Adaptive Load Balance and Handoff Management Strategy for Adaptive Antenna Array Wireless Networks

Size: px

Start display at page:

Download "Adaptive Load Balance and Handoff Management Strategy for Adaptive Antenna Array Wireless Networks"

Erica Benson
5 years ago
Views:

1 Adaptive Load Balace ad Hadoff Maagemet Strategy for Adaptive Atea Array Wireless Networs Jog-Shi Che 1 We-Chi Chiag 2 Neg-Chug Wag 3 Yug-Fa Huag Graduate Istitute of Networig ad Commuicatio Egieerig Chaoyag Uiversity of Techology, Taichug 413, Taiwa, R.O.C. 3 Departmet of Computer Sciece ad Iformatio Egieerig Natioal Uited Uiversity, Miao-Li 360, Taiwa, R.O.C. Abstract: Wireless etwors that employ adaptive atea array (AAA) have bee proposed to icrease the traffic-carryig capacity ad circuit quality. The AAA divided a cell ito several areas. Variatios i the traffic loads amog areas will lesse the traffic-carryig capacity. Moreover, the hadoff procedure usually taes place whe the call crosses the area boudary. A larger umber of areas will icrease the system overhead for hadlig the hadoff procedure. A ieffective maagemet will icrease the system overheads, such as code switch, data switch, ad eve etwor switch. The ivestigatio proposes a effective load balace ad hadoff maagemet strategy. These strategies are implemeted to solve traffic-adaptio problem that ca ehace the traffic-carryig capacity for variatios i traffic. For the maagemet of hadoff procedure, our strategy cosiders the mobility of mobile hosts ad the badwidth utilizatio. It ca decrease the umber of hadoff procedures ad lesse the system overhead. Furthermore, the simulatio results are preseted to cofirm the efficiecy of the proposed strategy. Key-Words: Adaptive atea array, wireless etwor, load balace, traffic-adaptatio, ad hadoff procedure. 1 Itroductio Techological advaces ad rapid developmet i hadheld wireless termials have facilitated the rapid growth of wireless commuicatios. Sice this tremedous growth of wireless commuicatio requiremets is expected uder the costrait of limited badwidth. The AAA framewors that ca provide more flexible to hadle the limited badwidth will be the maistream for wireless etwors [5, 15]. The geographical area of a system is covered by cells. I each cell, a base statio usig a AAA i the ceter supports the wireless commuicatios. Each AAA has a umber of sesors ad each sesor has a umber of the system codes used to establish commuicatios sessios i a part area of this cell. Accordigly, the umber of codes allocated to a sesor will affect the commuicatio quality i this service area ad the allocatios of system codes amog sesors will affect the traffic-carryig capacity of a system [2, 8, 11]. A reasoable allocatio should provide more codes to a sesor with heavy traffic tha a sesor with light traffic. Otherwise, it will experiece that the heavy traffic sesors do ot have sufficiet codes to carry their traffic loads but the light traffic sesors have may available codes. Thus, the traffic-carryig capacity of a system is reduced ad the call blocig probability arises. To cosider real-life etwors, the traffic distributios amog sesors should be chageable accordig to various coditios. I order to achieve higher code utilizatio, whe there are variatios i traffic, the code allocatios amog sesors should be effectively reallocated accordig to curret traffic profile [3-4, 13-14]. A cell is divided ito several areas accordig to the service areas of sesors. Two eighborig sesors ca ot cotais the same codes. Otherwise, the sesors that use the same codes to establish commuicatio sessios will iterfere with other. Whe a call arrives at a area, the codes of the targeted sesor ca be used to hadle the call. The problem i this assigmet is o high-mobility mobile hosts (MH) that usually the area boudary. It will icrease the umber of hadoff procedures. The hadoff procedure usually taes place whe the call crosses the cell boudary. I geeral, the hadoff procedure icludig data trasmissio, code switchig, ad eve etwor switchig taes tes or hudreds ms. A effective decrease of the umber of hadoff procedures that ca lesse the system overhead is meaigful [1-2, 6, 8-10, 12]. I light of above discussios, this study presets a adaptive load balace ad hadoff maagemet strategy for AAA wireless etwors. This strategy ca dyamically allocate codes amog sesors accordig to variatios i traffic to solve the ISSN: ISBN:

2 traffic-adaptive problem ad, accordigly, ehace the traffic-carryig capacity of cellular systems. For the hadoff maagemet, the assigmet to calls cosiders the mobility of MHs to decrease the umber of itra-hadoff. The rest of this paper is orgaized as follows. Sectio 2 itroduces the system model ad the correlatio research. I sectio 3, we describe the proposed strategy. The umerical results are give i Sectio 4. Coclusios are fially offered i Sectio 5. 2 Model & Correlatio Research Herei, first we itroduce the system model. The, use the defiitio of system model to describe the previous research cotets. 2.1 System Model The geographical area of a system is covered by cells. I each cell C i, a base statio usig a adaptive atea array (AAA) i the ceter supports the wireless commuicatios. Each AAA has m sesors ad each sesor ca charge a regio i C i to provide commuicatio service. For coveiece, C i,1, C i,2,, C i,m respectively represets the service areas of the m sesors i C i. The maagemet of the system badwidth is based o code divisio multiple access (CDMA) techique, where each cell ca acquire all the system badwidth B[7, 11]. Each sesor S i, has a part B(S i, ) of system badwidth B used to provide service of C i,. A commuicatio sessio (or a call) ca be established if available codes ca be allocated for supportig the commuicatio betwee the mobile host ad the sesor S i,. Two sesors caot cocurretly assig the same code to calls if their geographical distace is less tha the miimum reuse distace D mi ; otherwise, their commuicatio sessios will iterfere with each other. This situatio is referred to code iterferece [2, 12]. Defiitio 1: Give a sesor S i, i cell C i, the set of iterferig eighbors of S i,, deoted by IN(S i, ), is: IN(S i, ) = {S i, S i, C i ad D(S i,, S i, )<D mi }, where D(S i,, S i, ) is the geographical distace betwee C i, ad C i,. deotes the iterferig eighbors IN(S i, ) of a sesor S i,. Whe S i, assigs a code to a icomig call, it must esure that this code is ot cocurretly assiged to other calls i IN(S i, ). Code allocatio to sesors must esure if a code c is allocated to S i,, o other sesors i IN(S i, ) ca eep c as their code. Therefore, Defiitio 2 is the coditio of code allocatio. Defiitio 2 (The coditio of code allocatio): Give two distict sesors S i, ad S i, i cell C i, where S i, IN(S i, ), the coditio of code allocatio betwee S i, ad S i, is B(S i, )B(S i, )=. For istace i Fig. 1, the area of cell C i was divided to 12 subareas: C i,1, C i,2,, ad C i,12 accordig to 12 sesors: S i,1, S i,2,, ad S i,12. If D mi is 2 subareas away, IN(S i,3 )={S i,1, S i,2, S i,4, S i,5 }. Whe S i,3 was allocated to a B(S i,3 ), S i,1, S i,2, S i,4, ad S i,5, their allocated codes that respectively are B(S i,1 ), B(S i,2 ), B(S i,4 ), ad B(S i,5 ), ca ot have the same code with B(S i,3 ), i.e., ay S i, IN(S i,3 ), B(S i,3 ) B(S i, ) =. The trafic load λ(a) of a area a will determie its code allocatio. The traffic load is defied as Defiitio 3. Defiitio 3: The trafic load λ(a) of a area a is: λ(a)=t a a, where t a is the average call holdig time i a ad a is the average call arrival rate i a. I a cellular system, the Erlag B formula, as show i (1), ca be used to evaluate the call blocig probability of a area C i, with the umber of available resources (codes) ad the traffic load ( C i, ) (i erlags). 1 ( ) 1 ( C ) C EB (, ( C i, )) i,! i, 0! (1). The available badwidth B is divided ito two parts: B(S i ) ad BB(S i ). B(S i ) is termed as the cell-based resource ad is used by high-mobility MHs. BB(S i ) is termed as the subarea-based resource. The resource BB(S i ) will be divided ito several subsets ad allocated to sesors, where the allocated resource of a sesor S i, is deoted as B(S i, ). The allocated resource of each sesor is used by low-mobility MHs. The mobility is defied as Defiitio 4. Defiitio 4: The Mobility of a MH mh is M(mh), where M(mh) is the average umber of crossed areas (sesors) per uit time. Whe M(mh) is larger tha a threshold th H, the mh is termed as high mobility; otherwise, is termed as low mobility. Whe a call arrives at a area, the targeted sesor S i, will choose available codes from B(S i ) or B(S i, ). The available code c is defied as Defiitio 5. Defiitio 5: A(B')={c cb' ad c do't assig to ay mobile host}. ISSN: ISBN:

3 C i, 7 C i, 6 Ci, 5 Ci, 4 C i, 3 C i, 2 Code Assigmet: If A(B(S i, )) r The assig code(s) c to mh, where ca(b(s i, )) ad c =r Else bloc the request C i, 8 C i, 1 C i, 9 C i, 10 C i, 11 C i, 12 Fig. 1: Service areas of sesors 2.2 Correlatio Research The Correlatio Research divides ito two parts: sesor-based code allocatio ad call-based code assigmet. The sesor-based code allocatio meas the method to allocate each sesor S i, a part B(S i, ) of system resource B. Accordigly, whe a call arrives at C i,, the S i, assigs its allocated codes, usig a call-based code assigmet method, to assig codes for this call. Give a cell C i with m sesors {S i,1, S i,2,, S i,m } ad the system badwidth B. I geeral, each sesor S i, permaetly eeps the same umber of codes usig to serve the icomig call of this area. This code allocatio to sesors ca be formally described as follows. Code Allocatio: Step 1. Partitio the sesors {S i,1, S i,2,, S i,m } i cell C i ito G 1, G 2,,ad G cs disoit subsets, such that ay two sesors i the same subset are apart by at least a distace of D mi. Accordigly, partitio the available badwidth B ito B 1, B 2,, ad B cs disoit subsets, where the cs disoit subsets are as fair as possible. (cs is also termed as the cluster size). Step 2. The subset B is the allocated codes of sesors i G, where = 1, 2,, cs. For istace i Fig. 1, if the D mi is 2 subareas away, i.e., cs=3, sesors {S i,1, S i,2,..., S i,12 } ca be divided ito G 1 ={S i,1, S i,4, S i,7, S i,10 }, G 2 ={S i,2, S i,5, S i,8, S i,11 }, ad G 3 ={S i,3, S i,6, S i,9, S i,12 }. Accordigly, the badwidth B ca be divided ito B 1, B 2, ad B 3. Sesors i G ca acquire the codes B, where =1, 2, 3. After the area Ci, acquired the resource B(S i, ), whe a MH mh requests r capacity to establish commuicatio sessio, the sesor S i, will fid codes c from B(S i, ) ad the assigmet procedure is preseted as follows, where A(B(S i, )) is the umber of available codes i B(S i, ). 3 Subect Strategy The subect strategy is divided ito two parts: the dyamic code allocatio ad the code assigmet. Dyamic Code Allocatio: For a cell C i, the dyamic code allocatio first partitios a part B 0 resource from the system badwidth B. B 0 is used to serve the high-mobility MHs of the etire area of C i. The other resource (BB 0 ) the is partitioed ito several subsets ad allocates to the sesors. The allocatio is described as follows. Step 1. Partitio the sesors {S i,1, S i,2,, S i,m } i cell C i ito G 1, G 2,,ad G cs disoit subsets, such that ay two sesors i the same subset are apart by at least a distace of D mi. Accordigly, partitio the available badwidth (BB 0 ) ito B 1, B 2,, ad B cs disoit subsets. Step 2. The subset B is the allocated codes of sesors i G, where = 1, 2,, cs. The disoit subsets B 1, B 2,, ad B cs are determied accordig to the traffic loads of sesors i G 1, G 2,,ad G cs, respectively. For ay G, where =1, 2,, cs, the total traffic loads(g ) of sesors i G is described as (2). C i,. G C i, G (2). For ay B, where =1, 2,, cs, if the sesors i G, acquired B as their codes. The evaluated capacity, represeted by the call blocig probability, is described as (3). EBB, Ci, Ci, CI, JG PBG. G (3). For all sesors i cell C i, icludig G 1, G 2,,ad G cs with B 1, B 2,, ad B cs, the evaluatio is described as (4). PB C i cs PB 1 cs G G 1 G. (4). ISSN: ISBN:

4 Table 1: Traffic distributio Areas Traffic load Areas Traffic load C i,1 6.0 C i,7 5.0 C i,2 2.0 C i,8 0.5 C i,3 5.0 C i,9 4.0 C i,4 2.5 C i, C i,5 4.5 C i, C i,6 2.0 C i, Table 2: All allocatio forms B 1 B 2 PB(G 1 ) PB(G 2 ) PB(C i ) The (BB 0 ) is divided ito B 1, B 2,, ad B cs. The {B 1, B 2,, B cs } is a allocatio form, i.e., sesors of C i respects the allocatio to use their codes. Differet partitios have differet allocatio forms. Each allocatio forms ca acquire a value of evaluated capacity, show i (4). Sice the allocatio forms of a cell C i is coutable, the subect strategy is to choose a allocatio forms from all allocatio forms, where the evaluatio accordig to (4) is miimal. Let C i 1, C i 2,, C i be the all allocatio forms of C i, accordig PB(C i 1 ), PB(C i 2 ),, PB(C i ) be the evaluatios, respectively. The subect allocatio strategy is formalized as follows: Step 1. Partitio the sesors {S i,1, S i,2,, S i,m } i cell C i ito G 1, G 2,,ad G cs disoit subsets, such that ay two sesors i the same subset are apart by at least a distace of D mi. Step 2. Choose a C i from all allocatio forms {C i 1, C i 2,, C i }, where PB(C i ) is miimal amog {PB(C i 1 ), PB(C i 2 ),, PB(C i )}. Step 3. Accordig C i to allocated codes to sesors. For istace i Fig. 1 uder D mi is the oe subarea away, {S i,1, S i,2,,s i,12 } ca be divided ito G 1 ={S i,1, S i,3, S i,5, S i,7, S i,9, S i,11 } ad G 2 ={S i,2, S i,4, S i,6, S i,8, S i,10, S i,12 }. Suppose (BB 0 ) has 10 codes. Uder the traffic distributio of Table 1, all allocatio forms with the evaluatios are preseted as Table 2, where the allocatio form that B 1 =7 ad B 2 =3 presets best performace. Therefore, sesors S i,1, S i,3, S i,5, S i,7, S i,9, S i,11 ca respectively acquire 7 codes ad sesors S i,2, S i,4, S i,6, S i,8, S i,10, S i,12 ca respectively acquire 3 codes. Dyamic Code Assigmet: After each sesor S i, acquired a umber B(S i, ) of system codes, the dyamic code assigmet strategy is used to assig codes to MHs for establish commuicatio sessios. Whe a call requestig a umber c of codes with the targeted MH mh arrives at the area of C i,, the system first chec the mobility of mh ad the available codes of B 0. If mh is a high mobility mobile host, i.e., M(mh)th H, ad the available codes of B 0 are sufficiet, i.e., A(B 0 ) c, the system assigs available codes of B 0 to this call. Otherwise, the system checs the available codes B(S i, ) of sesor S i,. If the available codes of B(S i, ) are sufficiet, i.e., A(B(S i, )) c, the system assigs available codes of B(S i, ) to this call. The formal represetatio is show as Fig. 2. Call arrival at C i, with the targeted MH mh requestig codes c M(mh)>th H ad A(B 0 ) >c bloc NO A(B(S i, ) NO YES YES Assig a available codes of B 0 to mh Assig a available codes of B(S i, ) to mh Fig. 2: Code assigmet strategy ISSN: ISBN:

5 Fig. 3: Simulatio structure 4 Simulatio Results The simulatio eviromet, as show i Fig. 3, has 36 cells with 432 areas, arraged as 6-parallelogram structure, where the AAA divided each cell ito 12 areas. The reuse distace is 2 subareas apart, where cs is 3, respectively. The system has 64 codes, where each call requests a code. The total MHs are 10,000 ad are divided ito 70% high-mobility MHs ad 30% low-mobility MHs. The average stay times of a high-mobility MH ad a low-mobility MH respectively are 10 ad 30. The 432 areas iclude 20% hot areas ad 80% o-hot areas. Whe a MH moved from oe area to the eighborig areas, if the eighborig area is a hot area, there are 70% to move to the hot area. The call arrival rate of each MH is geerated accordig to the radom process from 1 to 8 calls/hour ad the average call holdig time is 3 miutes. I the simulatio, our allocatio strategy is first employed to determie the codes to sesors. Accordigly, our code assigmet strategy is the to assig codes to icomig calls. Fig. 4 describes the call blocig probabilities of fixed strategy ad our strategy with B 0 =0 uder overall calls, hadoff calls, ad ew calls. The results reveal that our code allocatio strategy ca mae the code assigmet strategy more efficiet to utilize the allocated codes. The reaso ca be described as follows. I fixed strategy, the code allocatio caot coform to the traffic distributios. Heavy sesors caot acquire sufficiet primary codes to serve the icomig calls whe other light traffic sesors still have some available codes. Our strategy ca adapt to the variatios i traffic. Therefore, heavy traffic sesors have a larger probability to acquire more codes tha light traffic sesors. Accordigly, the traffic-carryig capacity is larger. Fig. 5 describes the call blocig probabilities of fixed strategy ad our strategy with B 0 =3. B 0 =3 meas 3 codes are reserved for high-mobility calls ad if a call is assiged by this code, this call is uecessary to perform hadoff procedures i the same cell. The results reveal that our strategy icludig the code allocatio strategy ad the code assigmet strategy is efficiet to larger the system capacity. Fig. 6 (a) describes the average umber of hadoffs of a call uder B 0 =0, 3, ad 6. Fig. 6 (b) describes the hadoff decrease rates of B 0 = 3 ad 6 based o B 0 =0, the results reveal that uder B 0 = 3 (=6), the hadoff decrease rates are 6.54%~23.59% (10.71%~27.94%). Hadoff procedures will icrease the system overheads, such as code switch ad data switch. Therefore, our strategy ca decrease the system overheads. 5 Coclusios The ivestigatio proposed a effective load balace ad hadoff maagemet strategy. These strategies are implemeted to solve traffic-adaptio problem that ca ehace the traffic-carryig capacity for variatios i traffic. For the maagemet of hadoff procedure, our strategy cosiders the mobility of mobile hosts ad the badwidth utilizatio i sesors. It ca decrease the umber of hadoffs ad lesse the system overhead. Cosiderig ergoomic ad ecoomic factors to reallocate codes amog cells that ca satisfy ew treds i the telecommuicatio idustry is oe of our future wors. ISSN: ISBN:

6 (a) Overallcalls (a)overallcalls Callarrival rate (b) Newcalls (b)newcalls (c) Hadoffcalls Fig. 4: Call blocig Probability with B 0 =0. (c)hadoffcalls Fig. 5: Call blocig Probability with B 0 =3. ISSN: ISBN:

7 Number of hadoff Hadoff descrease rate (%) B 0 =3 B 0 =6 (a) Number of hadoff (b) Hadoff decrease rates B0=0 B0=3 B0=6 Fig. 6: Hadoff rate Refereces: [1] V. Abdulova ad I. Aybay, Hadoff Prioritizig Schemes i Cellular Wireless Networs, IEEE Iteratioal Symposium o Computer Networs, Ju. 2006, pp [2] G. Cao, Itegratig Distributed Chael Allocatio ad Adaptive Hadoff Maagemet for QoS-Sesitive Cellular Networs, ACM/Kluwer Wireless Networs (WINET), Vol. 9, Mar. 2003, pp [3] L. J. Cimii, G. J. Foschii, C. L. I ad Z. Milaic, Call Blocig Performace of Distributed Algorithms for Dyamic Chael Allocatio i Microcells, IEEE Tras. o Commuicatios, Vol. 42, Aug. 1994, pp [4] S. P. Chug ad J. C. Lee, Performace Aalysis ad Overflowed Traffic Characterizatio i Multiservice Hierarchical Wireless Networs, IEEE Tras. o wireless Commuicatios, Vol. 4, No. 3, May 2005, pp [5] W. Choi, J. G Adrews ad R. W Heath, Multiuser Atea Partitioig for Cellular MIMO CDMA Systems, IEEE Tras. o Vehicular Techology, Vol. 56, Sept. 2007, pp [6] C. Dazhog, X. Jig, ad W. Peg, Research o Improvemet i the Hadoff Performace for High-speed Mobile Service, IEEE Caadia Coferece o Electrical ad Computer Egieerig, May 2005, pp [7] G. J. Foschii, B. Gopiath ad Z. Milaic (1993), Chael Cost of Mobility, IEEE Tras. o Vehicular Techology, Vol. 42, Nov. pp [8] C. J. Huag, Y. T. Chuag, L. C. Che, W.K. Lai ad Y.H. Su (2005), Adaptive Resource Reservatio Schemes for Multimedia Hadoffs i Fourth-Geeratio Mobile Commuicatios System, Fifth Iteratioal Coferece o Iformatio Commuicatios ad Sigal Processig, Dec. pp [9] L. Huag, S. Kumar, ad C. -C.J Kuo, Adaptive Resource Allocatio for Multimedia QoS Maagemet i Wireless Networs, IEEE Tras. o Vehicular Techology, Vol. 53, Mar. 2004, pp [10] A. Iera, A. Moliaro ad S. Marao, Hadoff Maagemet with Mobility Estimatio i Hierarchical Systems, IEEE Tras. o Vehicular Techology, Vol. 51, Sept. 2002, pp [11] A. Kava, M. Karaoc, J. R. Clevelad ad H. E. Demiray, Dyamic Allocatio of OVSF Codes to Access Termials with a Adaptive Atea Array, IEEE 16th Iteratioal Symposium o Persoal Idoor ad Mobile Radio Commuicatios, Vol. 1, Sept. 2005, pp [12] I. Katzela ad M. Naghshieh, Chael Assigmet Schemes for Cellular Mobile Telecommuicatio Systems: A Comprehesive Survey, IEEE Persoal Commuicatios, Vol. 3, Ju. 1996, pp [13] A. N. Rousas ad D. N. Soutas, Maagemet of chaelizatio codes at the forward li of WCDMA, IEEE Commuicatios Letters, Vol. 9, Aug. 2005, pp [14] M. Seth ad A. O. Fapouwo, Adaptive Resource Maagemet for Multimedia Wireless Networs, IEEE 58th. o Vehicular Techology Coferece, Vol. 3, Oct. 2003, pp [15] Z. Xu, Y. Zaharov ad G. White, Atea Array Optimisatio Usig Semidefiite Programmig for Cellular Commuicatios from HAPs, Electroics Letters, Vol. 43, No. 2, Ja. 2007, pp ISSN: ISBN:

An Adaptive Load Balance Allocation Strategy for Small Antenna Based Wireless Networks

An Adaptive Load Balance Allocation Strategy for Small Antenna Based Wireless Networks JONG-SHIN CHEN 1 NENG-CHUNG WANG 2 ZENG-WEN HONG 3 YOUNG-WEI CHANG 4 14 Department of Information and Communication