What to Do About Multipath and Other Issues

Some potential issues and solutions

If you read the previous pages in this section, it should be clear that acoustic propagation is complicated. If someone tells you they can precisely predict the performance of a sonar system in water, you should be very cautious about purchasing Florida real estate from them. This page will tell you some things to look out for and to try, but the only real way to understand what kind of results you can get is by putting systems in the water and trying them out.

One important item is not strictly a multipath issue. For the same reasons you want to put radio antennas as high as is practical (i.e., better line of sight and fewer ground losses), you should do your best to use ROVL systems when they are as far distant as is practical from the surface of the water or the floor of your body of water. This is especially relevant to the topside unit which is often the receiver or transceiver. You want to operate topside units when they are at least a meter under water, and two or more meters underwater is better.

Operating in man-made pools is usually problematic. Pools often have nicely specular sides and bottoms. Often the sides are parallel to one another, and the bottom is often parallel (or nearly parallel) to the water's surface. Sonar pings reverberate like crazy in a pool. In our testing we have seen pool pings reverberate for up to 30mS. Reverberation is not the main problem with pools (but the long reverberation time betrays the highly reflective quality of the pool enclosure) -- the difficulty lies in getting the system components at least one ping length away from walls and floors.

It's easy to drop a receiver over the side of a boat when operating the system. Be sure to drop the receiver far enough to clear the boat hull by a ping length or more, or you may see errors in position estimates. Dropping off the side of a wharf may lead to similar issues, in which case the receiver should be positioned away from the wharf by a ping length or more. Think about the operating geometry and the Pythagorean Theorem, as in the figure below:

The receiver can be a just over a half-ping length away from an reflector (blue line). When the ROV is in position 1, the yellow reflected path (b) is more than a ping length longer than the direct green path (a). When the ROV is in position 2, the red reflected path (d) may be substantially less than one ping length longer than the green direct path (c), in which case there is a potential for interference. Also try to think about the same issue in all dimensions (i.e., transmitter and receiver operating next to a flat, reflective sea floor).

Submerged vegetation is difficult to penetrate with sonar. We have found that 10 meters of Eurasian milfoil can completely absorb a signal that can otherwise be easily heard 300 meters away. There's no good solution to problematic vegetation.

Stretches of shallow water (e.g., 1-2 meters deep) between the transmitter and receiver may cause the sonar signal to deform, making it seem like the signal is coming from the wrong direction. Underwater ridges that reach up near the surface can cause sonar system to diffract and dissipate as the sound flows over the ridge.

Flow noise may also be a problem with AUV/ROV-mounted Receivers, Transceivers, or Mk III Transponders. If there is turbulence around the transducer housing or AUV/ROV components, or if the thrusters are very loud, the ping signal may be washed out. A symptom of this is the ROVL working fine when the vehicle is stationary, but becomes more erratic as the vehicle speed increases. Note that this may also be a sign of electrical leakage from the thrusters as is described here. Note also that flow noise and electrical noise is rarely an issue with Mk II Transmitters mounted on the AUV/ROV.

For related issues and complications of multipath, see also the next page.

An example of the ping with ID from the previous page affected by multipath. A second reflected signal is combining with the direct signal, causing peaks and nulls (constructive and destructive reinforcement) as the two signals add together. The ROVL is able to detect range and direction of this combined signal, but the ID cannot be recovered.