Abstract
This paper proposes a superimposed training method for low probability of detection underwater acoustic communications. A long pilot sequence was superimposed to the message for equalization and synchronization purposes. A fast Hadamard transform (FHT) estimated the channel impulse response and compressed the pilot energy. A Wiener filter performed equalization. The interference signal was removed using hyperslice cancellation by coordinate zeroing. An inverse FHT decompressed the remaining sequence energy and the message was retrieved. Results from a shallow water experiment presented bit error rates
<
10
- 2 for signal-to-noise ratios
< - 8 dB.