Travel time estimation for ocean acoustic thermometry

Abstract

The basic datum of ocean acoustic thermometry experiments is the ray travel time, thus an accurate and precise estimator is of paramount importance. It is well known that a matched filter gives the best estimate if the transmitted signal is known exactly. At basin-scale ranges, the signal becomes a random process due to the limited coherence time and coherent bandwidth of the channel. A matched filter is no longer optimum in this case, but since the reception is still partially coherent, an energy detector is not optimum either. A better estimator is the RAKE correlator, which can account for the signal variability at the cost of increased receiver complexity and a reduction of resolution. In this talk, the performance of the RAKE correlator, the matched filter, and the energy detector versus coherence time, coherent bandwidth, and SNR is explored by simulation of realistic cases. The RAKE is then applied to a 4000-km range, 75-Hz sonar transmission and the performance improvement is remarkable. An added benefit of the RAKE is that there is less clutter, or fewer false alarms in the detection theory jargon, and thus tracking an arrival over multiple transmissions is much easier.