A Modified Profile-Based Location Caching with Fixed Local Anchor for Wireless Mobile Networks

Transcription

1 A Modiied Proile-Based Location Caching with Fixed Local Anchor or Wireless Mobile Networks Md. Kowsar Hossain, Tumpa Rani Roy, Mousume Bhowmick 3 Department o Computer Science and Engineering, Khulna University o Engineering and Technology (KUET) Khulna, Bangladesh auvikuet@yahoo.com tumparoy_st@yahoo.com 3 mousumicsekuet@yahoo.com Abstract Proile-Based Location Caching with Fixed Local Anchor (PCFLA) Strategy can reduce the requent access to the HLR. As a result, the total management cost is minimized in a wireless mobile network. However, i the s inormation is not ound at calling MSC then the HLR is accessed to get that inormation every time. Middle Location Register () is located between VLR and HLR. It helps to reduce the heavy load at HLR. Modiied PCFLA (MPCFLA) combines the with PCFLA to improve the perormance o PCFLA. The analytical results show that our proposed method reduces the overall cost than PCFLA strategy. Keyword- call delivery, location management, location registration, Middle Location Register, Proile- Based Location Caching with Fixed Local Anchor I. INTRODUCTION It is possible or a mobile terminal () to be called by other s, or to call other s, as it moves in wireless cellular networks such as Personal Communications Service (PCS) networks. When an calls another then the location o the called must be identiied beore the connection is established. Thereore, a location management strategy is necessary to keep track o the s eectively and locate a called when a call is initiated []. Location management strategy must handle two user location operations eiciently: location registration and call delivery. The ormer one is the process o inorming the network about the s current location inormation; whereas the latter one is the process o determining the serving VLR and the cell location o the called prior to connection establishment between the caller and called s. There are two common standards used or location management []: IS-4and GSM. Both are based on the two-level database hierarchy, which consists o Home Location Register (HLR) and Visitor Location Registers (VLRs). Fig. [9] shows the basic architecture o the wireless mobile networks under this two-level database hierarchy. The parameters used in Fig. are described in Table I.The location inormation and permanent user proile or each is stored and maintained by HLR, and the VLRs are distributed throughout the network to store location inormation o each currently residing in the Registration Area (RA). The task o Mobile Switching Centers (MSCs) is to provide switching unctions or the s in their associated RAs. In addition, the MSCs o dierent RAs are connected to a two-level hierarchy o Signaling Transer Point (STP). This STP is comprised o a Regional STP (RSTP) and Local STPs (LSTP). These STPs are responsible or routing o signaling messages based on their destination addresses. These network elements are interconnected by wireline links. As the number o the s increases, location management under the IS-4 standard has suered problem like increasing traic in the network. An extensive work has been done to overcome this problem under the IS-4 []-[9]. Among those strategies PCFLA strategy is one o the modern existing strategies [9]. In PCFLA, user proiles are eectively utilized to determine at which sites throughout the networks user s location inormation should be cached. In this approach, these site lists are prepared based on the long-term call related statistics maintained by the HLR rom the callee s user proile. These lists are used to store the callees location inormation to some o the most requently calling VLR caches o the corresponding callees.\ The location inormation o the called user is obtained at the same VLR i the calling and called users are within the same VLR area during call processing. I the two users are not in the same VLR area then the location query should reach HLR, which causes a serious bottleneck at HLR. To reduce the bottleneck problem, database is proposed in [0] which covers some VLRs. is a multi-stage system. So some s are managed by a next stage. MAIN database is the root o the hierarchy. I the calling and called users are not in the same VLR area then a location query is passed to the corresponding. I the has the ISSN : Dec 0- Jan 0 404

2 inormation o called user then the query is resolved there. Otherwise, the procedure repeats and inally the MAIN database is accessed. Thus, reduces the heavy load at HLR. c h HLR Cra Remote A-link D-link RSTP local A-link LSTP C d LSTP C v C la VLR MSC Fig.. Network cost model. In this paper, the system is embedded into the PCFLA strategy. In PCFLA, i the called FLA s inormation is not cached in calling MSC then HLR is accessed to get the called s inormation. As the numbers o s are increasing, this creates heavy load at HLR. I the system is used between HLR and VLR then the s location inormation can be ound at. So, the load at HLR will be reduced. As the PCFLA minimizes the total location management cost and system reduces the heavy load at HLR, by combining the with PCFLA, reduces the overall cost and the load at HLR more eectively than PCFLA and. TABLE I Description o symbols Symbol Description () Corresponding message number [] Cost or the particular signaling exchange {} Cost or accessing the particular database Exchange o the particular signaling message Acknowledgement o the signaling message The rest o the paper is organized as ollows. Section II provides an overview o the existing related research work. Our proposed approach is described in section III. Section IV provides the analytical modeling. Numerical results and comparison among dierent approaches based on some experimental results are described in section V. We provide a concluding remark in section VI. II. EXISTING RELATED WORK A large number o works have been done reported on location management to reduce the overall location management cost in terms o location registration cost and call delivery cost [] [9]. Some o them are basic scheme which are generally used to manage the location irrespective o all the wireless networks. Some others are based on reusing the user location inormation obtained during the previous call to the user to reduce the call delivery cost. On the other hand, some others manage the local hando locally, instead o inorming the centralized HLR. Thus the location registration cost is reduced. There are also some methods which use the s calling statistics rom the HLR and replicate its location inormation to these calling VLR cache. These also manage the local hando locally instead o accessing the heavily congested HLR. This reduces both the location registration and call delivery cost. Some existing location management strategies are shown in Fig Fig 7. The symbols used in Fig Fig 7 are described in Table I. The signaling and database access costs or the cost analysis is described in Table II. The location management procedures o these strategies are described in the ollowing subsections. ISSN : Dec 0- Jan 0 405

3 Parameter C la C d C ra C v C C h C m C mlr C tmlr TABLE II Description o cost parameters Description Cost or sending a signaling message through the local A-link Cost or sending a signaling message through D-link Cost or sending a signaling message through the remote A-link Cost or a query or an update o the VLR Cost or a query or an update o the FLA Cost or a query or an update o the HLR Cost or a query or an update o the Cost or sending a signaling message the Cost or sending a signaling message to the top rom A. IS-4 Standard In basic IS-4 standard [], each inorms its location inormation to the HLR during all type o hando procedures. When a call is initiated, the called s location inormation is searched in the HLR. New HLR Old [C la +C d +C ra ] () [C la +C d +C ra ] (3) [C la +C d +C ra ] (4) [C la +C d +C ra ] (5) [C la +C d +C ra ] {C v } {C h } {C v } Fig.. Location registration under the IS-4 standard. The location registration procedure o this strategy can be described using the ollowing steps (see Fig. ). An enters into a new RA and inorms its new location to the new MSC through the nearby BTS. () The MSC updates its associated VLR about this newly entered. Ater that, it sends a location registration message to the HLR. (3) Ater receiving the location registration message, the HLR updates the s record and sends back a registration acknowledgement message to the new VLR. (4) It also sends a registration cancellation message to the old VLR. (5) The old VLR removes the record o the immediately and sends back a cancellation acknowledgement message to the HLR. Calling HLR [C la +C d +C ra ] () [C la +C d +C ra ] (3) [C la +C d +C ra ] (5) [C la +C d +C ra ] (4) [C la +C d +C ra ] (6) Call Setup {C v } {C h } {C v } Fig. 3. Call delivery under the IS-4 standard. On the other hand, the call delivery procedure under this strategy is described as ollows (see Fig. 3). The calling initiates a call and sends a signal to its serving MSC through a nearby base station. () The calling MSC sends a request message to the HLR or the called s location inormation. (3) The HLR determines the called s current serving MSC and sends a location request message to that MSC. (4) The MSC allocates a Temporary Local Directory Number (TLDN) [9] to the and sends back a reply to the HLR together with the TLDN. (5) The HLR propagates this inormation to the calling MSC. (6) The MSC sends a call setup request message to the called MSC through the network shown in Fig.. ISSN : Dec 0- Jan 0 406

4 B. PCFLA Strategy PCFLA strategy [9] is a combination o LC (Location Caching) strategy and FLA (Fixed Local Anchor) strategy. It takes the advantages o both LC strategy and FLA strategy. In PCFLA strategy, each LSTP has a FLA (Fixed Local Anchor) which is actually a speciic VLR. The FLA has a table which keeps the current serving VLR or all s in its LSTP region. When an enters into a new RA then the MSC o the new RA registers the s location at the FLA. The HLR then gets the inormation about the new FLA. The FLA may be the current serving VLR or the other VLR within that RA. However, all the user proiles are stored and manage in the current HLR in the PCS systems which contains the user inormation such as a user s location, authenticate inormation, and calling activity log, etc. New Old () [C la ] (3) [C la ] (4) [C la ] (5) [C la ] {C v } {C } {C v } (a) Intra- LSTP movement New New () [C la ] ( 6.) [C la +C d +C ra ] HLR Caching MSC Old Old (6.3) [C la ] ( 6.) [C la +C d +C ra ] ( 6.4) [C la +C d +C ra ] ( 6.5) [C la +C d +C ra ] ( 6.6) [C la +C d +C ra ] (6.7) [C la ] (6.8) [C la ] {C v } {C } {C h } {0} {C } {C v } (b) Inter- LSTP movement Fig. 4. Location registration under the PCFLA strategy. The location registration procedure under the PCFLA strategy is described in the ollowing steps (see Fig. 4 (a)). An sends a location update registration signal to the new MSC through the nearby BS ater entering to the new RA. () The MSC sends a location registration message to its designated FLA in its LSTP region to check whether it has the s proile. (3) I the has just moved into a new LSTP region then s record will not be ound in the FLA. In that case, go to step (6). On the other hand, i the FLA contains the s record then, it updates the s previous record with the new one which indicates the associated new VLR. The FLA then sends a registration acknowledge message to the new MSC together with a copy o the s current record. (4) Ater that, The FLA sends a message to the old MSC to remove the s record. (5) The old MSC removes the move- out s inormation rom its associated VLR and sends an acknowledge message to the FLA to inorm that the deletion is complete. (Location registration by intra-lstp movement is complete. Do not continue to the next step). (6) I there is no s record in FLA then the ollowing steps are perormed (see Fig. 4 (b)). 6.) The MSC associated with the s new FLA sends a message to the HLR or location registration. 6.) As the moves to a new FLA, the HLR updates the s record with that inormation and sends a copy o the s updated proile to the new FLA. 6.3) The FLA updates its table by the s record to indicate that the associated new VLR and sends a copy o the s proile with a registration acknowledgment message. 6.4) All the MSCs those have location cache or the should be updated. For this reason, the HLR sends location cache update messages to those MSCs. ISSN : Dec 0- Jan 0 407

5 6.5) The HLR sends a message to the s old FLA to cancel the registration o the, which then removes the s record. 6.6) In response o the registration cancellation message rom the HLR, the old FLA sends a registration cancellation acknowledgment message back to the HLR. 6.7) The old FLA sends a registration cancellation message to the s old VLR in order to remove the s obsolete record. 6.8) Ater removing the s record, the old VLR sends a registration cancellation acknowledgment message back to the old FLA (Location registration by inter-lstp movement is complete). Calling () [C la ] or [C la +C d ] (3) [C la ] (4) [C la ] or [C la +C d ] (5) Call setup {C v } {C } {C v } (a) When calling MSC has location cache or the called FLA Calling HLR ( 6.) [C la +C d +C ra ] ( 6.) [Cla +C d +C ra ] (6.3) [C la ] ( 6.5) [C la +C d +C ra ] (6.6) Call setup ( 6.4) [C la +C d +C ra ] {C v } {C h } {C } {C v } (b) When calling MSC does not have location cache or the called FLA Fig. 5. Call delivery under the PCFLA strategy. The call delivery procedure under the PCFLA strategy is described as ollows (see Fig. 5 (a)). The calling initiates a call and sends a signal to its serving MSC through the nearby base station. () The calling MSC checks its cache or the location o the called s FLA. I the location is ound then it sends a location request message to the called FLA. Otherwise, go to the step (6). (3) The called FLA then orwards the location request message to the called MSC or the called. (4) The called MSC allocates TLDN to the and sends a reply back to the calling MSC together with the TLDN. (5) The calling MSC sends a message or call setup to the called MSC through the network show in Fig. (Call delivery is complete. Do not continue to the next step). (6) I the calling MSC s cache does not contain the location o the called FLA, the ollowing occurs (see Fig. 5 (b)). 6.) The calling MSC sends a request message to the HLR or the location o called. 6.) The HLR sends a location request message to the called FLA. 6.3) The called FLA then orwards this message to called MSC. 6.4) The called MSC allocates TLDN to the and reply back to the HLR together with the TLDN. 6.5) Ater receiving this inormation, the HLR orwards it to the calling MSC. 6.6) The calling MSC send a message or call setup to the called MSC through the network shown in Fig. (Call delivery is complete). C. Strategy Fig. 6 shows the hierarchical structure o system[0]. Here some VLRs are managed by a. This s are connected to top level which in turn is managed by immediate top level. MAIN is the top level database o the system. Actually is one kind o cache. Generally it manages the location inormation o users in the area. Even i the calling and called users are not in the same VLR area, the location query needs not to be connected to HLR since s are added to manage the location inormation in the middle stage and removes the search messages between VLR and HLR [0]. The total cost or the normal ISSN : Dec 0- Jan 0 408

6 scheme using HLR is calculated as *C VLR +C HLR. However, the total cost o the system can be calculated using *C VLR +C + P Cross *C. Where P Cross is the probability that the called and calling are not in the same area. Fig. 7 shows the two call delivery situations. and calling s are not in the same areas (at dotted circle containing two short arrows). () Two s are in the same area (thin dotted circle containing a long arrow). MAIN VLR VLR VLR VLR VLR VLR VLR VLR Fig. 6. Hierarchical structure o system. Location registration procedure in the system is given below which has been proposed in [0]. MSC sends location registration messages to HLR. () HLR sends the messages to MSC. The inormation in the messages is recorded at VLR. (3) VLR sends the messages to and Main. The messages have the inormation o mobile terminal, the location o mobile terminal and VLR. The inormation is recorded at s and Main. and Main are mapping each other. Calling Calling MAIN {C VLR } {C } {C MAIN } {C VLR } Fig. 7. Call delivery under the system. The call delivery procedure is as ollows (see Fig. 7). Calling mobile terminal sends a message to to know the location inormation o the called mobile terminal. () sends location messages o called terminal to MSC i it has the inormation. I the does not have the location inormation, the passes the messages to the next stage. At last, the messages are sent to the MAIN to know the location inormation. (3) MSC receives messages rom. In the reverse order, the call is connected to the called terminal. ISSN : Dec 0- Jan 0 409

7 III. PROPOSED STRATEGY HLR RSTP LSTP LSTP MSC Fig. 8. Network cost model or MPCFLA In the PCFLA strategy, there is VLRs and HLR database or location management. So when an calls another and i the calling and called are not in the same LSTP region then HLR database is the ultimate place to ind inormation about called. system is used into existing PCFLA strategy to reduce the load at HLR. I scheme is used then there is no need to go to HLR or inormation about called every time when the called and calling s are not in the same LSTP region. Here, the system is added between LSTP and RSTP region. As a result, the signaling cost is reduced during call delivery and total cost is reduced also. I inormation is not ound in a then there is a probability to ind the inormation at top which is indicated by P tmlr. The network cost model or Modiied PCFLA (MPCFLA) is shown in Fig. 8. New Old () [C la ] (3) [C la ] (4) [C la ] (5) [C la ] {C v } {C } {C v } (a) Intra- LSTP New associated S New HLR Caching MSCs Old associated S Old () ( 6.) [C la +C d +C ra ] (6.3) [C la ] (6.4) [C mlr + (n-)c tmlr ] ( 6.) [C la +C d +C ra ] ( 6.5) [C la +C d +C ra ] ( 6.6) [C la +C d +C ra ] ( 6.7) [C la +C d +C ra ] (6.8) [C mlr + (n-)c tmlr ] (6.9) [C la ] (6.0) [C la ] {C v } {C m } {C } {C h } {0} {C } {C m } {C v } (b) Inter- LSTP movement Fig. 9. Location registration under MPCFLA The location registration procedure under the MPCFLA strategy is described as ollows (see Fig. 9 (a)). An moves to the new RA and sends a location update registration to the new MSC through the nearby BS. ISSN : Dec 0- Jan 0 40

8 () The MSC sends a location registration message to its designated FLA in its LSTP region. (3) The FLA checks or the s proile. I there is not an s record in the FLA, which means that the has just to move into a new LSTP region, then go to step (6). Otherwise, it updates the s record to indicate the associated new VLR, and sends a registration acknowledge message to the new MSC together with a copy o the s proile. (4) The FLA sends a registration cancellation message to the old MSC. (5) The old MSC removes the record o an at its associated VLR and sends a registration acknowledge message to the FLA. (Location registration by intra-lstp movement is complete. Do not continue to the next step). (6) I there is no s record in FLA, the ollowings are perormed (see Fig. 9 (b)). 6.) The MSC associated with the s new FLA sends a location registration message to the HLR. 6.) The HLR updates the s record to indicate the s new FLA and sends a copy o the s proile to the new FLA. 6.3) The FLA also updates the s record to indicate that the associated new VLR and sends a registration acknowledgment message to the new VLR together with a copy o the s proile. 6.4) The FLA sends the inormation messages about the to all associated s. 6.5) The HLR sends a location cache update message to the MSCs which have location caches or the, which updates the location caches to indicate the s new FLA. 6.6) The HLR sends a registration cancellation message to the s old FLA, which remove the s record. 6.7) The old FLA sends a registration cancellation acknowledgment message back to the HLR. 6.8) The old FLA sends a registration cancellation message to the s old associated s, which remove the s record. 6.9) The old FLA sends a registration cancellation message to the s old VLR, which remove the s record. 6.0) The s old VLR sends a registration cancellation acknowledgment message back to the old FLA (Location registration by intra-lstp movement is complete) Calling () [C la ] or [C la + (C mlr +P tmlr *C tmlr )] (3) [C la ] (4) [C la ] or [C la + (C mlr +P tmlr *C tmlr )] (5) Call Setup {C v } {C } {C v } (a) When calling MSC has location cache or the called FLA Calling (6.) [C la + C mlr + P tmlr * C tmlr ] (6.5) [C la + C mlr + P tmlr * C tmlr ] Top (6.6)Call Setup {C v } {C m } {C m } {C } (6.) [C la + C mlr + P tmlr * C tmlr ] (6.3) [C la ] (6.4) [C la + C mlr + P tmlr * C tmlr ] {C v } (b) When calling MSC does not have location cache or the called FLA Fig. 0. Call delivery under MPCFLA The call delivery procedure under the MPCFLA strategy is described as ollows (see Fig. 0(a)). The calling sends a call initiation signal to its serving MSC through the nearby BS. () The calling MSC checks i it has location cache which indicates the FLA o the called. I yes, it sends a location request message to the called FLA. Otherwise, go to the step (6). ISSN : Dec 0- Jan 0 4

9 (3) The FLA orwards the location request message to the called MSC. (4) The called MSC allocates TLDN to the and sends it to the calling MSC. (5) The calling MSC requests a call setup to the called MSC through the through the network shown in Fig. 8 (Call delivery is complete. Do not continue to the next step). (6) I the calling MSC does not have location cache or the called FLA, the ollowing occurs (see Fig. 0(b)). 6.) The calling MSC sends a location request message to the. 6.) The sends a location request message to the called FLA. 6.3) The called FLA orwards a location request message to called MSC. 6.4) The called MSC allocates TLDN to the and sends it to the. 6.5) The orwards this inormation to the calling MSC. The calling MSC requests a call setup to the called MSC through the network shown in Fig. 8 (Call delivery is complete). In Fig. 0, the dotted circle indicates that i the inormation is not ound at calling then the inormation may be ound at top s. IV. ANALYTICAL MODEL In this section, a luid low mobility model [] is used to evaluate the perormance o our proposed strategy and PCFLA strategy. It is assumed that s are moving at an average speed o v in uniormly distributed direction over [0, π] with a view to crossing the LSTP region composed o N equal rectangular-shaped and sized RAs [9]. The ollowing parameters used or the s movement rates analysis. : the s movement rate out o an RA : the s movement rate out o an LSTP region : the s movement rate to an adjacent RA within a given LSTP region According to [], these parameters are calculated as ollows. 4v S 4v () NS (3) N Where v is the average moving speed o an, S is the size o the RA, and N is the number o RAs within a LSTP region. A continuous-time Markov Chain state transition diagram is used to show an s RA movement. It is shown in Fig. which represents the luid low mobility model. Each state i( i 0) deines as the RA number o a given LSTP region where an can stay and state 0 means the stays outside o this region. The state transition a j i ) represents an s movement rate to an adjacent RA within a given LSTP region, and ( i, a0, represents an s movement rate to an RA o that region rom another one. On the other hand, b i, 0( i ) represents an s inter-lstp region movement rate and it is assumed that there are maximum K number o such movements. Thereore, rom Figure, it is obtained that a i, j ( i ) and a 0, b i, 0, respectively. a 0, a, a k-,k K b,0 b,0 b k,0 Fig.. State transition diagram o an s RA movement On the other hand, assuming that is the equilibrium probability o state i, the ollowing equations can be obtained rom a continuous-time Markov Chain given in Figure. i ISSN : Dec 0- Jan 0 4

10 K 0 i i i i, i 0 i i, i K, i i i K According to the law o the total probability, the sum o the probabilities o all states is. So, K 0 K i i0 By substituting (8) into (4), it can be obtained the quilibrium probability o state 0, 0.So, 0 Finally, rom (5), (6), (7) and (9), is obtained as ollows. i i i i i i 0 i i K i i K (4) (5) (6) (7) (8) (9) (0) In this section, based on the user mobility model, we derive the location registration costs, the call delivery costs and the total location management costs o the MPCFLA strategy. Table III [9] shows the various parameters used or the cost analysis o MPCFLA strategy. P l P n P o P P c q m TABLE III Parameters used or the cost analysis. Probability (caller and callee are located within the same LSTP region) Probability (new VLR is the FLA) Probability (old VLR is the FLA) Probability (callee is located in the FLA area) Probability (calling MSC has location cache or the called FLA under the MPCFLA strategy) The s CMR (call-to-mobility ratio) The number o the MSCs which have location caches or the called FLA under the MPCFLA strategy (= the number o the most requently calling VLRs to the speciy) Table II and III describe dierent parameters that are used in order to calculate location management cost, signaling cost. The ollowing notations are also used to represent the cost o each strategy [9]. U X : The average location registration cost o the X strategy or an staying in an LSTP region. S X : The average call delivery cost o the X strategy or an staying in an LSTP region. T X : The average total location management cost o the X strategy or an staying in an LSTP region. Y U X : the average location registration cost o the X strategy generated by movement type Y or an staying in an LSTP region From Fig., the average number o unique RAs that an visits within a given LSTP or K movements which can be represented by the ollowing equation. ( K) K i 3 3 K K i i () The average location registration cost under the MPCFLA strategy can be expressed as [9]: U = int er int ra MPCFLA 0 * U + ( (K )-) * MPCFLA U MPCFLA ISSN : Dec 0- Jan 0 43

11 int U er MPCFLA int U ra MPCFLA int U er MPCFLA Where M and int ra and U MPCFLA can be expressed as ollows: p p0 M p p M p p M p p0 M p0 pn M pn p0 M3 pn p0 M M = n 0 n n 0 4 x = n 4 x (x ε {,, 3, 4}) mean the location registration costs or intra-lstp movement and inter- LSTP movement under MPCFLA strategy in case o (x) given in Table IV [9], respectively. And, these costs can be expressed as ollows (see Fig. 9(a) and 9(b)): M = M = ** Cla Cv C (5) M 4 4** Cla * Cv C M = M (6) = 4 ClaCd Cra4* Clam* ClaCd Cra* Cmlr( n) Ctmlr Cv C Ch ( ( n)) Cm (7) 4 C C C m* C C C * C ( n) C C C (( n)) C M 3 = la d ra la d ra mlr tmlr h m M = 4 ClaCd Cra8* Clam* ClaCd Cra* Cmlr( n) Ctmlr C v C Ch (( n)) Cm 4 (3) (4) (8) (9) When the inter-lstp movement o the occurs, the should update m location caches or that throughout the networks. In this case, we consider the signaling cost or sending cache update messages, but ignore the cache access cost since it is relatively too small as compared to the database access cost []. TABLE IV Four possible cases or location registration under MPCFLA strategy. Case Description New VLR is the FLA, and old VLR is not the FLA. () Old VLR is the FLA, and new VLR is not the FLA. (3) Both new VLR and old VLR are the FLAs. (4) Both new VLR and old VLR are not the FLAs. On the other hand, the average call delivery cost under the MPCFLA strategy is expressed as ollows [9]: S MPCFLA = P c * C cache MPCFLA + ( - P c) * C nocache MPCFLA cache Where C represents the call delivery costs under the MPCFLA strategy when the calling MSC has MPCFLA nocache location cache or the called FLA, and C means the call delivery costs when the calling MSC does not MPCFLA have location cache or the called FLA, respectively. There are our possible cases shown in Table V [9]. TABLE V Four possible cases or call delivery when the calling MSC has location cache or the called FLA under the MPCFLA strategy Case Description The caller and called are located within the same LSTP region, and the called is ound in the FLA area. () The caller and the called are located within the same LSTP region, and the called is ound in the other VLR area, not the FLA area. (3) The caller and the called are located in the dierent LSTP regions, and the called is ound in the FLA area. (4) The caller and the called are located in the dierent LSTP regions, and the called is ound in the other VLR area, not the FLA area. According to the our possible cases given in Table V, these costs are as ollows (see Fig. 0(a) and Fig. 0(b)): = p P MPCFLA N P * N Pl * P * N3 P C cache * N * where N x (x ε {,, 3, 4}) means the call delivery costs when the calling MSC has location cache or the called FLA under the MPCFLA strategy in case o case (x) given in Table V. These costs can be expressed as ollows (see Fig. 0(a)): N * C la C v C 4 (0) () ISSN : Dec 0- Jan 0 44

12 N N 3 3*C * C la v C * Cla * C mlr Ptmlr C tmlr C v C * C la * C mlr Ptmlr * C tmlr * C v C * C la * * N 4 * C nocache P * D D P * = MPCFLA Where D means the call delivery costs when the called is ound in its FLA area, and D represents the call delivery cost when called is ound in the other VLR area, respectively. These costs are as ollows (see Fig. 0(b)): D 4* C la Cmlr Ptmlr * Ctmlr C v C Cm D 4* Cla CmlrPtmlr* Ctmlr* Cv C Cm * Cla Finally, the average total cost o the MPCFLA strategy can be expressed as T = MPCFLA U + q * MPCFLA S MPCFLA V. NUMERICAL RESULTS AND COMPARISONS In this section, we compare the perormance o the MPCFLA strategy with that o the PCFLA strategy. For perormance analysis, we deine the relative cost o MPCFLA and PCFLA in the ollowing way Relative cost = T MPCFLA / T MIS-4 Relative cost = T PCFLA / T IS-4 For analysis purpose, N is set to 55. We set P n = P o = P = /55 = 0.08 as the number o the RA s per LSTP region is 55 [], [3], [5]. Similarly, since the number o the LSTPs per region is 60/7 = 3, the value o P l is /3= For simplicity we assume P c = 0.7, n= and m = 5 [9]. We also consider the total cost o IS-4 and MIS-4 is and P tmlr is 0.4.The value o v and S is considered as 5.6 km/h and 0 km []. The signaling cost, database access cost and total cost or MPCFLA strategy are compared to PCFLA strategy which is shown graphically in the ollowing subsection by taking values rom Table VI and Table VII. TABLE VI Signaling costs parameter set. Set C la C d C ra C mlr C tmlr (3) (4) (5) (6) (7) (8) (9) (30) (3) A. Signaling cost TABLE VII Database access costs parameter set. Set C v C C h C m (a) Set ISSN : Dec 0- Jan 0 45

13 (b) Set () Fig.. Signaling Costs At irst we determine the signaling cost which dominates by setting the database access cost parameters, C h, C v, C and C m to be 0. Set and () rom Table VI is used to calculate signaling costs or both MPCFLA and PCFLA. As system resides between VLR and HLR. So, C mlr and C tmlr must be less than C d. Here two is used. For this reason C mlr = 0.3 and C tmlr = 0.3 is considered. Fig. shows the comparison o Signaling costs between MPCFLA and PCFLA. It shows that the relative cost o MPCFLA is less than PCFLA. The dierence increases with increasing the CMR value. In MPCFLA, the inormation can be ound at and HLR access is not needed so our proposed approach results better output. B. Database access cost For calculating database access cost we assume signaling cost is zero. We consider the set (3) and set (4) rom Table VII to calculate Database access cost by setting C la, C d, C ra, C mlr and C tmlr to zero. (a) Set (3) (b) Set (4) Fig. 3. Database Access Costs Fig. 3 shows that the database access cost remains almost same or set (3) or all CMR values. But or set (4), it remains same at irst but with increasing CMR value the dierence is evident and it is clear that the cost o MPCFLA is less than PCFLA. The is one type o cache so the cost is very low or updating the so we set C m is 0.5 (set 3) and (set 4). As compared with HLR database the access cost o database is very small that is why our proposed approach gives better result than PCFLA. ISSN : Dec 0- Jan 0 46

14 C. Total cost The total cost is determined by considering set and set (3) in Fig. 4(a) and set () and set (4) in Fig. 4(b). In both cases MPCFLA shows better result than PCFLA. When the CMR value is small then the output o MPCFLA and PCFLA is approximately same. But MPCFLA perorms well than PCFLA with increasing the CMR value. The signaling cost and database access cost is reduced because o the system that is embedded in PCFLA strategy. Thus, the total cost is reduced in MPCFLA than PCFLA. (a) Set and Set (3) (b) Set () and Set (4) Fig. 4. Total Costs VI. CONCLUSION Now a days, Location management in wireless mobile networks is one o the most important and challenging issues in the current world. To perorm eicient location management an eective strategy called MPCFLA is proposed in this paper. This strategy uses system with PCFLA strategy. It reduces the access o HLR when the inormation o a called user is not ound at calling MSC. As a result, it minimizes the total location management cost in terms o location registration cost and call delivery cost regardless o the CMR. The numerical results show that proposed MPCFLA strategy outperorms the PCFLA strategy irrespective o the s calling and mobility pattern. REFERENCES [] Akyildiz, J. McNair, J. Ho, H. Uzunalioglu, and W. Wenye, Mobility management in next-generation wireless systems, Proc. IEEE, vol.87, no.8, pp , Aug [] R. Jain, Y.B. Lin, and S. Mohan, A caching strategy to reduce network impacts o PCS, IEEE J. Sel. Areas Commun., vol., no.8, pp , Oct [3] J. Ho and I. Akyildiz, Local anchor scheme or reducing signalling costs in personal communications networks, IEEE/ACM Trans. Netw., vol.4, no.5, pp , Oct [4] E. Bae, M. Chung, and D. Sung, Perormance analysis o ixed local anchor scheme or supporting UPT services in wireline networks, Comput. Commun., vol.3, pp , 000. [5] R. Jain and Y.B. Lin, An auxiliary user location strategy employing orwarding pointers to reduce network impacts o PCS, ACMBaltzer J. Wireless Networks, vol., no., pp.97 0, July 995. [6] G. Varsamopoulos and S. Gupta, Dynamically adapting registration areas to user mobility and call patterns or eicient location management in PCS networks, IEEE/ACM Trans. Netw., vol., no.5,pp , Oct [7] W. Ma and Y. Fang, A pointer orwarding based local anchoring (POFLA) scheme or wireless networks, IEEE Trans. Veh. Technol., vol.54, no.3, pp.35 46, May 005. [8] J. Li and Y. Pan, A dynamic HLR location management scheme or PCS networks, Proc. INFOCOM 004, vol., pp.66 76, March 004. ISSN : Dec 0- Jan 0 47

15 [9] K. Kong. Perormance analysis o proile-based location caching with ixed local anchor or next-generation wireless networks. IEICE transactions on communications, E9-B(): , November 008. [0] Eui-Jong Jeong, Goo-Yeon Lee, "Eicient Location Management Scheme using considering Local Usages," itng, pp.94-94, International Conerence on Inormation Technology (ITNG'07), 007. [] F. Baumann and I. Niemegeers. An evaluation o location management procedures. In Proceedings o the Third Annual International Conerence on Universal Personal Communications, USA, pp , September 994. [] K. Kong, M. Song, K. Park, and C. Hwang. A comparative analysis on the signaling load o Mobile IPv6 and Hierarchical Mobile IPv6: Analytical approach. IEICE Transactions on Inormation and Systems, E89-D:39 49, Japan, January 006. ISSN : Dec 0- Jan 0 48