ThischapterconsidersalargescaleSANETwithmultipleactuatorsassinksfordatagenerated
bythe sensors. Since many applications require to have each source node send all its locally
generated data to only one actuator for processing, it is necessary to optimally map each
sensorto itsactuator. Also consideringthe factthattheend-to-enddelaysinwireless
sensor-actuator networksisahardconstraint, wejointly optimizetheactuatorselection andoptimal
ow routing subject to energy and delay constraints with the global aim of maximizing the
networklifetime. Weproposedandevaluated(usingns-2)ouractuator-selection(LEAD-ADP)
and routing scheme (LEAD-RP) on top of a TDMA based MAC (LEAD-MAC) protocol.
We thenuse the Lagrangian dual decomposition method to devise a distributed primal-dual
algorithm tomaximizenetwork-lifetime inthenetwork. Thedeterministicdistributed
primal-dual algorithm requires no feedback control and therefore converges almost surely to the
optimal solution. The results show that the required optimal value of lifetime is achieved
for every node in the network by the distributed primal-dual algorithm. We also provide
a comparison to the analytical bound. Simulation results show that this approach has
near-optimalperformanceandispracticallyimplementableascomparedtoearlieranalytical studies
based onlyon numerical evaluations.
Thischapter also addresses the problem of isolated regionsin thesensingeld byletting
actuatorsexchangetheirCSITandjointlyperformbeamforminginordertodeliverscheduling
information to sensor nodes. The gains of cooperation were shown by simulating the
aver-age number of isolated sensors for the case of single actuator transmission and cooperative
transmission.
Inthenear-future,wewillconsiderareal-lifeSANETapplicationandsimulateitsbehavior
with the LEAD self-organizing framework to observe its performance. We will take into
consideration a dynamic actuator-assignment scenario to timely transport data in a mobile
wirelesssensor-actuator network.
Table 4.1: Notations
Symbols Denitions
N
The total numberofSensorsinthe networkM
The total numberofActuators inthenetworkB
The total numberofBaseStations inthenetworke i
Initial energy ofa sensornodei E i
Initial energy ofan actuator nodei
g i
The locallygenerated data rateat sensori P rx
Power consumption coecient for receivingdatac i,j
Power consumption coecient for transmitting datafromsensori
to sensorj α, β
Two constants termsinpowerconsumption for transmitting datad i,j
The geographicdistance between two nodesi
andj f s s i k ,s ,A j l
or f s s i k ,A ,A l
l
The ow ratefrom sensor
i
to sensorj
(oractuatorl
) withsource anddestination beingsensork
andactuatorl
F s,s (or F s,A )
The setof ows from onesensorto another (orActuator node)F s,s i
The setof ows coming into sensori
The datavolume (inbits)transferred from sensor
i
tosensorj
(orActuatorl
) withsource and destination being sensor
k
and Actuatorl
υ s,s (υ s,A )
The setof volume froma sensorto anothersensor(oran actuator)υ s,s i
The setof incomingvolume into sensori
υ s i ,s (υ s i ,A )
The setof outgoing volume froma sensori
to anothersensor(or anactuator)The ow ratefrom actuator
i
to actuatorj
(orBaseStationl
)withsourceand destination beingactuator
k
andBaseStationl
F A,A (or F A,B )
The setof ows from oneactuator to another (orBaseStation)F A,A i
The setof ows coming into an actuatori
F A i ,A (F A i ,B )
The setof ows goingout of actuatori
to other actuators(orBaseStation)G i
The locallygathered dataat actuatori
,G i = P
i g i , (1 ≤ i ≤ N ) λ A i ,B l
If thedatagathered at actuatori
willbe transmittedto BaseStationl
,then
λ A i ,B l = 1;
otherwiseλ A i ,B l = 0
The datavolume (inbits)transferred from actuator
i
to actuatorj
(or BaseStation
l
) withsource anddestination beingactuatork
and BaseStation
l
υ A,A (υ A,B )
The setof volumesfrom aactuator to another actuator (ora BaseStation)υ A,A i
The setof incomingvolume into actuatori
υ A i ,A (υ A i ,B )
The setof outgoing volumesfroma actuatori
to anotheractuator (or aBaseStation)
µ A i ,B l = λ A i ,B l T
inMILP-relaxTable 4.2: Useful states for the sensor node with associated power consumption and delay
(time to reach
S 4
from anygiven state)OperatingState StrongARM Memory ADC Radio PowerConsumption Delay(ms) NotationUsed
S 0
Sleep sleep O O 50(µ
W) 50E s node 0
S 1
Sleep Sleep On O 5(mW) 20E s node 1
S 2
Sleep Sleep OnRx
10(mW) 15E s node 2
S 3
Idle Sleep OnRx
100(mW) 5E s node 3
S 4
Active Active OnT x, Rx
400(mW) NAE s node 4
Table 4.3: The simulation area is such that there are atleast two sensors in each others
transmissionrange
Sensors Area (
m 2
) Actuators100 500*500 2
150 600*600 3
.
.
.
.
.
.
.
.
.
400 970*970 8
Cross-Layer Routing in UASNs
UASNs consist of sensors that are deployed to perform collaborative monitoring of tasks
over a given volume of water. These networks will nd applications in oceanographic data
collection,pollutionmonitoring,oshoreexploration,disasterprevention,assistednavigation,
tacticalsurveillance,and mine reconnaissance. The qualityof theunderwater acousticlinkis
highly unpredictable, since it mainly depends on fading and multipath, which are not easily
modeledphenomena. This inreturn severelydegrades theperformance at higher layers such
as extremely long and variable propagation delays. In addition, this variation is generally
larger inhorizontal links than invertical ones.
Inthis chapter, we rst analyzea modulation schemeand associatedreceiver algorithms.
Thisreceiver design take advantage of the TR and properties of spread spectrum sequences
known as Gold sequences. Furthermore, they are much less complex than receivers using
adaptive equalizers. This technique improves the signal-to-noise ratio (SNR) at the receiver
and reduces the biterror rate (BER).We then applied PC to the caseof network
communi-cation. We show thatthis approach can give almost zero BERfor a two-hop communication
mode compared to the traditional direct communication. This linklayerinformation isused
at thenetworklayerto optimize routingdecisions. Weshowtheseimprovementsbymeans of
analyticalanalysisand simulations.
5.1 Introduction
Acoustic signaling for wireless digital communications inthe sea environment can be a very
attractivealternativetobothradiotelemetryandcabled systems. However,time-varying
mul-tipath andoftenharshambientnoiseconditionscharacterizetheunderwateracousticchannel,
oftenmakingacoustic communications challenging. Thesensors mustbeorganized inan
au-tonomous network that self-congures according to the varying characteristics of the ocean
environment. Major challenges inthedesign ofUASNsare:
•
The channel isseverelyimpaired,mainly due to multipath.•
Temporary lossof connectivity mainly dueto shadowing.•
Thepropagation delay isve ordersofmagnitude higherthaninradiofrequencyterres-trial channels andis usuallyvariable[4 ].
•
Extremely low availablebandwidth.•
Limited battery energy at disposal.In this chapter, we present our analysis of a modulation scheme and associated receiver
al-gorithms. We also present the quantication of SNR and BER gains using PC in a single
transmitter-receiver setting. We then applied PC technique to a multi-hop communication
systesm. Thislinklayerinformationisusedatthenetworklayertooptimizerouting decisions.
Thiscross-layering improvesthenetworklifetimeofbatteryoperatedUASNsbyreducingthe
number ofretransmission attempts.
The organization of this chapter is as follows. Section 5.2details some of theinteresting
related work. In Section 5.3, we discuss the basic building blocks that contributed to the
proposed solution. We present the receiver algorithms that take advantage of TR and low
cross-correlation of Gold sequences for single-hop point-to-point communication and its
per-formance analysis in Section 5.4. We apply the idea of PC on a linear network to improve
someperformancemetricsinSection5.5. Wealsopresentadistributedrouting algorithm and
its performance analysis for a larger network size. In Section 5.6, we conclude the chapter
andoutlinethe future directions.