An Efficient Admission Control Algorithm for Load Balancing In Hierarchical Mobile IPv6 Networks

An Efficie nt Admission Cont rol Al gorithm for Load Balanci ng In Hiera rchical Mob ile IPv6 Network s Prof P.Harini HOD of IT Depa rtment StAnns Colle ge of En gineering & T echnolo gy Chirala, And hraPrad esh, India hariniphd@ yahoo.co m Dr. O.B.V.Ramanaiah Professor in CSE Depar tment JNT UCEA Ananthpur, AndhraPra desh, India Abstract — In h ierarchical Mobile IPv6 n etworks, Mobility Anchor Point (M AP) may become a single point of bottleneck as it handles m ore and more mob ile nodes (MNs). A n umber of schemes have been proposed to achieve load balancing am ong different MAPs. However, signaling reduction is still i mperfect because these schemes also avoid the effect of the number of CN’s. Also only the balancing of M N is performed, but not the balancing of t he actual traffic load, since CN of each M N m ay be different. This paper proposes an efficient admission control algorithm along wi th a replace ment mechanism f or HM IPv6 networks. The admission control algorit hm is based on the number of serving CNs and achieves actual load balancing among M APs. Moreover, a replacement mechanism is intro duced to decrease the new MN blocking probability and t he h andoff MN dropping pro bability. By si mulation r esults, w e show that, the han doff delay and packet loss are red uced in our scheme, when compared wi th the standard HMIPv6 based handoff. Keywords- Load balancing, Admissio n, Ha ndoff, Replacement, Blocking probab ility. I. I NTRODUCTIO N A. Mob ile IP In the ea rly da ys o f the In ternet’s d evelop ment a d ecision was made t hat Internet protoc ol (IP) addr esses would rep resent both the topolo gical loca tion and identit y of a n end -host (R FC 791) [Pos8 1b]. W hile this decision simplif ied the Inter net’s conceptual addressin g mode l and met the needs of earl y network deplo yments it has created difficulties for the develop ment and de ployment of tr uly mobile, IP-bas ed applicatio ns and ser vices. In the 1 990s a networ k-layer ‘Mobile IP’ solution was firs t developed in the context of IP version 4 ( IPv4) (RFC 3344) [Per02a] . During this sa me period of ti me, the Internet Engineering T ask Forc e (IET F) began work on a new version of IP that ha s beco me known as IP version 6 (IPv6) ( RFC 246 0) [DH98]. Despite the fact t hat IPv6 addr essing still maintains much of IPv4’s sema ntic link on location a nd identit y, exper ience with Mo bile IP v4 allowed the IET F to integrate better suppo rt for Mob ile IP1 into IPv6 [1]. The conventional I P a ddresses represent the host’s ide ntity and encode the host’s to pological locatio n o n the IP network [Pos8 1b] s imultaneousl y. This proble m is solved b y the mobile IP ad dress. By follo wing the p hysical move ment of the host’s attachment s, a host g ives the result of entering into a secondar y net work w ith respe ct to the networks IP topology. Once if this takes place, a new I P address must b e assigned to the host there fore the pa ckets m ay be c orrectl y routed to t he host’s new location. The transport layer connections are broken due to this type of moves which were acti ve for the duration of the move ment. This is beca use the hosts IP address is also used as a tr ansport level and end point id entity. So the packets are lo st which are sent to t he e arlier IP addr ess. I n addition to this, when the hosts se nds the IP pac kets at this moment then t heir prio r pee rs will not kno w the new add ress. When two IP addresses are establis hed for m obile hosts then the mobile IP assists in this regio n of this prob lem. They are static ‘ho me address and a tran sitory’ care of addresses. The static ‘home a ddress ’ of the host i s iden tified worldwi de. The transitor y ‘care of add ress’ identified temporar ily when it is attached to the d ifferent parts o f the network. H owever the mobile host is attached to the internet by its foreign address through dynamicall y managed IP-in-IP tunnels and particularl y e ncoded packet for warding rules and it is o perated from its home ad dress. As a result of t his, the net work layer mobility is suppor ted b y M obile IP to be transparent to wards all uppe r layers because j ust the a ddress is being u sed b y t he transport layer. The no n-mobile internet will function e ven i f the mobile hos t en vironment applicatio ns ar e de signed accord ing to the conven tional a ssumptions. B. Mob ile IPv6 Figure 1. Mobile I Pv6 Mobile IP v6 is designed to manage t he Mo bile Node (MN) movements i n t he IPv6 networ ks. E very MN i s availab le to all (IJCSIS) International Journal of Computer Science and Information Security, Vol.6, No. 2, 2009 291 http://sites.google.com/site/ijcsis/ ISSN 1947-5500 other ho sts with Mo bile IPv6 through its ho me id entity. In t he MN’s home network, a Home A gent (HA) is router. By maintaining the MN’s curre nt location infor mation, HA stores the mo bility of t he MN. The last ro uter i n the forei gn net wo rk is the Acces s Router ( AR) which can for ward the pac kets to MN. T he idea of Foreign Agent ( FA) does not exist i n the mobile I Pv6. AR takes p art as a subst itute for the F A e xcept assigning its IP address (the MN’s Care-of Address (CoA) would be the IP a ddress). A ne w CoA i s ob tained by the MN when it moves to a ne w sub net work. T he CoA is either configured using I Pv6 stateless ad dress auto co nfiguratio n or state ful addre ss configuratio n. Then t he MN register s this Co A to the HA. There are tw o possible ways for the MN and Correspo ndent Node (CN) to communicate when the MN is away from ho me: direct c ommunicatio n (Route Op timizatio n) and tu nneling via the H A. In Mob ile IP v6, route optimizati on is funda mental. MN sends binding upd ates to CNs for dir ect communicatio n. On the o ther hand, i f t he CN is a no n-Mobile IPv6 node, the CN can co mmunicate with the MN via tunneling throu gh the H A. Pac kets fro m a CN desti ned to the MN’s Home Address a re inter cepted by the H A a nd t hen tunneled to the M N’s CoA, and vice versa (p ackets from the MN CoA ar e encapsulated and tunneled to the HA and forwarded to the CN). C. Hierarchica l Mobile IPv6 The network overhead and the increased signalin g messages are introd uced by the basic mobile IP when the MN varies its point of attac hment to different network s freq uentl y. A loc alized mobilit y ma nagement pro tocol for MN is given b y the HMIP v6 which is an ext ension o f ba sic Mob ile IP v6. In order to handle mobility man agement a new co nceptual en tity called M obilit y Anchor P oint (MAP) is introduced b y this scheme. MAP is a r outer or a set of r outers. An excl usivel y reliable a dministratio n of a p articular do main i s mai ntained by this MAP. With t he help o f its co mmon routable IP address (Figure 1) , MAP connect s the do main and serves t he Intern et. A regional care-o f address (RCoA) and an on link care- of address (LCoA) are the t wo t ypes of the MN addr esses present in HMIPv6. The RCo A spec ifies a partic ular d omain of t he Internet and it is called as glob al addr ess. The MN transfor ms its LCoA whe n it i s m oved between the local networks inside a M AP do main. I t also requires the new LCoA to register in a MAP on the local link. T he MN changes both the address when it is moved fro m one M AP domain to the new MAP domain. Therefore , this nee ds registering ne w local LCoA and ne w RCoA to the new MAP. At t his instant, the new M AP registers global RCo A to t he MN’s HA. The MN has the ca pability of choosing bet ween the basic mode and the ex tended mode when it is moved inside a MAP domain. In ba sic mode, for rec eiving t he p ackets which are sent to MN inside the same do main, t he M AP acts as the same as the local HA. The packets whic h ar e intended to the RCo A are r eceived by the MAP and send them to the cor respondi ng LCoA of the MN. T he RCo A is the M AP’s add ress in the extended m ode. With the MN ho me add ress, the M AP keeps a binding ta ble with the current LCoA. W hen the M AP rec eive s the packets which are intende d to a MN then it D. Hand off in HIMPv6 Figure 2. Inter-domain Ha ndoff in HMIPv6 The following circums tances are the basis for the system for Inter-do main hando ff: a regional Care -of Addre ss RCoA1 and an on-lin k Care-of Addre ss LCoA3 (Fi gure. 2 ) are p resent in the M N. Whe n the CN se nds packet s to the MN, t he pack ets will be sent t hrough M AP1 to the MN’s LCo A3 [3]. E. Prob lems in HMI Pv6 Generally t he perfor mance of standard MI Pv6 is al ways better than HM IPv6. Decreased signali ng outside an MAP domain cannot co mpensate th e increased signali ng inside the MAP do main, when t he number o f CNs and the number of handoffs of each MN is low. T hus, an MN s hould decid e whether it is e ffective to use H MIPv6 under certain co nditio ns. Instead, an MAP may beco me a single p oint o f bottlenec k as it handles more and more MNs, A n umber of schemes have been proposed to achieve load balancing among d iffere nt MAPs. Amon g t hem, scheme s based o n multiple MAP lev els are proposed . T he m ultiple M AP le vels share the traf fic load and M AP se lection algorithm i s based o n the MN ’s sp eed. However, signaling re duction is still imperfect because t hese schemes also avoid the e ffect of the number o f CN’s. In some schemes, thresho ld- based admission contro l algorithms a re pro posed to avoid overloa d at par ticular MAP s. Each M AP set a threshold to l imit the number of ser ving M Ns. Such algorithms can only manage the balancing of MN, b ut not the balancin g o f the actua l traffic loa d, si nce CN of e ach MN may be d ifferent. This paper proposes an efficient load balancing scheme along with a replac ement mechani sm for HMIPv6 networks. It introduces an algorith m for enabling an M N to d etermine whether it i s suitable to use HM IPv6. T he admissio n contro l algorithm i s ba sed o n the nu mber o f serving CNs achie ves actual load balancing a mong MAPs. Moreover, the replacement mec hanism is intr oduced to decr ease the ne w M N blocking pro babilit y a nd the ha ndoff MN drop ping prob ability. (IJCSIS) International Journal of Computer Science and Information Security, Vol.6, No. 2, 2009 292 http://sites.google.com/site/ijcsis/ ISSN 1947-5500 II. R ELATED W ORK Nakaji ma.N [4], have pro posed a Robust Hierarc hical Mobile IPv6 (RH-MIP v6), which provides incor rect tolerance and stren gth i n mobile networ ks. In RH-MIPv6 [5], a mobile node (MN) register s p rimary (P -RCoA) and seco ndary (S- RCoA) re gional car e of ad dr esses to t wo differe nt MAPs (Primary a nd Seco ndar y) at the sa me time. The y have developed a method to p ermit the mobile node or corresp ondent node (CN) for the dete ction of the failure of primary MAP and modify their atta chment from primar y to secondar y MAP. Moon et al. [6] pr oposed a method to les sen the ha ndover delays. T his is done b y coope rating with t he in formation o f layer 2 and in addition b y per forming some functio ns of the mobile node o n the acc ess router. The y have reco mmended a slight more co mpetent and o ptimized mobility sche me. Thi s method has co mbined appro priately the advanta ges of earlie r proposed schemes. Pekka Nikander et al [7] discussed the design rationale behind the MIPv6 Route O ptimizatio n Securit y Desi gn.IPv6 (MIPv6 ) allo ws a Mobile Nod e to talk directly to its peers while retaining the abilit y to move around and c hange the currently used IP add ress. This mode of op eration is cal led Route Op timiza tion (R O), as i t a llows t he p ackets to travers e a shorter ro ute than t he defa ult one thro ugh the Ho me Agent. This is pote ntially d angerous, since a malicious ho st might be able to estab lish false bind ings, t hereby preventin g so me packets fro m r eachi ng their inte nded d estination, d iverting some traffic to the attac ker, or flooding third parties with unwanted traf fic. Aisha Ha ssan A bdalla Hashim et al. [8] descr ibed h ow mobile nod e ca n c hange its po int o f atta chmen t from one access r outer to a nother in Mobile IP v6 (MIPv6 ). Hierarchical Mobile IPv6 (HM IPv6) is one of them that are designed to reduce the amount of sign aling required and to improve handover spee d for mobile connectio ns. This is achieved by introducing a new network entity called Mobility Anchor Point (MAP). This presents a co mparative stud y of the Hierarchical Mobilit y IP v6 a nd Mobile IPv6 protoco ls and we have narrowed do wn the scop e to micro-mobilit y (int ra- domain). Md. Mo hiuddin K han et a l, [9 ] discussed about Mo bile IP and the pro posed handoff schemes, their pro s and cons and some areas where t hey ca n b e improved . T he reasons for handoff late ncy are e xamined and the sche mes are disc ussed by their genres. The y ob serve t hat combini ng differe nt methods resul ts in a better p erfor mance. T hey buffer i ssues like cross-la yer desig n and c ontext aware ness i mprove the handoff latenc y b ut also br ing overhead to them. Later th ose packets are deliver ed. This stud y de monstrated the fundamental po ints of concern related with efficient hand off designs. Pyung-Soo Kim, and Yon g Jin Kim [10] the new fast vertical handover scheme is proposed for the hierarchi cal Mobile IPv6 ( HMIPv6 ) to o ptimize and e nhance the exi sting fast vertical handover HMIP v6 (FVH-HMIPv6) in heterogeneo us wireless acc ess networks. T he recently standardized IEEE 802.21 Media Inde pendent Handove r Function (MIHF) is adopted for the propo sed FVH-HMIPv6. Firstly, the Media I ndep endent Infor mation Service ( MIIS) is extended b y includi ng new L 3 infor mation to provide d omain prefixes of hete rogeneo us neighborin g mobilit y a nchor po ints (MAPs), which i s critical to the ha ndover p er formance of proposed FV H-FMIP v6 with MIHF. Sec ondl y, the o peration proced ure for the p roposed scheme is describe d in detail. Li Jun ZHANG and Sa muel PIERRE [11] presents a comprehensive per for mance analysis of Fast h ando ver for Hierarchical Mobile IPv6 (F-HMIPv6 ) using the fluid -flo w and rando m walk mobilit y models. Loca tion upda te cost, packet deliver y cost and total co st functio ns a re for mulated based on the propo sed analytica l models. They investi gate the impact of several wire less s ystem facto rs such a s user ve locity, user densit y, mobilit y domain siz e, session -to-mobilit y r atio on these cost s, and pre sent some nu merical re sults. In o ur previous w ork we [ 12] recommend a novel cross- proto col de sign appr oach for the DAD pro blem. It uses information from current r outing proto col traffic. T his utilize s Passive Duplicate Address Detection ( PD AD) to id entif y duplicate ad dresses. Usi ng PDAD, a node exa mines inco ming routing proto col packets to de velop hints about address controversies. Since it is independent of the r outi ng proto col, it p roduces nearly n o protocol overhead an d reduces i nitial delay. In our p revious work we [ 13] propose an Enhan ced Hierarchical M obile IP v6 (E -HMIP v6) arc hitecture b ased on a novel cros s-layer/cross -protoco l design a pproa ch. A new node , called Help er node, is adop ted in o ur (E-HMIP v6) architecture which can a ssist the Mobile Station ( MS) to obtain the (regional ca re of ad dress) RCoA a nd LCo A (local c are o f address) and execute the DAD procedure f or proving t he unique. The MS can still transmit da ta to the Corresp ondent Node (CN) d uring the p re-handof f pr ocedure . Thus it is evident that the latenc y o f handoff in the proj ected proto col is lower than in the traditio nal HMIPv6 and MI Pv6. For the DAD pro blem, we use the P assive Duplicate Address Detection (PDAD) mecha nism to id entify duplicate addre sses. In order to develop the hints a bout the add ress controversies , a node exami nes the inco ming ro uting proto col packets u sing PDAD. Being indep endent of the ro uting pr otocol the P DAD does n ot c reate any proto col over head and red uces initial delay. III. A DMISSI ON C ONTROL Let us co nsider the scenario depicte d in Fi gure.3 as an example. T hree MAPs MAP1 , MAP2 and M AP3 are placed. Each MAP consists o f a ccess routers. Let MAP2 contains two access routers AR1 a nd AR2. MN2 is movi ng from AR1 to AR2, resultin g in handoffs and the process o f admission control. Here, we assume tha t the M APs t hat t he t wo M Ns registered p reviousl y are no lo nger valid. AR2 ac hieves M AP o ptions t hrough route ad vertise ments. When M N2 a ttaches to AR2’s li nk, it is ad mitted a s pe r t he admission co ntrol al gorithm. (IJCSIS) International Journal of Computer Science and Information Security, Vol.6, No. 2, 2009 293 http://sites.google.com/site/ijcsis/ ISSN 1947-5500 Figure 3: Architecture of t he Proposed Scheme A. Classificatio n of MN In the prop osed sche me, incomi ng MNs are cla ssified into two types: • new MN • handoff MN (1) New MN: The M N performing t he initial bindi ng updates BU ) to the MAP (e.g. whe n an MN is turned o n). (2) Handoff MN: Whe n an ongoi ng MN moves i nto a ne w MAP d omain, t he MN sends a local BU message to wards t he new M AP to complete local re gistration. Since the ha ndoff MN is registered , it is probab ly communicati ng with o ne or more CNs. T hus, t he ha ndoff MN should have hi gher prior ity than the ne w MN. An M AP sh ould deter mine w hether the received BU message co mes from a handoff M N or a ne w M N usi ng the following ste ps: 1. A flag A is ad ded to the existin g BU message, a nd it i s set if the MN i s in the r eady state when it se nds a BU message. 2. I f the flag A in B U i s se t, the MN is re garded as a handoff MN. Ot herwise, it is regard ed as a new MN. 3. A ready state timer is contro lled by MN to kno w i f it is in read y or idle state. I f an MN in id le state sends or receives data, the rea dy timer is in itialize d and the MN cha nges to the ready state. E very ti me the M N sends or r eceives data , the ready ti mer is re set. If the MN does not send o r receive data until the read y timer e xpires, the MN returns to the idle s tate. In the prop osed a dmission c ontrol algorithm for the two types of MN ’s, two thre shold value s are maintained as • thr N - threshold val ue for ne w MN • thr H - threshold val ue for handoff M N Let cn tot denotes the total nu mber of CNs that a M AP currently serve s. Normally thr H is equal to the capa city o f a M AP which is the maximu m number o f CNs that t he M AP can serve. The ad mission contro l algorith m is defined as: Admission Co ntrol Alg orith m 1. cn tot Determine 2. then thr cn N tot if ≤ hand MN and Admit MN both new 3. then N thr thr cn H tot if ≤ < new hand MN reject and MN Admit 4. then cn thr tot H if < hand new MN and MN both Reject Where admit MN indica tes t he MNs which ca n be admitted under the ab ove conditions, hand MN d enotes th e handoff MN s and new MN denotes the ne w MNs B. Rep lacemen t Mechanism When a new MN o r a handoff MN cannot be accep ted, it should not be b locked or drop ped dir ectly. The M AP will choose an M N fro m e xisting MN s to be replaced. If the number of c onnecti ng CN s ) ( cn con o f an MN is equal to o r larger than that o f the i nco ming MN, it beco mes a candidate to be repla ced. T he MAP sends b inding ackno wledgement (B A) to the chosen M N. T he BA message co ntains an e rror code with the reason “Insu fficient resourc es”. T hen, the r eplac ed MN performs the following MAP selectio n al gorithm to choose another suitable M AP among the re maining M APs. In pro posed scheme, MN maintains a table co ntaining information on availab le M APs until new router advertise ment arrives. T hus, it enab les the MN to choose different M APs. MAP Selectio n Algorithm 1. i MAP each for 1.1 Y ratio th e Calculate cn cn tot con Y / = 1.2. W measure combined the Calculate ) ( S Y W + = α constant - and MN of speed - S α Where 1.3. then T W map < if i MAP Choose value T map threshold Where − 2. for end IV. E XPERIMENTAL R ESULTS A. Simu lation Setu p We use NS2 to si mulate our p roposed (E-HMIP v6) architecture . In o ur simulation, the c hannel capacity of mobile hosts is set to the same value: 2 Mbps. We use the distrib uted coord ination fun ctio n ( DCF) o f I EEE 802.1 1 f or w irele ss LANs as the MAC la yer p rotoc ol. I t has the functionality to notify the network la yer about link breakage. T he follo wing table (Table. 1) summarizes the simulation settings. The CBR (IJCSIS) International Journal of Computer Science and Information Security, Vol.6, No. 2, 2009 294 http://sites.google.com/site/ijcsis/ ISSN 1947-5500 traffic is estab lished from C N to MS, and the bandwidth and latency for every link between every two co mponents are also specified in t his scenar io. T ABLE I. S IMULATION SET TINGS No. of No des 20 Area Size 1000 X 1000 Mac 802.11 Simulation T ime 50 sec Traffic Sourc e CBR Packet Size 512 Speed 5,1 0,15,20 Trans mission range 75m Routing P rotoco l AODV For the simula tion, we make use o f Hie rarchica l Mobile IP (HMIP ) implementation, which has implemented in Columbia IP Micro-mobility So ftware (CIMS). It suppo rts micro mobility proto cols for insta nce Ha waii, cellular IP , and HMIP extension meant for the ns-2 network simulato r b ased on version 2.1b 6. We have add itionally include d MAP functionalit y to provide re gional re gistration with the existing CIMS imple mentatio ns. The simulatio n ha s carried out using the net work to polog y shown in Fig ure 4 . It co nsists o f M AP1, MAP2 and M AP3. Each M AP contains t wo acces s routers. Figure 4. Netw ork Topology Initially t he m obile node MN1 9 w as in MAP3 in t he domain AR6. Duri ng t he simu lation we perfor m intra and inter domain hando ff on MN19 . Initially, at ti me t1, the mobile node p erforms intra do main handoff b y moving from AR6 to AR5 within M AP3. Next at time t2, it start moving to wards AR4 fro m AR5, thus b y performing inter do main handoff. At time t3, it moves from AR4 to AR3, within MAP2 . Finall y at ti me t4 , it moves bac k to AR1, once a gain perfor ming inter d omain handoff. We e valuate the perfor mance o f our sc heme based o n the following par ameters.  Handof f Lat ency: T he hando ff latenc y is d efined as the time in terval from last p acket re ceived for m serving BS to and ne w pac ket re ceived fro m target BS.  Throughp ut: T he n umber of p ackets received at the MS. B. Results A. Varying Ra te Initially we vary the ra te of tr affic as 0 .1Mb, 0.2Mb ….0.5Mb. Figure 5 shows the packets r eceived for AC-HMIP v6 and E-HMIPv6 sche mes. Fro m the figure, we ca n see that the throughput increa ses when the rate increases. Due to its load balancing sche me, AC-HMI Pv6 has more throughp ut when compared with E-HMIP v6. Figure 6 sho ws the hando ff dela y for E-HMIP v6 a nd AC- HMIPv6 based schemes. Clearly the ha ndoff d elay for (AC- HMIPv6) is significantl y less when compared with E - HMIPv6. Throughput 0 200 400 600 800 1000 1200 0.1 0.2 0.3 0.4 R ate ( M b) P ackets A C - H M I P v6 E -H M I P v 6 Figure 5. Rate Vs Throughp ut De l a y 0 0.1 0.2 0.3 0.4 0.5 0.1 0.2 0.3 0.4 R ate ( M b) D e la y ( s ) A C - H M I P v 6 E - H M I P v 6 Figure 6. Rate Vs Handof f Delay B. Var ying Speed Next we var y the speed of the M N as 5, 10, 15 and 20 m/s. Figure 7 shows that t he throughput decrease s when the speed o f the mob ile incre ases si nce it ha s to p erfor m t he admission co ntrol a nd M AP selectio n a lgorith ms withi n a short perio d. From Fig 7, we can see that the thro ughpu t is once agai n le ss in the c ase of E-HM IPv6 scheme when compared with our AC-HMIPv6. (IJCSIS) International Journal of Computer Science and Information Security, Vol.6, No. 2, 2009 295 http://sites.google.com/site/ijcsis/ ISSN 1947-5500 Figure 7 s hows tha t the dela y d ecreases as t he speed of the mobile increases since it has to perfor m the MAP selecti on algorithms within a short per iod. We can see that AC -HMIP v6 scheme has lo w de lay whe n co mpared with E-HMIP v6 scheme. Throughput 0 200 400 600 800 1000 5 10 15 20 S pe e d( m /s ) P ackets A C - H M I P v 6 E -H M I P v 6 Figure 7. Speed Vs Throug hput H and o f f D e lay 0 0.1 0.2 0.3 0.4 5 10 15 20 Spe e d ( m /s ) D e l a y ( s ) A C - H M I P v 6 E - H M I P v 6 Figure 8. Speed Vs Handof f Delay Thus, the AC-HMI Pv6 based arc hitecture with our implementatio n strategie s adap ts all scenarios p erfectl y. V. C ONCLUSION In this pap er we have proposed an efficie nt ad mission control algorit hm along wi th a replace ment mechani sm for HMIPv6 net works. In the propo sed scheme, inco ming MNs are classi fied as new MN a nd handoff MN. It co nsists o f a technique b y whic h a M AP can dete rmine whet her the received B U message co mes fro m a hando ff MN or a ne w MN. T he admission co ntrol algo rithm is b ased on t he numb er of servi ng CNs and achie ves act ual load bala ncing among MAPs. Moreo ver, a rep lacem ent mec hanism is i ntroduced to decrease the ne w MN b locking p roba bility and the hando ff MN dropping probabilit y. B y simulation results, we have shown that, the handoff delay and p acket loss are reduced in our scheme, when compared with the standard HMIPv6 based handoff. As a future w ork, w e li ke to i nclude r ecover y mechanis ms for pac ket loss o ccurred due to ha ndoff. Also we need to develop a rate c ontrol mechanism based on th e bandwidth avai labilit y of the n ew access po int. R EFERENCES [1] “Mobile IPv6 Overview ”, Cisco Systems, Dec 2004 [2] D.Johnson et al, “Mobility Support in IPv6”, Interne t Draft, June 2003 [3] Hesham Soliman, Flarion et al, “Hierarchical Mobile IPv6 Mobility Management”, I nternet Draft, June 2003 [4] Nakajima, N. Dutta. A, Das, S. Schulzrinne, H. “Handoff Del ay Analysis a nd Meas urement fo r SI P based mobility in IPv6”, I EEE International Confer ence, May 2003 [5] Taewan You, Sangheon Pack, and Yanghee Choi “Robust Hierarchical Mobile IPv6 (RH-MI Pv6)” Vehicular Te chnology Conference, 2003 [6] Moon, Seung Wook Lee, Jong Hyup “Red ucing Handover Delay in Mobile I Pv6 by cooper ating with L ayer 2 an d Laye r 3 Handovers”, IEEE, 2008, ICA CT 2008 [7] Pekka Nikander et al. “Mobile IP version 6 (MIPv 6) Rou te Optimization Security Design, I EEE, 2003. [8] Aisha Hassan Abdalla Hashim, Fauzana Ridzua n, and Nazreen Rusli, "Evaluation of Ha ndover L atency in Intra-Domain Mobility", proceedings of wo rld academy of science, 2005 [9] Md. Mohiuddin Khan, Md. Arifur R ahman Bhuyan, Fiash Kiswar, A. S. M. Ashique Mahmood, "Overview a nd Compa rison of Methods for Minimizing Handoff L atency in Mobile IP" , IJCSNS ,2008 [10] Pyung-Soo Kim, an d Yong Jin Kim,” Hierarchical Mobile IPv6 Based Fast Vertical Handover usi ng I EEE 802.21 M edia Independent Handover Function”, I EEE, 2007 [11] Li Jun ZHANG an d Samuel Pierre,” Evaluating t he Perfor mance of Fast Handover For H ierarchical MIPv6 in Cellular Ne tworks”, Jour nal Of Networks, 2008 [12] P. Harini and Dr.O. B.V.Ramanaiah,” A n Efficient DA D Scheme f or Hierarchical Mobile I Pv6 Handoff”, IJCSNS, 2008 [13] P. Harini and Dr.O .B.V.Ramanaiah, ”An Enhanced A rchitecture To Improve Handoff Perf ormance In Hierarchical Mobile IPv6 “ Prof. P. Harini M.Tech.(Remote Sensing),M.Tech. (CSE), [Ph. D. (Mobile Computing)]. I obtained my M.Tech. (Remote Sensing) in 1997 & M.Tech. (C SE) in 2003 from JNTU, Masab Tank, Hyderabad. I worked as a Research Asso ciate in JNTU, Ma sab Tank, Hyderabad in Remote Sensing De partment for 01 year, 05 years worked as a Assistant Professor in QIS College of Engineering, Ongole an d 01 year worked as a Asso ciate Professor in Rao & Naidu Engineering College , Ongole. At present I am working as Professor & Head of th e Computer Science and Engin eering Department in St. Ann's Col lege of Engineering & T echnology, Chirala. Dr. O.B.V. Ramanaiah received Ph. D. in Computer Science from University of H yderabad i n 2005, M. Tech in C omputer Science and B . T ech in Computer Science & Engineering . My Total Te aching Experience is16 Years with Total Research Experience of 12 Years. I h ave 3 Public ations in International Conference Proceedings and 2 I nternational Journals . I ha ve Visited USA for p aper presentation in an IEEE Conference, IT CC 04, held du ring April 5-7, 2004. I am providing Research Guidance to 8 st udents and h ave organize d many Refresher Courses. (IJCSIS) International Journal of Computer Science and Information Security, Vol.6, No. 2, 2009 296 http://sites.google.com/site/ijcsis/ ISSN 1947-5500