DTIC ADA459345: Spatial Reuse and Collision Avoidance in Ad Hoc Networks with Directional Antennas

6 Pages·0.12 MB·English

by  Defense Technical Information Center

Spatial Reuse and Collision Avoidance in Ad Hoc Networks with Directional Antennas
YuWang J.J.Garcia-Luna-Aceves
DepartmentofComputerEngineering
UniversityofCaliforniaatSantaCruz
SantaCruz,CA95064
U.S.A.
{ywang,jj}@cse.ucsc.edu
Abstract—Thequestforefficientmediumaccesscontrol(MAC) However, these schemes usually force all those neighbors
protocolsformulti-hopadhocnetworkshasarousedgreatinter- who overhear the transmissions between the sending and re-
estinusingdirectionalantennas. SomeMACprotocolsusingdi-
ceivingnodestodeferaccesstothesharedchannelduringthe
rectionalantennashavebeenproposedinthepast,whichtradeoff
whole handshake. This is required because any transmission
spatialreuseandcollisionavoidanceviaacombinationofomni-
directionalanddirectionaltransmissionmodes.Inthispaper,itis oftheseneighborscaninadvertentlycollidewiththeon-going
arguedthatthebenefitofspatialreuseachievedbyaMACproto- handshake. Thus,theseschemesreducesubstantiallythepos-
colthatusesdirectionalmodeinalltransmissionscanoutweigh sibilityofspatialreuseenabledbymulti-hopnetworksandcan
the benefit of a conservative collision avoidance MAC protocol
limitthemaximumachievablethroughput.
thatsendssomeomni-directionalcontrolpacketstosilencepoten-
tialinterferingnodes.Detailedsimulationexperimentsofthepop- Smartantennasoradaptiveantennashavebeenusedincel-
ular IEEE 802.11 MAC protocol and its variants that make use lular networks to narrow transmitting or receiving to certain
ofdirectionaltransmissionmodeinsufficientlyrandomnetworks directions while filtering out unnecessary interferences from
are presented. It is concluded that, in contention-based MAC
other directions and thus greater throughput can be achieved.
protocolsformulti-hopnetworksinfestedwithhiddenterminals,
Asaresult,someresearchers[4,5]haveconsideredusingthese
theaggressivechannelaccessschemefeaturedbyall-directional
transmissionsindeedoutperformsotherconservativeschemesin antennasinthedesignofMACprotocolsforadhocnetworks.
termsofenhancedthroughputandreduceddelay. Alltheproposedschemesusedirectionaltransmissionmodein
transmittingdataandacknowledgmentpackets,andthemajor
differences among them lie in how control packets (RTS and
CTS) are sent and how transmitting and receiving with direc-
I. INTRODUCTION
tionalantennasaremodeled.
To address the inherent “hidden terminal” problem [1] in For example, Ko et al. [4] propose two schemes. In one
multi-hop ad hoc networks that can degrade throughput dra- scheme, nodes use directional transmission of RTS packet
matically, some coordination between a pair of sending and and omni-directional transmission of CTS packet for colli-
receivingnodesandotherpotentialinterferingnodestoavoid sion avoidance. In the other scheme, nodes use both direc-
collisionsismandatory.VariouscollisionavoidanceMACpro- tionalandomni-directionaltransmissionofRTSpacketsalter-
tocols[2,3]havebeenproposedintherecentpast,andmostof natively. Theomni-directionalmodeisusedwhenthelocation
thememployafour-wayhandshakebetweenapairofsending ofthereceiverisnotwellknownorallofthetransmittingan-
andreceivingnodes. Thatis, theactualdatapackettransmis- tennasareunblocked. Itisclearthatinthesetwoschemesthe
sionanditsacknowledgmentareprecededbyshortrequest-to- authors are weighing the tradeoff between the increased pos-
send(RTS)andclear-to-send(CTS)packetsfromthesending sibility of simultaneous transmissions by neighboring nodes
and receiving nodes respectively. The RTS and CTS control (scheme one) and the reduced possibility of collision of con-
packets are used to silence those neighboring nodes that can trolpackets(schemetwo).
overhear these packets. These protocols differ in the lengths
Nasipuri et al. [5] also propose a different model for MAC
ofcontrolpackets,whetherpacketsensingorcarriersensingis
protocol enabled by directional antennas. In the authors’
used,andthechoiceofbackoffschemes. Thoughnotallpro-
model, each node is equipped with M antennas whose orien-
tocolscanensurecollision-freetransmissionsofdatapackets,
tationscanbemaintainedallthetime,regardlessofthenodes’
generallytheyareshowntoalleviate,ifnoteliminate,thehid-
movement. It is also assumed that nodes have directional re-
den terminal problem and thus perform much better than the
ceptioncapability,i.e.,nodescanactivatetheantennapointing
carriersensemultipleaccess(CSMA)protocols[1].
to the direction of the desired source while deactivating an-
tennas in otherdirections. Thus, the receiving node isnot in-
ThisworkwassupportedinpartbytheDefenseAdvancedResearchProjects
fluencedbysimultaneoustransmissionsfromotherdirections.
Agency(DARPA)underGrantNo. DAAD19-01-C-0026andbytheUSAir
Force/OSRunderGrantNo.F49620-00-1-0330. This is different from the model assumed by Ko et al. [4], in
Report Documentation Page Form Approved
OMB No. 0704-0188
Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and
maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information,
including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington
VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if it

does not display a currently valid OMB control number.
1. REPORT DATE 3. DATES COVERED
2002 2. REPORT TYPE 00-00-2002 to 00-00-2002
4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER
Spatial Reuse and Collision Avoidance in Ad Hoc Networks with
5b. GRANT NUMBER
Directional Antennas
5c. PROGRAM ELEMENT NUMBER
6. AUTHOR(S) 5d. PROJECT NUMBER
5e. TASK NUMBER
5f. WORK UNIT NUMBER
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION
University of California at Santa Cruz,Department of Computer REPORT NUMBER
Engineering,Santa Cruz,CA,95064
9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR’S ACRONYM(S)
11. SPONSOR/MONITOR’S REPORT
NUMBER(S)
12. DISTRIBUTION/AVAILABILITY STATEMENT
Approved for public release; distribution unlimited
13. SUPPLEMENTARY NOTES
14. ABSTRACT
15. SUBJECT TERMS
16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF 18. NUMBER 19a. NAME OF
ABSTRACT OF PAGES RESPONSIBLE PERSON

a. REPORT b. ABSTRACT c. THIS PAGE 5
unclassified unclassified unclassified
Standard Form 298 (Rev. 8-98)
Prescribed by ANSI Std Z39-18
whichantennasarealwaysactiveforreceivingandthustrans- GROUP 1 GROUP 2
missionstodifferentantennasresultinfailedreception. Inthe
proposedMACprotocol,omni-directionalRTSandCTSpack-
E
etsarefirstexchangedbetweenapairofsendingandreceiving D
B
nodes and then antennas of these two nodes are directed to-
A C
wardseachothertoreceiveensuingdataandacknowledgment
F
packets. Itcanbeinferredthatcollisionavoidanceisweighed
moreinthisschemebyusingomni-directionalmodeintrans-
mitting control packets, though it also has the advantage that
the locations of the sending and receiving nodes need not be
knownbyeachotherinadvance.
Theseschemeshavebeenshowntoperformbetterthanthe Fig.1. SpatialReusevsCollisionAvoidance
existing omni-directional IEEE 802.11 MAC protocol on top
ofwhichmostadhocroutingprotocolsarebuilt. However,the
performanceevaluationsaredoneinrelativelysimple,regular hopneighbors1 ofthenodespartakinginahandshakearedis-
networktopologiesanddonotprovidemuchinsightinthein- couraged from initiating handshakes with one-hop neighbors
teractionbetweenspatialreuseandcollisionavoidance,where of the nodes. The effect is especially prominent when an ad
theformeroptsfordirectionaltransmissionwhilethelateropts hoc network is composed of loosely coupled groups which
foromni-directionaltransmission. Itisalsonotclearwhether are connected by a few “hub” nodes, as shown in Fig. 1.
thecombinationofomni-directionalanddirectionaltransmis- In Fig. 1, the dashed lines show where inter-group commu-
sionmodesfeaturedbytheproposedschemesindeedachieves nications take place. It is clear that, when omni-directional
theoptimaltradeoffbetweenspatialreuseandcollisionavoid- RTS/CTSschemeisenforced,theinter-groupcommunications
ance. Therefore, in this paper, we investigate the interaction can almost prevent other inner-group communications from
betweenspatialreuseandcollisionavoidancewhendirectional taking place. Similarly, certain inner-group communications
antennasareutilized. will also prohibit inter-group communications. Hence, the
achievable throughput can be unduly low. The other problem
Wefirstshowthattheuseofomni-directionaltransmission
is that, with the increase of one-hop and two-hop neighbors,
mode for control packets can largely nullify the spatial reuse
itismoredifficulttogetallthenodescoordinatedwelldueto
benefitofdirectionalantennasinSectionII. Wethenpropose
therandomnatureofthechannelaccessandthelimitedinfor-
that a scheme in which all transmissions are directional may
mationavailabletoeachnode. Notknowingothernodes’con-
perform better than other conservative schemes that transmit
tentionstatus(busy,idle,deferringorbackingoff),nodesmay
some control packets in omni-directional mode for collision
havetochooseartificiallyverylargebackoffvaluestoshiftthe
avoidance. InSectionIII,weelaborateontheantennaandnet-
timesoftheirattemptstoaccessthechannelandthusmuchof
workmodelstobeusedinoursimulations.TheninSectionIV,
thevaluablechannelresourceiswastedinwaiting.
simulationresultsofthreeMACschemesbasedonthepopular
IEEE802.11MACprotocolarepresented. Itisshownthatthe The above discussion may lead to the conclusion that col-
MACschemeinwhichalltransmissionsaredirectionalindeed lision avoidance becomes ineffective in relatively crowded
achievesthebestperformanceamongthethreeintermsofen- multi-hop networks. In fact, the conclusion is supported by
hanced throughput and reduced delay. Section V concludes priorwork[6,7].
thispaper. Some researchers have realized the benefit of spatial reuse
enabled by directional antennas and their proposed schemes
have been introduced in Section I. In these schemes, omni-
II. SPATIALREUSEORCOLLISIONAVOIDANCE? directionaltransmissionofsomecontrolpackets(mostnotably,
CTS packets) are unanimously used for collision avoidance.
Itisapopularbeliefthatcollisionavoidanceisofparamount
The effort to have a balanced tradeoff between spatial reuse
importance in the design of MAC protocols for multi-hop ad
andcollisionavoidanceisconspicuous. However,herewewill
hoc networks due to the adverse effects of hidden terminals.
show that the omni-directional transmission of CTS packets
Collision avoidance schemes have been shown to alleviate, if
can largely nullify the spatial reuse, even if other packets are
noteliminate,thehiddenterminalproblem. Theseschemesre-
transmitteddirectionally.ReferringbacktoFig.1,considerthe
quire coordination from those overhearing neighboring nodes
fournodesA–D.SupposeAandBhavepacketsforeachother
whothenhavetodeferaccesstothesharedchannelduringthe
andthesameisthecaseforC andD. Afteranexaminationof
wholehandshake.
this scenario, it is clear that only flows A → B and C → D
However,therearepotentiallytwoproblemsassociatedwith
cancoexistatthesametime,becausetheomni-directionalCTS
these schemes. One is that spatial reuse is greatly reduced.
Not only are the direct neighbors (or one-hop neighbors) dis-
1Herewerefertothosenodesthathaveatleastonecommonneighborwitha
couraged from initiating any transmissions, but also the two- nodebutarenotdirectneighborsofthenodeasthenode’stwo-hopneighbors.
from either A or C prevents the flow at the other side from In contrast with the rather regular network topologies used
taking place. Even under the assumption of the capability of previously, where nodes are placed on a uniform grid, nodes
directionalreception[5],thosenodeswhooverhearaCTSstill aredistributeduniformlyinplanarcircles(orrings)inournet-
have to defer access to the channel for the whole handshake workmodel. Inthisway,nodesaredistributelymuchlessreg-
accordingtothespecificationoftheprotocol. Thedirectional ularly, because nodes can have different number of one-hop
reception capability just helps to filter out some interference neighborsandtwo-hopneighborsindifferentdirections. Tobe
from others nodes that cannot receive the CTS clearly and it specific,weplacenodesinconcentriccirclesorrings. Thatis,
doesnotimprovespatialreusemucheither. Thus,itisdubious given that a node’s transmitting and receiving range is R and
iftheomni-directionalCTSschemeperformsmuchbetterthan that there are on average N nodes within this circular region,
thebasicschemeinnetworkswithsufficientlyrandomtopolo- weplaceN nodesinacircleofradiusR,subjecttoauniform
gies. distribution. Becausethereareonaverage22N nodeswithina
However, in a scheme where all transmissions are direc- circleof radius2R, we place22N −N = 3N nodesoutside
tional,flowsA→BorB →AcancoexistwitheitherC →D thepreviouscircleofradiusRbutinsidetheconcentriccircle
or D → C and spatial reuse is maximized. Even though the ofradius2R,i.e.,theringwithradiiRand2R,subjecttothe
useofdirectionaltransmissionsdoesnotforceallneighboring sameuniformdistribution. Then32N−22N =5N nodescan
nodestodeferaccesstothesharedchannel,collisionswillnot beplacedinanouterringwithradii2Rand3R,andsoon.
necessarilyhappeniftheseneighboringnodesdonottransmit Tofilteroutboundaryeffectsthatsomeboundarynodesface
inthedirectionofeitherthesenderorthereceiver. Forexam- lesscontentionandmayattainhigherthanaveragethroughput,
ple,inFig.1,nodesEandF donotdeferaccessforBandthe wejustfocusourattentionontheaverageperformanceofthe
handshakebetweenthemwillnotaffectB ifthetransmission innermost N nodes. According to our experiments, conclu-
beamwidthissufficientlynarrow. Obviously,theuseofomni- sionsdrawnfromacircularnetworkofradiusofmorethan3R
directionaltransmissionofCTSpacketcanbequitedisruptive do not affect the conclusion to be drawn in the next section.
inthesecases. Therefore,wepresentonlytheresultsforacircularnetworkof
Based on the above arguments, we can reason that an all- radius3R.
directionaltransmissionscheme2 mayachievebetterthrough- Toavoidsomeextremecases,weonlyusenetworktopolo-
putthanthoseconservativecollision-avoidanceschemeswhen giesthatsatisfythefollowingrequirements:
the benefit of spatial reuse outweighs that of collision avoid- • For the inner N nodes, each node should have at least 2
ance,whichweinvestigatehereafter. neighborsandatmost2N −2neighbors.
• For the intermediate outer 3N nodes, each node should
III. SIMULATIONMODELS haveatleast1neighborandatmost2N −1neighbors.
Different assumptions can be made regarding the function- Tobespecific,evenwhennodesaredistributeduniformly,the
alitiesofdirectionalantennas.Wemakethefollowingassump- numberoftheneighborsofanynodeinsuchnetworkscanstill
tionsinourmodel: vary considerably. It is not uncommon for some innermost
nodestohavenoneighborswhilesomeothernodeshavemore
• Anodeisequippedwithantennassuchthat,whendirec-
than3N neighbors,ifwedonotenforcesuchrequirements.
tionaltransmissionwithabeamwidthofθ isused,nodes
In our simulation, we investigate three typical MAC
outside the beamwidth will not receive any signal from
schemes. In the first one, all packet transmissions are omni-
thenode.
directional, which is just the scheme commonly used in tra-
• When a node is transmitting with one of its antennas, it
ditional MAC protocols that emphasizes collision avoidance.
appears “blind” in other directions. This is also the case
Forthesakeofsimplicity,wecallthisscheme“ORTS-OCTS.”
when each node is equipped with only one steerable an-
Though there are quite a few “dialects” of this scheme, we
tenna. When a node is transmitting, it cannot sense any
choose the IEEE 802.11 MAC protocol as the example. In
otherchannelactivityatall.
the second case, an RTS packet is transmitted directionally
• Wedonotassumedirectionalreceptioncapability,which
andCTSpacketistransmittedomni-directionallyandthendata
is a node’s ability to receive transmissions from one di-
packet and acknowledgment packet are transmitted direction-
rectiononly,whiledeactivatingantennaspointingtoother
ally. This is the scheme that tries to strike a balance between
directionstoavoidunnecessaryinterference.
collision avoidance and spatial reuse, and it is called “DRTS-
Therationalebehindtheseassumptionsisthatthemajorityof
OCTS” scheme. The third case is called “DRTS-DCTS,” in
mobile nodes to be deployed in ad hoc networks in the near
whichallpackettransmissionsaredirectional. Obviously,this
future are unlikely to be equipped with multiple powerful di-
schemeemphasizesspatialreuse.
rectionalantennasonaparwithbasestationsincurrentcellular
networks.
IV. SIMULATIONRESULTS
2Despitethename’ssimilaritytoomni-directional,hereitreferstothespe-
In this section, we investigate the performance of the three
cificschemewhereallpacketsaretransmitteddirectionallyjustforthesakeof
simplicity. schemesintroducedinSectionIII. WeuseGloMoSim2.0[8]
TABLEI
way of channel access to achieve spatial reuse, DRTS-DCTS
IEEE802.11PROTOCOLCONFIGURATIONPARAMETERS scheme also enjoys on average less delay than the other two
schemes, especially when N is large. In addition, it is also
RTS CTS data ACK DIFS SIFS
desirable to use narrower beamwidth in the DRTS-DCTS
20-byte 14-byte 1460-byte 14-byte 50µsec 10µsec
scheme,evenwhenitdoesnotaffectthethroughputmuchfor
contentionwindow slottime sync.time prop.delay
thecaseofsmallN,becausenodesarelessaffectedbythesur-
31–1023 20µsec 192µsec 1µsec
roundingsendingandreceivingnodesandthuscanspendless
timewaiting.
We also experiment with some configurations in which the
as the network simulator. The traditional IEEE 802.11 MAC
numberofnodesintheinnermostcircleofradiusR isdiffer-
protocol just uses ORTS-OCTS scheme. We implement the
entfromN andwedenoteitbyi.Bychoosingdifferentvalues
other two schemes based on the existing IEEE 802.11 imple-
ofi,wecanvarythecontentiondensitiesaroundthesenodes.
mentation for fair comparison. Direct sequence spread spec-
When i is more than N, the transmissions to and fro these
trum(DSSS)parametersareusedthroughoutthesimulations,
nodes loosely correspond to the inner-group communications
whichareshowninTableI.Therawchannelbitrateis2Mbps.
showninFig.1. WheniissmallerthanN,thenthetransmis-
Inoursimulation,eachnodehasaconstant-bit-rate(CBR)traf-
sions loosely correspond to the inter-group communications
fic generator with data packet size of 1460-byte, and one of
showninFig.1. Duetospacelimitation,Theresultsshowwe
its neighbors is randomly chosen as the destination for each
onlyshowtheresultsforthecasewhenN =5,i=3,5,10in
packetgenerated. Allnodesarealwaysbackloged.
Fig.4andmoreresultsareavailablefromtheauthorsuponre-
We run simulations with N = 3,5,8 with beamwidth
θ = 30◦,90◦ and 150◦. We generate 50 random topologies quest.ItisclearthattheDRTS-DCTSschemestilloutperforms
undisputablytheothertwoschemes,evenwhenthecontention
that satisfy the uniform distribution and then obtain averaged
densitiesarenotuniform.
throughput and delay for the N nodes in the innermost cir-
cle ofradius R for eachconfiguration. The results areshown
in Figs. 2 and 3. In these figures, the vertical lines show the V. CONCLUSION
range of throughput achieved by each scheme, that is mean In this paper, we have discussed the tradeoff between spa-
± standard variance. These lines are shifted a bit for clarity. tial reuse and collision avoidance in contention-based MAC
ItcanbeseenthatthethroughputoftheDRTS-DCTSscheme protocols for multi-hop ad hoc networks. It is shown that,
doesnotdegrademuchinarelativelargerangeoftransmission whendirectionaltransmissionisused,itismuchbettertouseit
beamwidthswhenN issmallandthethroughputoftheDRTS- throughoutthewholehandshake. Theomni-directionaltrans-
OCTS scheme degrades very little regardlessof the transmis- missionofsomecontrolpacketsforcollisionavoidancewillin
sion beamwidth. This can be explained as follows. When a factdefeatthepurposeofusingdirectionalantennastoachieve
node has few neighbors, it usually does not make much dif- betterthroughput. Simulationresultsofsomesufficientlyran-
ferenceifthenodetransmitswitheitheranarrowerorawider domnetworksvalidatethisargumentandshowthattheDRTS-
beamwidth. For example, if a node has three neighbors that DCTS scheme indeed outperforms the other two conserva-
are distributed around it, it can transmit with beamwidth of tive collision-avoidance schemes in terms of both enhanced
either 30◦or 90◦. However, in reality, it is usually more de- throughputandreduceddelay.
sirable to transmit with narrower beamwidth, because signal
energyismoreconcentratedandahighersignal-to-noiseratio
REFERENCES
canbeachieved, thoughphysicallayerimpairmentotherthan
[1] F.A.TobagiandL.Kleinrock,“PacketSwitchinginRadioChannels:Part
Gaussianwhitenoiseisnotmodeledinthesimulations. Fig.3 II-theHiddenTerminalProbleminCarrierSenseMultiple-accessModes
alsoshowsthatnarrowerbeamwidthtransmissionscanleadto andtheBusy-toneSolution,”IEEETrans.onCommunications,vol.23,
no.12,pp.1417–1433,1975.
reduceddelayintheDRTS-DCTSscheme.
[2] J.J.Garcia-Luna-AcevesandC.L.Fullmer, “FloorAcquisitionMulti-
The results reported in Figs. 2 and 3 clearly show that pleAccess(FAMA)inSingle-channelWirelessNetworks,”ACM/Baltzer
the DRTS-DCTS scheme outperforms the other two MAC MobileNetworksandApplications,vol.4,no.3,pp.157–174,1999.
[3] IEEE Computer Society LAN MAN Standards Committee, ed., IEEE
schemes when beamwidth is narrow. In addition, it is also
Standard for Wireless LAN Medium Access Control (MAC) and Physi-
shownthattheDRTS-OCTSschemeperformsonlymarginally calLayer(PHY)Specifications. IEEEStd802.11-1997,TheInstituteof
betterthantheORTS-OCTSschemeintherandomtopologies ElectricalandElectronicsEngineers,NewYork,1997.
[4] Y.-B.Ko,V.Shankarkumar,andN.H.Vaidya,“MediumAccessControl
investigatedhere. Thisshowsthattheunwittinguseofomni- ProtocolsUsingDirectionalAntennasinAdHocNetworks,”inIEEEIN-
directional transmission of CTS packet can make almost all FOCOM2000,Mar.2000.
[5] A.Nasipuri, S.Ye, J.You, andR.E.Hiromoto, “AMACProtocolfor
overhearing but non-interfering hidden terminals defer access
MobileAdHocNetworksUsingDirectionalAntennas,”inProceedingsof
tothechannel,andthuscannullifyalmostallthespatialreuse theIEEEWirelessCommunicationsandNetworkingConference(WCNC)
benefitthatdirectionaltransmissionsbringforth. Theseresults 2000,(Chicago,IL,U.S.),Sept.2000.
[6] Y.WangandJ.J.Garcia-Luna-Aceves, “CollisionAvoidanceinMulti-
confirmtheconjecturesmadeinSectionII.
HopAdHocNetworks,”inProc.ofIEEE/ACMMASCOTS2002,(Forth
It is also clear from Fig. 3 that, with a more aggressive Worth,Texas,U.S.A.),Oct.2002.
Throughput comparison (N=3) Throughput comparison (N=5) Throughput comparison (N=8)
0.9 0.8 0.7
DRTS−DCTS DRTS−DCTS DRTS−DCTS
DRTS−OCTS DRTS−OCTS DRTS−OCTS
0.8 ORTS−OCTS 0.7 ORTS−OCTS 0.6 ORTS−OCTS
0.7 0.6
0.5
0.6
0.5
Throughput00..45 Throughput0.4 Throughput00..34
0.3
0.3
0.2
0.2 0.2
0.1 0.1 0.1
00 1/6 1/3 Beamwidth1 /(2x p radians) 2/3 5/6 1 00 1/6 1/3 Beamwidth1 /(2x p radians) 2/3 5/6 1 00 1/6 1/3 Beamwidth1 /(2x p radians) 2/3 5/6 1
(a)N=3 (b)N=5 (c)N=8
Fig.2. ThroughputComparison–Set1
7800 Delay comparison (N=3) DDORRRTTTSSS−−−DOOCCCTTTSSS 120 Delay comparison (N=5) DDORRRTTTSSS−−−DOOCCCTTTSSS 116800 Delay comparison (N=8) DDORRRTTTSSS−−−DOOCCCTTTSSS
100
60 140
80 120
50
Delay (ms)40 Delay (ms)60 Delay (ms)10800
30
40 60
20
40
20
10 20
00 1/6 1/3 Beamwidth1 /(2x p radians) 2/3 5/6 1 00 1/6 1/3 Beamwidth1 /(2x p radians) 2/3 5/6 1 00 1/6 1/3 Beamwidth1 /(2x p radians) 2/3 5/6 1
(a)N=3 (b)N=5 (c)N=8
Fig.3. DelayComparison–Set1
00..78 Throughput comparison (N=5, i=3) DDORRRTTTSSS−−−DOOCCCTTTSSS 00..78 Throughput comparison (N=5,i=5) DDORRRTTTSSS−−−DOOCCCTTTSSS 0.04.54 Throughput comparison (N=5, i=10) DDORRRTTTSSS−−−DOOCCCTTTSSS
0.6 0.6 0.35
0.3
0.5 0.5
Throughput0.4 Throughput0.4 Throughput0.02.52
0.3 0.3
0.15
0.2 0.2
0.1
0.1 0.1 0.05
00 1/6 1/3 Beamwidth1 /(2x p radians) 2/3 5/6 1 00 1/6 1/3 Beamwidth1 /(2x p radians) 2/3 5/6 1 00 1/6 1/3 Beamwidth1 /(2x p radians) 2/3 5/6 1
(a)N=5,i=3 (b)N=5,i=5 (c)N=5,i=10
Fig.4. ThroughputComparison–Set2
[7] Y.WangandJ.J.Garcia-Luna-Aceves,“PerformanceofCollisionAvoid- SimulationofLarge-scaleWirelessNetworks,”inProc.ofthe12thWork-
anceProtocolsinSingle-ChannelAdHocNetworks,”inProc.ofIEEE shoponParallelandDistributedSimulations,May1998.
ICNP2002,(Paris,France),Nov.2002.
[8] X.Zeng,R.Bagrodia,andM.Gerla,“GloMoSim: aLibraryforParallel