Wireless Underwater Communications and Networks: Current Achievements and Future Research Challenges

littleparsimoniousΚινητά – Ασύρματες Τεχνολογίες

21 Νοε 2013 (πριν από 3 χρόνια και 11 μήνες)

89 εμφανίσεις

WirelessUnderwater
CommunicationsandNetworks:
CurrentAchievementsandFuture
ResearchChallenges
MilicaStojanovic
MassachusettsInstituteofTechnology
millitsa@mit.edu
Overview
Introduction
Channelcharacteristics
Signalprocessing:bandwidth-efficient
underwateracousticcommunications
Underwaternetworks:channelsharingand
multiple-access
Futureresearchandopenproblems
Background
Motivation:
Majorscientificdiscoveriesof
thepastdecade(e.g.,hydro-
thermalvents):cabled
submersibles
Cablesareheavy,deploymentis
expensive
Wirelessinformation
transmissionthroughtheocean:
Remotecontrolofrobots,
vehicles
Remotedataretrieval
Wirelesscommunication:
radio(30Hz-300Hz,veryhigh
attenuation)
optical(shortdistance,pointing
precision)
acoustic
History:
underwatertelephone(analog
modulation,8-11kHz).
DSPtechnology:acoustic
modems(fewkbps/fewkm).
Researchgrowingwithnew
applications(andviceversa).
Applications:oceanobservation
Environmentalmonitoring
climaterecording
pollutioncontrol
predictionofnaturaldisasters
oil/gasfields
harborprotection
Underwaterexploration
discoveryofnaturalresources
marinephenomena
deep-seaarchaeology
Scientificdatacollection
oceanography
geo-sciences(physics/chemistry)
marinebiology
Searchandsurvey
,militaryandnon-
military
detectionofobjects
oceanbottomimagingand
mapping
Underwaterwirelessnetworks
Today:point-to-pointacousticlinks
Future:autonomousnetworksforoceanobservation
Examplesoffuturenetworks:
adhocdeployablesensornetworks
autonomousfleetsofcooperatingAUVs
Typesofnodes:
fixed,slowlymoving,mobile
sensors,relays,gateways
Typesofsignals,systemrequirements:
low/highrate(~100bps-100kbps)
real-time/nonreal-time
high/moderatereliability
Configurations:
standalone
integrated(e.g.,cabledobservatories)
NSFITR:Acousticnetworks,navigationandsensing
formultipleautonomousunderwaterroboticvehicles.
Communicationchannel
Physicalconstraintsofacousticpropagation
:
limited,range-dependentbandwidth
time-varyingmultipath
lowspeedofsound(1500m/s)
Systemconstraints
:
transducerbandwidth
batterypower
half-duplex
Worstofbothradioworlds
(landmobile/satellite)
0
2
4
6
8
10
12
14
16
18
20
0
10
20
30
40
50
60
70
5km
10km
50km
100km
frequency[kHz]
relative SNR [dB]
tt(1±v/c)
ff(1±v/c)
A(d,f)~d
ka(f)
d
N(f)~Kf
-b
B>1/T
mp

frequency-selective
fading
Summaryofchannelcharacteristics
Synchronization(phase,
delay)
Channelfeedbacklatency
Throughputefficiencyof
networkprotocols
Typicalv/cvaluehigherby
severalordersofmagnitude,
severeDopplerdistortion
Verylongpropagationdelay
Speedofsound
Adaptiveequalization,array
processing
Needfordynamicprotocols,
x-layeroptimization
ISIspanstensofsymbols
Fading/outagemodelsnot
known
Fading/multipath
Needfordatacompression,
bandwidth-efficient
modulation
Networktopologyselection,
resourcesharingandreuse
Bandwidthseverelylimited
Bandwidthdependson
distance
Bandwidth/SNR
Implicationsforsignal
processing,networkdesign
Acoustic/radio
difference
Channelparameter
Signalprocessingforhigh-rate
underwateracousticcommunications
Bandwidth-efficientmodulation(Rb/B>1bps/Hz):PSK,QAM
Coherentdetection:
decision-feedbackequalization
phaseanddelaysynchronization
(IEEEJOE94,withJ.Proakis,J.Catipovic).
x
+
decision
a
b
v(t)2/T
d(n)
^
d(n)
~
d(n)
trainingdata
-
e
j(n)
datadecision
dataestimate
sampling
phasecorrection
feedforward
filter
feedback
filter
inputsignal
(baseband)
.
..
.
JointMMSEadaptationofequalizer,synchronizer
+
-
error
inp.K
com-
biner
forward
forward
+
_
decision
feedback
adaptationalgorithm
inp.1
inp.2
dataout
sync.
filter
coefficients
trainingdata
dataest.
Multi-channel
combining
Example:
NewEnglandContinentalShelf
(JASA95,withJ.Proakis,J.Catipovic)
Currentachievements
Point-to-point
(2/4/8PSK;8/16/64QAM)
mediumrange(1km-10km~10kbps)
longrange(10km100km~1kbps)
basinscale(3000km~10bps)
vertical(3km~15kbps,10m~150kbps)
Mobilecommunications
AUVtoAUVat5kbps
Multi-usercommunications
fiveusers,eachat1.4kbpsin5kHzband
WHOImicro-modem
:
FixedpointDSP
lowrateFSK(~100bps)w/noncoherentdetection
Floatingpointco-processor
highratePSK(~5000bps)w/coherentdetection
(adaptiveDFE,Dopplertracking,coding)
4-channelinput
10-50Wtx/3Wrx(active)
1.75inx5in.
Researchinsignalprocessing
Goals:
lowcomplexityprocessing
improvedperformance
betterbandwidthutilization
Specifictopics:
spreadspectrumcommunications(
CDMA,LPD
)
multipletx/rxelements(
MIMO
)
multi-carriermodulation(
OFDM
)
Multi-carriermodulation/
orthogonalfrequencydivisionmultiplexing
Availablebandwidthdivided
intonarrowsub-bands:
channelappearsideal(flat)
ineachsub-band
OFDM:efficient
implementationviaFFT
+Low-complexityequalization
(frequency-domain)
-Highsensitivitytofrequency
offset
High-rateacousticsystem
isinherentlywideband(UWB):
Dopplerdistortionisnot
uniform
acrosssub-bands
f0
f1
fK-1
mod
mod
mod
SP
+
IFFT
demod
f0
dec
demod
f1
dec
demod
fK-1
dec
PS

FFT
0
10
20
30
-2
0
2
4
top:subchannelK-1
bottom:subchannel0
phaseestimates[rad]
blockindex
0
10
20
30
-4
-2
0
2
x10
-5
blockindex
a-estimate
0
500
1000
0
0.2
0.4
0.6
0.8
1
subchannelindex
channelestimates(abs.)
Systemparameters:
M=12receiverelements
K=1024subchannels(32blocks)
nooverlapadd
pilotchannels:0
phasedifferencefiltering:0
channeltracking:0.99
MSE:-16.3dB
SER:0
-2
-1
0
1
2
-2
-1
0
1
2
Re
Im
outputscatterplot
0
10
20
30
-30
-20
-10
0
blockindex
MSE-time[dB]
0
200
400
600
800
1000
-30
-20
-10
0
subchannelindex
MSE-frequency[dB]
FFT
FFT
c
o
m
b
i
n
e
r
c
o
m
b
i
n
e
r
.
.
.
.
.
.
in1
inM
1
K
1
K
dec
dec
out1
outK
Experiment:
BuzzardsBay05
Adaptivechanneltracking,
Dopplercompensation(v/c)
30+kbps/2.5km
@minimalcomplexity
(workinprogress;IEEEOceans06)
Futureautonomousunderwater
systems:networktopologies
CentralizedDecentralized
Nodescommunicateviaacentralstation(cellularnetwork
paradigm).
Channelmustbesharedregulationofmultipleaccess.
Centralstationsareconnectedthroughaseparate
infrastructure(cableonthebottom,radioonthesurface).
Nodescommunicatethroughneighbors(adhocnetwork
paradigm).
Messagesmustberelayedtoreachdestinationregulationof
channelaccess.
Theremaybeanendnodetogateway.Nodesmayformclusters.
Openproblems(Therearenooperationalautonomousunderwaternetworks,onlyisolatedexperimentaldemonstrations):
Capacityofanacousticnetwork?(Energyand
bandwidthdependoninter-nodedistance.)
Efficientandscalablechannelsharingprotocols?(Speedofsoundisfiveordersofmagnitude
lessthanspeedofelectro-magneticwaves.)
Networkarchitecture
Eachlayerassumesperfectoperationofthelayerbelow.
Theremaybeatotaloffiveormorelayers(transport,application).
Network
Physical
Datalink
layer
layer
layer
Network
Physical
Datalink
layer
layer
layer
Network
Physical
Datalink
layer
layer
layer
NodeANodeBNodeC
Routingpathselection
(end-to-endthroughput,delay)
AutomaticRepeatRequest(ARQ)
MediumAccessControl(MAC)
(throughput,packeterrorrate)
Modulation/detection
(bitrate,biterrorrate)
Channelsharing(accessregulation)
Deterministic/scheduling
Multipleaccess:frequency,time
codedivision(FDMA,TDMA,
CDMA
);alsospacedivision.
Continuoustraffic,fixednumber
ofusers
Complex,lowoverhead
Scalability?Spatialreusefor
efficientresourceallocation
(power,bandwidth).
Random/contention
Aloha,CarrierSensingMultiple
Access(CSMA),Multipleaccesswith
CollisionAvoidance(MACA
)
Burstytraffic,variablenumberof
users
Simple,highoverhead(RTS/CTS)
Scalable.
Thetwostrategiescanbecombined:reservation-basedprotocols.
U1
U2

UN
U1
XProbeU2Data2(n)Guard
B
A
C
Ex:N<5for10sec.updating,1kmx1km,1sec.packet
AsendsRTS,includingpacketlength.
Bhears,replieswithCTSincludingpacketlength;
ChearsB,deferstransmissiontoletBreceivedatapacket.
MediumAccessControl(MAC)
Acousticchannel:extremely
longdelay
collisionsshouldbeavoided
Aloha:doesnotavoidcollisions
CSMA:goodonlywhendelay<<packetduration
MACA:toavoidcollisionsmusthaveRTS/CTS>delay
(controlpacket>datapacket!)
Collisionswastetimeand
energy
SlottedMACA:slot
>delay+RTS/CTS
Q:Whattxpowertouse?
Criterion:max.throughput
(packettime/averagetime
forsuccessfultransmission)
Simulation:16AUVs,5kmx5km,5kt,
randomdirectionchange,1kbps.
A:Thereexistsoptimaltxpower,which
canbecalculatedforagivensystem.
(IEEEOceans06,withM.Molins)
Reliablepackettransmision
DatalinklayerARQ
Acousticmodems:half-duplexARQbasedonStop&Wait
Acousticchannel:varyingBER,longwaitingtimepoor
throughputefficiency=packettime/averagetimeneededfor
successfulpackettransmission
longerpacket:betterutilizationofwaitingtime,
butgreaterchancesofpacketerroroptimalpacketlength
SelectiveS&WwithMpackets/group:
0
500
1000
1500
2000
2500
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Pe=1e-3
solid:S&W-1
dashed:S&W-2,M=16
dotted:S&W-3,M=16
toptobottom,lR[m-bps]:
5e+4
5e+5
5e+6
packetsize[symbols]
throughput efficiency
-5
-4.5
-4
-3.5
-3
-2.5
-2
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
log(Pe)
optimalpacketsize
solid:S&W-1
dashed:S&W-2,M=16
dotted:S&W-2,M-->inf.
lR=5e+5m-bps
lR=5e+4m-bps
Rate:100bps1000bps
Distance:500m5km
-5
-4.5
-4
-3.5
-3
-2.5
-2
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
log(Pe)
maximalthroughputefficiency
solid:S&W-1
dashed:S&W-2,M=16
dotted:S&W-2,M-->inf.
lR=5e+5m-bps
lR=5e+4m-bps
Analyticalresults:
OptimalpacketsizedependsheavilyonBER,rate-distanceproduct.
ByincreasingthegroupsizeM,theproblemofdelaycanbeovercome.
MaximalthroughputefficiencyremainslimitedbyBER.
Futurework:
AdaptiveARQfortime-varyingchannels.
Practicalimplementations.
(IEEEOceans05)
Experimentalacousticnetworks
Nowidelyacceptedchannelmodelsexperimentaldemonstrationremainsde-factostandard.
Needeasilyprogrammableanddeployablenetworkingplatforms.
Reconfigurablemodem
:MatlabSimulink+TIDSP=flexible,modular(softwareradioconcept).
Currentsoftware
:
Packetizingandpacketsynchronization
PSK/DPSK,QAMmodulation
Multi-channeldecisionfeedbackequalization
Convolutionalcoding,Viterbidecoding
Inprogress
:
Dopplercompensation,OFDM,FSK
MACAprotocol
Hardware:poweramplifier,transducer,housing.
A/DandD/Aboard:250kHzsampling,
100kHzmax.frequency,
4in/outchannels.
Micro-lineboard:
TMS320C6713floating-pointDSP,
XilinxVirtex-IIFPGA
Powersupplycarrierboard(12-36V).
5x2.5in
MatlabtoCcode:RealTimeWorkshop
Compiling,downloading:CodeComposerStudio
Futurework
(withE.Sozer):
Firstfieldtest:summer06?
Goals:opentestingplatformforsignalprocessingalgorithms,networkprotocols
in-houseacousticnetwork.
Openproblemsandfutureresearch
Fundamentalquestions
:
Statisticalchannelmodeling
Networkcapacity
Researchareas
:
Datacompression
Signalprocessingforcommunications
:
adaptivemodulation/coding
channelestimation/prediction
multiplein/outchannels(tx/rxarrays)
multi-usercommunications
communicationsinhostileenvironment
Communicationnetworks:
networklayout/resourceallocationandreuse
networkarchitecture/crosslayeroptimization
networkprotocols:alllayers
Experimentalnetworks
:
Systemspecification
:
typicalvs.application-specific(trafficpatterns,performancerequirements)
optimizationcriteria(delay,throughput,reliability,energyefficiency)
Conceptdemonstration
:
simulation
in-water
prototypes
Underwateropticalcommunications
:
blue-greenregion(450-550nm)
+muchhigherbandwidth(~Mbps)
+negligibledelay
-shortdistance(<100m)
Systemintegration
:
Cabledobservatories
Integrationofwirelesscommunications:
cabledbackbone+mobilenodes=extendedreach
Wirelessextension:acousticalandoptical
complementary
toacoustics