Background information

I graduated with a Ph.D. in electrical and computer engineering (signal and image processing) in UC San Diego 2014. My advisors are Prof. William Hodgkiss, Adj. Prof. Peter Gerstoft and Dr. Caglar Yardim. My Ph.D. goal is to develop new ideas, algorithms and mathematical models for geoacoustics inversion problems using mobile and bistatic single source and receiver configurations in low signal to noise ratio (SNR). Broadband synthetic aperture geoacoustic inversion is the initial result of this work. The synthetic aperture is formed when the pressure field caused by a moving source and receiver trajectory is modeled instead of approximating it with a static-point pressure field model.

I have returned to Singapore. I now do research at DSO National Laboratories, 14, Science Park Drive, Singapore 118226. Email: tbienaik@dso.org.sg

What is geoacoustic inversion?

Assuming the signal measured at the receiver is some distance away from a source, the general idea in geoacoustic inversion is to optimize the forward model parameters by minimizing the difference between the measured and replica acoustic signal. In doing this, we estimate sediment properties without resorting to costly direct measurements such as sediment coring. Knowing the seafloor acoustic properties is important for various applications such as sonar performance prediction, source localization and, detection and classification of underwater man-made objects. However, matched field inversion typically uses large aperture arrays and powerful transmissions to reduce parameter estimation uncertainty.

Why single sensor approach?

In contrast, broadband synthetic aperture geoacoustic inversion has two advantages. Firstly it is operationally attractive. This bistatic mobile single source/receiver method computes the waveguide Doppler trajectory field due to a moving source and hydrophone, instead of an approximation with a static point field. This method is well suited for rapid environment assessment by autonomous underwater vehicles (AUV). The source or receiver may be towed horizontally by a ship or an AUV. Alternatively, battery powered acoustic source may be dropped onto the ocean bottom to aid AUV-based geoacoustic inversion. Secondly, it works in low SNR or low source power condition. It coherently exploits repeated transmission to reduce parameter uncertainty. The long observation time improves the SNR and creates a synthetic aperture due to relative source-receiver motion.

However, the inversion performance degrades when source/receiver acceleration exists. Coherently processing a train of pulses all-at-once is impractical as it prevents a recursive way to improve or assess the parameter estimation uncertainty as new data is made available. Therefore, a recursive Bayesian estimation that coherently processes the data pulse-by-pulse is my current approach. In addition, source/receiver acceleration is approximated by assuming piecewise constant but linearly changing velocities pulse-to-pulse.