The Multipath Problem in Sonar Imaging of the Seafloor
March 8, 2012 in Sonar
Sonar
Sonar or sound navigation and ranging is based on using sound to detect and locate objects.
In passive sonar we listen to the sound of the environment and detect and possibly locate objects from this.
In active sonar we transmit a sound pulse of some kind and use the echo to detect and locate objects. In principle this is a simple process where we emit sound and measure the amount of time it takes for the sound to travel to the target, reflect off the target and travel back to the receiver. This time-delay tells us how far away the target is. The strength of the received echo indicates how strong the reflection of the target is. By plotting this information, an image of the seafloor can be created.
If you want to learn more about sonar systems I recommend the book Sonar Systems, which is distributed under a creative commons licence and is free to download and use. If you reside in Oslo, there are a few courses at the University of Oslo that are relevant; Namely Imaging, Digital signal processing, Digital signal processing II and Signal processing in space and time.
Multipath
Sound does not just reflect back to our receiver, but scatters in all directions. Sometimes this scattering hits the bottom and surface a number of times, and then makes it back to the receiver. This is illustrated in the figure below. These signals that have taken other paths than the direct path is added as noise to the received signal. These multiple paths vary in strength depending on a number of factors, and are referred to as multipath.
The notation for multipath used here is “b” for bottom and “s” for surface. Capital letters indicate a single non-specular reflection, while a small letter indicates a specular (mirror-like) reflection.
Several methods exist for reducing the effect of multipath and I have grouped them into two groups by where in the process they work and mentioned the most important ones. See table 1.1 for an overview.
Data collection
Using multiple sensors for data collection has been shown to improve the range significantly. (article) This is, however, at the expense of resolution (as the bandwidth is limited) and the system complexity. It is not an adaptive method, and thus it might not perform well in very rough and varying terrain.
With an interferometric sonar system it is possible to use the spatial coherence as a measure of data quality, and thus get a measure of obtained imaging range. This coherence is mainly a measure of multipath interference, and by modeling the ocean environment it is possible to simulate what could have been done better. For instance we could have sent some of the energy in a different direction, or listened in a different direction on reception. If the system has an adjustable transmit beam in the vertical direction, e.g. by electronic steering of the beam, it would be possible to adapt the transmit beam while in the water in such a way that the strongest multipath interference is reduced. This is the main subject of my master thesis work.
Data processing
Traditional delay-and-sum (DAS) beamforming, super-resolution techniques that improve DAS and image enhancement are all methods that are used to improve the available data. Thus they may be easier to implement, but can not improve imaging beyond the actual maximum data available in the signal. The data collection methods can, since they are either collecting more signals or adapting before collecting the signal to improve the amount of data available in the signal.
Future work
It is important to improve data during collection rather than trying to improve flawed data. Data processing methods can even be applied as a second layer of improvement after improving the data collection.
Among the data collection methods, the multiple sensor setup has been shown to work, but it imposes some limitations on the sonar system design. I will look into adjusting the transmit beam and other sonar parameters to reduce multipath in my thesis entitled Environmentally Adaptive Sonar on an Autonomous Underwater Vehicle.
Please contact me or leave a comment if you are interested in discussing this subject and reading the complete essay, in which I look into the different methods mentioned in table 1.1.
