Dynamic Bandwidth Allocation among Authentic Mobile Nodes in ...

International Journal of Applied Engineering Research, ISSN 0973-4562 Vol.7 No.11 (2012)© Research India Publications; http://www.ripublication.com/ijaer.htmTABLE-3: Pseudo code of BRM_sig_dis function{ MN_k i :=number of MNs in the coverage areaofBS k_iComputes MN_K i from List_BS k_i__Computes MN_BRM k as ∑(MN_K )P_BS k_i :=arrival probability of MNs in thecoveragearea of BS k_iComputes P_BS k_i as MN_k i /MN_BRM ktotsig_BS k_i :=total signature assigns to BS k_iComputes totsig_BS k_i asP_BS k_i *totsig_BRM k }2.1.2 Signature assignment technique:BS k_i generates Sig j for New_ MN j in the form of (j, BRM k ,BS k_i , t3, Sig_j) and assigns it to New_MN j . The size ofsignature packet (Size_SPAC) islog 2 (tot_MN CAM )+log 2 (tot_BRM)+log 2 (tot_BRM_BS)+(Size_TS)+(Size_Sig_j) bits, where Size_TS is the size of timestampfield and Size_Sig_j is the size of signature. BS k_i inserts Sig_jin List k_i and sends an authentication message (AM) in theform of Sig_j to BRM k . BRM k increases New k by 1, insertsSig_j in List_BS k_i and sends an authentication message in theform of Sig_j to CAM. CAM inserts Sig_j in List_BRM k . Thesize of authentication message (Size_AM) is (Size_Sig_j) bits.So the communication overhead of the signature assignmenttechnique (COOH_SA) for New_MN j is(2*Size_AM+Size_SPAC) bits.2.1.3 Signature verification technique:The verification of signature associated with Intra_MN j andInter_MN j is considered for discussion in this section.Signature verification for Intra_MN j : Let Sig j is of the form of(j, BRM k , BS k_p , t2, Sig_j). BS k_i detects an intra handoff ofMN j from BS k_p and sends an intra traceability message in theform (BS k_p , Sig_j) to BRM k . The size of this intra traceabilitymessage (Size_Intra_TM) is log 2 (tot_BRM_BS)+(Size_Sig_j)bits. BRM k searches List_BS k_p for Sig_j.If not found Intra_MN j is not authentic. BRM k resets aflag and sends intra revocation message in the form of (Sig_j,flag) to BS k_i and CAM. CAM inserts Sig_j in the revocationlist.Otherwise Intra_MN j is authentic. BRM k increases Intra k by1, sets a flag and sends an intra acknowledgement message inthe form of (Sig_j, flag) to BS k_i . The size of intraacknowledgement message (Size_Intra_ACK) is(Size_Sig_j)+1 bits. BRM k also starts to wait for the deletionlist which contains Sig_j as one of the deleted signatures fromBS k_p under its coverage area. The size of the deletion list(Size_DL) for MN j is Size_Sig_j bits. BRM k inserts Sig_j inList_BS k_i and deletes Sig_j from List_BS k_p after receivingthe deletion list from BS k_p . BS k_i inserts Sig_j in List k_i andassigns the same signature to Intra_MN j by replacing BS k_p byBS k_j and t2 by the current time stamp. BS k_i enters into thesession initiation phase as discussed in section 2.2.So the communication overhead of the signatureverification technique for authentic Intra_MN j(COOH_SV intra )is(Size_Intra_TM+Size_Intra_ACK+Size_DL+Size_SPAC)bits.Signature verification for Inter_MN j : Let Sig j is of the form of(j, BRM q , BS q_t , t1, Sig_j). BS k_i detects an inter handoff ofMN j from BRM q to BRM k and sends an inter traceabilitymessage in the form of (BRM q , BS q_t , Sig_j) to BRM k . Thesize of this inter traceability message (Size_Inter_TM BS ) islog 2 (tot_BRM)+log 2 (tot_BRM_BS) + (Size_Sig_j) bits. BRM ksends an inter traceability message in the form of (BRM q ,Sig_j) to CAM. The size of this inter traceability message(Size_Inter_TM BRM ) is log 2 (tot_BRM)+(Size_Sig_j) bits.CAM searches List_BRM q for Sig_j.If not found, Inter_MN j is not authentic. CAM resets theflag and sends inter revocation message in the form of (Sig_j,flag) to BS k_i through BRM k . CAM also inserts Sig_j in therevocation list.Otherwise Inter_MN j is authentic, CAM sets a flag, sendsinter acknowledgement message in the form of (Sig_j, flag) toBRM k and also starts to wait for the deletion list whichcontains Sig_j as one of the deleted signatures from BRM qunder its coverage area. CAM inserts Sig_j in List_BRM k anddeletes Sig_j from List_BRM q after receiving the deletion listfrom BRM q . BRM k increases Inter k by 1 and sends interacknowledgement message to BS k_i in the form of (Sig_j,flag). The size of inter acknowledgement message(Size_Inter_ACK) is identical to Size_Intra_ACK. BRM kinserts Sig_j in List_BS k_i . BS k_i inserts Sig_j in List k_i ,assigns the same signature to Inter_MN j by replacing BRM qby BRM k , BS q_t by BS k_i and t1 by the current time stamp.BS k_i enters into the session initiation phase. So thecommunication overhead in the signature verification phasefor authentic Inter_MN j (COOH_SV inter ) is(Size_Inter_TM BS +Size_Inter_TM BRM +2*Size_Inter_ACK+Size_DL+Size_SPAC) bits.2.2 Session initiation phase: BS k_i uses this phase to establishthe requested session of MN j after verifying its authentication.The pseudo code of this phase of operation is shown inTABLE-4.TABLE-4: Pseudo code of session initiation phase{ MN j sends new session request message to BS k_ifor p thclass of traffic//Functions of BS k_iComputes the desired bandwidth to establish thesessionof MN jExecutes session establishment function forchecking theavailability of desired bandwidthIf desired bandwidth is available{Sends positive acknowledgement to MN jExecutes bandwidth reservation functionUpdates details of ongoing sessions of MN j inTABLE-5Computes bandwidth utilizationSends bandwidth utilization to BRM kGo to L}Else

International Journal of Applied Engineering Research, ISSN 0973-4562 Vol.7 No.11 (2012)© Research India Publications; http://www.ripublication.com/ijaer.htm{Sends negative acknowledgement to MN jExecutes bandwidth reservation functionComputes the number of blocking sessions ofMN jUpdates TABLE-5Go to L}}L: Starts to process the session in TABLE-5 usingprocessing new session function2.2.1 Bandwidth reservation function: BS k_i maintains 3counters to count the number of new session requests pertraffic class. The counter C k_i_p is used to count the number ofnew session requests for p th class of traffic. BS k_i increasesC k_i_p by 1 when it receives a new session request message forp th class of traffic from MN j . It also computes the total numberof new session request (tot_n k_i ) asC __ and arrivalprobability of p th class of new session request (AP k_i_p ) asC k_i_p /(tot_n k_i ).Let the available bandwidth at BS k_i is BW k_i kbps which isassigned by BRM k as discussed in section 2.2.4. BS k_i reservesa fraction of BW k_i for new session request per traffic classdynamically depending upon their arrival probability. So thereserved bandwidth for p th class of new session request (R k_i_p )is AP k+i_p *BW k _ i kbps. It also computes its total unusedbandwidth as (BW k_i -R __ ) kbps and maintains thisbandwidth in a bandwidth reservation pool (BR_Pool k_i ).2.2.2 Details of ongoing sessions:BS k_i maintains a list of ongoing sessions within its coveragearea in tabular form as shown in TABLE-5. Each row inTABLE-5 has a few records and it is for a particular MNwhich is identified by MN_id attribute of the records. Thenumber of row in TABLE-5 depends up on the number ofMNs within the coverage area of BS k_i (MN_k i ). The numberof records in the row of a particular MN_id depends upon thenumber of different types of traffic classes for which thecorresponding MN_id has ongoing sessions. For example, j1 thMN (MN j1 ) has ongoing sessions for all the 3 classes of trafficwhereas j2 th MN (MN j2 ) has ongoing sessions for class 1 andclass 3 type of traffic as observed from TABLE-5. B j1 and B j2are the total number of blocking sessions of MN j1 and MN j2respectively.TABLE-5: List of ongoing sessionsMN_id TrafficclassMN j1 123MN j2 13Number ofpacket pertraffic classP 1j1P 2j1P 3j1P 1j2P 3j2Totalnumber ofblockingsessionUpdation of TABLE-5: Let BS k_i receives a new sessionrequest message from MN j for class 3 type of traffic in theform of (j, 3, NO_OF_PAC j ). BS k_i uses session establishmentB j1B j2function as discussed in section 2.2.3 to verify the availabilityof the desired bandwidth (=RC 3 *NO_OF_PAC j ) for thetransmission of NO_OF_PAC j number of class 3 type ofpackets. BS k_i searches TABLE-5 for MN j . If not found itinserts a new row in TABLE-5 for MN j . In the new row thevalue of the MN_id attribute is inserted as MN j , traffic classattribute is inserted 3, number of packets per traffic classattribute is inserted as NO_OF_PAC j if the desired bandwidthis available otherwise 0 and total number of blocking sessionattribute is inserted as 0 if the desired bandwidth is availableotherwise 1. Otherwise BS k_i searches the existing records ofMN j in TABLE-5 for class 3 type of traffic. If found BS k_iincreases the number of packet per traffic class attribute byNO_OF_PAC j if the desired bandwidth is available and totalnumber of blocking session attribute by 1 if the desiredbandwidth is not available. Otherwise BS k_i inserts a newrecord in the existing row of MN j in TABLE-5. In the newrecord the value of the traffic class attribute is inserted as 3,number of packet per traffic class attribute is inserted asNO_OF_PAC j if the desired bandwidth is available otherwise0 and total number of blocking session attribute is inserted as0 if the desired bandwidth is available otherwise 1. BS k_ideletes a row from TABLE-5 if the MN corresponding to thatrow goes out from the coverage area of BS k_i . It deletes arecord from a particular row of TABLE-5 if the ongoingsession corresponding to that record is over.Computation of unused bandwidth:BS k_i consults with TABLE-5 and computes the used reservebandwidth for class 1 (UR k_i_1 ), class 2 (UR k_i_2 ) and class 3(UR k_i_3 ) type of new session request as RC 1 *(P 1j1 +P 1j2 ) kbps,RC 2 *P 2j1 kbps and RC 3 *(P 3j1 +P 3j2 ) kbps respectively. It alsocomputes the unused bandwidth for new session request of p thclass of traffic (UU k_i_p ) as (R k_i_p -UR k_i_p ), where UR k_i_p bethe used reserve bandwidth for p th class of traffic.Computation of bandwidth utilization:It computes the bandwidth utilization (BU k_i )as UR __ / R __ . It sends BU k_i to BRM k .Each BRM receives bandwidth utilization from all thetot_BRM_BS number of BSs under its coverage areaperiodically. Let the bandwidth available at N k is B k . So BRM kallocates a fraction of B k as__(BW k_i =(BU k_i / BU _ )*B k ) to BS k_i under it. Itrepeats the same operation for all the tot_BRM_BS number ofBSs under its coverage area.Computation of user satisfaction:BS k_i computes the blocking probability (BP) of MN j1 (BP j1 )and MN j2 (BP j2 ) as B j1 /(P 1j1 +P 2j1 +P 3j1 ) and B j2 /(P 1j2 +P 3j2 )respectively. It also computes user satisfaction (US) of MN j1(US j1 ) and MN j2 (US j2 ) as 1 - BP j1 and 1 - BP j2 respectively.2.2.3 Session establishment function:BS k_i uses this function to verify the availability of the desiredbandwidth after receiving any new session request messagefrom a MN which is authentic. If BS k_i receives a new sessionrequest message for class 1 type of traffic from MN j , it

International Journal of Applied Engineering Research, ISSN 0973-4562 Vol.7 No.11 (2012)© Research India Publications; http://www.ripublication.com/ijaer.htmcomputes the desired bandwidth (D 1 ) as the product of RC 1and the number of packets in the requested session. BS k_iestablishes the session if UU k_i_1 >D 1 else if BR_Pool k_i >D 1else if UU k_i_3 >D 1 else if UU k_i_2 >D 1 else block the session. IfBS k_i receives a new session request message for class 2 typeof traffic from MN j , it computes the desired bandwidth (D 2 ) asthe product of RC 2 and the number of packets in the requestedsession. BS k_i establishes the session if UU k_i_2 >D 2 else ifBR_Pool k_i >D 2 else if UU k_i_3 >D 2 else block the session. IfBS k_i receives a new session request message for class 3 typeof traffic from MN j , it computes the desired bandwidth (D 3 ) asthe product of RC 3 and the number of packets in the requestedsession. BS k_i establishes the session if UU k_i_3 >D 3 else ifBR_Pool k_i >D 3 else block the session. In case of block ofsession BS k_i increases a block count counter (Block_Count k_i )by 1.2.2.4 Processing new session function:BS k_i computes the user satisfaction of the MNs that entersinto its coverage area and arranges their user satisfaction in theascending order in a list. Let MN j1 and MN j2 are the two MNsin the coverage area of BS k_i having user satisfaction US j1 andUS j2 respectively and they have ongoing sessions as shown inTABLE-5. If US j1

International Journal of Applied Engineering Research, ISSN 0973-4562 Vol.7 No.11 (2012)© Research India Publications; http://www.ripublication.com/ijaer.htmFig.3 Bandwidth utilization vs. Traffic loadAt low traffic load both the average session loss and averagebandwidth utilization are minimum as most of the bandwidthat all the 3 networks are free. At heavy traffic load all the 3networks are saturated. So at heavy traffic load the averagesession loss increases but the average bandwidth utilizationbecomes almost constant as most of the bandwidth at all the 3networks is utilized. It can be observed from Fig.2 and Fig.3that the average session loss is minimum and averagebandwidth utilization is maximum for WiMAX network dueto its high data rate whereas average session loss is maximumand average bandwidth utilization is minimum for GPRSnetwork due to its low data rate. The data rate of IEEE 802.11network is higher than GPRS network and lower thanWiMAX network. So its performance in terms of session lossand bandwidth utilization is better than GPRS network andworse than WiMAX network.IV.CONCLUSIONThe present work is a dynamic bandwidth allocation schemeamong MNs in heterogeneous networks. It considers theintegration of 3 access networks. The available bandwidth ateach access network is distributed dynamically among theMNs for establishing their sessions. The proposed work can beextended by considering the QoS factor normalization and theoverall architecture implementation.REFERENCES[1] C.Perkins, “IP Mobility Support for IPv4”, IETFRFC 3344, August 2002.[2] D.Johnson, C.Perkins, and J.Arkko, “MobilitySupport in IPv6”, IETF RFC 3775, June 2004.[3] C.Ahlund, R.Brannstrom, K.Andersson andO.Tjernstrom, “Port-based Multihomed Mobile IPv6for Heterogeneous Networks”, Local ComputerNetworks, pp.567-568, November 2006.[4] E.Nordmark and M.Bagnulo, “Shim6: Level 3Multihoming Shim Protocol for IPv6”, Internet-Draft,draft-ietf-shim6-proto-10.txt, February 2008.[5] R.Atkinson, S.Bhatti and S.Hailes, “A proposal forunifying mobility with multi-homing, NAT, andsecurity”, pp. 74-83, MobiWac ACM 2007.[6] http://en.wikipedia.org/wiki/Key_generation