Geo-Encryption Protocol For Mobile Networks

Transcription

1 Geo-Ecrypio Proocol For Mobile Neworks Ala Al-Fuqaha, Omar Al-Ibrahim, Ammar Rayes * Deparme of Compuer Sciece Weser Michiga Uiversiy Kalamazoo, MI {alfuqaha, oalibrah Absrac We propose a geo-ecrypio proocol ha allow mobile odes o commuicae securely by resricig he decrypio of a message o a paricular locaio ad ime period. Our proocol will hadle he exchage of moveme parameers, so ha a seder is able o geo-ecryp messages o a movig decrypio zoe ha coais a mobile ode s esimaed locaio. We also prese mehods for esimaig he ode s moveme parameers o allow for geo-ecrypio. Fially, we evaluae our model by measurig he iduced overhead o he ework ad is performace i erms of decrypio raio. Idex Terms geo-ecrypio, locaio-based securiy, GPSbased ecrypio G I. INTRODUCTION PS-based ecrypio (or geo-ecrypio) is a iovaive echique ha uses GPS-echology o ecode locaio iformaio io he ecrypio keys o provide locaio based securiy. GPS-based ecrypio adds aoher layer of securiy o op of exisig ecrypio mehods by resricig he decrypio of a message o a paricular locaio ad ime period. Applyig his echique o a mobile evirome, wih a dyamically chagig opology, requires a proocol o hadle he disribuio of moveme iformaio so ha commuicaig hoss ca keep rack of each ohers locaios. Exisig GPS-based ecrypio echiques ([3] ad [4]) have limied suppor for mobile odes, herefore, we propose a mobiliy model for exisig geo-ecrypio echiques o allow mobile odes o exchage moveme parameers so ha a seder is able o geo-ecryp messages o a movig decrypio zoe ha coais a mobile ode s esimaed locaio. We also simulae his proocol o fid ou is performace ad scalabiliy i a muli-hop ework. The paper is orgaized as follows: Secio II iroduces he geo-ecrypio model. Secio III proposes a mobiliy model by iroducig some moveme parameers ad he proocol required o se up ad maiai a mobile geo-ecryped sessio. Secio IV preses he model equaios. Secio V shows how o esimae ad updae he mobiliy parameers ad how o achieve he model s objecives. Secios VI I prese he simulaio deails. Secio cocludes he paper. II. DENNING S MODEL OF GEO-ENCRYPTION Figure : Geo-ecrypio forces he receivig ode (B) o be iside a predefied locaio L o be able o decryp he message. Deig s model ([], [] ad [3]) for addig securiy o rasmissios uses locaio-based ecrypio o limi he area iside which he ieded recipie ca decryp messages. Fig. shows he geeral idea. A geo-lockig fucio is employed durig he ecrypio process o combie a ecrypio key wih he recipie s geographic locaio (L) o produce a geo-secured key for rasmissio alogside a ecryped message; he message ca oly be decryped if he geo-secured key ca be recovered, which ca oly be doe if he recoverig machie is physically posiioed a locaio L. The seder also rasmis parameers which defie he shape of he area where decrypio is permied (he decrypio zoe ), ad he ime period durig which decrypio ca be accomplished. Deig s model is effecive whe he seder of a message kows he recipie s locaio L ad he ime ha he recipie will be here, ad ca be applied especially effecively i siuaios where he recipie remais saioary i a wellkow locaio. Deig recogizes ha geo-ecrypio is also desirable i siuaios where he recipie is mobile, wihou a pre-plaed iierary. All he seder eeds is he recipie s curre/upcomig locaio ad he ime ha he recipie will be here, ad Deig oes ha a velociy parameer (he recipie s velociy) ca be added o he geolockig fucio. However, we have o see he deails of mobiliy suppor i Deig s geo-ecrypio model, ad herefore we propose a model o provide for mobiliy whe usig GPS-based ecrypio. * Cisco Sysems, Ic. rayes@cisco.com

2 III. THE MODEL We propose a mobiliy model based o he geo-ecrypio echique i [] i which boh seder ad receiver are mobile, wihou preplaed iieraries, ad ca securely deliver heir curre locaios o oe aoher wheever ecessary. I order o do his, each mobile ode ha will be receivig geoecryped messages eeds o iform poeial seder odes abou is ieded moveme i order for a seder ode o esimae he mobile ode s expeced locaio a ay poi i ime. This is doe by sedig iformaio regardig he mobile ode s moveme, which we call mobiliy parameers, o he seder via a sequece of message exchages. The ex wo subsecios explai hese parameers, he proocols ad he messages ha eed o be exchaged bewee a mobile receiver ode ad a saioary or a mobile seder ode. A. Mobiliy parameers Figure : Diagram illusraig he four mobiliy parameers: velociy, direcio, speed maeuverabiliy, ad breadh maeuverabiliy. Le A be a mobile saio (he movig age) ad le B be a saioary (base) saio i a ework usig Deig-syle geo-lockig for a added layer of securiy. I our model, he geo-lockig fucio akes shape, ime, velociy, direcio, ad wo maeuverabiliy parameers. The shape parameers defie a ellipse as he decrypio zoe. A ellipse is suiable for he shape of our decrypio zoe because i has a legh ad breadh, ad whe boh are equal, he ellipse becomes a circle ha provides uiform coverage i all direcios. (A recagle also has a legh ad breadh, bu whe boh are equal, i forms a square, wih o-uiform coverage.) The ime parameer specifies he period durig which decrypio is possible. Whe A is i moio, B will eed o calculae a ime parameer ha represes a fuure ime whe A will acually be i he decrypio zoe whe a geo-ecryped message arrives for decipherme a A. Figure shows he four mobiliy parameers ha a mobile ode uses o adverise is moveme iformaio. The velociy parameer, v, describes he recipie s speed. This is he average speed a which he recipie is expeced o ravel. Velociy ( v ) is deermied from observig he disace raveled durig a specified ime ui i is auomaically calculaed from rece moveme, o specified by a user. The direcio parameer,θ, describes he direcio i which he recipie is ravelig ad is measured as he posiive agle bewee he posiive x-axis ad he velociy vecor o a Caresia coordiae sysem. The firs maeuverabiliy parameer,α, is a idicaio of how frequely he movig recipie migh eed o chage speeds while ravelig o he ew desiaio (how much leeway, i erms of speed chages, ha should be buil io he size of he decrypio zoe). This speed-maeuverabiliy parameer iflueces he legh of he ellipse-shaped decrypio zoe. For ravel o a commercial airlier a a fixed speed, his facor would be small, while for ravel o a highway, where uexpeced delays migh crop up, i would probably be larger. The secod maeuverabiliy parameer, β, defies how much he movig recipie migh deviae from a sraigh lie while ravelig o a ew desiaio (how much wiggle room is deemed ecessary, based o a assessme of he errai beig followed). This parameer affecs he breadh of he ellipse-shaped decrypio zoe, ad we refer o i as he breadh-maeuverabiliy parameer. O a sraigh highway, or i a airplae, he breadh maeuverabiliy facor will probably be small. Bu i mouaious errai, or while ravelig o horseback or o a widig river, he breadh maeuverabiliy facor will be larger. I may make more sese i he real world for a mobile saio o defie large maeuverabiliy values for a sigle, log, widig ad upredicable move, raher ha frequely compuig a series of small direcioal ad speed chages while followig a crooked rail. A mobile saio mus deermie is ow velociy ad maeuverabiliy parameers, based o is rece moveme ad a evaluaio of he errai i quesio, ad commuicae hem o oher saios for use i geo-lockig messages back o he movig saio. The decrypio zoe oly eeds o be large eough for A o exrac he geo-secured decrypio key wihi he specified ime period, o for A o decryp he accompayig message. I he desig of our model, we make several securiy ad reliabiliy assumpios: ha rouig is secure; ha auheicaio is assured by proocols oher ha geoecrypio; ha he GPS hardware works flawlessly, is amper-proof ad uspoofable; ha rasmissios use some sor of spread specrum mehod i order o couer riagulaio aemps by rivals searchig for our saios; ad ha rival elecroic couermeasures do o jam our mobile saios rasmiers (presumably i lapop compuers). B. Moveme Updaes I he descripios below, he oaio E ({ C}, L) meas a message wih coes C ha are geo-ecryped o he geographical area L. Also, ode A s curre locaio is labeled LA while ode B s curre locaio is labeled LB i he descripios:

3 3 Oe of he mos impora corol messages i our mobiliy model is he moveme updae message. I order for he seder o keep rack of he locaio ad moveme of he mobile receiver ode, he receiver iself mus accumulae iformaio abou is ow ogoig moveme ad adverise i o he seder whe ecessary. Usig he example where A is he mobile ode ad B is he saioary ode, each ime A s chage i velociy ad direcio is greaer ha a cerai hreshold (discussed i Secio V), A seds is curre moveme parameers (v, θ, α, ad β ), LA, ad he curre ime o B i a moveme updae message: E({ v, θ, α, β, LA, }, LB) Based o his iformaio, ode B ca predic ode A s fuure locaio uil A seds aoher moveme updae message o B. Because B will be esimaig A s locaio based o his moveme iformaio, here will be errors bewee he esimaed ad acual locaio of A. Therefore, LA ad are se i he moveme updae message from A o B so ha B ca deermie A s fuure locaio kowig ha A was a locaio LA a ime. Similarly, B has o kow A s iiial locaio a he sar of he mobile geo-ecrypio sessio (i.e., a ime = ). IV. THE MODEL EQUATIONS Suppose he mobile ode A sars a ime a a locaio whose logiude ad laiude values are LA (,Y ), which are assumed o be iiially kow o ode B. This could be achieved, for example, by usig he geo-ecrypio model i [], or by ay oher meas. Periodically, ode A collecs GPS locaio saellie readigs LA (, Y ) a ime wih =,, 3,K such ha i = + id where d is a fixed ime ui ierval whose value is arbirary bu kow. y-axis decryp regio C (C, C Y ) Lie of moveme θ values i Eq.. The lie of moveme makes a agle θ wih he posiive direcio of he laiude. As ime progresses, he decrypio regio for ode A moves alog ha lie a a cosa velociy V (Figure 3). The moveme of ode A iself is arbirary i ay direcio ad ay velociy bu oherwise resriced o he decrypio regio a all imes. The parameers of he ceer of he decrypio regio cosaly chage wih ime bu o he shape. The parameers of he shape of he regio remai fixed ad are oly allowed o chage whe a predeermied fixed umber of ime uis has passed. The ceer (C,CY ) of he decrypio regio a ime is give by C CY = = Y + ( + ( ) V cosθ ) V siθ Thus, a ay ime ode B eeds oly he iiial parameers ad he ime value o locae he ceer of he decrypio regio. O he oher had, he shape of he regio is deermied by he maeuverabiliy parameers α ad β as well as he moveme direcio θ. If we assume he regio has he bivariae ormal disribuio wih ceer (C,CY ) ad if we adop he 3-sigma rule [3] he he equaios relaig he shape parameers of he regio wih he maeuverabiliy parameers are give by x y = C = CY + ( α C ( α C )cosθ + )siθ + ( β CY )siθ ( β CY )cosθ Hece, a ime, he decrypio regio is defied by: C Y CY R ( ( ) ( ), Y ) = + Y (3) ( C ) ( Y CY ) ρ c Y where ρ = cosθ ad c is a cosa deermied from values of α ad β. () () LA (,Y ) x-axis V. PARAMETER ESTIMATION AND UPDATE A. Esimaig he Mobiliy Parameers Figure 3: Moveme of he decrypio regio depeds o he iiial values V, θ ad LA. To defie he decrypio regio for he mobile ode A, i is assumed ha some iiial values are available for he mobiliy parameers α, β V ad θ a ime. Give hese iiial values for he mobiliy parameers ad LA (,Y ) as he iiial values for he ceer of he ellipse, he decrypio regio for ode A is defied iiially by subsiuig hese Alhough he GPS readigs were o eeded o locae ad deermie he decrypio regio i he above equaios, everheless hey mus be foud o esimae ad updae he mobiliy parameers. The GPS readigs LA (, Y ) a ime =,, 3,K are foud ad cosaly used o calculae ad updae he velociy V, ad he agle θ, of he decrypio regio. V ad θ are he velociy ad agle for moveme alog he lie of moveme. These values will be used o updae he

4 4 iiial values V ad θ each ime he regio chages is direcio or velociy. I such a eve, we also updae he iiial values α, β usig he las GPS readigs (, Y ) usig he Gauss-Markov model give by Z = γ Z + ( γ ) µ + γ ε =,, 3,L where γ is a uig parameer represeig differe levels of radomess, µ is he asympoic mea of Z, ad ε are ucorrelaed saioary radom Gaussia process wih zero mea ad ukow sadard deviaio. Noice ha whe γ equals o he model is ideical o radom walk model, ad whe γ equals o zero he model is he cosa velociy fluid flow model. I applicaios oe may selec he value of γ arbirary or could be esimaed from he daa. The model is described more fully i [9] ad []. Assumig ha boh he velociy ad he agle of he mobile ode follow he model, ad akig he asympoic meas of velociy ad agle equal, respecively, o heir iiial values V ad θ, he esimaes of velociy ad he agle from he k h period are obaied as Vˆ k ˆ θ k where Vˆ ˆ θ = γ Vˆ = γ ˆ θ k k = ' = = arca + ( γ ) V + ( γ ) θ d Y k =,, 3, (4) Y Y + d Y = Noice ha he k h period sars a ime ( k ) ad eds a ime k. We obai he formulas for esimaig α ad β by iverig he formulas i (equaios ) o ge ˆ α = C ˆ β = CY where ˆ x = + ( ˆ ˆ θ ˆ ˆ x C )cos ( y CY )siθ () + ( ˆ C )si ˆ θ + ( ˆ CY )cos ˆ θ = x ( ), = y = (5) ˆ y = ( Y Y ), Y = Y (7) = B. Updaig he Mobiliy Parameers = Each ime he mobiliy parameers are esimaed, he mobile ode mus decide wheher or o o replace he old values of he parameers wih he ew values ad wheher or o o adverise hem. Typically, he old values are replaced wih he ew values ad he updaes are adverised oly whe hey are sigifica i.e. whe he differece bewee he old ad he ew values of a parameer exceeds some predeermied hreshold se by he mobile ode. Oherwise, he old values are kep ad ohig adverised. I addiio o he mobiliy parameers, he iiial locaio parameers (, Y ) of he mobile ode mus also be updaed oce eiher or boh of V or θ are foud sigifica. This is because he geo-ecrypio process depeds o deermiig he ceer (C, CY ) ad, as oed from Eq., he recipie eeds (, Y ) o esimae he ceer. If a ime * a sigifica V or θ is deeced he o oly he four mobiliy parameers are adverised bu also he ew value for which is esimaed * by ˆ =. Give he values of V ˆ, ˆ,ad θ ˆ he recipie will use Equaio o esimae he updaed iiial locaio ( ˆ, Y ˆ ). The smaller he hreshold values for he parameers he more ofe he parameers are updaed ad adverised. So, choosig opimal hreshold values is he key i opimizig he decrypio zoe o achieve a balace bewee he probabiliy he mobile ode falls iside he decrypio zoe agais he frequecy of moveme updaes ad adverisemes. C. Opimizig he Decrypio Zoe The proposed model aemps o achieve he followig goals: ) capure he locaios of oher mobile saios wihi a give decrypio regio wih high probabiliy, ) keep saios locaios secre from rivals, ad 3) permi he saios o be as mobile ad maeuverable as possible. These hree goals are i coflic ad hus we eed o make cerai compromises i he desig. For he prese model, more mobiliy is gaied a he expese of posiio secrecy ad area of he decrypio regio. I order o miimize he frequecy wih which odes adverise heir movemes ad a he same ime opimize he size of he decrypio zoe, we propose he followig: A mobile ode may fall io oe of hree regios as show i Figure 4. Regio represes he adveriseme-free zoe, meaig ha a mobile ode will o adverise moveme updaes whe hey fall wihi his regio alhough i cosaly updaes hem. The size of his zoe is deermied by he -sigma rule similar o he 3-sigma rule used o defie he decrypio zoe R(x,y) i equaio 3. Hece, i is obaied

5 5 by replacig by 4 i equaios. Based o he bivariae ormal disribuio assumpio, he probabiliy of fallig i his zoe is abou 95% provided ha he ceer of he regio is deermied accordig o equaios. Regios ad ogeher make up he decrypio zoe. I regio, he mobile ode is abou o leave he decryp zoe ad eers he odecryp zoe of regio 3. I his zoe, he ode eeds o rasmi is mobile parameers updaes if hey are sigifica. They are declared sigifica whe he updaes i he mobiliy parameers exceed he parameers hresholds. Noe ha by resricig adveriseme of updaes o regio we effecively reduce he frequecy of adverisemes. Regio Regio Regio 3 Figure 4: Regios,, ad 3 for opimizig he decrypio zoe. VI. PROTOCOL OVERVIEW Our proocol builds o op of exisig wireless muli-hop rouig proocols, hus we will o address he rouig issues of mobile muli-hop eworks. We evaluaed a simplified versio of he Geo-ecrypio proocol by simulaig a modified DSR proocol [] usig s- uder seleced scearios (secio VII). Our proocol will hadle he commuicaio of moveme iformaio bewee mobile odes ad he updaig of his iformaio wheever odes move uexpecedly. I aims o allow mobile odes o commuicae heir moveme iformaio accuraely while a he same ime reducig he overhead o he ework. Whe a mobile ode, say MB, wishes o commuicae wih aoher ode MB, i broadcass a message o discover a roue o MB. Whe such a roue exiss, ode MB will receive a roue reply specifyig he sequece of hops o reach ode MB. I our model, i will also eed o kow he posiio of he desiaio ode. This is mea o simulae a Geo-ecrypio model i a mobile ework. Node MB will eed o keep a able of he posiios of he odes i ieds o sed daa o. The able will be kep curre wih posiio updae messages from hese odes. Each ode ca obai ad sed is coordiaes o correspode odes usig a posiio updae message. The desiaio ode MB, upo receip, will map is rue posiio o he ieded locaio of he message usig he shape parameers (discussed i []). If he resul of he mappig maches he locaio se by MB, he decrypio will be cosidered successful else he message will be dropped. We will simplify his model o es if he ode is wihi a give square ceered o he received coordiaes. VII. SIMULATION MODEL We decided o make he posiio updae aspec of he proocol be reacive similarly o DSR rouig. To simulae geoecrypio usig DSR we made he followig assumpio:. Auheicaio is prese ad free: he esablishme of commuicaio bewee wo odes is assumed auheicaed ad ayime ecrypio fails he auheicaed relaioship is assumed reesablished auomaically by he posiio updae message.. Decrypio of he message is assumed par of aoher layer. Our simulaios are oly cocered wih posiio updaes; if he message coais he righ posiio parameers x ad y o a cerai olerace +/- Tolerace, he message is cosidered decryped. 3. Flow sae i DSR was disabled o allow every message o carry is source roue. 4. DSR corol messages like roue discovery messages are o subjec o ecrypio ad are o esed for i. Whe a message is received by is desiaio ode, if oe of is proocol flags are se (RoueReques, RoueReply, PosiioUpdae), he ode ges is ow real ad Y values (GPS) ad compares hem o he x ad y values i he packe header. If x [ real ± Tolerace] ad y [ Yreal ± Tolerace] he message is cosidered decryped oherwise he message is cosidered o decryped ad he packe is passed o GeoHadler, a fucio ha respods o decrypio failures. GeoHadler is also called if he message is decryped bu is coordiaes are o wihi half he olerace (x [ real ± Tolerace/] ad y [ Yreal ± Tolerace/]) his is mea o preemp fuure decrypio failures. GeoHadler cosrucs a message, pus he real ad Y values i he x ad y fields, ses he Posiio Updae Flag ad seds he message o he reverse roue ha he received message arrived o. Whe a message is received ad i has is Posiio Updae Flag se, i is o esed for decrypio bu is used o updae he able eries correspodig o he source ode ha se he message. This makes our Geo ecrypio proocol oally o demad ad should oly have overhead whe decrypio failures occur. To evaluae he proocol we added lies o he race file idicaig he followig eves: A message was successfully decryped. A message failed decrypio. These would give a merics of he proocol performace. A Posiio Updae Message was se. A posiio Updae Message was received. These would allow us o gauge he overhead of he proocol o regular DSR.

6 he ework icludig DSR messages as his is a modificaio o DSR. VIII. MOBILITY FILE AND SIMULATION RUNS: We used a subse of bus roues from he Seale area as our mobiliy file []. Firs we ploed he daa for oe of he files a our disposal ad, afer proper ui coversio, deermied a 5x5 m area ha preseed dese raffic. We he seleced he movemes of buses wihi ha area durig a 5 miue period (simulaio arge ime). Furhermore we excluded from ha selecio he buses ha came o close o he edge of our area (5 m) o avoid sudde ode disappearaces durig he simulaio. Ad fially, from he remaiig daa, we seleced he 5 buses wih he mos umber of updaes i he give period (see Figure 5). From ha we creaed a iiial posiio file ad movemes file wih TCL commads o iclude i our simulaio file. We also icluded differe pause imes bewee moveme updaes o creae several moveme files wih decreased mobiliy; he pause imes are, 5, 5, 75,,, 4, 5 ad 9 secods. Movemes updaes ha exceeded 9 secods were purged from he files, hus i effec he 9 secod pause ime correspods o zero mobiliy whe all he odes say a heir origial posiios. I. SIMULATION RESULTS Wih he olerace fixed o m ad ruig for, ad 3 seders ad receivers respecively, we oe ha he geeral red is, as expeced, ha he decrypio raio falls wih a icrease i mobiliy (i.e. he raio icreases wih bigger pause imes). See Figure. This is due o he fac ha higher mobiliy meas ha odes move more ofe away from heir perceived posiios a he sedig odes. As a resul more messages are o decryped. Decrypio Raio (%) Decrypio Raio vs Mobiliy Seders Seders 3 Seders Pause Times Figure : Decrypio raio vs. mobiliy for,, ad 3 CBR sources Figure 5: Fial simulaio daa We used s-.8 for our simulaios, ruig DSR wih flow sae disabled ad he modificaios described above. For each oe of he mobiliy files we did hree rus wih seder receivers, seder receivers ad 3 seder 3 receivers. Every seder se daa a a Cosa Bi Rae CBR of 4 packes per secods wih a packe size of 5 byes. For each ru we recorded he decrypio raio ad he proocol overhead. We measured our decrypio raio as he raio of successfully decryped messages amogs hose ha were received. Thus he raio does o reflec he delivery raio of DSR. We measure he proocol overhead for posiio updaes as he raio of posiio updae messages o he oal umber of daa messages (CBR), decryped or o ha were received. A beer measure would have bee he raio of geeraed posiio updae messages o he oal umber of messages o Also from Figure we ca see ha a icrease i CBR seders ad receivers resuls i a decrease i decrypio. Tha ca be explaied by he icreased delay i message delivery due icrease ework cogesio. If a message is buffered ofe alog he way, i allows ime o he desiaio ode o move furher away from is curre posiio ad hus icrease he chage of a decrypio failure. This is cofirmed by he fac ha a low mobiliy, he gap bewee he raios of he hree cases is reduced. A oiceable feaure of he graph is also he fas dip for pause imes bewee ad secods. We have ye o explai i. O he oher had, overhead decreases wih icreased pause imes. See Figure 7. This behavior is ypical of a proocol ha is reacive o moveme. If here is o moveme he here is o eed for moveme updaes. The overhead does o quie go o zero a zero mobiliy because i our simulaio, odes do o iiially kow he posiios of heir desiaios bu have o lear hem by sedig he defaul coordiaes (-,-) ad causig a posiio updae message.

7 7 4 Overhead Seders Seders 3 Seders This red ca be furher see i Figure 9 where we show he resuls of icreasig he olerace hrough 5,,, 5, 7 ad meers for he maximum mobiliy case ad wih CBR sources. As expeced, decrypio icreases wih more slack oleraces, ha raio would o reach % hough due o he eed o esablish he kowledge of he iiial posiios. Packe Icrease (%) 8 Decrypio Raio vs Tolerace Pause Times Figure 7: Proocol overhead vs. mobiliy for,, ad 3 CBR sources Decrypio Raio (%) 95 9 We ca also see ha i geeral, overhead icreases wih a icrease i CBR seders ad receivers. This mus follow from he drop i decrypio raio discussed previously as every failed decrypio or ear failed (o wihi half olerace) geeraes a updae message ad hus icreases he overhead of he proocol. By he same oke, he iiial icrease i overhead for small pause imes compared o zero is a resul of he equivale behavior of he decrypio raio i Figure. We do o cosider he drop of he overhead of he CBR below ha of 3 CBR, o be sigifica. We believe i o be a arifac ha would disappear if we ra more simulaios. For our secod series of simulaios, we used our base case of seders ad receivers o es he effec of chagig he olerace. Firs we compared he decrypio raio of our referece olerace of o ha of a more resricive olerace of 3 meers. The resuls over he usual pause imes goig from o 9 secods ca be see i Figure 8. Decrypio Raio vs Mobiliy Tolerace (m) Figure 9: Decrypio raio vs. olerace for CBR sources I a similar behavior, he overhead drops rapidly wih icreased olerace dow o he miimum required for he iiial posiio messages ad remais cosa afer a cerai poi. Tha behavior is mosly depede o he rage of moio of he buses durig he simulaio. If i does o exceed half of he give olerace he o moveme updaes are required for he life of he simulaio. 4 Overhead Decrypio Raio (%) 95 9 Packe Icrease (%) Tolerace (m) Tolerace= Tolerace= Pause Times Figure 8: Decrypio raio vs. mobiliy for olerace of 3 ad The value of he olerace ses he size of he decrypio square. Thus a more resricive olerace will resul o a smaller square ad lower decrypio raios as see i he figure. Figure : Proocol overhead vs. olerace The expoeial shape ca be explaied by he possibiliy ha a lo of buses have a limied rage of moio ad he slighes icrease i olerace is sufficie o cover heir eire or mos of heir pah ad make updaes uecessary while a few buses have larger rages of moio ad will be affeced oly by a large icrease i he olerace.

8 8. CONCLUSION Usig geo-ecrypio adds a sigifica layer of securiy o ework rasmissios. Mobile eworks should be able o ake advaage of his echique. We believe ha our model serves a source for furher work o locaio-based securiy for mobile eworks. I our proposal, mobile odes which sray from heir adverised locaios ca reesablish a secure saus wihi he ework a he same ime do o sacrifice he secrecy of heir locaios. We evaluaed a simplified versio of he Geo-ecrypio proocol by measurig he iduced overhead o he ework ad is decrypio performace by simulaig a modified DSR proocol usig s- uder seleced scearios. Our resuls proved some of our expecaios abou decrypio declie wih icrease i mobiliy ad a equivale icrease i overhead. We also saw some resuls we did o predic such as he decrease i decrypio raio wih a icrease of ework raffic due o icreased message queuig delay. Fially we ca poi cerai seps o improve he performace of he proocol as ex posiio predicio a he seder or he receiver based o hisory of moveme, or he sedig of moveme parameers such as speed ad direcio by he receiver o he seder. Ad for improved securiy we ca exed our proocol o a muli-hop ecrypio scheme ha would require he seder o have kowledge of he posiio of all he forwardig odes. ACKNOWLEDGEMENT We hak Dorohy Deig for providig us wih a paper deailig her work o geo-ecrypio. We also hak Joe Baird ad Weg Liog Low for heir assisace i he preparaio of a early versio of his work. We are idebed o Aasasios Giaoulis ad George Abi Nader for heir help wih he s- simulaios. REFERENCES [] D.E. Deig, Geo-ecrypio, GeoCodex LLC ad Naval Posgraduae School, Jue 4. [] D. Deig ad P. MacDora, Locaio-Based Auheicaio: Groudig Cyberspace for Beer Securiy, i Compuer Fraud ad Securiy, p.elsevier Sciece Ld, 99. [3] L. Sco ad D. Deig, Geo-ecrypio: Usig GPS o Ehace Daa Securiy, GPS World, April 3. [4] Trimble Iformaio Services, Powerig he Trasformaio of Locaio Daa io Locaio Iformaio, [Olie docume]. [5] M. Kim, B. Noble. Mobile Nework Esimaio. I ACM Mobicom,. [] E. Gabber ad A. Wool, O Locaio-Resriced Services, IEEE Nework, Nov. /Dec., 999. [7] J. S. Warer ad R. G. Johso, GPS Spoofig Couermeasures, [Olie docume] Dec. 3. [8] Aheros Commuicaios Ic., Buildig a Secure Wireless Nework: How Aheros Defies Wireless Nework Securiy Today ad i he Fuure, [Olie docume] Apr. 4. [9] B. Liag ad Z. Haas, Predicive Disace-Based Mobiliy Maageme for PCS Neworks. I Proceedigs of he Joi Coferece of he IEEE Compuer ad Commuicaios Socieies (INFOCOM), March 999. [] V. Toley. Load Reducio i Ad Hoc Neworks usig Mobile Servers. Maser s hesis, Colorado School of Mies, 999. [] David B. Johso ad David A. Malz. Dyamic Source Rouig i Ad hoc Wireless Neworks. I Mobile Compuig, edied by Tomasz Imieliski ad Hak Korh, chaper 5, pages Kluwer Academic Publishers, 99. [] Jorjea G. Jecheva, Yih-Chu Hu, Saashil PalChaudhuri, Ami Kumar Saha, ad David B. Johso. Desig ad Evaluaio of a Meropolia Area Muliier Wireless Ad Hoc Nework Archiecure, Proceedigs of he 5h IEEE Workshop o Mobile Compuig Sysems & Applicaios (WMCSA 3), pp. 3-43, IEEE, Moerey, CA, Ocober 3. [3] Weser Elecric. Saisical Qualiy Corol Hadbook. Weser Elecric Corporaio, Idiaapolis, Ic.,95.