In Situ Performance Prediction of a Coherent Acoustic Modem in a Reverberant Environment

A channel sounding and communications experiment was performed in the Oslofjord, using eight bottom-mounted instrument units deployed in a network configuration. Five units were equipped with a software-defined 4-8-kHz acoustic modem, programmed to transmit probe signals and communication packets in...

Full description

Saved in:
Bibliographic Details
Published in:IEEE journal of oceanic engineering 2022-01, Vol.47 (1), p.236-254
Main Authors: van Walree, Paul A., Colin, Mathieu E. G. D.
Format: Article
Language:English
Subjects:
Acoustic communication
Ambient noise
Communication
Computational modeling
Delay
Delays
Doppler sonar
Errors
Links
Modems
Noise prediction
Noise propagation
Packet transmission
Performance prediction
Predictive models
Receivers
Reverberation
Sea measurements
signal fluctuations
Signal to noise ratio
terminology
Waveforms
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:A channel sounding and communications experiment was performed in the Oslofjord, using eight bottom-mounted instrument units deployed in a network configuration. Five units were equipped with a software-defined 4-8-kHz acoustic modem, programmed to transmit probe signals and communication packets in a round-robin fashion. All transmitted waveforms were recorded by all units over 35 horizontal links and 5 vertical links. The channels reveal a reverberant environment with long and dense multipath arrival patterns. Measured power-delay profiles and delay-Doppler spread functions are used to predict receiver output signal-to-noise ratio (SNR) over a signaling period of 26 h. To this end, the channel quantities are first calibrated for the propagation loss. The prediction examines the effect of ambient noise, reverberation of previously transmitted packets, the packet's own reverberation, and Doppler spread. The delay profile can be used under calm conditions, and results in a mean prediction error (averaged over all links) of about 3 dB, even hours after the measurement of the profiles. The mean error on individual links (averaged over time) is reduced to 1-2 dB by using up-to-date channel information. The relevance of predicting output SNR is finally illustrated by establishing a relationship between output SNR and the probabilities of bit and packet error.
ISSN:0364-9059
1558-1691
DOI:10.1109/JOE.2021.3085942