between two mediums. The transmitted ray has three conditions: First, ray penetrates to other media. Second, reflection ray travels between two  Bahrami) 111 media interface (critical angle), the Equation (3) represents the formula. Third, the ray comes back to its incident media again (total reflection), when this phenomenon occurs which transmitted ray angle is higher than critical angle. The Figure 1 shows the total reflection. A wave path between two points will be named an acoustic ray [2].
When the media layer is more than two then the law becomes like as Equation (4):  (4) The refraction relation along coordinate z can be generalized into: The Figure 2 shows the refraction waves with discontinuous and continuous sound speed distribution with depth [2].

Signal Attenuation and Loss in Underwater
Signal attenuation is due to the viscous absorption phenomenon, it refers to fluid resistance against flow. The boric acid and magnesium sulfate salt are the two important seawater materials, which are causing this phenomenon. The signal attenuation is heavily dependent on employed frequency. If the carrier frequency is higher than hundred hertz, the attenuation can be described by Thorp equation [2,3].
The propagated signal is faced to surface agitation, this occurrence is reason of signal dispersion and energy loss. The surface loss is computable by many models such as: Eckhart, Schulkin-March is [2,3].
The signal after hitting to seafloor deflected and loses its intensity as well. The bottom loss is dependent on seafloor coverage and topography, and based on utilized frequency is calculable by LFBL, HFBL and HFEVA model [3].

Underwater Noise
Generally, underwater noise divided into four sections: turbulence, shipping, waves and thermal noise [4]. The extent of the frequency in each group is different for example: turbulence noise is extended for f < 10 Hz and this range for shipping noise is the frequencies between 10 to 100 Hz. The most important noise that can cover acoustic band is surface wave noise, which is spread on 100 Hz until 50 KHz. The equation (7) described wave noise [3,5].

Underwater channel
The significant of underwater channel characteristic is time-varying behavior. Every phenomenon which is able to change medium such as surface agitation, seafloor undulation and physical properties of water, are the reasons of channel time-varying property [6]. Underwater channel is consist of many ducts that separated by different sound speed. This difference is the result of changes in physical water properties (temperature, salinity and depth) [2,3]. To simulate accurate underwater channel model, the remote sensing information from network nodes positions and CTD (conductivity, temperature, depth) data are essential.

Sound speed
Sound velocity in water is function of pressure, temperature and salinity. It means the sound speed between surface and sea bottom has variation (depend on depth), and it will modify with climate changes and seasons as well. There are many models such as Mackenzie, UNESCO, Chen and Millero for SSP calculation, based on the water physical properties range. This paper selected Mackenzie formula regarding to case study CTD measurement. This equation is accurate for the temperature 0 to 35 °C, salinity 0 to 40 parts per thousand and depth 0 to 1000 m. The equation (8) shows the Mackenzie's equation [2,3].
Where T is temperature in degrees Celsius, S is salinity in parts per thousand and D is equal to depth in meters.

Underwater node
Nowadays, the researches focus on sensor nodes with smaller size, ultra power saving, more storage to buffering, high process capacity and more bandwidth efficiency. In fact, underwater nodes are modems which sensors are connected to them. The node task is information collection from sea and transmitted to other nodes, generally. This information may include data from Oxygen, sediment, CTD sensors or processed data such as positioning, detection and tracking, exploring procedures [7].
The node maintenance and installation process in seafloor due to nodes nonavailability are expensive and costly. So, the battery replacement in short time is impossible. This problem will be solving with reliable design of communication link and best node placement. In this paper, all simulations are based on acoustic releases modem by 867-A model from Teledyne Benthos Company. The Table 1 shows features of this modem [8]. propagation in water columns with consideration of sound speed profile, top and bottom boundaries, reflections and Tx, Rx qualifications. For study on theoretical background refer to reference [9]. The Figure 3 depicts the input and output files for the toolbox. All information from channel quality, Tx and Rx characteristics, altimetry and bathymetry data are coded in environmental file as an input (file.env) [9][10][11].
Bellhop is able to analyze and calculate signal transmission loss in different ranges and depths, over the channel. The acoustic toolbox outputs are: sound speed profile, transmission loss, Impulse response, Eigen rays and time series in receiver based on simulated channel. This paper focuses on overall transmission loss achievement by underwater channel simulation through Bellhop toolbox [12]. In this section, the paper explains all simulation procedures, the Figure 4 shows methodology chart. The simulation refers to Table 2 assumption has been done.  After determination of the case study area, the GeoMapApp program is utilized to collect seafloor geographic data [13]. The research is employed Aviso+ online database, to obtain surface wind speed [14]. The salinity, temperature and depth of position are measured on November 2013 as an experimental measurement. The SSP is determined by Mackenzie model, and all these information are coded to bellhop toolbox as an environmental file, then the TL evaluation process runs to determine the best node placement. Finally, the link budget is calculated.

Environmental Data
The environmental records are accessible from online databases such as: National Oceanic and Atmospheric Administration (NOAA) [15] or Woods Hole Oceanographic Institution (WHOI) [16], Bedford Institute of Oceanography (BIO) [17]. But, this paper, utilized self measurement of case study on November 2013 in Desaru beach on eastern shore of Johor in Malaysia, on the South China Sea. The link starts on (104.875° E, 1.625° N) until (104.625° E, 1.625° N). The distance between nodes is near to 0.25 degree longitude, around 28 Km. Figure 6 shows nodes location, Node1 duty is broadcast signal to Node2, by means of simplex communication link.

Altimetry and Bathymetry Data
To get altimetry data, the paper employed Aviso+ online database. This database collects surface agitation data from Aviso+ satellite that is based on radar technology [14]. Ocean bathymetry and topography obtained by GeoMapApp program [13].

Results and Discussion
In this section, the channel simulation with transmission loss determination in different ranges and depths are considered. All simulated results are based on real data that collected from mentioned sources. So, the results are very accurate and reliable.

Sound Speed Profile
The Table 3 is shown physical properties of case study points, based on experimental measurement on November. The Figure 6 represents SSP graph, which is obtained with Mackenzie formula. Based on Figure 5 the maximum sound speed is on 30 meter depth with 1540.267 (m/s) and minimum is 1540.127 on surface.

Sound Propagation
According to our scenario the Tx is placed on seafloor on 50 meter depth. Based on Figure 8, signal travelling start from 50 meter and go directly until 15 Km and due to seabed gradient, the signal bent upward to surface in 22 Km approximately. After this point the signal travels path from bottom to surface, repeatedly. Based on acoustic propagation graph and network application, the decision to choose the best node position for network designer is possible. The aims of paper are node placement and link budget calculation. Then the graph shows 22 until 22.5 Km range is suitable for buoy position. With transmission loss determination the network designer is able to find out the best range and receiver depth to maximize signal to noise ration and more power efficiency.

Transmission Loss in Range
The Figure 9 represents distribution of acoustic signal transmission loss, while signal is travelling in channel. To focus on transmission loss on surface, the paper simulates TL with the receiver node placement on 2 and 5 meter below than surface, supposedly. According to Figure 10, the minimum TL for 2 and 5 meter depths are 22.15 and 22.25 Km range. Figure 9. Signal TL in channel Figure 10. TL versus depth

Transmission Loss in Depth
After range determination, the paper should finds out the best depth position. So, the Figure 11 shows the TL for Two found ranges. Finally, excellent depth position for 22.15 Km is 2.92 meter with 50 dB TL and this value for 22.25 Km range is 2.1 m with 61.5 dB TL. So, the link budget calculation for designer is possible. Figure 11. TL versus depth

Link Budget
With calculated overall transmission loss which is included signal attenuation, surface and backscattering loss, the signal to noise ratio by Equation (9) is assessable [2][3][4][5].

SNR=SL-TL-NL+DI ≥ DT
Where SL is transmitted signal level, NL is noise level, DT is detection threshold and DI refers to directivity index that is calculated from intensity of a directional source divide to intensity of an Omni directional source of equal power. The Figure 12 shows SNR values for 0 to 28 Km based on (Table 1) and wave noise calculation (Equation 7). Where the transmitted power is 192 dB, Noise level is 47.14 dB and DI is 4.7dB and detection threshold is 60 dB. According to Figure 12 the receiver node on surface, with the best mentioned position has 30 dB received power more than detection threshold.
The Figure 7 shows, the sound speed is increased until 30 meters, and between 30 to 40 m is decreasing. The Figure 8 demonstrates, if the velocity increases enough, an acoustic path inclined relative to the straight can undergo total reflection. On the other hand, a decrease in speed will increase the grazing angle of the signal. Based on Figure 10 around 22 Km the signal has two peaks near to surface, the stranger peak has TL values between 50 until 61.5 dB. The graph 11 determines the best node position in depth. As a result the best receiver node position as a buoy stands on 22.15 Km range and 2.92 meter depth with the minimum overall TL with 50 dB value. So, based on link budget calculation the received power has 30 dB power more than detection threshold in optimum point, but after that, receiver faced to many multipath signal which are upper than DT.

Conclusion
This paper focuses on underwater channel analysis for acoustic signal travelling by Bellhop acoustic toolbox. The outcomes results of this research are: The signal crossing between ducts is most depend on sound speed changing rate, acoustic signal after transmitting prefer to travels in low speed ducts unless the environmental condition is changed. Based on the TL simulation, the optimum receiver position on surface is determined. According to 867-A modem features; the link budget analysis demonstrates the node placement design has 30 dB safety margin. To have higher directivity index value to better signal to noise ratio, the designer should choose directional transmitter with narrow fan beam characteristic.