Phase unwrapping of signals propagated under the Arctic ice crust: a statistical approach

The acoustic source generates a narrowband sequence with a phase that experiences random fluctuations. At the receiving hydrophones, the measurements are corrupted by wideband noise. Discrete optimal nonlinear filtering techniques are used to design the phase unwrapping algorithm. The scheme presented proves less sensitive to noise than other nonstatistical unwrappers, supporting less stringent prefiltering constraints. Under the conditions of the Arctic experiment, it accommodates processing of each individual sensor's output data, avoiding the need for beam forming. The prefilters are designed with large bandwidths in order to achieve the shorter duration impulse responses needed to discriminate between intrinsic phenomena and transients induced by the prefilters. An examination of real signals that traveled several hundred kilometers through a highly unstable channel shows that nonlinear statistical designs are viable and useful in many practical problems of underwater acoustics. >