Coexistence Evaluation of LTE with Active Antenna System

Transcription

1 Communcatons and etwork Publshed Onlne September 013 ( Coexstence valuaton of LT wth Actve Antenna System Fe Xue Qme Cu Tng Fu Jan Wang Key Laboratory of Unversal Wreless Communcaton Mnstry of ducaton Bejng Unversty of Posts and Telecommuncatons mal: xue_fe0515@163.com Receved June 013 ABSTRACT Due to the new rado archtecture of Actve Antenna System (AAS) the LT BS equpped wth AAS wll have more flexble RF plannng than that equpped wth passve antenna. owever the new RF characterstc of AAS wll have potental mpacts on the exstng BS RF requrements whch s mperatve to be evaluated. In ths paper general antenna pattern and spatal Adjacent Channel Leakage Rato (ACLR) are ntroduced consderng the new RF features of AAS and the mpacts of AAS on BS performance are analyzed based on the AAS fundamental applcatons. System-level coexstence smulaton s conducted to evaluate the RF features of transmtter and recever of AAS on BS performance. Smulaton results show that RF feature of spatal ACLR of AAS transmtter has lttle mpact on throughput loss n vctm and ACLR of 45 db per transmtter of AAS can fully meet the coexstence requrement; RF feature of n-band blockng level of ndvdual recever of AAS s hgher than that of recever of BS equpped wth passve antenna around 1-4 db whch mposes harder requrement to desgn the channel selecton flter n the ndvdual recever of AAS thus n-band blockng requrement needs to be redefned for AAS BS n the exstng 3GPP specfcaton. Keywords: Coexstence valuaton; AAS; LT; Spatal Characterstcs 1. Introducton In the progress of the specfcaton of 3GPP Release 11 lots of new characterstcs and studes are ntroduced to further enhance the performance of LT-Advanced [1]. Gven that rado network are beng requred to support multple technologes and mprove the capacty n the foreseeable future there s a growng need to ntegrate the antenna array and the transcevers n a typcal base staton to mnmze ste footprnt and lower costs []. The new rado archtecture of AAS s rased n ths demand. Besdes brngng n smaller ste footprnt and lower costs AAS can also support a host of advanced electronc beam-tlt features that can enable mprovements n network capacty and coverage. A typcal mplementaton of advanced antennas tlt features of AAS s vertcal cell splttng n a sngle sector cell [6]. Thus AAS s ntroduced as an alternatve antenna from the one nstalled n the conventonal BS. A study tem and work tem have been approved n 3GPP TSG RA WG4 to study the characterstcs of the AAS transmtter and recever and nvestgate the mpacts on the coexstence performance wth other s based on un-coordnated deployment [3]. Introducng of AAS may brng specal changes n RF performance. In the tradtonal passve antenna there s only one transcever connected wth passve antenna array through coaxal cable. owever as shown n Fgure 1 there are several ndvdual transcevers n the AAS whch s connected wth ndvdual antenna element separately []. Ths modfcaton n rado archtecture wll lead the two RF features ACLR and n-band blockng beng dfferent from that n passve antenna. As the correlaton level vares wth dfferent RF sgnals (e.g. n channel sgnals and adjacent channel sgnals) nduced on the antenna element n the physcal devce composte antenna pattern for dfferent RF sgnals should be dstngushed. ACLR of AAS wll be spatally dstrbuted whch s dfferent from flat dstrbuted ACLR of tradtonal passve antenna [45]. Detaled theoretcal analyss of spatal ACLR s shown n Secton. In the desgn of rado ACLR requrement s ndspensable for power amplfer n any rado transmtter to guarantee an acceptable adjacent channel nterference caused by ts nonlnear characterstc. Thus t s mperatve to nvestgate the spatal ACLR of AAS and defne the reasonable ACLR requrement for AAS. Meanwhle due to new rado archtecture of AAS n-band blockng of RF recever of AAS whch s used to study the recever s ablty to detect a wanted sgnal on ts assgned channel wth an unwanted sgnal nsde the Copyrght 013 ScRes. C

2 18 F. XU T AL. operatng band should be evaluated at the nterface between antenna element and ndvdual recever to defne ts flter specfcaton of channel selecton flter not at the nterface between recever and coaxal cable n the tradtonal passve antenna. Thus n-band blockng requrement of the ndvdual recever s also mperatve to be evaluated to defne the flter specfcaton to guarantee the recever functon well. owever untl rght now there are no research papers analyzng the RF requrements of AAS n detal to ensure practcal applcaton. In ths paper based on the new rado archtecture of AAS general antenna pattern and spatal ACLR s ntroduced consderng the correlaton level of dfferent sgnals nduced at the antenna element. Then we analyzed the mpacts of AAS on BS performance based on the AAS fundamental applcatons. System-level coexstence smulaton s conducted to evaluate the spatal characterstc of transmtter and recever of AAS on BS performance extendng our work n [11]. Smulaton results show that spatal ACLR of AAS transmtter wth dfferent correlaton levels has lttle mpacts on throughput loss n the vctm especally for ndvdual transmtter of AAS wth hgh ACLR. ACLR of 45 db per transmtter of AAS can fully meet the coexstence requrement and n-band blockng level of ndvdual recever of AAS s hgher than that of recever of BS equpped wth passve antenna around 1-4 db thus n-band blockng requrement for AAS ndvdual recever needs to be redefned n the exstng 3GPP specfcaton [3]. The rest of ths paper s organzed as follows: the general antenna pattern and the spatal ACLR of AAS s ntroduced n Secton. The mpacts of AAS on BS performance and evaluaton results are descrbed and analyzed n detal separately n Secton 3 and 4. Conclusons are drawn n Secton 5.. General Antenna Pattern and Spatal ACLR of AAS.1. General Antenna Pattern Rado archtecture of AAS s shown n Fgure 1. ach radaton element of AAS s connected wth an ndvdual transcever unt. owever actve components connected wth each antenna element shall make the RF sgnals nduced on the antenna element from several transcevers unt n the transcever array partally correlated. For example the dentcal sgnals are appled to the nput of the transcever array unwanted sgnals generated on the radaton element s mplementaton specfc and hence unknown n nature (e.g. amplfed thermal sgnal nose s random and hence wll have no smlarty n dfferent transcevers) [9]. Gven ths consderaton the composte antenna pattern of AAS should nclude both radaton element pattern and correlaton of the RF sgnals from dfferent transcevers. The sgnal from the drecton U s actng on the antenna array and elevaton angle of the sgnal drecton s denoted as and azmuth angle of the sgnal drecton s denoted as. The composte antenna pattern of AAS conssts of three parts: radaton element pattern array factor for antenna array and correlaton matrx of dfferent transcever paths whch wll be descrbed as the followng. 1) Antenna lement pattern: The horzontal pattern of radaton element s descrbed as [] A ( ) mn[1( ) Am ] (1) 3dB where 3 db 65 s the horzontal 3dB bandwdth of antenna element and Am 30dB s the front to back rato. The vertcal pattern of radaton element s descrbed as [] 90 AV ( ) mn[1( ) SLAv ] () 3dB where 3 db 65 s the vertcal 3dB bandwdth of antenna element and SLAv 30dB s the attenuaton lower lmt of sde lobe. The antenna model of antenna element can be modeled as [1] 0log 10 ( P ( )) (3) G mn{[ A ( ) A ()] A } Max V m where P ( ) s denoted as the radaton gan of antenna element GMax 8dB s the maxmum drectonal gan of antenna element. ) Array factor for sngle column: Due to specfc array placement and electrcal steerng down-tlt the sgnal radated from each radaton element wll experence dfferent phase shft before t arrved at the User qupment (U) and the power strength of receved sgnal wll vary wth the specfc phase shft. ere V s the phase shft due to the array placement as shown n Fgure. Fgure 1. Rado archtecture of AAS. Copyrght 013 ScRes. C

3 F. XU T AL. 19 v 1 v v 3 v 4 v 5 Z d d d d 3 4 P( x y z) 1 X w 1 w w 3 w 4 w 5 Z d d d d 3 4 P ( xyz ) Fgure. Geometry for calculaton of the phase shft for each dpole n the far feld. V [ v1 v... v ] T n 1... (4) dv vn exp(- n ( 1) cos( )) (5) W s the steerng factor whch s used to control the basc physcal characterstcs of composte antenna pattern for example 3dB beam-wdth sde lobe attenuaton electrcal down-tlt [10]. Steerng vector W manly provdes the mnmum power level of sde lobe the electrcal down-tlt and sde lobe attenuaton for composte antenna pattern whch can be denoted as W [ w1 w... w ] T n 1... (6) 1 dv wn exp(- ( n 1) sn( etlt )) (7) the electrcal down-tlt etlt s set to be 9 that optmzes the throughput [7]. The combned phase shft for each dpole n the far feld can be denoted as W W V (8) 3) Correlaton matrx of RF sgnals for sngle column: The RF sgnals nduced on all antenna elements s St () [ s() t s()... t s ()] t (9) 1 The complex output of antenna array at the far feld becomes y( t) s ( t) w ( ) P ( ) St () W n n n n1 X (10) where n ( ) s the radaton gan of antenna element together wth phase shft due to array placement at the far feld and we assume that radaton gan of dfferent antenna elements s dentcal. The combned radaton pattern s the mean output power of antenna array at the far feld whch can be obtaned by calculatng condtonal expectaton over y( t) [9]. P( ) y( t) P ( ) W S ( ( t) St ( )) W P ( ) W R W (11) R s the correlaton matrx of the RF sgnals n dfferent transcever paths represented as R11 R1 R1 R1 R R R ( S ( t) S( t)) (1) Rnm R1 R R Correlaton of RF sgnals n dfferent transcever paths s denoted as a correlaton coeffcent 0 R nm 1 defned as the smlarty of the unwanted sgnals generated n dfferent transcever paths when an dentcal sgnal s appled at the nput of ndvdual transcever unt. Unwanted sgnals under dfferent crcumstances can be regarded as correlated Rnm 1 (e.g. unwanted sgnals generated by Crest Factor Reducton wll be generated dgtally and hence dentcal n each path [10]) or uncorrelated Rnm 0 (e.g. amplfed thermal nose n each amplfer []). Gven that the fast fadng between dfferent antenna elements s spatally correlated correlaton coeffcent can be denoted as [] R nm K snk s mk k 1 K K snk smk k1 k1 (13) where K s the number of samplng pont n a dstngushed tme slot. For the sake of complexty and suffcent correlaton level analyzed n the coexstence study the same level correlaton level s assumed whch s a value between 0 and 1 []. 1 1 R ( S ( t) S( t )). (14) 1 1 The radaton pattern of composted antenna array can be smplfed as [] AA( ) 10log10 1 wn vn 1 n1 (15) 0log ( P ( )).. Spatal ACLR Model 10 In the exstng specfcaton [3] ACLR s defned as the rato of the fltered mean power on the assgned channel frequency to the fltered mean power at an adjacent Copyrght 013 ScRes. C

4 0 F. XU T AL. channel frequency. Due to the fact that the dfferent RF sgnals (e.g. n channel sgnals and adjacent channel sgnals) n dfferent transcever paths nduced on the antenna element have dfferent correlaton level n AAS BS antenna pattern should be appled separately for dfferent RF sgnals. Spatal ACLR of AAS can be descrbed as s () t s () t s () t (16) In_ Ch Adj_ Ch ( yin _ Ch ( t) ) ACLRlement 10 log 10 (17) ( yadj _ Ch ( t ) ) where sin _ Ch () t represents RF sgnal nduced at the antenna element on the assgned channel frequency and sadj _ Ch () t represents RF sgnals nduced at the antenna element on the adjacent channel. We assume the RF sgnals n dfferent transcever on the assgned channel frequency are fully correlated namely 1 and the RF sgnals n dfferent transcever on an adjacent channel are partally correlated namely 0 1. The spatal ACLR n db can be denoted as w v 1 ACLR( ) 10log 10 ) 1 w v 1 1 ACLR lement 3. Impacts of AAS on BS Performance 3.1. Impacts of Transmtter Characterstcs (18) As shown n Fgure 3 ACLR of AAS BS s spatally dstrbuted n the vertcal plane and ACLR of tradtonal passve antenna s flat. Compared wth the tradtonal passve antenna some of the areas covered by adjacent may suffer more nterference from AAS BS whle some others may suffer less. Therefore spatal ACLR of AAS wll have potental mpacts on cell average and cell edge throughput loss. 3.. Impacts of Recever Characterstcs As mentoned n Secton 1 there s only one recever connected wth antenna array through coaxal cable n tradtonal passve antenna. Therefore n-band blockng whch s the RX power level from Us wthn the s at the adjacent channel should be evaluated at the nterface between recever and coaxal cable. In-band blockng for the recever of tradtonal passve antenna experenced the composte antenna pattern gan and cable loss. owever n the AAS there are several ndvdual recevers whch s connected wth ndvdual antenna element separately. Thus n-band blockng for ndvdual recever should be evaluated at the nterface between antenna element and ndvdual recever. In-band blockng wll just experence the antenna element pattern gan whch s qute dfferent from that of tradtonal passve antenna. 4. Performance valuaton 4.1. Coexstence Scenaro In ths paper -UTRA Macro to -UTRA Macro coexstence scenaro s evaluated by -level smulaton for the purpose of nvestgatng the spatal characterstcs of AAS BS as shown n Table 1 and Table. The channel frequency of the aggressve s located tghtly besdes that of the vctm. Table 1. Coexstence scenaro for ACLR. Case Aggressor Vctm 1a 60 1b c(Baselne) ACLR (db) 30 0 ro=1 ro=0.8 ro= ro=0.4 ro=0. ro= Theta (deg) Fgure 3. Spatal ACLR of AAS. Table. Coexstence scenaro for n-band blockng. Case Aggressor Vctm 1a 1b 1c(Baselne) Copyrght 013 ScRes. C

5 F. XU T AL etwork Layout The layout of the vctm and aggressor network s dentcal and aggressor network s stes are located at the vctm network s cell edge wth worst ste shfts. The nter-ste dstance s 750 m. Detaled network layout can be found n [3] Large Scale Channel Model The channel path loss model s defned as [3] PathLoss = max{l(r) Free_Space_Loss} + shadowfadng where Free Space Loss s defned as Free_Space_Loss = *log 10 (R) (R n klometre) L(R) s defned as L(R) = log 10 (R) The fnal couplng loss s defned as Coupl_Loss_Macro = max{pathlossfree_space_loss}-g_tx-g_rx where G_Tx s the transmtter antenna gan and G_Rx s the recever antenna gan Smulaton Parameters General smulaton parameters for coexstence study of AAS BS are lsted n Table 3 [8]. The Power Control (PC) scheme n LT s performed for the evaluaton of n-band blockng and detaled power control parameters and lnk to throughput mappng are descrbed n [3]. Both PC set 1 and PC set are adopted for studyng the spatal recever characterstcs of AAS BS. Table 3. General smulaton parameters. Parameters Carrer frequency System bandwdth Mnmum dstance U<->BS Log normal shadowng Shadow correlaton coeffcent umber of actve Us U max Tx Power U mn Tx Power BS max Tx Power Schedulng algorthm Antenna confguraton at U The heght of BS The heght of U umber of antenna elements 10 ACS of LT U Cable Loss (Legacy/AAS) Values Gz 10 Mz 35 m Standard Devaton of 10 db 0.5 (nter ste) / 1.0 (ntra ste) UL: 3Us(16RBs /U) DL:1U(50RBs/U) 3 dbm -40 dbm 46 dbm Round Robn Full buffer Omn-drectonal 30 m 1.5 m 33 db 1/0 db 4.5. Smulaton Results Smulaton results of Case 1a and Case 1b n Fgure 4 and Fgure 5 show that dfferent correlaton levels of spatal ACLR have lttle mpact on cell average and cell edge throughput loss. The reason s that the adjacent channel nterference n the downlnk manly depends on adjacent channel selectvty of U especally for AAS BS wth hgh ACLR. Comparng Case 1a Case 1b and Case 1c the throughput loss s almost consstent wth the same ACLR assumpton. The small gap between Case 1a and Case 1b s due to the cable loss of passve antenna n Case 1a resultng n lower transmt power and hence hgher throughput loss n vctm. For AAS BS smulaton results ndcate that ACLR of 45 db per transcever for specfc coexstence scenaro studed s suffcent to fulfll the coexstence requrement. o matter the vctm s equpped wth passve antenna or AAS the throughput loss of vctm n Case 1a and Case 1b s lower than 5%. Downlnk cell average throughput loss [%] Downlnk cell edge througput loss [%] Case1b Case1c Case1a case1a-0.0 case1a-0. case1a-0.4 case1a-0.6 case1a-0.8 case1a-1.0 case1b-0.0 case1b-0. case1b-0.4 case1b-0.6 case1b-0.8 case1b-1.0 case1c ACLR lement [db] Fgure 4. Downlnk cell average throughput loss. case1b case1c case1a case1a-0.0 case1a-0. case1a-0.4 case1a-0.6 case1a-0.8 case1a-1.0 case1b-0.0 case1b-0. case1b-0.4 case1b-0.6 case1b-0.8 case1b-1.0 case1c ACLR lement [db] Fgure 5. Downlnk cell edge throughput loss. Copyrght 013 ScRes. C

6 F. XU T AL. Uplnk n-band blockng [dbm] PC1 PC recever of BS equpped wth tradtonal passve antenna around 1-4 db thus t s necessary to redefne n-band blockng requrement for the ndvdual recever of AAS n 3GPP LT exstng requrement. 6. Acknowledgements The research work s supported by atonal Scence and Technology Major Project of Chna (01ZX ZX ) Bejng Cty Scence and Technology Project (D ) Case 1a Case 1b Case 1c Fgure 6. Uplnk n-band blockng. Comparng Case 1a and Case 1c n Fgure 6 the uplnk n-band blockng nterference sgnals presented at the AAS ndvdual recever are hgher than that of the recever of BS equpped wth tradtonal passve antenna for both PC set 1 and PC set around 1-4 db. The blockng nterference sgnals of Case1a are a lttle hgher than that of Case 1b. The reason s that the aggressve BS n Case 1a equpped wth tradtonal passve antenna has hgher cable loss than AAS whch leads to hgher transmt power of aggressve U and hence hgher blockng nterference sgnals n vctm. The smulaton results ndcate that t s necessary to redefne the blockng requrement for AAS ndvdual recever to desgn the channel selecton flter ensurng performance and stablty under dfferent coexstence scenaros. 5. Conclusons In ths paper the spatal transmtter and recever characterstcs of AAS are nvestgated and evaluated through level coexstence smulaton. Smulaton result show that dfferent correlaton level has lttle mpact on throughput loss caused by spatal ACLR of AAS and ACLR of 45 db per transmtter of AAS s suffcent to meet the coexstence requrement. And n-band blockng level of AAS ndvdual recever s hgher than that of the RFRCS [1] 3GPP Overvew of 3GPP Release 11 V0.1.4 Mar.013. [] 3GPP TR V0.3.0 Study of AAS Base Staton Dec.01. [3] 3GPP TR V UTRA Rado Frequency (RF) System Scenaro June.01. [4] 3GPP TR V8.0.0 UMTS 900 MZ Work Item Techncal Report Sept.009. [5] 3GPP TR V9.0.0 Further Advancements for -UTRA Physcal Layer Aspect Mar.010. [6] O.. C. Ylmaz S. amalanen and J. amalanen System Level Analyss of Vertcal Sectorzaton for 3GPP LT I 6 Internatonal Symposum On Wreless Communcaton System (ISSWCS) pp Oct.009. [7] 3GPP R On the Down-tlt for 3D Coexstence Smulaton 3GPP TSG RA WG4 Meetng#64. [8] 3GPP R Text Proposal for Smulaton Assumptons for AAS 3GPP TSG RA WG4 Meetng#63. [9] 3GPP R Correcton on Composte Array Radaton Pattern for AAS 3GPP TSG WG RA4 Meetng#64bs. [10] 3GPP R Dscusson for Composte Radaton Pattern of AAS 3GPP TSG RA WG4 Meetng#64. [11] P. C. Kang Q. M. Cu S. Chen and Y. J. Lu Performance valuaton on Coexstence of LT wth Actve Antenna Array System I3rd Internatonal Symposum on Personal Indoor and Moble Rado Communcaton (PIMRC) pp Sept. 01. Copyrght 013 ScRes. C