Mobile Agent as an Approach to Improve QoS in Vehicular Ad Hoc Network

IJCA Special Issue on “ Mobile Ad-ho c Netw orks ” MANET s, 20 10 67 Mobile Agent as an A pproach to Improve QoS in Vehicular Ad Hoc Net w ork Rakesh Kum ar, Dr. Mayank Dave Ass istant Pro f es sor, M. M. Un iversity, Mul lana Ass ociate Profess or, N.I.T. Ku ruks hetra Harya na, Ind i a, Haryana, In dia ABSTRACT Vehicul ar traff i c is a foremo st pr o blem in moder n cities. Hug e amo unt of time and resource s are wasted whi le trave ling due to traff ic c ong est ion. Wi t h the introduc t ion of sophis ticated t raf fic manag ement sy stems, such as t hose inco r po rating dy namic traff ic assignm ents , mo re string ent demand s are being placed upon the availab l e re al time traffic da t a. I n thi s pap er we have p r o posed mo bil e ag ent as a mec hanism to handle the traff i c problem on road. Mobile sof t ware agents can b e used to prov ide the better Qo S (Quali ty of S ervic e) in v ehicular ad ho c netwo r k to improv e the safe ty applicatio n and d r ive r co mfort. Keywords : Qo S, Mob il e Age nt, VAN ET 1. INTRODU CTION A Vehicular Ad hoc Network (VA NET) i s a form of wirele ss ad hoc network to prov ide co mmunications amo ng vehic l es and ne arby roadside equipm ent s . I t is emerging as a new techno logy to i ntegrate the cap abil ities of ne w genera t ion wireless netwo rking t o vehic l es. The majo r purpose of VA NET is t o pro vide (1) ubiqu i tous co nnectivity whil e on the road to mo bil e us ers, who are otherwi se co nnected to t he o utside wo r ld throug h other network s at ho me or at t he wo rk pl ace , and (2) efficie nt vehic l e- to -vehic l e com munications tha t enable the I nt elligent T ranspor t atio n Sy stems (I TS). ITS inclu des a varie ty of applica tions suc h as coo perat ive traff i c mo ni toring, co nt rol of traffic flow s, blind crossing ( a cro ss ing witho ut lig ht contro l), prev e ntio n of collisio ns, nearby info r mation serv ices , and r eal-time de tour routes com put atio n. I n this paper nodes a nd vehicles are u s ed intercha ngeably . 2. NETWOR K ARCHITECTU RES AND CHARA CTERIS TICS Wire less ad hoc networks genera l ly do not rely on fixed infrast ructu re for com munication and di ssem ination o f info rmation . VA NETs follow the s ame pr inci ple and apply it to the h ighly d y namic enviro nment of surfac e transpor t a tion. A s shown in Fig. 1, the arc hitec ture of VA NETs mainly falls with in three catego ries: pur e cel l ula r /WL AN , pure ad hoc , and hy brid. VA NETs may use fixed cellula r gateway s and WL A N/WiMax access po i n ts at t raf fic intersec t ions to co nnect to the I nt ernet , gather tr aff ic info r mat ion, or fo r routing pu rposes . The n et wo r k arc hitectu r e unde r this scenar io is a pure ce llular or WLA N st ructure as s ho wn i n Fig. 1(a). VA NETs can combine both cellular netwo rk and WL A N t o fo rm the netwo rks so tha t a WL AN is used where an a c cess point i s ava ilable and a 3G co n nection otherwi se. (a ) Cellular /WLAN (b) Ad Hoc (c) Hy brid Figure 1: N etwork arc hitecture s for VANETs Stationa ry or fixed gateway s around the sides of roads co uld pr o vide co nne c tivity t o mobile nodes (vehicles) , but are even t ual l y unfeasible co ns idering the infrastruc t ure co st s i nvo lved. I n such a scenar i o, all veh i cles and ro ad - side wireless devic es can fo rm a pure mo bile ad hoc network (F i g . 1(b )) to perf orm vehicl e to v ehicle com munications and achi eve certa i n go al s, such a s bl i nd cro ss ing. Hy br id arc hitec ture ( F ig. 1(c)) of com bining i nfra struc ture network s and ad hoc networks tog ether has also been a possib le solut ion fo r VA NETs. Nambo odiri et al . [ 9] propo sed such a hy brid archi tecture , w hi ch uses s o me vehic l es w ith both WL AN and cellular c apabi l it ies a s the gateway s and mobile netwo r k route rs so t hat vehicles with only WL AN capabi lity c an comm unicate with them throug h multi-hop links to remain connec te d t o the worl d. The hy bri d architec ture can prov ide better cov erage, b ut also causes new problem s, s uch as the seaml ess transi tio n of the co mmunic at ion amo ng differe nt wire l ess sy ste ms. 3. CHARA CTERIS TICS OF VANET I n additio n t o the similar i ties to ad ho c net wo rks, such as short rad io transmi ssio n range, self-org ani zation an d self- manag ement, a nd l ow bandwi dth, VA NETs can be disting ui shed from other kind s of ad hoc networks as fo l low s[15]: Highly dynam i c topo l o gy: Due to high speed o f mov ement between v ehic l es, the to pology of VA NETs is alway s c hanging . Freque ntly disc o nne cted ne twork : Due t o the same reaso n, the co nnectivity of t he V A NETs could als o be chang ed frequently . Especia l ly when the vehicle dens i ty is low, it has highe r pro babil ity that the network is disco nnected . IJCA Special Issue on “ Mob ile Ad-ho c Netw orks ” MANET s , 20 10 68 Mobil ity mo deling and predi catio n: Du e to high l y m obile node mo vem ent and dy namic topo logy , mo bi lit y model and predic ation pl ay an i mpo r tant role i n netwo rk proto col desig n fo r VA NETs. Geograp hical type o f comm unic atio n: Co mpared to other network s that use unicas t or mult icast wher e t he com munication end points are defined by ID or gro up I D, the VA NETs often ha v e a new ty pe of comm unication t h at addres ses g eog raphical areas wh ere packe t s need t o be fo rwarde d Variou s com munic ations env i ronm ents: VA NETs are usually operated in two t y pi cal c ommunications enviro nment s. I n hi ghw a y traffic sc enario s, the enviro nment is r elativ el y simpl e and straigh tforward (e.g., co nst rained one-dimen si o nal m ov ement), wh ile i n c ity co ndit ions i t bec omes muc h mo re co mplex. Suffic ient en er gy and st o rage: A co mmon characte r ist ic of nodes in VAN ETs i s that nodes have ample energ y a nd com put ing power (i nc lud i ng both storage and processi ng), since nodes are cars ins t ead o f small h a ndh eld dev ices. Hard delay constraints: I n so me VANE Ts applicatio ns, the ne t wo r k does not requ ire high dat a r at es but has hard delay constrain ts. Interac tion with on-bo ard sensors : I t is assume d tha t the nodes a re equipp ed w it h on-bo ar d sensors to prov ide info rmation t ha t can be used to fo rm comm unication li nks and fo r rout i ng pu r poses . 4. MOBILE A GENT The term “mo bile agent” was intro duced by Telescr ipt, which suppor t ed mo bility at the prog ramming l a nguag e level. A mobile agent is a pr o gram , whic h represen t s a user in a com put er netwo rk, and is cap able of mig r ating autono mously from node to no de, to perfo rm som e com put atio n on behalf o f the use r . Mobile ag ents are define d as o bjec ts t hat have behav i o ur, stat e, and loc at ion [4] . I t s tasks are determ i ned by the ag ent applicat i on, and can range fro m online shopp i ng t o real - time d evic e co ntr ol to distr ibuted sc ient ific c om put ing Mobile a g ents are co ns idered as a prog ram, tho ugh they have so me special prope r tie s that disting ui sh them fro m t he standa rd pro grams: mandatory and ortho go nal (optiona l) proper ties [5 -7]. Th e manda tory propert ies ar e as fo l low s: Auto nomy: agents o perate wi t ho ut the d irec t in terv entio n of humans or o t hers, and hav e some kind o f contro l ov er their ac tio ns and intern al sta t e. Decisio n mak i ng: reactiv e o r proac t ive deci sion making . Reac tive deci s ion making i s implem ente d in some form of direc t mapping from sensors input (sen sing the enviro nment ) to actions using som e r ule s. Proac tive decis ion m aking m ay use Belief Desi re I nt entio n (BDI ) archi tecture [6] . The bel iefs of an a g ent are its mo del of the dom ai n, its desires prov i de som e s ort of ordering between states, and i t s inte ntions ar e thing s it has dec ided to do . Tempo ral cont i nu i ty: ag ents are co nt inuo usl y r unn ing proc es ses (either running activ e in the f o regro und o r sleep ing/pa ssiv e in th e bac kgro und); Goal orie nted: an agent i s capa ble o f handling a t ask to meet i ts des i red g o al; The or t hog onal prope rties are as fo ll ow s : Mobil ity: agent s are c apable o f roam ing ar ound in a n electro ni c ne t wo r k; Co mmunicative: ag ents interac t with ot her agent s and (possibly ) hum ans via some kind o f agent c omm uni cat i on languag e. Mutual : an agent s ho uld be c a pa ble o f com puting the desired task s of the use rs/ proc ess by cooperat ing with other agents , som etime s it may reject execut i o n of certain t ask s, because (f o r i nsta nce) they wo ul d put an o b j ect i onabl e hig h load on the network reso urce s or b ecause it wo uld c ause damag e to other u s ers; Learn i ng: a g ents can learn the env ironment facto rs, u ser prefe rences , et c. and dev elop certa in degree of reasoning to take i ntell i g ent decisions /a c tions tha t i mproves t he eff iciency of the sy stem. 4.1 Arch itectu ral ov erview & its co mpo nents Mobile agen t pro vides a new design model fo r applic atio ns as co mpared to the tradit ional cl ient serv er mode l . First and fo re mo st , t he mobi le agen t blow s ap art the very no t ion o f client and serv er. With mobil e ag ents, the flow of co ntrol actually moves acro s s the network, ins tead of usi ng the reques t /respo nse architect ur e o f client -s erver .I n effec t, every n ode is a server in the agent netwo r k, and t he agent mov es t o t he lo cation where it may find the s erv i c es i t needs to run at eac h point i n its execut ion [8] . T he vario us co mpo nent s o f t he a r chi t ectu r e ar e: - Ag ent Manag er : The A gent Manag er prov i des the com munication infras t ruct ure u si ng the TCP/I P stac k fo r agent transmis s ion . I t abstracts the net wo r k int er fac e in order t hat agen t prog r amm ers need not know any network s p ecif ics nor nee d to prog ra m an y netwo r k interfac es. Sec ur ity Manager: The Security Manager is respons ible fo r ident ify ing us ers, authent i cating their agents, p rotec ting server r eso urces an d ens uring t he secur ity and integ r ity of agents . Pers istenc e Manag er: The Per sistenc e Manag e r is com pl etely transparen t and m ainta ins the state of agents i n transit around the netwo rk. As a side benef i t, it a llo ws for the chec kpo int and restart of agents in the event of s y stem failure . Ev ent M anag er: The Event Manager handles the registr atio n, posting and notif ication of events to a nd from agents . The Ev ent Manage r can pass event notifica t ion to agents on any node i n t he netwo rk t hus suppo rting agent co ll aborat ion. IJCA Special Issue on “ Mob ile Ad-ho c Netw orks ” MANET s , 20 10 69 Fig 1: Architec ture of a mo bi le agent Queue Manag e: The Queue Manag er is responsi ble for the schedul ing and po ssibly retry ing the mov ement of a g ents between systems whic h include main t enanc e of a g ent a nd persiste nce o f agent sta t e. Direc tory Manager: The Dir ecto r y manager prov ides naming servic e in the n etwork . The Direc tory M anager may co nsul t a loc al name servic e or may be set up to pass reques t s to o t her , ex i sting na me serv ers. Servic e man ager: The Serv i ce manag er pro vides the interfac e from agents to the services ava ilable at the v ario us mac hi nes in the ne two rk. I t comprises a set of pro g ramming extens ions t o prov ide access t he native A PI ‟s a nd interfac ing. 4.2 Life Cy cle of a Mob ile Agent & its l ife state log stru c tu r e The mo bile agen t paradig m is based on t h e migrat i ng wo r kflo w sy s tem mo del [9][10] . Wo rkflow o riented life cyc le model co nsists o f five states, (creat i ng, running , deleting , suspend i ng, r esum ing ) and a number of tr an sitio n states. The wo rkflow oriented life cy cle model is sho wn i n the fig . 2 To accomplish its task , the mobi l e agen t can transpo r t itse lf to another serv er in search of t he needed res o urce / serv ice, spawn new ag e n ts, or interac t w i th o ther stationa ry age nts. Upo n co mpletion, t he mo bi le ag ent de liv e rs the r esults to the send ing clien t or to ano t her serve r . 1. I n t he creating st ate , the agen t is crea t ed but not activate d y et. 2. I n the runn ing stat e, the agent is running, perfo rming ac tio ns and solv e it p urpose . 3. I n the del eting sta te, the agent is te r mi nated . 4. I n the suspend ing state , the agen t cannot run and still w ithin the agen t serv er. 5. I n the resumi ng st ate , the agent is travel ling between two serve r instanc es . Fig 2: Lif e cy cle of a mob i le agent The life cyc l e of m obile a g ent begins at the mo ment w hen it is creat e d . When PA is m igra ting from one host to anothe r host i n ord er to achieving its go al s; an d the PA returns its se rver on wh ich it was c rea t ed. Two or mo re than t wo states, in l ife cy cle of PA , may be occ ur red at the differen t time o r pl ace . The mo bil e agen t life state lo g structur e can be defin ed in f our tuple: 4.3 Im plementatio n strategies Ag ent m obility and nam ing are t he tw o major issu es i n implem enting the mobi l e agen ts i n VA NET. A n agent should be unique ly named, so that its owner can com municate with or contro l it wh ile it travel s on its itinera ry [11]. W e have i dent i fie d fo ur i mp l eme nt at i on strateg i es: Sequent i al CS This i s bas ed on the tradit ional cli ent-serve r paradig m. The client make s a reques t to the first serv er and after proc es sing the reply , make s a r eque st t o the seco nd server and s o o n, ti l l t he list of server s t o be v i sit ed is exhaust ed. T his st r ateg y is illus trated in fig ur e 3 (a) . Sequent i al MA I n thi s ca s e a s ingl e M A mov es from i ts so ur ce of orig in (client) t o the fi rst site (s e rver) in its itinera ry . I t then mo ves to the next site and so on, till it has visited all the s ite s in its itinerary . This strateg y is illust rated in fig ur e 3 (b). Paral lel CS This al so bas ed on the c lient-serve r parad igm . Ho wev er, instead of sequen tial requests , th e cl i ent i nitiat es paral lel threads of e xecut i o n where eac h thr ead co n currently m akes a reques t to o ne of t he serv e rs and proc es ses the reply . This str ategy is illust rated in figure 3 (c). Creatin g Running Suspend Dele t ing Re sumin g Active Suspend Terminatin g Dispatch Resuming IJCA Special Issue on “ Mob ile Ad-ho c Netw orks ” MANET s , 20 10 70 Clie n t Serve r Mob il e age n t Mess age Exc hange 123456 Numb ers along th e arrow s indic a te the sequence of m essages/ M A movem ent. Fig. 3 Imp lementatio n strateg ies Parall el MA I n t his case the clien t initiate s multiple MAs, each of which visits a subset of the serve r s in the itinera ry . The MA s then return to the client and co l late th eir r esu lts to com pl ete the task. Th is strat egy is i llustra ted in fig ur e 3 (d) . 5. MANA GING Q o S USIN G MO B ILE AG ENT Quality of service (QoS ) i s the m easur e of a servic e off er ed by the network to the user. The mo re determ i nistic network perfo rmanc e co uld be achiev ed with the mobile agent s , so that inform atio n co n ceded by the network c an be better deliv ered and netwo r k resourc es can be effec t ively uti lized. A net wo rk or a service prov i der can put forward different kinds of servic es to the users. A service can be charac terized by a s et of requiremen ts such as mi ni mum bandw i dth, m aximum delay , maximum delay variance (jitter) , a nd ma ximum pac ket loss rate. Qual i ty of Servic e (Qo S) managem ent parame te rs a re vital for fac i lit at i ng multimed ia servic es in a netwo rk. A ty pical QoS archi tecture sho ul d s u pport the fo ll o wing: conf i g urati on, predictio n , and manag ement of QoS at all the levels of abstract i on (user, s y stem and netwo r k level ); manag ement, co ntr o l, and proc essing o f a flow must be d istinc t activ i tie s; applicat i on must be transpa r ent fro m estab lishme nt and manag ement; as y nchro nous r esource manag ement of differen t co mponents; and pe rfo rmanc e enhanc emen t . A n agent based Qo S architec t u re do supports all these featu r e s. [12] Qo S off ers flexibi lity , scalabili t y , eff iciency , adaptab ility , sof t ware reusa bility and m ainta inabil i ty . Fig 4: Age n t base d Qo S manageme n t. Ag ent -base d s chem es has seve ral adv antages as c ompared to traditiona l appro a c hes: r educed lat ency , wo rks in hetero geneo us netwo rks, reduced ne t wo r k traff ic, encapsu l ate s protoc ols, flex i bil ity , ada p tabili ty , software reusab i lity and main tain abil ity , and fac ilitates the creat ion of custom i sed dy namic sof tware ar c hitectu r es [1,2,3]. Ho wever, mobile ag ent techno l ogy is still in its infancy and has certa i n pro bl ems that hav e t o be resolv ed. Even thoug h it is diffic ult to quantify th ese features , we expla in how Qo S is achiev ed t hroug h mo bile agent s . Flexib ility : The ag ents allow learning capab ilities to be integ rated in a natural way to suppo rt delay predictio ns , bandw i dth p r edic tions, and play out decis i on-making based o n th e hos t architec ture a nd netwo r k lo ads. Reusabi l ity : Mobile A gent so ftware can be reused by differen t ty pes of multi me dia applic a tio ns by making slight mo difications t o the sof t ware. I t is possib le because of a utono mous o pe ration of the agents in agent-bas ed sy st ems. Maintai na b ility : The softw are c a n be easily maintained since e v ery a gent of an ag ent -based sy st em i s develo ped on a m odula r app r oac h. A dapt ability : Mobile ag ent can eas i ly adapt t he rapid chang es i n the netwo r k co nditions (cong est ion/fai l u re) and us er requ i rem ents . Eff i ciency : The use of a mobile ag e nt inc reases n etwork resource u tilisatio n efficienc y because of its adaptab ility to network and use r r equir emen t s and exchang e of minimal inf orm ation d uring task exec ution and dec isio n making with mu l tip le r eso urce info rmation . Scalab i lity: The scalability can be achiev e d b y using the mo bi le agent at a cl i ent to e xecut e si m i lar tasks of severa l user s . The schem e uses only certai n deg ree neighbo ur ‟s info rmatio n to co mpute the mu l tip le paths . A s according to literat ure, a l o t of concepts hav e alre ady been app l ied on VA NET to pr ovide better Qo S t o the user but still som e o f t he i ssue s ar e not y et an swered and they are: - L ack of real-t i me traff i c info r mat ion. L a ck o f access t o travel info rmation and 24 hour re al -tim e alterna te rout e info rmation. Better altern ate route guid ance. IJCA Special Issue on “ Mob ile Ad-ho c Netw orks ” MANET s , 20 10 71 L a ck of readil y av ailable transi t inform atio n t o increase rid er shi p. Mobile agent s c an bette r ad dress t hese issues i n VAN ET. I n this tec hnique, we propo s e an i n tellig ent agent based network that can pro vide five dif fer ent ty pes of mess ages in the fo rm of ag ents: - 1. Directio n -Find i ng A gent (DFA) 2. Portable Ag ent (PA) 3. I n Vehicular A gent (I VA) 4. Observan t A gent ( OBA ) 5. I nformatio n Finding A g ent (I FA). A l l vehicles mov ing in t he ne twork c onside red to be part of the netwo rk fo r traff i c mo nitoring and contro l. The co re of the sy stem is creat ion of mobile agent of i ntel ligen t node s . Ag ent s are sta tic nodes p l aced at intersect i ons and along the str eets, which maintai ns t he databa s e of traff i c info rmation and routing informat i on. Ro ad side sensors mo ni tor the traff ic situatio n and it is prov ided to nearby Ag ent s. The ne t wo r k c onsist s o f mo bil e no des in an ad -ho c enviro nment . Th e Portab le A gents ( PA ) are strateg i cal l y placed such that PA is well co nnected t o at least one neighbo r ing PA . Desirab l e c harac t erist i cs o f PA no de are: Pro cessing pow er to sus tain dist ribution of databa se and node man agement. More than av erag e norma lized li nk cap a c ity PA to PA and PA to m obile no de co nnectivity and rel i abil ity L ar g e buff er capacity to mainta i n ubiqu i tous databa se and ro uting ta ble. DFA s are responsi ble f or ro ut e disc overy , route main tenanc e an d distr ibut ion of omnipresent databas e. Po rtable A gent keeps the route inform ation and presenc e info rmation . Each DFA fi nds the p at h to neighbo ri ng DFA s using PA s. PA s co ll ect network behav i o r i nfo rmation which c ontain network reso ur ces lik e bandw idth and buffer availabi l ity at a node and pass it to a DFA . The nex t task is to fo rm routing t abl e at DFA node. To fo rm routing table, paths fro m a DFA node to its ne ighbo rs are e stab lished. Path d is co very i s c arr ied out by PA s. A DFA node sends Fo rward Po rtable Ag ent s (FPA s ) in netwo r k to dis co ver the paths betwee n itself and neighbo r ing DFA s. Since any DFA node c an be connec ted t o ne arest D FA node within maxima l number of hops. This avo ids floo ding. Betw een each pair of unique DFA n ode, many paths are disco vered and all paths are re co rded in routing ta ble. Alg orithm 1 explains working o f A P A. Algorit hm 1 . A dvance Po rtable A gent (A P A) If ( hop co unt== 1) then , Mobi l e ag ent is dele ted else if DFA node r eac hed to nex t node then C onv e y all the co ll ected info rmati o n to DFA , i.e., inf orm ation reg ar d ing path fo llowed and resources av ailabl e on t hat path . Create R ever s e Mobi l e Ag ent w i th p ath info rmation . else Dec rease the hop c ount of mo bi le ag ent Co l lect the netwo rk inform atio n needed fo r rou t ing & subm er ge t he mo bile ag ents to neig hbor no des. end if Wo rki ng of RPA is expl ained in A l go ri thm 2. Algorit hm 2 . Reve rs e Po rtable A gent ( R M A ) if DFA no de reache d to next node the n Co nve y all t h e co ll ected information t o DFA , i.e., info rmation regarding path f o llowed a nd resourc es availabl e on t hat path to update ro uting table. Remo ve mo bi le agent . else Give a l l the inf orm at ion c o llected to node . Travel to nex t hop. end if In Vehicul ar A gent (IVA) : IVA is static ag ent r esi de s i n vehic l e whic h co mmunic at es with the DFA to acquire/ spread t he re l evan t i nfo rmati o n. I VA co ll ects the status (mo v ing or st at i onary ) and loc ation i nformat i on of vehicle fro m s ensors equi pped i n a vehic le. Obse rvant A gent ( OA) : OA is a mo bile ag ent t hat travels around the netwo rk by creating its clo nes to propagat e t he decis ive info rmation during the cr itical situatio ns. Exampl es of crit ica l situa ti o n a re acc ident, traffic j am, bad weather c ondit ions, tracing a veh icle i nvo lved i n crime or traffic r ule vio lation etc. I t also i nfo rms I VA and upda tes the v ehi cl e da tabase. O A is sent by D FAs to t he vehicles mov i ng in the ne t wo r k. Informa t ion F inding Age nt (I FA) : I F A travel s in the network to searc h fo r the re qu isite i nfo r mation as des ired by vehic le user. IFA is s en t by t he DFA in the netwo rk on the reque s t issued by user or DFA its elf to get traffic info rmation . [11] 7. Conclusion and Future Scope Mobile ag ent techno l o gy has been highl i ghted as a very interes t ing approac h to build applic a tions fo r mo bi le enviro nment s. How e v er, it is hard t o find p ract ical applicat i ons wi t h real proto t y pes and using the avai lable mo bi le agen t p latfo rms. O ne reaso n is pro bab l y that suc h platfo rms have been deve l oped with a fixed distribu ted enviro nment in mind , and not co nsider ing t he fea t u res t hat may be of s pecial interes t in a mo bi le env i ronm ent (e .g., relianc e against s ecur ity t hreats, adap t ation to the netwo rk techno l ogy , and se r v ice/node disco very ). Mobile agent scales effec t ively as t he siz e o f the data to be o bt ained increase s. IJCA Special Issue on “ Mob ile Ad-ho c Netw orks ” MANET s , 20 10 72 We hope that fu t ure r esea rch and dev elopment eff orts wi ll eventua lly l ead to co nsolidate a go od relatio nship betw een mo bi le agent s and mo bil e dev ices a nd it will also co n centrate on the rea l tim e aspects . 8. Refer ence s [1] Chess, D. , Ha r riso n, C ., and Kers henbaum , A . : „Mobi le a g ents: ar e they a goo d i dea?‟. I BM Researc h Divisio n, T.J. Watson Res earc h Center, Yo r ktow n Heights , New York , March 1995 . [2] Lang e, D.B., and Oshima , M.: „Seven goo d r easons fo r mobile agents ‟ , Comm un. AC M , 1999 , 42, pp. 88 – 89. [3] S.S. Manvi a nd P. Ven kataram “Mobile ag ent ba s ed approac h fo r QoS r outing ” The I nstit u tio n of Eng ineering a nd Techno logy 2007, 1, (3) , pp. 430 – 439. [4] A l fonso Fu gg et ta, Gian P ietro Pic co and G i ov anni Vigna, "Unders t andi ng Code Mobil ity ", I EEE Transac tions on Software Engine ering , vo l. 24(5), 1998. [5] S.S. Manv i , P. V enkataram , “ Appl i cat i ons o f agent techno l ogy in comm uni cations: a review” , Co mput er Co mmunications 27 ( 2 004), 1493 – 1508 . [6] G. Weiss, Multiagen t sy stems a mo dern appro ach to distribu ted artif icial intellig ence, MIT Pr ess, Boston, 2001. [7] R. Chadh a, G. Lapioti s, S. Wrig ht , “Po licy -B ased Netwo r king ”, I EEE Netwo r k speci al i ssu e, Marc h/A pril 2002 , Vol . 16 No . 2, guest e dito rs. [8] Danny B. L a ng e and Mitsuru Os hima, “S even G oo d Reasons for Mobile A gents " , Co mmunicat i ons of A CM, vo l . 42, no . 3 , Marc h 1999 . [9] Shinichi Moto mura, Takao Kawamura, Kaz unori Sug ahara, “Pers i ste ncy for Jav a based Mobi l e A g ent Sy stems”, Proc . Third Interna ti ona l Co nference on I nt ernet and Web A pplicatio ns and Serv ices . Pp. 470 - 475,2008 , [10] Yang Go ng -ping , Zeng Gua ng- z h ou, ”Mob ile A gent L ife State Man agem e nt”, I M A CS Multi -co nference on Co mputational E ngin eering in Sy stems A pplicatio ns (CESA ), 2006 , Beijing, China [11] Neeran M. Karnik and A nand R. Tripa thi “ Des i g n I ss ues in Mobile- A gent Prog ramming Sy stems”, I EEE Co ncurrency , J u ly -September , 19 98 pp 52-61 . [12] T. Bheemar j una Reddy , I . Kart higey an, B.S . Ma noj, C. Siva Ram Murthy “Quality of service prov isioning in ad ho c wireless netwo rks : a survey o f iss ues and solutio ns” Ad Ho c Netwo rks VOL . 4, 8 3– 124, (2006 ). [13] Chess, D., Harriso n, C . , and Ke rshenb aum, A.: „Mobi le a g ents: ar e they a goo d i dea?‟. I BM Researc h Divisio n, T.J. Watson Res earc h Center, Yorktow n Heights , New York , March 1995 . [14] Sada f M o meni, Mahm o od Fat hy , “VAN ET' s Co mmunication”, I EEE 2008 . [15] Nai shadh K . Dav e and Vanaraj B . Vaghela “Veh icular Traff ic Control: A Ubiquito us Co mputing Appro ach”, I C3 2009, CCI S 40, pp . 336 – 348 , 2009 . [16] O. Urra , S. I lar ri and E. M ena, “A gents Jump ing in the A i r: Dream or Reality ?”, , Part-I LNC S 5517, pp.62 7- 634, I WAN N 2009