An underwater acoustic voice communications system can provide a vital communication
link between divers and surface supervisors. There are numerous situations in which a
communication system is essential. In the event of an emergency, a diver's life may depend
on fast and effective action at the surface. The design and implementation of a digital
underwater acoustic voice communication system using a digital signal processor (DSP) is
described. The use of a DSP enables the adoption of computationally complex speech
signal processing algorithms and the transmission and reception of digital data through an
underwater acoustic channel. The system is capable of operating in both transmitting and
receiving modes by using a mode selection scheme. During the transmission mode, by
using linear predictive coding (LPC), the speech signal is compressed whilst transmitting
the compressed data in digital pulse position modulation (DPPM) format at a transmission
rate of 2400 bps. At the receiver, a maximum energy detection technique is employed to
identify the pulse position, enabling correct data decoding which in turn allows the speech
signal to be reconstructed.
The advantage of the system is to introduce advances in digital technology to underwater
acoustic voice communications and update the present analogue systems employing AM
and SSB modulation. Since the DSP-based system is designed in modular sections, the
hardware and software can be modified if the performance of the system is inadequate. The
communication system was tested successfully in a large indoor tank to simulate the effect
of a short and very shallow underwater channel with severe multipath reverberation.
The other objective of this study was to improve the quality of the transmitted speech
signal. When the system is used by SCUBA divers, the speech signal is produced in a mask
with a high pressure air environment, and bubble and breathing noise affect the speech
clarity. Breathing noise is cancelled by implementing a combination of zero crossing rate
and energy detection. In order to cancel bubble noise spectral subtraction and adaptive
noise cancelling algorithms were simulated; the latter was found to be superior and was
adopted for the current system.
A Doctoral Thesis. Submitted in partial fulfilment of the requirements for the award of Doctor of Philosophy of Loughborough University.