ADVANCES IN THE ROBLINKS PROJECT ON LONG-RANGE SHALLOW-WATER ROBUST ACOUSTIC COMMUNICATION LINKS
Martin van Gijzen, Paul van Walree1,
俄罗斯装甲车Daniel Cano, Jean-Michel Pasrieux2,
酒神2009Andreas Waldhorst, Rolf Weber3
1 TNO Physics and Electronics Laboratory, P.O. Box 96864, 2509 JG The Hague,
The Netherlands
2 Thomson Marconi Sonar SAS, BP 157 0690
3 Sophia-Antipolis Cedex, France
3 Ruhr Universität Bochum, Lehrstuhl für Signaltheorie, Bochum, Germany.
Abstract
Within the ROBLINKS project waveforms and algorithms have been developed to
establish robust underwater acoustic communication links with high data rates in
shallow water. To evaluate the signalling schemes, a wide range of experiments has
要挟的读音
been performed during a a trial that has been held in May 1999, in the North Sea,
off the Dutch coast. The analysis of the resulting data t shows that the original
aims of ROBLINKS with regard to data rate and transmission range are achieved
and in some respects even surpasd.
INTRODUCTION
The importance of underwater acoustic communication is steadily increasing. This communication is of importance for underwater activities ranging from data transfer between underwater and surface platforms, remote control applications such as underwater vehicles and robots, communication with divers, etc. With regard to underwater acoustic communication, the main difficulty associated with a large range/depth ratio is time spreading due to pronounced multipath propagation. This occurs in addition to high temporal (pha) and spatial variability.
The scientific innovation of the ROBLINKS project is that it focuss on continuous parallel identification of the channel respon, to provide lf-adaptive algorithms innsitive to channel fluctuations. Two competing strategies are investigated: identification with parallel monitoring and blind identification. Identification with parallel monitoring is established by transmitting a superposition of a known reference signal and a communication signal that is taken from an alphabet of signals orthogonal to the reference signal. By monitoring the reference signal one can estimate the respon of the channel and correct for its detrimental effects. This approach has the disadvantage that only part of the energy is devoted to the communication signal. The blind approach does not utilize a reference signal. To correct for adver channel conditions a lf-trained, decision-directed equalization algorithm is ud. The two approaches are evaluated on basis of data collected during a a trial in a coastal part of the North Sea, which took place from April 30 to May 7, 1999. Further information on the project can be found in [1] and on the project homepage l/instit/fel/roblinks/.
OBJECTIVES
Prent communication systems have good performances but they either have poor bandwidth efficiency (noncoherent methods), are limited by a time spread that is less than the symbol duration (
differentially coherent), or require operator assistance to adjust the receiver parameters to the channel. The specific objectives of ROBLINKS are:
1.To develop new signal concepts and algorithms for optimal coherent signal processing in
the time domain to achieve reliable long-range underwater acoustic communication in shallow waters at a large range/depth ratio (! 100). The aim is to achieve this at reasonable data rates (! 1 kbit/c), and within the frequency band 1-15 kHz. The propod algorithms should be lf-adaptive with regard to environmental variations. The word "robust" in the project title is ud in this particular n.
2.To evaluate experimentally the performance of the waveforms and processing algorithms
with data acquired during a shallow-water a trial. Selected waveforms and processing algorithms are implemented in a real-time system and the real-time performance is evaluated with data recorded during the a trial.
THE SEA TRIAL
Experimental t-up
A trial has been executed in the North Sea, approximately 10 km off the Dutch coast near the coastal resort Noordwijk. A data t was collected to evaluate the communication waveforms and processing algorithms and to asss the propagation conditions. The water depth at the location of the trial is approximately 18 m. The bottom is relatively flat with sand rims reaching heights of up to one meter. Two platforms were involved in the trial, HNLMS Tydeman and Meetpost Noordwijk.
HNLMS Tydeman, an oceanographic rearch vesl of the Royal Netherlands Navy, acted as the transmitter platform. The acoustic source ud to emit the signals had a source level between 185 dB (re 1 "Pa @ 1m) and 195 dB over the frequency band from 1-15 kHz. The source was deployed at a depth of 9 m.
Meetpost Noordwijk, a fixed rearch and monitoring platform owned by the Dutch Directorate-General for Public Works and Water Management, was the receiver station. A vertical array of 20 hydrophones, 60 cm apart and thus covering the greater part of the water column, was vertically fixed between a beam connected to the platform and a weight on the bottom of the a.
Fig. 1 displays the t-up of the acoustic experiments. It also indicates the main problem encountered in shallow-water acoustic communications, namely multipath propagation. Sound scatte
ring off the a bottom and water surface is also indicated. This gives ri to reverberation. Further, reflections off moving surfaces, such as waves, contribute to the Doppler spread of the signals.
Fig. 1. Set-up of the experiment
Measurements
Communication signals were transmitted along two different tracks: a primary track from west to east, and a condary track along the coast (more or less south # north). The first part of the experiments took place in a fixed point to fixed point configuration with moorings of HNLMS Tydeman at 1 km, 2 km, 5km and 10 km distance from Meetpost Noordwijk on the primary track and
at 2 km distance on the condary track. A moving point to fixed point configuration was ud in the cond part of the experiments in which the HNLMS Tydeman was sailing along the primary and the condary track.
To asss the propagation conditions, each communication signal (with a typical length of about 13 minutes) was preceded by
$60 conds of noi recording to determine ambient noi levels;
$ A CW (10 s) at the passband centre frequency, to determine the Doppler spread;
$Two linear FM sweeps (10 s over the frequency band 1-14 kHz and 0.2 s over the signal passband) to determine multipath propagation.
All project partners defined a t of communication signals. TMS created quences containing a reference signal, RUB concentrated on blind equalization algorithms and TNO confined itlf to more conventional modulations like BPSK to enable a comparative evaluation with the newly developed waveforms.
The acoustic experiments were complemented by a range of environmental measurements, such as
CTD probes to determine sound velocity profiles and XBT casts to record temperature profiles, and echo soundings to determine the bathymetry of the tracks.
The trial has resulted in a vast, high-quality data t. The most illustrative and interesting part of the data t will be made available to the EU rearch community via the IFREMER/SISMER data banking centre. Information about this lected data t can be found in [2] and at the ROBLINKS archive www.ifremer.fr/sismer/program/roblinks/.
RESULTS OF THE DATA ANALYSIS
Analysis of the channel respon
The 10-s FM and the CW signals have been analyd to estimate the time spread and Doppler spread, respectively. This is reported in detail in [3]. Typical values for the time spread are between 14 ms at shorter ranges (1 and 2 km) and 8 ms at the longer ranges (5 and 10 km).The structure of the different arrivals is rather stable at lower frequencies, whereas at higher frequencies (above 6 kHz) the temporal and spatial variability is significant. A typical value for the relative Doppler spread is 4102/%&'(f f .
Performance of high data-rate BPSK-signals
For the comparative evaluation of the newly developed waveforms with conventional modulations, a number of BPSK signals were broadcasted. The signals consist of standard Nyquist raid cosine puls, supplemented with a pudo random learning signal and two displaced carriers. The learning signal rves to estimate the channel respon, and the displaced carriers are uful for timing recovery. Processing at the receiver includes linear least-squares equalization, adaptive beamforming and decision-directed equalization. After some fine-tuning quite good results can be obtained. For example, a signal of 4 kbit/s transmitted in a moving point to fixed point configuration is successfully demodulated to yield a zero bit error rate. The details are reported in [4].
Transmission using parallel monitoring
Results of the analysis of the communication signalling schemes with a reference signal at ranges of 2 and 5 km are prented in [5]. The transmitted signal is bad upon Gold or Oppermann ts of quences and consists of a superposition of one fixed reference quence for the purpo of channel identification and up to 16 quences with arbitrary indices and phas to carry the uful information. First analys, using long signals (2 minutes) from a single hydrophone channel and robust and simple reception algorithms (without fine operator tuning and/or delay for channel identification), have already shown that transmission with very low bit error rate (BER ~ 10-4) is poss
汤生ible for data rates up to 800 bit/s.
Transmission using blind identification
To investigate the feasibility of a completely lf-recovering equalization and synchronization method, a system which us offt-quadripha-shift keying (OQPSK) signals was designed first. The number of message bits in each of the signals was approximately 105.By exploiting the special OQPSK modulation structure and by jointly processing the outputs of 3-7 hydrophones, a relatively simple symbol-spaced linear adaptive multichannel-equalizer could be applied with great success to the entire medium-range (2-5 km) experimental data. It could be demonstrated that bit error rates of approximately 10-4 were attainable without channel coding and with data rates ranging from 500-4000 bit/s. The results also show that virtually the entire number of bit errors per signal occur during the initial acquisition pha of the equalizer which lasts for a few thousand symbols at most. Additional details are prented in [6].
蓝牙耳机设置CONCLUDING REMARKS
ROBLINKS continues until the end of November 2000, and the receiver algorithms will continue to be improved until that day. Yet, the results more than half a year before the project end already eclip
the original objectives in veral respects, viz.,
$The reported bitrate of 4 kbit/s surpass the 1 kbit/s minimum objective;
$The transmission range of 5 km exceeds the 2-km distance that corresponds to a range/depth ratio of 100:1;
$Successful mobile underwater acoustic communication at high data rate has been demonstrated.
During the remaining project duration, further algorithm improvements will focus on the aspect of robustness. Furthermore, a real time implementation of the ‘best’ algorithms is being made. Acknowledgements:The technical and logistical support of Joost Kromjongh and Adri Gerk (both TNO) was esntial for the trial success. Guus Goons (RWS) and Edmond Flayosc (TMS-SAS) are acknowledged for their assistance on board Meetpost Noordwijk. The colleagues of the SWAN project are thanked for the esntial back-up arrangements during the trial. The commander and crew of HNLMS Tydeman are acknowledged for their enthusiastic collaboration.
The European Commission is thanked for supporting the ROBLINKS project through the MAST program. TNO is sponsored by the Royal Netherlands Navy.
衍加心读什么>雨的散文
REFERENCES
[1] D. Cano, M. B. van Gijzen, A. Waldhorst, Long Range Shallow Water Robust Acoustic
Communication Links ROBLINKS, In Third European Marine Science and Technology Conference, Lisbon, Volume III, pp. 1133-1136, 1998.
[2]M.B. van Gijzen, P.A. van Walree, D. Cano, J.M. Pasrieux, A. Waldhorst, C.
Maillard. The ROBLINKS underwater acoustic communication experiments, To appear in The proceedings of the fifth European Conference on Underwater Acoustics, 2000, Lyon, 2000
[3]P.A. van Walree, M.B. van Gijzen, D.G. Simons, Analysis of a shallow-water acoustic
communication channel, To appear in The proceedings of the fifth European Conference on Underwater Acoustics, 2000, Lyon, 2000
[4]M.B. van Gijzen and P.A. van Walree, Shallow-water acoustic communication with high
bit rate BPSK signals, To appear in The proceedings of the Oceans 2000 Conference, Providence (USA), 2000.
[5]J.M. Pasrieux and D. Cano, Robust shallow water acoustic communications bad upon
orthogonal quences and real-time channel identification, To appear in The proceedings of the UDT Europe Conference, 27-29 June 2000, London (UK), 2000
[6] A. Waldhorst, R. Weber and J.F. Böhme, A Blind Receiver For Digital Communications
in Shallow Water, To appear in The proceedings of the Oceans 2000 Conference, Providence (USA), 2000.带舞的成语
