Keywords LTE, Uplink, Power Control, Fractional Power Control.

Transcription

1 Volume 3, Issue 6, June 2013 ISSN: X Internatonal Journal of Advanced Research n Computer Scence and Software Engneerng Research Paper Avalable onlne at: Uplnk Power Control Schemes for LTE P Sravya, Rajendra Prasad K, Purnachand S Assstant Professor Department of Electroncs and Communcaton Engneerng K L Unversty, Guntur, Inda Abstract In mult user envronment number of users share the same rado resources. A consequence of the lmted avalablty of rado channels n the network s that the same channel has to be assgned to many users. Thus a sgnal ntended for a certan user wll reach other users and ntroduce nterference to ther connecton, and degrade the qualty. Uplnk power control s a key rado resource management functon. It s typcally used to maxmze the power of the desred receved sgnals whle lmtng the generated nterference. Ths paper presents the 3GPP long Term Evoluton (LTE) power control mechansm, and compares ts performance to two reference mechansms. The LTE power control mechansm consttutes of a closed loop component operatng around an open loop pont of operaton. The open loop component compensates path loss and shadowng through fractonal power control enablng a tradeoff between cell edge throughput and mean cell throughput. The closed loop component allows further mprovement n the performance of the system by compensatng fast varatons n channel. Ths paper presents the performance analyss of LTE power control schemes. Smulaton results ndcate that fractonal power control s advantageous compared to the conventonal open loop power control n terms of mean cell throughput. Keywords LTE, Uplnk, Power Control, Fractonal Power Control. I. INTRODUCTION Power control s a crucal rado network functon n cellular systems. Ths paper descrbes the LTE power control for the Physcal Uplnk Shared Channel (PUSCH), dscusses dfferent applcatons of t, and evaluates ts performance for dfferent parameter settngs. The focus s on the beneft of fractonal pathloss compensaton, frst proposed n. A performance comparson to an SINR balancng power control scheme s also ncluded. Implementaton of LTE s based on new multple access schemes on the ar nterface: OFDMA (Orthogonal Frequency Dvson Multple Access) n downlnk and SC-FDMA (Sngle Carrer Frequency Dvson Multple Access) n uplnk. Usage of SC-FDMA n uplnk elmnates ntra-cell nterference. But as 3GPP LTE s desgned for frequency reuse 1 the exstence of nter cell nterference cannot be neglected. Snce both data and control channels are senstve to nter cell nterference there should be Power Control (PC) functonalty n uplnk to mnmze the effect of nter cell nterference. In LTE, the standardzed uplnk power control formula contans an open loop component and a closed loop component. In open loop power control (OLPC), the transmttng power s set at the user equpment (UE) usng parameters and measures obtaned from sgnals sent by the base staton. In ths case no feedback s sent to the UE regardng the power to be used for transmsson. The closed loop component s consdered to mprove the performance of FPC by compensatng fast varatons n channel. In closed loop power control (CLPC) the base staton sends feedback to the UE, whch s then used to correct the transmttng power. Qualfyng the power control technque as open loop and closed loop helps to have an antcpated dea of the mplementaton complexty and expected level of performance. For example, t s presumed that a closed loop power control scheme would requre hgh sgnal overhead of transmsson but at the same tme t would provde wth a fast mechansm to compensate for nterference and channel condtons. On the other hand, an open loop power control would result n smpler mplementaton and low sgnalng but would be unable to compensate for channel varatons for ndvdual users. The rest of the paper s organzed as follows: Secton II provdes a detaled descrpton of open loop power control component. Secton III brefly descrbes the closed loop power control component. Secton IV gves the detals of smulaton setup and results followed by conclusons and future work n secton V. II. OPEN LOOP POWER CONTROL Ths secton focuses on the open loop component of the LTE standardzed power control scheme. The power control n LTE UL has an open loop and a closed loop component. The open loop component s meant to compensate the slow varatons of the receved sgnal, that s, path loss plus shadowng. The closed loop component s meant to further adjust the users' transmsson power so as to optmze the system performance. A. Power Control Scheme n LTE UL The settng of the UE transmts power P tx for the uplnk transmsson n a gven subframe s defned n Equaton (1), n db scale. P tx = mn{p max, P log(m) + α PL + δ mcs + f(δ )} (1) 2013, IJARCSSE All Rghts Reserved Page 1668

2 Where: P max : Maxmum power allowed by the trasmsson n for uplnk. It depends on the UE. M: The number of allocated Physcal Resource Blocks (PRBs) per user P 0 : The power to be contaned n one PRB. It s cell specfc parameter and measured n dbm/prb α : Path loss compensaton factor. It s a cell specfc parameter n the range [0 1] PL : Estmated uplnk path loss at the UE δ mcs : MCS dependent offset. It s UE specfc f(δ ) : Closed loop correcton functon The parameters P 0 and α are same for all cells and sgnaled from the BS to the UEs as broadcast. Path loss s measured at the UE based on the reference symbol receved power (RSRP). Ths nformaton enough to let the UE ntally set ts transmttng power and thus they are called as open loop parameters. δ mcs s a UE-specfc parameter dependng on chosen modulaton and codng scheme. However, δ mcs s not ncluded n ths study. Δ s a closed correcton value and f s a functon that permts to use absolute or cumulatve correcton value. Δ s sgnaled by the BS to any UE after t sets ts ntal tranamt power.e., Δ have no contrbuton n the settng of ntal transmt power by UE. B. Fractonal Power Control Concept The expresson, based on whch a UE sets ts ntal transmttng power can be obtaned from Equaton (2) by gnorng δ mcs and closed loop correcton factor. Whle power lmtaton can be negelected snce t corresponds to the UE to respect t. P tx = P log(m) + α PL [dbm] (2) The power assgnment for the transmsson at the UE performed n such a way that each PRB contans equal amount of power. Thus the expresson used by the UE to assgn power to each PRB can be obtaned by neglectng M, and s gven by PSD tx = P 0 + α PL [dbm/prb] (3) Then Equaton (3) can be rewrtten n terms of path gan as Equaton (4) n db and as n Equaton (5) n lnear. PSD tx = P 0 α PG [dbm/prb] 4) psd tx = p 0 pl α mw/prb 5) Where, PL s the path loss of the user to the servng Base Staton. To explore the open loop power control concept, frst the effect of the parameters P 0 and α on PSD tx s studed. Note that the PSD tx s lnearly dependng on P 0, whle α weghts ts dependency wth the path loss. P 0 s constant for all users whle the term α PL vares for each UE accordng to ts experenced path loss. Attenton s drawn to ths, snce t s the element that wll dfferentate a user's performance. Fg. 1 PSD tx Vs. Path loss (PL) for α = 1 and α = 0.6 Fg. 1 shows the effect of α on PSD tx for a wde range of PL values. The case α = 1 results n a PSD tx that ams to compensate the degradaton caused by the path loss. The compensaton s done allowng user to transmt wth more power f such path loss s hgher. The second case, α = 0.6, shows the same tendency for the result but wth a less spread dstrbuton.e., wth dfferent slope and the slope s equal to α when the plot s seen n db. For example, the dfference on PSD tx values for the two α values around 75dB of path loss s less than that of around 125dB of path loss. It can be noted that the user more path loss (.e., cell edge user) s transmttng more power wth ncrease n α. The case of α = 0 represents no PC, snce all users transmt wth the same power, whle wth α = 1, they transmt wth a power that ntends to totally compensate for ther path loss, referred to as full compensaton also known as Conventonal power control scheme. 2013, IJARCSSE All Rghts Reserved Page 1669

3 Values of α between 0 and 1 are cases to compromse between the full compensaton and no PC where only a fracton of the path loss s compensated to the user. Thus, the scheme s known as Fractonal Power Control scheme. C. Effect of P 0 and pathloss compensaton factor α on SINR The SINR s one of the factors that determne the performance. Therefore, a dscusson on the mpact of the OLPC parameters P 0 and α on SINR would be very helpful for the operator. The SINR for a user s gven by s = psd rx I + n (6) Where s denotes the SINR of user, psd rx s the receved psd of user at ts servng BS. I s the nterference densty level, whle n s the thermal nose densty level both receved at the BS servng user. The receved power densty, psd rx can be gven as psd rx = psd tx pl mw/prb (7) Where psd tx, s the transmtted power densty of user and pl s the total path loss from user to ts servng BS. From Equatons (5) and (7) psd tx s further smplfed to psd rx = p 0 pl (α 1) [mw/prb] 8) It s mportant to note that n conventonal PC scheme.e., when α=1 the receved power densty at the BS s equal to P 0, whch s same for all users. For 0 < α < 1 the receved power densty depends on path loss of user. So psd rx wll be dfferent for each user n the case of Fractonal PC scheme. By replacng the receved power densty n Equaton (6), the SINR of user s gven by Rewrtng the above Equaton n db as s = p 0 pl (α 1) I + n (9) S = P log M + α 1 PL IoT N [db] 10) Where, IoT s the Interference over Thermal, s calculated as the rato of nterference plus thermal nose over thermal nose n lnear doman, and N s the thermal nose. Assumng a constant level of nterference and nose, a hgher P 0 means shftng the SINR curve to the rght and hence an overall SINR ncrease. But n a real system, an ncrease n P 0 wll rse the power of all users and hence the level of nterference. Thus the ncrease n overall SINR s lesser than the expected ncrease n SINR. For example, as shown n Fg. 2(a) an ncrease of 7dB n P 0 results approxmately 1dB rse n SINR dstrbuton. Smlarly, a change of α changes each user transmttng power, makng t lowers for lower α values. A lower α not only decreases the SINR, but also spreads the curve whch leads to a hgher dfferentaton n terms of SINR between cell edge and cell center users. In SINR terms, P 0 controls the mean SINR and α controls the varance of SINR. Fg. 2 CDF of SINR per user (a) for two dfferent values of P 0 and a fxed α (b) for two dfferent values of α and a fxed P , IJARCSSE All Rghts Reserved Page 1670

4 Fg. 3 Cell edge user throughput Vs. P 0 for α = 0.8 and α = 1.0 The cell edge user throughput s defned as the 5 t percentle pont of the Cumulatve Dstrbuton Functon (CDF) of user throughput. It s an ndcator of the outage performance. Fg. 3 gves the dependency of the cell edge throughput wth P 0 for a gven α. Both α cases show an ncrease of cell edge throughput up to certan P 0, after whch the cell edge throughput shows a sgnfcant drop. Snce, an ncrease n P 0 wll ncrease the power of users, cell edge users wll reach the maxmum power lmt beyond certan P 0 and contnue to transmt same power. Furthermore, the users wth good rado condtons wll boost ther power as P 0 ncreases tll the maxmum lmt reaches and cause more nterference. Ths degrades the cell edge performance consderably beyond certan P 0 value. It can be observed n the Fg. 3 that peak cell edge throughput pont for dfferent α values corresponds to dfferent P 0 values. Ths shows that both the OLPC parameters need be tuned to acheve the better performance. III. CLOSED LOOP POWER CONTROL Ths secton focuses on the closed loop term of the LTE standardzed PC scheme to analyze the performance of conventonal closed loop power scheme. A. Closed Loop PC Concept The In a closed loop power control system, the uplnk recever at the BS estmates the SINR of the receved sgnal, and compares t wth the desred SINR target value. When the receved SINR s below the SINR target, a Transmt Power Control (TPC) command s transmtted to the UE to request for an ncrease n the transmtter power. Otherwse, the TPC command wll request for a decrease n transmtter power. The 3GPP specfcatons allow 2 types of TPC commands: 1) Absolute: the user apples the offset gven n the PC command usng the ntal transmt power n OLPC as reference. 2) Cumulatve: the user apples the offset gven n the PC command usng the latest transmsson power value as reference. In LTE, closed loop power control operates around an open loop pont of operaton. The ntal power s set usng open loop power control. The ntal power s further adjusted usng closed loop correcton value. Equaton (11) defnes the closed loop power control expresson. P tx = mn{p max, P OL + f(δ )} [dbm] 1) P OL s the uplnk power set n the open loop pont of operaton and f(δ ) s the closed loop correcton functon. f(δ ) s defned by the expresson f(δ ) = f(δ 1 ) + Δ [dbm] 2) Δ s the correcton value, also referred as TPC command. The TPC commands are sent after the OLPC has set the ntal transmt power usng desred α and P 0 values. The TPC commands are generated based on the dfference between SINR target and receved SINR. The possble values transmtted by TPC command are Δ = [ 1,0,1,3]. The closed loop correcton value s obtaned from the SINR dfference as: If dfference[db] <= 1 then 1 s sent, else f 1 < dfference[db] <= 1 then 0 s sent, else f 1 < dfference[db] <= 5 then 1 s sent, else f dfference[db] > 5 then 3 s sent B. CLPC wth Constant SINR Target To understand the behavor of CLPC, average receved SINR s nvestgated for closed loop and fractonal power control operatons. In conventonal closed loop power control the SINR target s kept same for all users. Fg. 4 gves the SINR 2013, IJARCSSE All Rghts Reserved Page 1671

5 dstrbuton for CLPC and FPC. It can be seen n the plot, some of the users are not able reach the target SINR because of maxmum power lmt. Those users, who are already transmttng wth maxmum power cannot ncrease ther transmt power, and hence, the SINR. The fractonal power control allows users wth good rado condtons (users close to the base staton) to acheve hgh receved SINR, resultng n hgh mean user throughput whle keepng reasonable cell edge throughput. Whereas conventonal closed loop power control steers all users to acheve equal receved SINR, as a consequence of ths, users wth good rado condtons whch can acheve hgh receved SINR are affected, thus resultng n lower mean user throughput. CLPC allows cell edge users to reach better SINR, t provdes better cell edge throughput. Settng a hgh closed loop SINR target means users need to transmt more power to acheve target SINR. Due to power constrant some users may not reach such hgh SINR target whch results n low cell edge throughput though t provdes hgh mean user throughput. Whle lower SINR target leads to low mean and hgh cell edge throughput. Thus, settng of the closed loop SINR target s a trade-off between the cell edge and mean throughput. It s desred to desgn a closed loop power control scheme that can provde a reasonable mprovement n cell edge throughput and allowng users wth good rado condtons to acheve hgh receved SINR, thus hgh mean user throughput can be acheved by consderng dfferent SINR targets for dfferent users. Fg. 4 CDF plot of receved SINR for FPC usng α = 0.8 and CLPC wth SINR target = 3dB IV. SIMULATION SETUP AND PERFORMANCE ANALYSIS A. Smulaton Model To analyze the performance of uplnk power control schemes n LTE a smple system model s needed. For ths purpose, a smplfed statc smulaton approach has been used whch focuses manly on power control by assumng deal channel, path loss and nterference estmatons. The approach conssts prmarly n takng a certan nstance of the system where a confguraton of users transmts wth a certan power, and proceeds to calculate the nterference and sgnal dstrbutons. In ths paper, the performance analyss s done by consderng uplnk receved SINR and transmttng power, average cell throughput and cell edge user throughput as the performance ndcators. The scope of usng dfferent performance ndcators s to provde wth a relatve measure of the gan of a specfc power control scheme n terms of system as well as user performance. TABLE I SIMULATION PARAMETERS Parameter Value Carrer frequency 2.4 GHz Doppler Spread 7Hz Cell layout 19 cell No. of BSs 19 No. of Sectors per BS 3 Users per Sector 10 Number of strong nterferer 8 Number of antennas at the BS 2 Number of antennas at the UE 1 Recever structure MRC FFT sze 1024 System Bandwdth 10 MHz UE Bandwdth 900KHz [5 PRBs] Scheduler Round Robn 2013, IJARCSSE All Rghts Reserved Page 1672

6 Thermal Nose per PRB -116 dbm Base staton nose fgure 5 db Maxmum UE Transmttng 23dBm Power B. Results and Performance Analyss Fg. 5 shows the SINR dstrbuton performance of FPC wth α = 0.8 and conventonal open loop power control. It can be observed the range of receved SINR values s more wth α = 0.8 than that of wthα = 1.0. When α = 1 (full compensaton) the receved power densty of all the users s same because of total compensaton of path loss. Ths reduces the varance n SINR dstrbuton. Whle a lower α means the receved power densty s dfferent for dffernt users dependng on the path loss of the user. Thus, a lower α leads to a hgher dfferentaton n terms of experenced SINR between cell edge and cell center users. A lower α decreases the perceved path loss of the users located at the cell edge more than those located close to the cell center. Ths leads ncrease n average cell throughput as cell center users experence a hgher SINR. However, such mprovement s at the cost of a decrease n power of cell edge users, and hence, cell edge throughput. Fg. 6 shows that the cell edge throughput s slghtly better wth α = 1.0 than wth α = 0.8. But n case of average throughput, FPC wth α = 0.8 features better performance. Fg. 5 CDF plot of receved SINR for FPC wth α = 0.6, α = 0.8 and α = 1.0 Fg. 6 CDF of user throughput for FPC wth α = 0.6, α = 0.8 and α = 1.0 TABLE II PERFORMANCE OF FPC FOR DIFFERENT PATH LOSS COMPENSATION FACTORS α P 0 [dbm /PRB] Average cell Throughput [Mbps] Cell edge Throughput [Kbps] Table. II gves the performance of fractonal power control wth dfferent path loss compensaton factors. The FPC algorthm ams at decreasng the perceved path loss of the users located at the cell edge more than those located close to the cell center. Thus, lower α means hgher dfferentaton n SINR of cell edge and cell center users. FPC scheme allows cell center users to acheve hgher SINR, and hence, hgher throughput. However, such SINR mprovement s at the cost of a decrease n power of cell edge users, whch means lower SINR, resultng n a poorer performance. As α gets close to 2013, IJARCSSE All Rghts Reserved Page 1673

7 the value 1 the spreadness n SINR dstrbuton decreses whch leads to decrease n average cell throughput and ncrease n cell edge throughput. V. CONCLUSIONS AND FUTURE WORK A. Conclusons Ths secton summarzes the man conclusons of ths work and presents further practcal consderatons along wth related future work. Ths paper s focused on the power control for EUTRAN LTE cellular system. The power control s specfed to functon both wth open loop and closed loop mechansms. The open loop functonng s based on the Fractonal Power Control technque whch s desgned to allow full or partal compensaton for the path loss. On the other hand, the algorthms used to mplement the closed loop term are vendor specfc and stll under research. The uplnk power control n LTE s flexble, smple and robust. It conssts of a closed loop component operatng around a reference obtaned by parameterzed open loop. It enables a varety of mplementatons wth dfferent objectves supportng dfferent deployment scenaros and servces. A capacty mprovng feature s the fractonal path loss compensaton of the open loop. It enables a trade-off between cell edge btrate and cell capacty. It has clear advantages compared to tradtonal full compensaton open or closed loop. Smulaton results ndcate that the fractonal compensaton can Improve the cell-edge btrate wth up to 20% for a gven average btrate Improve the average btrate wth up to 20% for a gven cell-edge btrate Improve the capacty wth up to 20% at the same tme the power consumpton s reduced. The fractonal compensaton s confgurable wth a smple broadcast factor used by the UE n the open loop algorthm. B. Future work In ths paper, a comparatve analyss of open loop power control schemes has been done. The closed loop power control concept ntroduced by consderng same SINR target for all the users. Instead of usng same SINR target for all users, who are havng dfferent rado condtons, t s worthy to consderng closed loop power control scheme wth dfferent SINR target for each user based on rado condtons of the users. Furthermore, the power control schemes were analyzed by assumng a fxed bandwdth allocaton for each user. Most of the Rado Resource Management (RRM) functonaltes are neglected to focus the study on power control. Thus, the RRM functonaltes are stll open aspects that could be studed. LTE offers dfferent Modulaton and Codng Schemes (MCS), and these should be ncluded n further study. REFERENCES [1] [2] 3GPP TS V9.1.0, E-UTRA Physcal layer procedures [3] 3GPP TS V8.8.0, Evolved Unversal Terrestral Rado Access(EUTRA); Physcal Channels and Modulaton [4] R , Intra cell Uplnk Power Control for E-UTRA -Evaluaton of Fractonal Path Loss Compensaton [5] R , Uplnk Power Control for E-UTRA - Range and Representaton of P 0 [6] A. Smonsson and A. Furuskar, Uplnk Power Control n LTE Â Overvew and Performance, IEEE Transactons on communcatons,2008 [7] Anl M. Rao, Reverse Lnk Power Control for Managng Inter-cell Interference n Orthogonal Multple Access Systems. [8] La Kng (Anna) Tee, Cornelus van Rensburg, Jann-An Tsa and Farooq Khan, Uplnk Power Control for Next Generaton Moble Broadband Wreless Access Systems, IEEE Transactons on communcatons, 2006 [9] Moray Rumney, LTE and the Evoluton to 4G wreless: Desgn and Measurement Challenges [10] Stefan Parkvall and Davd Astely, The Evoluton of LTE towards IMT-Advanced 2013, IJARCSSE All Rghts Reserved Page 1674