Wireless Sensor Networks - Part 2 - Apple

brainybootsΚινητά – Ασύρματες Τεχνολογίες

21 Νοε 2013 (πριν από 3 χρόνια και 8 μήνες)

61 εμφανίσεις

WirelessSensorNetworks,Spring2008138
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
WirelessSensorNetworks,Spring2008140
•ArbitraryTrafcModel
Eachnodechoosesdestination,trafcload
•NodeXi
haslocationofXi
=(x
i
,y
i
)
•distancebetweenXi
andXj
(Euclidean)
|Xi
−Xj
|=
r
(x
i
−x
j
)2
+(yi
−yj
)2
3.Successfulreceptionmodels
•Whenisapacketreceivedsuccessfully?Thisis
WirelessSensorNetworks,Spring2008137
Chapter8,NetworkCapacity
1.Thischapterlooksatthetheoreticallimitofbitsthat
canbesentwithinanetworkregioninunittime.
2.Networkmodel
•nnodesona1m
2
area
•channeldatarateisWbps
•Arbitrarytransmissionranges
WirelessSensorNetworks,Spring2008139
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
WirelessSensorNetworks,Spring2008142
preventsneighboringnodetocollideatreceiver
allowsimprecisionoftransmissionrange
•Physicalmodel
P
k
:transmissionpowerofsendingnodeXk
{Xk
}:setofalltransmittingnodes
P
i
|X
i
−X
j
|
ω
N+
P
k6=i
P
k
|X
k
−X
j
|
ω
≥β
β:thresholdofminimumsignal-to-interference
WirelessSensorNetworks,Spring2008144
onebithasbeensent(1-hopormulti-hop)the
distanceofonemetertowarddestination
•Knuth'snotation
f(n)=O(g(n)),asymptoticupperbound,
0≤f(n)≤cg(n)
g(n)=O(f(n))
f(n)=Θ(g(n)),asymptotictightbound,
0≤c
1
g(n)≤f(n)≤c
2
g(n)
WirelessSensorNetworks,Spring2008141
importantforsuchcapacityanalysis.
protocolmodel
physicalmodel
•Protocolmodel
Consideringnodeisendingtonodej
Forallsendersk6=i,
|Xk
−Xj
|≥(1+Δ)|Xi
−Xj
|
Δ:guardzone
WirelessSensorNetworks,Spring2008143
ratio(SIR)forsuccessfulreception
ω:pathlossexponent
N:noiselevel
4.Capacityofwirelessnetwork
•Twomeanings:
bitsofinformationtransmitted
distancethattheseinformationtravel
•unitofbit-meterorbit*meter
WirelessSensorNetworks,Spring2008146
•throughputnotlimitedbytheformationofhot-spot
•Physicalmodel,λ(n)=Θ(
cW

nlogn
)
6.Capacityofarbitrarynetwork
•capacityS(n)=Θ(W

n)
optimalnodeplacement,trafc,range
capacityS1
(n)=Θ(
W

n
)eachnode
•upperbound


W
Δ

n
optimalspatialandschedulingstrategies
WirelessSensorNetworks,Spring2008148
•AreasizeA
capacitymultipliedby

A
•Onlylocaldestination
neighbordistanceO(
1

n
)
•basestations?
basestationsareconstrainedwiththesame
capacity,unlesswired
λ(b)=Θ(
W

blogb
)
WirelessSensorNetworks,Spring2008145
5.Capacityofrandomnetwork
•considersthethroughputofeachnodeper
second:λ(n)
randomlychosendestination
couldbemulti-hopaway
•λ(n)=Θ(
W

nlogn
)
asnincreases,λ(n)approaches0
totalthroughputofentirenetworknλ(n)
WirelessSensorNetworks,Spring2008147
•upperbound
W
1+2Δ
n

n+√

namultipleoffour
nodes,trafcpatterns,ranges,schedules
appropriatelychosen
•Physicalmodelcapacity
cW

nisachievable
c

Wn
(ω−1)/ω
isnot
7.Discussions
WirelessSensorNetworks,Spring2008150
Chapter9,SensingCoverage
1.Coverage
•Areaofinterestmustbecoveredbysensors
•Coveredbyk>1sensors
•k-coverage,k-covered
2.Differentcoveragemodels
(a)Range-basedcoverage
WirelessSensorNetworks,Spring2008152
•Theorem1:Asetofsensorsthatcoveraconvex
regionformsaconnectednetworkgraphif
r
t
>2r
s
.
WirelessSensorNetworks,Spring2008149
8.Mobility
•increasingpossibilityofpktcollision
•store-and-movepktsbutnotstore-and-forward
•capacityincreased!
•longerdelay
•Neverunderestimatethebigpassingtruckthat
carriesdigitaltapes!
WirelessSensorNetworks,Spring2008151
•CircularcoverageR
s
(similartotransmission
range,R
c)
•EventscanbedetectedwithinacircleofR
s
ofa
sensor
•Coverageproblem:howdoweplacethesensors
suchthateachpointiscoveredbysensors(or
k-covered)?
(b)Sensingcoverageandnetworkconnectivity
WirelessSensorNetworks,Spring2008154
(c)AngularandVolumetricCoverage
•Eachsensorhasaniteangle/viewcoverage
•Camera,radar
•Coverageproblem:howtoplaceandorient
sensorstocoveraregion
(d)Steerable/Re-targetablesensors
•unxedvieworlocation
•differentdelity
WirelessSensorNetworks,Spring2008156
sensors
thepolygonalcellisgivenbythehalf-planes
denedbythebisectors
•VoronoiDiagram
WirelessSensorNetworks,Spring2008153
•Theorem2:Asetofsensorsthatk-covera
convexregionformsak-connectednetwork
graphifr
t
>2r
s
.
•Theorem3:AconvexregionAisk-coveredbya
setofsensorsif
Sensorsintersect;sensorsandA'sboundary
intersect
allintersectpointsarek-covered
WirelessSensorNetworks,Spring2008155
3.Nearest-neighborinterpolation
•Goals:ndthenearestsensorforalllocations
•Input:asetofsensorsinthe2-Dplane
•Output:aplanarsubdivisionintopolygonalcells.
Eachcellcontainsallpointsforwhichthesingle
sensorinsideistheclosest.
•Construction:
drawbisectorbetweenasensorandallother
WirelessSensorNetworks,Spring2008158
•Autonomouslimitedmobilityisintroduced
•Guideline:movetowardthefailedsensorto
coveritsarea
•maintainingcoveragewithitsneighbors
(b)Whichsensorshouldmoveandhowfar?
•heuristicapproach
5.Force-basednodemovement
WirelessSensorNetworks,Spring2008160
WirelessSensorNetworks,Spring2008157
•Drawalinebetweeneachpairofsensorswith
theirbisectorshowninVoronoiDiagram
•DelaunayTriangulation:everycircumcircleofa
triangleisemptyofnodes(nosensorsinsideany
triangle)
4.DynamicCoverage
(a)Fault-tolerant
•Nodefailureleadstoareasnotbeingcovered
WirelessSensorNetworks,Spring2008159
(a)Forcebetweenelectrons
F=
q
1
q
2
4πǫ
0
d
2
(b)Drivingthemapartwhenq
1
andq
2
havethesame
sign
(c)Flargerforsmallerd
(d)Multipleneighbors=⇒multipleforces(vectors)
WirelessSensorNetworks,Spring2008162
6.Coveragewithsleepingsensors
(a)whensensorsgotosleepwhilemakingsurethat
allareashave100%coverage
(b)Innernodeshavemoreneighborsandgettosleep
longer.
(c)Bordernodesstayawakelonger,runningoutof
energysooner.
(d)onioneffect(peeledoffafterawhile)
WirelessSensorNetworks,Spring2008164
•differenttoregularrouting
insteadofone-to-onetransmission
multiplesourcessendtooneDS
Noneedtoxintermediatenodes(routersor
forwarders)
2.Trajectorybasedforwarding(TBF)
•forwarddatapacketsalongatrajectory
speciedinthepacket
WirelessSensorNetworks,Spring2008161
(e)Idea:usecombinedforcetondthedirectionand
distancetomoveasensor
WirelessSensorNetworks,Spring2008163
Chapter10,DataDeliveryand
Aggregation
1.Backgroundsondatadelivery
•sensingresultssenttoDS
•goals
forwarddataefciently
lowcontroloverhead
WirelessSensorNetworks,Spring2008166
•Usage
Canhelpoodingofmessages
tradeoffcommunicationwithcomputation
3.Directeddiffusion
•datacentricrouting
application-aware
efcientaggregation
datadeliverytoinquiringobserver
WirelessSensorNetworks,Spring2008168
•broadcastinginterest
•exploratorygradients
gradienthigherthanthesourcerebroadcasting
theinterest
resultssentbacktowardnodeswithlower
WirelessSensorNetworks,Spring2008165
couldbecurvesorstaightlines
lookingforroutersclosesttothetrajectory
•Assumption:
locationinformationknown
trajectorycurveknown
WirelessSensorNetworks,Spring2008167
•query-responsemodel
observersendsquerytothenetwork
nodesrespondtoquery
WirelessSensorNetworks,Spring2008170
•Selectionofcoordinators
criterion
∗higherresidualenergy
∗higherutility
Coordinatorcontentionprocess
∗Anodebecomesacoordinatorwhentwo
neighborsarenotconnected
∗delayimposedtoavoidunnecessary
WirelessSensorNetworks,Spring2008172
5.Dataaggregation
•Aggregationcategories
max:ndsmaximumvalueamongallsensor
readings
min:ndingminimumvalueamongallsensor
readings
average:
count:numberofsensorsobservingtheevent
WirelessSensorNetworks,Spring2008169
gradients
•Reinforcedpaths
datapassingthroughthesepaths
4.SPAN
•RedundancyinWSNs
sleepingnodes
managedbycoordinators
onlycoordinatorsforwardtrafc(CH?)
WirelessSensorNetworks,Spring2008171
coordinators





1−
E
r
E
m


+



1−
C
i
N
i
2




+R



N
i
T
∗N
i
:numberofneighborsofnodei
∗C
i
:connectedpairsofneighborswithias
coordinator
∗T:timetotransmit/receiveashortmessage
WirelessSensorNetworks,Spring2008174
dataaccuracy
latency
•Dataaggregationinatnetworks
Pushdiffusion
∗datasourceactivelypushdata(diffusion)
towardsink
∗datasinksubscribetothesourcesthrough
enforcement
WirelessSensorNetworks,Spring2008176
∗Sensorssenddatatotheirclosestneighbors
towardLeadernode.
∗Leadernodesendsaggregatedresulttosink.
Tree-based
Grid-based
WirelessSensorNetworks,Spring2008173
region:regioninwhichsensorsobservingthe
event
•Dataaggregationintheprocessofdatagathering
Datagathering:sysmaticcollectionofsensed
datafrommultiplesensorstobetransmittedto
datasink.
•Dataaggregationgoals:
networklifetimeextension
WirelessSensorNetworks,Spring2008175
∗dataisdescribedwithmeta-data
∗schemetermedSPIN
Two-phasepulldiffusion(DirectedDiffusion)
•Dataaggregationinhierarchicalnetworks
Cluster-based
∗allsensorssenddatatoclusterhead.
∗clusterheadsendsaggregatedresulttosink.
Chain-based