Simplified sound velocity profile for the Middle Atlantic Ocean (left), sound diffusion with varying radiation angle (right) 

Contexts in source publication

Context 1

... the vehicle acts autonomously, it has to be equipped with a hierarchical system for guidance, navigation and control (GNC) to fulfill the respective missions. The main application of the USV is to update the positions of the underwater vehicles, even in heavy sea states when the communication capability is reduced due to large roll and pitch angles of the surface vehicle. The GNC system is based on a dynamic model of the vehicle’s motion, which is e.g. used for predicting states, synthesis of heter o- geneous low-level and high-level controllers and diverse mission planning tasks. For conventional operations of USV for hydrographic surveys or special applications like Search-and- Rescue, Majohr and Buch (2006), Kurowski and Lampe (2014) , it is sufficient to consider the motion of the vehicle in three degrees of freedom (3DoF). In contrast to that, the total motion of the SMIS USV has to be considered in 6DoF as it has been addressed in several publications, e.g. Krishna- murthy et al. (2005) . During the design phase, calculations based on potential theory and computa- tional fluid dynamics (CFD) can be made to set-up or change significant design parameters of the vehicle, Kutz (2015) . Furthermore, they can be used to calculate extreme values in motion, which can be applied in a predictive way in the guidance system of the vehicle. In practice, it is a cumbersome task to identify the unknown parameters of such nonlinear models, due to strong couplings of the motion variables, measurement noise and unknown disturbances. Hence, simplifications should be made to identify the basic characteristics of the dynamic behavior of the vehicle, especially in roll and pitch motion. In order to parameterize the resulting models, special maneuvers have to be carried out to de- couple the motions and identify the corresponding parameters. Afterwards, these simpler models can be used to update the mission settings and ensure the reliable operation as communication node. This section starts with a short introduction to acoustic underwater communication and its physical properties. Afterward some of the difficulties in this field are investigated, supported by results of field experiments. The acquired insight is then used to discuss some design specifications of the USV. Compared to terrestrial communication, the underwater channel is a harsh environment in terms of data transfer. Electromagnetic waves, used in common radio communication, will be absorbed by seawater within less than 100 m. Only with frequencies lower than 300 Hz it is possible to communi- cate over larger ranges; however this requires big antennas and excessive transmission power, Peach and Yarali (2013) . The only operational solution for long range communication in underwater environments is acoustics. Underwater acoustic systems however, yield some particularities that need to be addressed. The sound velocity in water mainly depends on three factors: pressure, temperature and salinity. While pressure primarily correlates with the water depth, temperature and salinity can vary widely from one sea to another and changes over the water column. Due to the change of sound velocity, acoustic signals will be deflected from their original propagation direction. A popular example of this phenomenon is the SOFAR channel, where, through refraction effects, sound waves can propagate thousands of kilometers before dissipating, Lerch et al. (2009) . These refraction effects also concern the communication capabilities of the USV. Fig. 1 shows how sound waves get deflected for different radiation angle when transmitting a signal from the surface. Here, the computed sound ray refraction is solely based on Snell’s Law , Jensen et al. (2011) , neglecting all other possible influences like shad- owing and air bubbles. The illustration utilizes a typical sound velocity profile for the Middle Atlantic Ocean, showed in simplified terms on the left side. As shown by the diagram, sound waves that propagate vertically from top to bottom are less affected by refraction, than those signals with a horizontal angle of propagation. Considering this, the communication beam has a curved body. For simplifica- tion reasons, this can be neglected when assuming a reduced radiation angle, which leads to the approximation of a communication cone. Due to divergence, acoustic signal intensity decreases quadratic with distance to the source. Furthermore, frequency dependent relaxation attenuation impairs signal energy and thus communication range. In order to identify the maximal communication range under real conditions, multiple experiments were conducted in the Middle Atlantic Ocean, Neumann et al. (2015) . The experiments have shown that communication was still possible with a slant range of more than 9000 m, however the success ration of delivered packages dropped rapidly at ~8500 m slant range from sender to receiver, Fig. 2. Like in every communication channel, noise is almost ubiquitous. Natural acoustic noise is introduced mainly at the water surface by wind, streams, rain and waves. In addition to the volume, air bubbles, caused by waves, produce big signal absorption at the water surface. This reduces the communication range of surface to surface transmissions, depending on the weather conditions. For a transceiver at the water surface, this can even result in a total communication break down during heavy sea. However, the risk of communication loss decreases rapidly when placing the acoustic transceiver further away from the surface. Also an acoustic baffle, that mutes the surface noise, can help to increase the signal to noise ratio for a surface transceiver. These considerations where taken into account during the design of the USV. In addition to the natural noise sources, artificial noise like motor noise leads to a significantly in- creased noise pollution. The impact on noise level of a passing ship has been investigated experimen- tally by measuring the noise with an acoustic modem that was attached to a GPS equipped buoy. De- tails to the experimental set-up can be found in Neumann et al. (2015) . The modem was located at a depth of around two meters and the research vessel passed by several times with a constant velocity of 8 knots. Fig. 3 shows the impact of the passing ship on the noise level. Besides noise of other vehicles, the propulsion system of the USV produces its own noise. These effects and their implication to the communication capabilities of the USV could not be investigated so far and need further analysis. During a SMIS team operation, the USV acts as a communication node between the underwater seg- ments and the control station (operator), which can be placed on a ship in the operation area or on- shore. Furthermore, it provides geo-references (global satellite based position data) to the underwater vehicles by utilizing acoustic USBL transmissions to correct the resulting drift of the position due to dead reckoning. The surface communication is realized by different redundant radio and satellite te- lemetry links, depending on distance and data volume. While the underwater vehicles perform measuring and monitoring tasks in up to 6000 m depth, the USV has to contact them in cyclic intervals to transmit position references and control commands or to receive state information. Fig. 4 shows a SMIS team, consisting of an USV, several AUVs performing a lawn-mowing maneuver and a SBS. In addition, a cone indicates the acoustic communication and localization area. Based on the experiments described in Section 2, the sound beam is approximated by a cone, whose lateral surface depicts the maximal underwater communication range. The USV positions itself above the respective operation area to assure the cyclic communication with the underwater vehicles. Depending on the depth of the respective communication partner, the communication area can be defined to be the area at the base of the cone. Using the experimental determined maximum slant ranges, the angle of beam spread and the radius of the base of the communication cone can be calculated by trigonometrical relations, Fig. 5. Hence, the acoustical outshined volume or rather the swept area is obtained by the circular surface. Table I shows the results of the experiments in the Atlantic Ocean. The respective rows represent the data at the time of communication dropout for several long distance ...

Context 2

... acoustic signals will be deflected from their original propagation direction. A popular example of this phenomenon is the SOFAR channel, where, through refraction effects, sound waves can propagate thousands of kilometers before dissipating, Lerch et al. (2009). These refraction effects also concern the communication capabilities of the USV. Fig. 1 shows how sound waves get deflected for different radiation angle when transmitting a signal from the surface. Here, the computed sound ray refraction is solely based on Snell's Law, Jensen et al. (2011), neglecting all other possible influences like shad- owing and air bubbles. The illustration utilizes a typical sound velocity ...

Context 3

... point. In order to validate the structure and to obtain the parameters for the acoustic relevant degree of freedom in roll and pitch, negative step response trails have been carried out in the towing tank of the Technical University of Berlin. The parameters have been determined from the step responses by using the prediction error method. Fig. 10 illustrates this test in case of exciting the roll motion. In order to handle this effect, the nonlinear model (4) has been considered, which is adapted from Perez (2005). In that description the parameters (5) are functions of the model states. Fig. 11 shows the identification results of both models. The dashed line depicts the ...

Context 4

... parameters have been determined from the step responses by using the prediction error method. Fig. 10 illustrates this test in case of exciting the roll motion. In order to handle this effect, the nonlinear model (4) has been considered, which is adapted from Perez (2005). In that description the parameters (5) are functions of the model states. Fig. 11 shows the identification results of both models. The dashed line depicts the measurement. As can be seen, the amplitudes of the linear model (solid line) cannot be fitted to the roll motion of the vehicle, especially if larger roll angles occur. The nonlinear model (dotted line) shows satisfactory results even in cases of larger roll ...