Joint Centralized Power Control and Cell Sectoring for Interference Management in CDMA Cellular Systems in a 2D Urban Environment

Size: px

Start display at page:

Download "Joint Centralized Power Control and Cell Sectoring for Interference Management in CDMA Cellular Systems in a 2D Urban Environment"

Loraine Armstrong
5 years ago
Views:

Wreless Sesor Network, 010,, 599-605 do:10.436/ws.010.8071 Publshed Ole August 010 (http://www.scrp.

1 Wreless Sesor Network, 010,, do:10.436/ws Publshed Ole August 010 ( Jot Cetralzed Power Cotrol ad Cell Sectorg for Iterferece Maagemet CDMA Cellular Systems a D Urba Evromet Abstract Mohamad Dosaraa Moghadam 1, Hamdreza Bakhsh, Gholamreza Dadashzadeh 1 Departmet of Electrcal Egeerg, Qazv Islamc Azad Uversty, Qazv, Ira Departmet of Electrcal Egeerg, Shahed Uversty, Tehra, Ira E-mal: m_dmoghadam@qau.ac.r, {bakhsh, gdadashzadeh}@shahed.ac.r Receved May 15, 010; revsed May 6, 010; accepted Jue 7, 010 The terferece reducto capablty of cell sectorg ad power cotrol algorthms have bee cosdered separately as meas to decrease the terferece Code Dvso Multple Access (CDMA) cellular systems. I ths paper, we preset Swtched-Beam (SB) ad Rotatable Equal Sectorg (RES) techques for CDMA cellular systems a D urba evromet. I the SB techque by usg a umber of fxed, depedet, or drectoal ateas we crease the dowlk capacty of the CDMA systems. Also the RES method, the equal sectors of the base statos are rotatg together to decrease the ter-cell ad tra-cell terfereces. Also ths paper we use cetralzed power cotrol to overcome the ear-far problem. Smulato results dcate that the proposed techques cosderably crease the capacty of the CDMA cellular systems compared to ordary Equal Sectorg (ES) method. Keywords: CDMA, Cetralzed Power Cotrol, Equal Sectorg, Forward Lk, Rotatable Equal Sectorg, Swtched-Beam 1. Itroducto Code-dvso multple access (CDMA) for cellular commucato etworks requres the mplemetato of some forms of adaptve power cotrol. I uplk of CDMA systems, the maxmum umber of supportable users per cell s lmted by multpath fadg, shadowg, ad ear-far effects that cause fluctuatos of the receved power at the base stato (BS). Depedg o the locato where the decso o how to adjust the trasmtted powers s made, the power cotrol algorthms ca be dvded to two groups: cetralzed ad dstrbuted techques [1-6]. I cetralzed power cotrol, a etwork ceter ca smultaeously compute the optmal power levels for all users. However, t requres measuremet of all the lk gas ad the commucato overhead betwee a etwork ceter ad base statos [7]. Dstrbuted power cotrol, o the other had, uses oly local formato to determe trasmtter power levels. It s much more scalable tha cetralzed power cotrol. However, trasmtter power levels may ot be optmal, resultg Ths work was fuded by the Islamc Azad Uversty of Qazv, Ira. performace degradato [8]. I ths paper a cetralzed power cotrol algorthm s used to compesate for ear-far effects. Dversty ad power cotrol are two effectve techques for ehacg the sgal to terferece plus ose rato (SINR) for wreless etworks. Dversty explots the radom ature of rado propagato by fdg depedet (or, at least, hghly ucorrelated) sgal paths for commucato. If oe rado path udergoes a deep fade, aother depedet path may have a strog sgal. By havg more tha oe path to select from, the SINR at the recever ca be mproved. The dversty scheme ca be dvded to three methods: 1) the space dversty; ) the tme dversty; 3) the frequecy dversty [1]. I these schemes, the same formato s frst receved (or trasmtted) at dfferet locatos (or tme slots/ frequecy bads). After that, these sgals are combed to crease the receved SINR. The atea array s a example of the space dversty, whch uses a beamformer to crease the SINR for a partcular drecto [9,10]. I ths paper we preset swtched-beam (SB) techque wth atea arrays CDMA cellular systems. Also, we propose a method called rotatable equal sectorg (RES)

600 M. D. MOGHADAM ET AL. that s as smple as the equal sectorg (ES) method, but t s able to rotate the sectors based o slow varato of users dstrbuto wth a day.

I Secto 3, the system model ad formulato are troduced. I Secto 4, we preset the RES method. Secto 5 descrbes the SB techque.

. Propagato Model Because of usg D urba structure ths paper, for computg yeld for path betwee base stato ad a user, propagato model urba evromets are dramatzed.

$reflecto pheomea. I codto that radated beam s coflcted to a obstacle edge, the dffracto pheomea s happeed ad dffractg pot s dffusg ew beams to all drectos lke a trasmtter.$ All reflected beams, wll stay the evromet tll the tme ther power are ot reduced to a threshold lmt.

All reflected beams, wll stay the evromet tll the tme ther power are ot reduced to a threshold lmt.

$We lst the propagato parameters for ths smulator Table 1. Fgure 1. Dffracto pheomea ad reflecto pheomea (LoS ad No-LoS paths) for a D urba evromet forward lk. Table 1. Propagato parameters USPS.$ I mult-cell CDMA systems, a moble set a gve sector may be exposed by both terfereces ts ow base stato ad surroudg sectors ad base statos.

I mult-cell CDMA systems, a moble set a gve sector may be exposed by both terfereces ts ow base stato ad surroudg sectors ad base statos.

esurg SINR requremets for each user. Fgure shows a urba area used our smulatos that s part of Caada Toroto cty.

2 600 M. D. MOGHADAM ET AL. that s as smple as the equal sectorg (ES) method, but t s able to rotate the sectors based o slow varato of users dstrbuto wth a day. The orgazato of the remader of ths paper s as follows. I Secto, propagato model a D urba evromet ad also fuctoal state of urba sgal propagato smulator (USPS) are descrbed. I Secto 3, the system model ad formulato are troduced. I Secto 4, we preset the RES method. Secto 5 descrbes the SB techque. Fally, smulato results ad coclusos are gve Sectos 6 ad 7, respectvely.. Propagato Model Because of usg D urba structure ths paper, for computg yeld for path betwee base stato ad a user, propagato model urba evromets are dramatzed. I a propagato model of urba evromets ad forward lk (dowlk), base stato atea s radatg beams whch are dffusg all drectos ad parts of beams reach to base stato. I urba evromet, delvered beam from base stato by the tme of collso to a obstacle lke a wall surface or a buldg, reflects to a ew agle ad cotues ts path, ths s called reflecto pheomea. I codto that radated beam s coflcted to a obstacle edge, the dffracto pheomea s happeed ad dffractg pot s dffusg ew beams to all drectos lke a trasmtter. All reflected beams, wll stay the evromet tll the tme ther power are ot reduced to a threshold lmt. Fgure 1 shows both pheomea forward lk ad for le of sght (LoS) ad o-los paths. I ths paper, the software USPS s used to mplemet a D urba evromet [9,11]. We lst the propagato parameters for ths smulator Table 1. Fgure 1. Dffracto pheomea ad reflecto pheomea (LoS ad No-LoS paths) for a D urba evromet forward lk. Table 1. Propagato parameters USPS. Parameter Amout Resoluto 1 Path Loss Mmum receved power USPS 0.1 db/m 100 dbm 3. System Model ad Formulato 3.1. System Model I ths paper, we maly focus o sectorg a mult-cell CDMA system ad oly the dowlk of cellular system s cosdered. I mult-cell CDMA systems, a moble set a gve sector may be exposed by both terfereces ts ow base stato ad surroudg sectors ad base statos. Our objectve here s to fd the best sectorg of the cells at ay tme, to maxmze system capacty ad mmze the total trasmtted power for the whole system, whle at the same tme esurg SINR requremets for each user. Fgure shows a urba area used our smulatos that s part of Caada Toroto cty. The area cotas fve base statos that are used for studyg dowlk capacty of the CDMA systems. For smplcty, Fgure. Part of D plae of Caada Toroto cty ad placg fve base statos. we lst the otatos for ths paper Table. 3.. System Capacty Cosder a CDMA cellular system (see Fgure ) where each of M base statos use N drectoal ateas for sgal trasmsso. The receved SINR for the user the sector from the base stato j ca be represeted by the sgal to terferece plus ose rato ( SINR, j, )

3 M. D. MOGHADAM ET AL. 601 Table. Lst of otatos. Symbol M N Quatty Total umber of base statos/cells Total umber of sectors ES method Nose power G K j, P h Agle for rotatg sectors RES method Voce actvty of the user (coected to the sector of the base stato j ) Processg ga of the user. G W / R, where W s the total badwdth ad R s the bt rate of the user Number of users the cell (base stato) j ad the sector Trasmtted power by the sector of the base stato to the user Chael ga betwee the sector of the base stato ad the user The SINR target of the user (coected to the sector of the base stato j ) whch s gve by [1] SINR K j, k1, k Gh () P h ( k) P I k, j, k, j, where I s out-sector ad out-cell terferece power for the user that s the sector of the base stato/cell j ad ca be show to be I M N K m, k, m,, m, k, m, m1 1 k1 m, j, j j (1) h ( k) P () where hm,, ( k) s the chael ga betwee the sector ' of the base stato m (coected to the user k) ad the user (see Fgure 3). Let bt rate R = R for all users, thus processg ga for all users s G = G (see Table ) Cetralzed Power Cotrol Algorthm A major lmtg factor for the satsfactory performace of CDMA systems s the ear-far effect. Power cotrol s a tellget way of adjustg the trasmtted powers cellular systems so that the total trasmtted power s mmzed, but at the same tme, the user SINRs satsfes the system qualty of servce (QoS) requremets [13-15]. The goal of power cotrol ths paper s to fd all P, j, wth cetralzed power cotrol techque, such as the total trasmtted power of each base stato s m- Fgure 3. Chael ga betwee base statos ad users forward lk [1]. mzed whle a certa requred QoS s guarateed for all users all cells. Ths algorthm defes as follows [16]. j N K j, k, j, ; 1,..., M (3) 1 k1 P P j The QoS for the user the sector of the base stato j ca be defed by meas of ts receved SINR as Gh () P K (4) j, h ( k) P I k1, k where m max k, j, k, j,, max ad m are maxmum ad mmum receved SINR for each user, respectvely. Wthout loss of geeralty, we assume: M N K, 1 m P I h ( k) k, m,, m, N m1 1 Km, k1 m, j, where P s maxmum trasmtted power for each base stato. The (4) ca be rewrtte as Gh () P K j, k1, k where s defed as h ( k) P k, j,, j, k, j,, j, I matrx form, (6) ca be expressed as where I (7) H(h,g )p (8) p j, j, j, j, [ P,..., P ] T j, 1, j, Kj,, j, T j, [ 1, j,,..., K,,, ] j j (9)

4 60 M. D. MOGHADAM ET AL. are the K j, 1 vectors of the trasmtted powers ad out-sector ad out-cell terferece plus ose power, respectvely. Also (8), s the vector cotag all chael gas [ 1, j,,..., K j ] j,,, T [ H(h,g )] j, j, l,, ad h j, h () l for, l 1,..., kj,, g, Gh () h () l for l l, j, j for l Usg (8), the optmal trasmtted powers ca be computed as [16] 1 p j, [ H(h j,, j, ] j, g ) (11) 4. Rotatable Equal Sectorg Method A way of reducg the terferece betwee users s to sectorg the cells usg drectoal ateas. Whle ths approach stll utlzes the spatal doma to troduce orthogoalzato to the system, t s fudametally dfferet tha beamformg. Beamformg method combes the receved sgals from multple ateas a uque way for each user to suppress the terferece that the user sees. Sectorg merely employs drectoal ateas ad each users sgal s receved at oly oe of these ateas. Sce oly a subset of the users s receved at each atea, the terferece that each user sees s less compared to a sgle atea system. No spatal comber s used. Whle ths s a more rgd scheme tha beamformg, the smplcty of the resultg recevers s appealg ad sectorg ca be qute beefcal especally for statc systems, f the users do ot have to hadoff from sector to sector very frequetly [14,15]. It has bee show that sectorg creases the umber of users admssble a system. However uder hghly o-uform traffc loads, covetoal sectorg,.e., ES method, whch dvdes the cell to equal wdth sectors mght fal to brg much capacty mprovemet. To crease the capacty of CDMA cellular systems, ths paper, the RES s proposed to reduce rejectg of call requests stuatos lke as traffc jams ad crowded hours etertamet places. Ths method s capable of rotatg the patter β degrees each drecto, dscretely. Hece, there are 360/(βN) possble dfferet states, where N s the umber of sectors for each base stato. I the RES method, agle rotato of each base stato s chose such that the SINR of each user remas betwee γ m ad γ max uder mmzg the trasmtted power of each base stato. 5. Swtched-Beam Techque Oe smple alteratve to the fully adaptve atea s the swtched-beam archtecture whch the best beam s chose from a umber of fxed steered beams. Swtchedbeam systems are techologcally the smplest ad ca be mplemeted by usg atea arrays ad or a umber of fxed, depedet, or drectoal ateas [17]. We lst the SB techque codtos for ths paper as follows. 1) Accordg to Fgure 4, beams coverage agle s 30 ad overlap betwee cosecutve beams s 0. Thus each base stato has 36 beams. ) Each user ca use max beams for ts each path to commucate wth a base stato at ay tme. I the SB techque, umber of beams s chose such that SINR of each user remas betwee γ m ad γ max uder mmzg the trasmtted power of each base stato. I Fgure 5, we show select of beams for three users wth the SB techque for max =. It should be metoed that usg max = meas maxmum two beams are payg servce to users form of ut beam. 6. Smulato Results I ths secto we preset the smulato results to compare the performace of the RES, SB, ad ES methods. The smulated ES method uses the dstrbuto of users whch ca be see Fgure 6 for patter cofgurato of M = 5 base statos (see Fgure ). Sce the ES method does ot chage ths cofgurato accordg to the chages of the dstrbuto, some codtos ths method s ot able to hadle may of call requests. But case of usg the RES method, we ca mprove the performace, as ths method rotates the patters of base statos accordg to the dstrbuto of the users at each momet. I our smulato, we use the followg parameter settg: Total umber of base statos M = 5; total umber of sectors the ES method N = 3 (sector agle s 10 ); processg ga G = 51; voce actvty for all users 1; ose power ; agles for rotatg sectors the RES method β = 10 ad β = 30 ; maxmum trasmtted power for each base stato P = 10W; γ max = 9 db ad γ m = 7 db. I Fgure 7, the umber of actve users dfferet hours of day ad ght a day off for the RES, SB, ad ES methods s compared. I such a day, we assume users are spedg ther spare tme places such as park, cema ad sport clubs ad are gathered two dfferet places ad aroud dfferet hours (aroud 17 ad hours). As you ca see especally aroud metoed tme rego we face crses, creasg capacty for the RES method s very cosderable comparso wth the ES method. Ths creasg capacty for the RES method ad for β = 10 s hgher tha β = 30, because the system has more degrees of freedom. I addto, we see that the umber of actve users metoed tme wth the SB

5 M. D. MOGHADAM ET AL. 603 techque s hgher tha the RES ad ES methods, because the SB techque the terferece s lower tha the RES ad ES methods. It should be metoed that creasg the umber of actve users the SB techque, wll lead more complexty recever comparso wth the RES ad ES methods. Also t ca be see that the umber of actve users wth the SB techque for max = 3 s hgher tha the other cases ( max = 1, max = ). Fgure 8 shows the comparso of the umber of actve users dfferet hours of a day ad ght a workg day for the RES, SB, ad ES methods. Ths study s assumed where hghest user demads for cocetrato o coversato are happeed aroud 7.30, 1.30 ad (startg ad fshg tmes for educatoal cetres, offces, ad ). Smlar to Fgure 7, we observe that the umber of actve users the RES method s much hgher tha the ES method ad also s lower tha the SB techque. Also t ca be see that the umber of users the RES method ad for β = 10 s hgher tha β = 30. Fgure beams each base stato wth swtched-beam techque. Fgure 7. Number of actve users (Normalzed) for the RES, SB, ad ES methods a day off. Fgure 5. Select of beam for three users wth swtched-beam techque. Fgure 8. Number of actve users (Normalzed) for the RES, SB, ad ES methods o a workg day. Fgure 6. Dstrbuto of users the ES method. Fgure 9. The average of actve users (Normalzed) for the RES, SB, ad ES methods a day off.

6 604 M. D. MOGHADAM ET AL. actve users ad complexty of recever compared to the RES ad ES methods. 8. Refereces Fgure 10. Percetage of mprovemet the system capacty the SB techque versus the RES ad ES methods a day off. Fgure 9 shows the average of actve users dfferet hours of a day ad ght a day off for the RES, SB, ad ES methods (see Fgure 7). Accordgly, we observe that the umber of users metoed hours s pad servces by the RES method whch s much hgher tha the ES method. It s otced that wth β = 10, the RES method has better performace tha β = 30. Also observe that the average of the umber of actve users the SB techque s hgher tha the RES ad ES methods. Also t ca be see that the umber of users the SB techque for max = 3 s hgher tha the other cases. Fgure 10 shows the percetage of mprovemet the system capacty for the SB techque versus the RES ad ES methods a day off (see Fgure 7). For example, aroud hour, the umber of users allowed the system for the SB techque ad for max = 3 creases by approxmately 41.5% comparso wth the ES method, whle versus the RES method ad for β = 30 ad β = 10 ca provde a capacty cremet 5% ad 1.5%, respectvely. 7. Coclusos I ths paper we propose the swtched-beam ad rotatable equal sectorg techques that are sutable to crease the capacty of a mult-cell CDMA system. I the SB techque, each user ca use max beams for ts each path to commucate wth a base stato at ay tme. Also the RES method s dscretely capable of rotatg the patter β degrees each drecto. I both techques, umber of beams ad agle rotato of each base stato are chose such that SINR of each user remas betwee γ m ad γ max uder mmzg the trasmtted power of each base stato. Smulato results dcated that besdes the RES method has the smplcty of the ES method, t creases the capacty of the CDMA cellular systems compared to the ES method. It has also observed that usg the SB techque wll crease the umber of [1] J. T. Wag, Admsso Cotrol wth Dstrbuted Jot Dversty ad Power Cotrol for Wreless Networks, IEEE Trasactos o Vehcular Techology, Vol. 58, No. 1, Jauary 009, pp [] A. Abrardo ad D. Seat, O the Aalytcal Evaluato of Closed-Loop Power-Cotrol Error Statstcs DS-CDMA Cellular Systems, IEEE Trasactos o Vehcular Techology, Vol. 49, No. 6, November 000, pp [3] L. Carrasco ad G. Femeas, Reverse Lk Performace of A DS-CDMA System wth both Fast ad Slow Power Cotrolled Users, IEEE Trasactos o Wreless Commucatos, Vol. 7, No. 4, Aprl 008, pp [4] L. Qa ad Z. Gajc, Varace Mmzato Stochastc Power Cotrol CDMA System, IEEE Trasactos o Wreless Commucatos, Vol. 5, No. 1, Jauary 006, pp [5] M. Rtamak, H. Kovo ad I. Hartmo, Adaptve Closed-Loop Power Cotrol Algorthms for CDMA Cellular Commucato Systems, IEEE Trasactos o Vehcular Techology, Vol. 53, No. 6, November 004, pp [6] F. Rashd-Farrokh, L. Tassulas ad K. J. Ray-Lu, Jot Optmal Power Cotrol ad Beamformg Wreless Networks Usg Atea Arrays, IEEE Trasactos o Commucatos, Vol. 46, No. 10, October 1998, pp [7] S. Gradh, R. Vjaya ad D. Goodma, Cetralzed Power Cotrol Cellular Rado Systems, IEEE Trasactos o Vehcular Techology, Vol. 4, No. 4, November 1993, pp [8] J. Zader, Dstrbuted Co-Chael Iterferece Cotrol Cellular Rado Systems, IEEE Trasactos o Vehcular Techology, Vol. 41, No. 3, August 199, pp [9] M. Dosaraa-Moghadam, H. Bakhsh, G. Dadashzadeh, ad P. Rahmat, Adaptve Beam Formg Method Based o Costraed LMS Algorthm for Trackg Moble User, IEEE Global Moble Cogress, Shagha, October 009, pp [10] J. Chag, L. Tassulas ad F. Rashd-Farrokh, Jot Trasmtter Recever Dversty for Effcet Space Dvso Multaccess, IEEE Trasactos o Wreless Commucatos, Vol. 1, No. 1, Jauary 00, pp [11] H. Zamr-Jafara, M. M. Mrsaleh, I. Ahad-Akhlagh ad K.N. Plataots, Moble Stato Postog Usg Radal Bass Fucto Networks, IEEE Iteratoal Symposum o Persoal, Idoor ad Moble Rado Commucatos, Vol. 4, September 004, pp [1] J. Zhag, J. Lu, Q. Zhag, W. Zhu, B. L ad Y. Zhag, A Effcet Algorthm for Adaptve Cell Sectorg CDMA Systems, IEEE Iteratoal Coferece o

7 M. D. MOGHADAM ET AL. 605 Commucatos, Achorage, Vol., May 003, pp [13] S. Kadukur ad S. Boyd, Optmal Power Cotrol Iterferece-Lmted Fadg Wreless Chaels wth Outage Probablty Specfcatos, IEEE Trasactos o Wreless Commucato, Vol. 1, No. 1, Jauary 00, pp [14] M. Rtamak, H. Kovo ad I. Hartmo, Adaptve Closed-Loop Power Cotrol Algorthms for CDMA Cellular Commucato Systems, IEEE Trasactos o Vehcular Techology, Vol. 53, No. 6, November 004, pp [15] R. Zhag, C. C. Cha ad Y. Lag, Jot Beam Formg ad Power Cotrol for Multatea Relay Broadcast Chael wth Qos Costrats, IEEE Trasactos o Sgal Processg, Vol. 57, No., February 009, pp [16] A. F. Almutar, S. L. Mller, H. A. Latchma ad T. F. Wog, Power Cotrol Algorthm for MMSE Recever Based CDMA Systems, IEEE Commucatos Letters, Vol. 4, No. 11, November 000, pp [17] B. Alle ad M. Beach, O the Aalyss of Swtched- Beam Ateas for the W-CDMA Dowlk, IEEE Trasactos o Vehcular Techology, Vol. 53, No. 3, May 004, pp

A Novel Bandwidth Optimization Manager for Vehicle Controller Area Network, CAN, System

A Novel Bandwidth Optimization Manager for Vehicle Controller Area Network, CAN, System A Novel Badwdth Optmzato Maager for Vehcle Cotroller Area Network, CAN, System Y WANG, Z-y YOU, L-hu HUI Guzhou Normal Uversty, Guyag, Guzhou 550001, Cha Abstract Ths paper cosders the badwdth lmtato of