The Impact of Different Radio Propagation Models for Mobile Ad hoc NETworks (MANET) in Urban Area Environment

Transcription

1 The Impac of Differen Raio Propagaion Moels for Mobile A hoc NETworks (MANET) in Urban Area Environmen Ibrahim Khier Elahir Communicaion Sofware an Swich Cener,Dep of Elecronic an Informaion Sysems Huazhong Universiy of Science an Technology,P.R.China-Wuhan-37 E_mail: ibrahimchina8@yahoo.com Absrac Nework simulaion ools are frequenly use o analyze performance of MANET proocols an applicaions. They commonly offer only simple raio propagaion moels ha neglec obsacles of a propagaion environmen. The raio wave propagaion moel has a srong impac on he resuls of he simulaion run. This work shows ifferen raio propagaion moels ino a simulaion ool. The moel is base on aa of he simulaion area in urban area. Consequenly, we obain ifferen performance evaluaion resuls. This paper gives insighs on he effec of ifferen propagaion moels for MANETs in inoor an ouoor environmens an presens he parameers an he resuls of simulaing using nework simulaor. Key wors: Mobile Ahoc Nework, Propagaion moels, Mobiliy Moels, Simulaion. 1. Inroucion Mobile a-hoc neworks (MANETs) are creae sponaneously by wireless communicaion peers, wihou relying on a fixe infrasrucure. The evices communicae irecly wih each oher when hey are in ransmission range. Nework simulaion ools [1], [] are frequenly use o analyze he performance of MANET proocols an applicaions. These ools moel he applicaions running on mobile evices, he wireless nework proocol sack, raio signal propagaion, an he mobiliy of he nework users. The raio propagaion moels use in common MANET simulaors assume an obsacle-free area an a free line-of sigh beween all communicaing parners. As a consequence, he communicaion range is moele by a simple circle aroun he mobile evice. However, his poorly reflecs raio wave propagaion in a ypical ouoor scenario, like a ciy cener, in which builings significanly affec he communicaion beween noes. Neverheless, he vas majoriy of publicaions ha invesigae MANET proocol an applicaion behavior sill rely on such simple moels. Due o he naure of self-organizaion, he ynamic opology cause by mobiliy an ransmission power conrol, an he muliple-hop rouing in MANETs, i is ifficul o buil a complee analyical moel o suy he nework performance. On he oher han, a real esbe is expensive. Therefore, he simulaion suy of MANETs is imporan. In his paper, we suy ifferen raio propagaion moels using ns- [] because i is open source an is wiely use in boh acaemia an inusry. In ns- he raio propagaion moels have he following feaures: he Friss-space moel is use for shor isances an he approximae wo-ray-groun moel is use for long isances. The shaowing moel is employe o characerize he probabilisic muliple pah faing uring raio propagaion. These moels are consiers aa of he simulaion area, which are available from urban environmen. Raio propagaion waves are necessary propagaion characerisics for any configuraion. The environmens sysems are inene o be insalle are ranging from inoor up o ouoor environmen. Hence wave propagaion moels are require covering whole range incluing inoor an ouoor environmen scenarios. The phenomena which effec raio wave propagaion can generally be escribe by five mechanisms as following; Reflecion: is he abrup change in irecion of a wave fron a an inerface beween wo issimilar meia so ha he wave fron reurns ino he meium from which i originae. Reflecing objec is large compare o wavelengh. Scaering: is a phenomenon in which he irecion (or polarizaion) of he wave is change when he wave encouners propagaion meium isconinuiies smaller han he wavelengh resuls in a isorere or ranom change in he energy isribuion. Diffracion: is he mechanism ha he waves sprea as hey pass barriers in obsruce raio pah (hrough openings or aroun barriers).diffracion is an imporan when evaluaing poenial inerference beween erresrial an saions sharing he same frequency. Absorpion is he conversion of he ransmie EM energy ino anoher Broaban an Ulra Wieban Communicaions (AusWireless 7) /7 $5. 7

2 form, usually hermal. The conversion akes place as a resul of ineracion beween he incien energy an he maerial meium, a he molecular or aomic level. One cause of signal aenuaion ue o walls, precipiaions (rain, snow) an amospheric gases. Refracion: is reirecion of a wave fron passing hrough a meium having a refracive inex ha is a coninuous funcion of posiion or hrough a bounary beween wo issimilar meia. For wo meia of ifferen refracive inices, he angle of refracion is approximae by Snell's Law known from opics peneraion. Figure.1 shows hese mechanisms. We use an exising implemenaion of he propagaion moel from a specialize ool. We prove ha he usage of raio propagaion moels change simulaion resuls consierably. The remainer of his paper is srucure as follows. In Secion II, we presen mobiliy moels. Secion III gives an overview of ifferen raio propagaion moels. Secion IV escribes he scenarios in urban area. Secion V presens he escripion of simulaion. Secion VI illusraes he resuls.secion VIIconclues his paper. Diffracion Scaering Reflecion Fig.1, The phenomena effec raio wave propagaion. Mobiliy Moels A mobiliy moel is a represenaion of a cerain real or absrac worl ha conains moving eniies. A mobiliy moel is usually use o escribe he mobiliy of an iniviual subscriber. Someimes i is use o escribe he aggregae paern of all subscribers. The following iscussion aemps a brief overview of he wo commonly use mobiliy moels o analyze esign sysems in wireless a hoc neworks, each wih a specific goal an suiable for a specific scenario. Ranom Way Poin mobiliy moel (RWP) [3] [] [5] [6] is a simple, wiely use, moel in he many simulaion suies of a hoc rouing proocols. In his moel each noe is assigne an iniial posiion uniformly isribue wihin a region (recangular region). Then, each noe chooses a esinaion uniformly insie he region, an selecs a spee uniformly from [minspee, maxspee] inepenenly of he chosen esinaion. Tha means he isribuions of noes spees an locaions are saionary. To avoi he ransien perio from he beginning, one soluion is o choose he noes iniial locaions an spees accoring o he saionary isribuion; anoher one is o iscar he iniial ime perio of simulaion o reuce he effec of such ransien perio on simulaion resuls. The noe hen moves owar he chosen esinaion wih he selece spee along a sraigh line saring from curren waypoin. Afer reaching he esinaion, he noe sops for uraion calle pause ime, an hen repeas he proceure. All noes move inepenenly of each oher a all imes. Manhaan moel is use o emulae he noes movemen on srees efine by maps [] [5]. The map is compose of a number of horizonal an verical srees. Each sree has wo lanes, one in each irecion (Norh an Souh for verical srees, an Eas an Wes for horizonal ones). Each noe is only allowe o move along he gri of horizonal an verical srees. A an inersecion of horizonal an verical srees, a mobile noe can urn lef, or righ, or go sraigh wih probabiliies.5,.5, an.5, respecively. The spee of a mobile noe is emporarily epenen on is previous spee If wo mobile noes on he same freeway lane are wihin he Safey Disance (SD), he velociy of he following noe canno excee he velociy of preceing noe Mobiliy moels capure he geographic resricions. The spee of a noe s () is upae accoringo: s ( + 1 ) = min( s max, max(, s() + a() * X )) where X Uniform [ 1, 1], an a () is Acceleraion Spee. 3. Differen Raio Propagaion moels Raio channels are much more complicae o analyze han wire channels. Their characerisics may change rapily an ranomly. There are large ifferences beween simple pahs wih line of sigh (LOS) an hose which have obsacles like builings or elevaions beween he sener an he receiver (Non Line of Sigh (NLOS)). To implemen a channel moel generally wo cases are consiere: large-scale an Broaban an Ulra Wieban Communicaions (AusWireless 7) /7 $5. 7

3 small-scale propagaion moels. Large scale propagaion moels accoun for he fac ha a raio wave has o cover a growing area when he isance o he sener is increasing. Small scale moels (faing moels) calculae he signal srengh epening on small movemens or small ime frames. Due o mulipah propagaion of raio waves, small movemens of he receiver can have large effecs on he receive signal srengh. In he following, four frequenly use moels for he ns- nework simulaor are escribe in more eail Free Space Moel This is a large scale moel. The receive power is only epenen on he ransmie power, he anenna s gains an on he isance beween he sener an he receiver. I accouns mainly for he fac ha a raio wave which moves away from he sener has o cover a larger area. So he receive power ecreases wih he square of he isance. The free space propagaion moel assumes he ieal propagaion coniion ha here is only one clear line-of-sigh pah beween he ransmier an receiver. H. T. Friis presene he following equaion o calculae he receive signal power in free space a isance from he ransmier [7] [8] [9][1]. P GGrλ Pr ( ) = (1) ( π ) L Where P is he ransmie signal power. G an Gr are he anenna gains of he ransmier an he receiver respecively. L( L 1) is he sysem loss, anλ is he wavelengh. I is common o selec G = G r = 1 an L = 1 in ns simulaions. 3.. Two Ray Groun Moel The Two Ray Groun moel is also a large scale moel. I is assume ha he receive energy is he sum of he irec line of sigh pah an he pah incluing one reflecion on he groun beween he sener an he receiver. A limiaion in ns- is ha sener an receiver have o be on he same heigh. I is shown ha his moel gives more accurae preicion a a long isance han he free space moel [7] [8] [9][1]. The receive power a isance is preice by: P GGrh hr Pr ( ) = () L Where h an hr are he heighs of he ransmi an receive anennas respecively. To be consisen wih he free space moel L is ae here. The above equaion shows a faser power loss han Eqn. (1) as isance increases. However, he wo-ray moel oes no give a goo resul for a shor isance ue o he oscillaion cause by he consrucive an esrucive combinaion of he wo rays. Insea, he free space moel is sill use when is small Ricean an Rayleigh faing moels These wo moels are faing moels, meaning ha hey escribe he ime-correlaion of he receive signal power. Faing is mosly cause by mulipah propagaion of he raio waves. If here are muliple inirec pahs beween he sener an he receiver, Rayleigh faing occurs. If here is one ominan (line of sigh) pah an muliple inirec signals, Ricean faing occurs [1]. 3.. moel The shaowing moel of ns- realizes he lognormal shaowing moel. I is assume ha he average receive signal power ecreases logarihmically wih isance. A Gaussian ranom variable is ae o his pah loss o accoun for environmenal influences a he sener an he receiver. The shaowing moel consiss of wo pars. The firs one is known as pah loss moel, which also preics he mean receive power a isance, enoe by P r ( ) I uses a close-in isance as a reference. P r ( ) is compue relaive o P r ( ) as follows. ( ) β Pr = (3) Pr ( ) β is calle he pah loss exponen, an is usually empirically eermine by fiel measuremen. From Eqn. (1) we know ha β = for free space propagaion. Table.1 gives some ypical values of β larger values correspon o more obsrucions an hence faser ecrease in average receive power as P r isance becomes larger. ( ) can be compue from Eqn. (1). The pah loss is usually measure in. So from Eqn. (18.) we have P ( ) ( ) r = 1β log () Pr The secon par of he shaowing moel reflecs he variaion of he receive power a cerain isance. I is Broaban an Ulra Wieban Communicaions (AusWireless 7) /7 $5. 7

4 a log-normal ranom variable ha is; i is of Gaussian isribuion if measure in. The overall shaowing moel is represene by P ( ) ( ) r = 1β log (5) Pr Where X is a Gaussian ranom variable wih zero mean an sanar eviaion σ. σ is calle he shaowing eviaion, an is also obaine by measuremen. Table. shows some ypical values of σ (). Eqn. (5) is also known as a log-normal shaowing moel. The shaowing moel exens he ieal circle moel o a richer saisic moel: noes can only probabilisically communicae when near he ege of he communicaion range [7] [8] [9][1]. Table.1, Some Typical values of pah loss β Ouoor In builing Environmen Free space β Shaowe urban area.7 o 5 Line-of-sigh 1.6 o 1.8 Obsruce o 6 Table. Some Typical values of shaowing eviaionσ Environmen σ () Ouoor o 1 Office, har pariion 7 Office, sof pariion 9.6 Facory, line-of-sigh 3 o 6 Facory, obsruce 6.8. Descripion of Urban Area Scenarios To evaluae he impac of he raio wave propagaion moel on he performance of a Mobile A Hoc Nework he hroughpu an elay of muliple consan bi rae (CBR) sreams is aken as an inicaor. Measuremens conuce by several researchers show ha mos simulaors give oo goo values for hese merics. So any preicion erive from his simulaion ha concerns real neworks is base on false assumpions. In his work, wo scenarios are simulae in eail. They represen very ifferen working environmens. One is an inoor scenario wih low mobiliy. The secon one is an ouoor scenario simulaing peesrian walking hrough a ciy. This scenario is characerize by hosile environmen for raio waves. Boh scenarios share some similariies: Nework raffic is creae by saring CBR connecions beween ranomly selece noes. The simulaion uraion is 9 sec.this secion inrouces scenarios of urban area environmen which is ivie in wo pars as following: - scenario The inoor scenario is conuce on a simulaion area whose layou shown in fig.1. For his layou a map wih D an some measuremens of he raio signal srengh exiss on an area of 1m 1m. The movemen for he noes is creae using ranom waypoin moel. Noes moving insie builing have a very low mobiliy. Their pause ime is equally isribue. The movemen spee is isribue uniformly. The number of noes in a scenario is en. The maximum number of CBR connecions is se o en; he offere loa per connecion is 51Bye/s. -Ouoor scenario The ouoor scenario is base on he sree map of ciy, when he noes are on he sree, hey move as Manhaan mobiliy moel movemen paern. The Builings ac as obsacles for he raio waves an narrow srees may ac as wave guies. Builings have high aenuaion bu o no compleely block he signal. The movemen noes are ivie in wo groups epening on heir spee a peesrian group wih a low spee an a vehicular group wih a higher spee. The peesrian group of users is moving wih a normal isribue spee wih a mean of 3 km/h an a sanar eviaion of.3 km/h []. The vehicular group of users has also a normal isribue spee bu wih a mean of 5 km/h an a sanar eviaion of.5 km/h. A each cross-roa, users of boh groups have can eiher coninue sraigh wih he probabiliy Pr (sraigh) =.5 or urn lef/righ wih he probabiliy Pr (righ) = Pr (righ) =.5. In his paper he area is wrappe aroun Norh-Souh an Wes-Eas an he gri is compose of 3 by 3 builings. The builings are 3x3 m an he sree has wo opposie lane, he isance beween lane 1 m an he wih of lane 6 meer. The movemens of a noe swich from one mobiliy moel (Manhaan or Ranom Waypoin) o anoher base on is locaion in urban environmen.fig. shows he movemen of noes in simulaion area using Manhaan an Ranom Waypoin mobiliy moels. Broaban an Ulra Wieban Communicaions (AusWireless 7) /7 $5. 7

5 environmen Environmen Environmen Ouoor environmen environmen Environmen environmen Fig., Layou of Urban environmen environmen RWP environmen environmen Ouoor environmen Simulaor (NS-) environmen wih CMU wireless ahoc neworking exension [7] an we exrac he useful aa from race file using mobiliy race analyzer ool (version 1. bea) [8], hen he graphs are generae using Malab.Simulaion environmen consiss of 1 wireless noes forming an a hoc nework, moving over a 1 X 1 fla space, AODV rouing proocol for 9 secons of simulae ime. Each run of he simulaor acceps as inpu a scenario file ha escribes he exac moion of each noe an he exac sequence of packes originae by each noe, ogeher wih he exac ime a which each change in moion or packe originaion is o occur. We have generae ifferen scenario files wih varying movemen paerns an raffic loas (CBR), he raffic consis of cbr ype wih 1 connecions, aa rae 1 packe/sec, packe size 51 bye an he ransmission range 5m. An hen ran agains each of hese scenario files. When Noes in sree move accoring o manhaan moel oherwise hey move accoring o Ranom Waypoin moel. The movemen scenario files we use for each simulaion are characerize by a max spee. Each simulaion ran for 9 secons. We ran our simulaions wih movemen paerns generae for 6 ifferen maximum spees, 1,, 3,, 5, 6 m/sec wih consan pause ime. sec. 6. Simulaion Resuls The analysis of simulaion resuls is performe base on he sanar merics of number of sen packes, hroughpu, number of roppe packes, packe elivery raio an packe rouing overhea beween ifferen raio propagaion moels. We conuce our simulaions on changing he parameers for mobile noes' movemen scenarios an heir connecion paern files. We suppose ifferen spee an pause ime for movemen scenarios files. The resuls are ivie ino wo subsecions: performance wih varying pause ime an performance wih varying velociy. Fig.3, Movemen of Noes in Simulaion Fiel using Manhaan an Ranom Waypoin 5. Simulaion Descripion A variey of marices have been use for he MANET environmen. In his secion, we suy he mos popular propagaion moels, Free Space moel, moel, an moel in he environmens where MH an RWP exis. Our evaluaions are base on he simulaion using Nework Performance wih varying velociy: Figure.3 inicaes ha shaowing moel sens very few packes an free space moel sen high packes. Figure.; shows ha free space an wo ray groun moels are beer performance han shaowing moel when hroughpu is consiere as meric.in conras, shaowing raio propagaion moel rops few packes as shown in figure.5, alhough he resuls inicae ha is hroughpu is lower han oher raio propagaion moels. Figure.6 evaluaes he reliabiliy of packe elivery raio. In comparison o free space an wo ray Broaban an Ulra Wieban Communicaions (AusWireless 7) /7 $5. 7

6 groun raio propagaion moels, he resuls in Figure.6 inicae ha shaowing raio propagaion moel is low wih respec o he measure elivery rae over a variey of velociies. Our resuls in Figure 7 show ha free space an wo ray groun perform consisenly well wih respec o rouing overhea over a variey of velociies. Performance wih varying pause ime: Figure.8 gives approximaely he same resul as in figure.3 for all propagaion moels. Figures.9, 11 show ha free space an wo ray groun moels relaively o he same performance when hroughpu an elivery rae are consiere as merics over a variey of velociies. Figure.1 inicaes ha free space rops more packes over a variey of pause imes. Our resuls in Figure1 show ha free space an wo ray groun perform he same resul as for ifferen spee wih respec o rouing overhea bu shaowing moel has more rouing overhea. packe sen (packe) 7 x Max Spee (m/sec) Fig.3, Packe sen vs. spee Throughpu packe roppe (packe) 18 x Max Spee (m/sec) Fig., Throughpu vs. spee 16 x Max Spee (m/sec) Fig.5, Packe roppe vs. spee Broaban an Ulra Wieban Communicaions (AusWireless 7) /7 $5. 7

7 Packe elivery raio (%) packe sen (packe) Max Spee (m/sec) 5 Fig.6, Packe elivery raio vs. spee x Fig.8, Packe sen vs. pause ime Rouing overhea (packe) hroughpu Max Spee (m/sec) Fig.7, Rouing overhea vs. spee Fig.9, Throughpu vs. pause ime Broaban an Ulra Wieban Communicaions (AusWireless 7) /7 $5. 7

8 packe roppe (packe) Rouing overhea (packe) Packe elivery raio (% ) Fig.1, Packe roppe vs. pause ime Fig.11, Packe elivery raio vs. pause ime Fig.1, Rouing overhea vs. pause ime 7. Conclusion This work presene a raio wave propagaion moels showe how hese moels affec he performance of Mobile A Hoc Neworks in urban area. We presen he opions available an provie he parameers use in he creaion of scenarios for inoor an ouoor environmens in an urban environmen. We showe he effec of raio propagaion moels for wave propagaion ino ns-. We emonsrae ha he usage of more accurae raio propagaion moel changes simulae opologies consierably beween commonly use propagaion moels. Consequenly, we obain ifferen performance evaluaion resuls. We compare raio propagaion moels performance variey of merics, Packes sen, hroughpu, roppe packes, Packe Delivery Raio, an packe rouing overhea. For movemen scenarios case, we suppose maximum spee an pause ime. Researchers mus be aware of significan ifference beween he real connecion opologies an he opologies obaine wih simple moels offere by MANET simulaion ools. For obaining quaniaive performance evaluaion resuls in he arge area, more accurae raio propagaion moels nee o be use. 8. References [1] hp://pcl.cs.ucla.eu/projecs/glomosim. [] hp://nile.wpi.eu/ns/. [3] J. Yoon, M. Liu, an B. Noble, Soun Mobiliy Moels, in Proc. ACM/IEEE In l Conf. Mobile Compuing Broaban an Ulra Wieban Communicaions (AusWireless 7) /7 $5. 7

9 an Neworking (MOBICOM 3), pp. 5-16, Sepember3. [] F. Bai, N. Saagopan, an A. Helmy, IMPORTANT: A framework o sysemaically analyze he Impac of Mobiliy on Performance of RouTing proocols for Ahoc NeTworks, in IEEE INFOCOM 3, San Francisco, March/April, 3. [5] F. Bai an A. Helmy, A Survey of Mobiliy Moeling an Analysis in Wireless Ahoc Neworks,chaper 1,Universiy of souhern California,USA. [6] W. Navii an T. Camp, Saionary Disribuions for he Ranom Waypoin Mobiliy Moel, IEEE Transacions on Mobile Compuing ITMMC-3 (), pp Ahoc Neworks, Book Chaper in submission o Kluwer Acaemic Publishers [7] Ingo Gruber, Oliver Knauf an Hui Li, Performance of A Hoc Rouing Proocols in Urban Environmens, In Proceeings of European Wireless (EW', Barcelona, Spain,February - 7,, Barcelona, Spain. [8]hp:// l-ns/noe196.hml [9] [1] Arne Schmiz, Marin Wenig, The Effec of he Raio Wave Propagaion Moel in Mobile A Hoc Neworks,, Torremolinos,Malga,Spain,MSWiM 6,Ocober -6,6 Broaban an Ulra Wieban Communicaions (AusWireless 7) /7 $5. 7