Close Proximity Acoustic Communication Analysis Utilizing PCSWAT

In order to determine the performance capabilities of high-speed underwater acoustic communication, it is necessary to study some of the parameters involved in detail. These parameters include signal to noise ratio (SNR), salinity, marine life density, and sea state, to name a few. The aim of this project is to study and predict SNR values for water channel geometries likely to be encountered by autonomous underwater submarines (AUV) while scanning shallow water for mines.

The analysis was performed using a software program developed by the Navy called PC SWAT. The software was used to create a grid of the SNR for specific water channels with specific sea state and salinity values. PC SWAT does not have full field capabilities for this sort of calculation, so the software was run multiple times with the locations of the transducers in different places to make up a grid. The significance of this is that the SNR data below is only valid for the situation where the sending and receiving transducers are at the same elevation, as that is how the data was created in PC SWAT. So when looking at the resulting plots, you must look at one horizontal slice at a time to see the SNR that would be encountered in an acoustic communication.

PC SWAT uses a time series forward method to calculate SNR values. Because of the computational magnitude of this problem and because of an apparent buffer size limit of PC SWAT, the simulations could not be run for sufficient time with the signal frequency at the desired 26 kHz. Instead it was run at 10 kHz; however the results obtained can be scaled to be valid for a 26 kHz signal if the depth and range values are divided by 2.6 and this has been done in the following plots. The data cannot be scaled in this way when modeling the effect of the salinity of the water because the resulting absorption is proportional to the distance that the signal travels, but for the relatively short distances in consideration the absorption has not been found to be significant. In addition, salinity will only make the SNR larger because the multi-path travels farther than the direct path. The signal transmitted was a sine wave in shape, and it was transmitted for .7 seconds. A macro was used to run the program and save the output to disc.

The PC SWAT simulation output is an array containing the complete signal that the receiving transducer receives in amplitude, the time data, and the Hilbert transform. In addition to running a grid of SNR for each of the test cases of interest, a single set of data was created to use to subtract from the SNR grid PC SWAT data so that the reverberation alone could be extracted. This was done by running one simulation for each range with the depth at 10000m and the transducers placed at mid depth so that there would be no reverberation and so that the time increment would match that of the SNR raw data. Once all of the raw data was created in PC SWAT as explained, then MATLAB was used to subtract the signal from the signal plus reverberation that PC SWAT output, leaving only the reverberation. This analysis does not concern the ramp up and down of the reverberation, only the stationary value, so to this end only the middle third of the signal was extracted for the calculations. The standard deviation of a waveform is approximately equal to its RMS value, so the next step was to calculate the standard deviation of the extracted portion of the input signal and the reverberation amplitudes. To calculate the SNR, the following equation was used:

The results are shown in the following plots:

Fig1. Plot of SNR for a water channel 25m deep. Only reverberation from the surface and bottom was considered.

Simulations have also been done with directional transducers rather than monopoles. The following plot represents a sender with a 200dB source level and a 13dB side lobe level, and a 26 degree half angle. The receiver also has a 26 degree half angle. As in the previous simulations, the sender and reciever are at equal elevations for each communication, so when looking at the plot you must look at one horizontal slice at a time. As anticipated, the results are much better with directional transducers.

Other inputs for these simulations:
All ray options set to highest precision
APL/UW Surface and bottom loss model used
Bistatic surface and bottom scattering model used
Volume scattering set to low
Water temp 20 °C
Zero: salinity, wind speed, wind direction, rain rate, air-sea temp difference, ambient noise