Sunteți pe pagina 1din 5

A Novel Transmit Power Control Bit Estimation Technique for FDD-UMTS Downlink

Ozgur Ekici, Abbas Yongacoglu


School of Information Technology and Engineering University of Ottawa Ontario, Ottawa K1N 6N5 Email: {oekici,yongacog}@site.uottawa.ca

Mohamed Siala
SUPCOM Cit Technologique e des Communications Ghazala Ariana, Tunisia mohamed.siala@supcom.rnu.tn

Abstract A novel transmit power control (TPC) bit estimation technique for FDD-UMTS downlink is proposed. Conventional FDD-UMTS downlink TPC bit estimation algorithm is based on coherent detection, which utilizes pilot bits for channel estimation and uses this estimate to remove the wireless channel effect on TPC bits. However, joint utilization of pilot and TPC bits can improve the detection performance as well as minimize the inherent overhead caused by excessive pilot bits usage in the transmission. In the proposed algorithm, pilot and TPC bits are used collectively for non-coherent detection of the power control command. Power increase and decrease orders are represented by two orthogonal vectors which are made up of pilot and TPC bits. TPC bit error rate (BER) performance comparisons of conventional and proposed algorithms are presented for various multi-path channels and slot formats. It is shown that in all the slot formats and different channel conditions, the proposed algorithm provides superior performances in terms of TPC bit error rate. This performance gain can go up to 15 dB at practical TPC BER (102 ) with certain channel conditions and slot formats.

Keywords-UMTS, non-coherent detection, TPC I. I NTRODUCTION The 3G mobile systems are based on code division multiple access (CDMA) technology. Tight and fast power control is perhaps the most important aspect in wideband-CDMA (WCDMA) in order to alleviate fast fading problem. In order to have a balanced uplink user powers at base stations (BTS), the reliability of TPC bits in downlink has to be improved. This issue is considered during the 3GPP development process [1]. Two possible solutions (equivalent from a performance point of view) proposed in the working group meetings are: i) increasing the power of TPC eld; or ii) increasing the number of TPC bits to be transmitted in downlink. In this paper, we propose an alternative way to improve TPC detection performance, which combines pilot and TPC bits for joint power control command prediction and channel estimation. After the publication of some papers (e.g. [2], [3]) favouring coherent detection performance in CDMA uplink connection with the help of pilot channels over non-coherent detection methods, researchers mainly focused on improving the channel estimation performance of the WCDMA and using this coherent estimate to determine the TPC bits as well as data

bits. Channel estimation techniques utilizing a separate pilot channel [4], time multiplexed pilot symbols [5] and data aided methods [6] have all been considered in the literature. In the conventional UMTS power control algorithm, coherent detection of the power control bits is utilized, where pilot bits are used to sound the channel and the effect of hostile wireless channel is removed from TPC bits by the information obtained from this channel sounding. In coherent detection, channel estimation can be performed by a separate common pilot channel (CPICH) (i.e. Primary CPICH in entire cell or Secondary CPICH in a part of a cell) as well as time-multiplexed pilot symbols in downlink dedicated channel (DCH). However, if a user-specic beamforming method is employed, in which each user is assigned a specic beam pointing to user direction, utilizing CPICH for channel estimation would degrade the performance signicantly because CPICH that is broadcasted to entire cell would experience a different radio channel than the DCH [7]. Therefore, mobiles operating in a UMTS network that use user specic beamforming have to use so called dedicated pilot bits for channel estimation. The estimation performance increases as the number of pilot bits (or power) increases. However, using many pilot bits for coherent channel estimation introduces considerable amount of communication overhead. This overhead can go up to 60% (e.g. Frame # 0 and # 1 in [8]) in low bit rate connection in hostile channel environments. The proposed method of this paper alleviates this problem by detecting power change commands non-coherently by combining TPC and pilot bits together which necessitates less control bits (i.e. pilot & TPC) for the same BER performance or improves TPC reliability for the same amount of control bits. Noncoherent detection of bits is not a new application in CDMA systems and it is utilized for uplink bit estimation of IS-95 systems [9]. In this system, information bits are decoded by correlating Walsh-Hadamard orhogonal codes at the receiver. In this paper a novel TPC bits decoding algorithm is presented. Instead of dividing the tasks of pilot and TPC bits as channel estimation and power control information delivery, it is proposed to use them jointly for TPC bit and channel estimation. The structure of the paper is as follows. A

1-4244-0063-5/06/$20.00 2006 IEEE

Channel amplitude (dB)

general description of UMTS downlink transmission is given in Section II. Section III explains the multi-path wireless channel environment. Conventional and proposed channel estimation and TPC bit detection algorithms are presented in Section IV. Simulation results including TPC bit error rate and performance comparison of conventional and proposed algorithms are presented in Section V. Finally, conclusions and future ideas are given in Section VI. II. FDD-UMTS D OWNLINK OVERVIEW Downlink frame structure, position of bits in the frame, number of pilot, TPC and data bits are all important for TPC bit detection. Therefore, in this section FDD-UMTS downlink communication properties are mentioned briey. The downlink physical channel type that is relevant to the subject of this paper is the dedicated physical channel (DPCH). The estimation of TPC bits is performed by only using the dedicated pilot symbols because of the reasons indicated in the previous section. There is only one type of downlink DPCH in FDD-UMTS. Within DPCH, dedicated data generated, control information (i.e. pilot, TPC and optional transport format combination indicator [TFCI] bits) are transmitted in a timemultiplex manner [8]. A single transmission frame (10 ms) is made up of 15 slots and each slot is made up of 2560 chips. Within each slot the power is the same. The number of chips is constant for each slot. Therefore data transmission rate depends on the spreading factor (SF). SF may change from 512 down to 4 in the downlink, which denotes symbol rates ranging from 7.5 to 960 ksps. In 3GPP standard, there are several slot formats for downlink channel which have different number of pilot, TPC and TFCI bits. The relative number and position of pilot and TPC bits in a slot effects the TPC and data bit correct detection probabilities. For example, if we increase the number of pilot and TPC bits, the reliability of both TPC and data bits would increase with a penalty of reduced capacity. The exact number of bits in the downlink DPCH (Npilot , NT P C , Ndata ) that are considered in this study is given in Table I. The channel bit rates given in Table I are the rates immediately before spreading.
TABLE I S IMULATED UMTS SLOT FORMAT PROPERTIES Format 8 12 13 14 Sym Rate (ksps) 30 60 120 240 SF 128 64 32 16 N-total 40 80 160 320 N-data 34 68 148 296 N-tpc 2 4 4 8 N-pilot 4 8 8 16

ITU Pedestrian A Channel

0 10 20 30 0 40 0 100 200 0.5 1 Time (sec) 300 1.5 400 500 2

Delay Profile (ns)

Fig. 1. Power delay prole of simulated ITU Pedestrian A channel that has four multi-paths with user speed 30km/h

fading at the receiver side. This fast fading effect is represented as Rayleigh distributed random variable. On the other hand if the reecting objects are located very far from each other, the multipath prole can have large delays between arrived signals. If the time difference between arriving multi-path signals is more than one chip duration, the RAKE receiver can be utilized to collect the signal energy arriving from multi-path signals. In the simulations, four different propagation conditions are implemented according to International Telecommunications Union (ITU) recommendations [10]. The power delay prole and Doppler frequency properties of the simulated ITU channels are given in Table II. Table II indicates the multi-path proles of the channels as well as the Doppler effect of the mobiles by giving mobile user speed. As an illustration of a power delay prole of a simulated channel, ITU Pedestrian A channel with mobile user speed of 30 km/h is plotted in Figure 1 with corresponding relative path powers and delay prole. In the considered multipath fading channel, the received signal is represented as follows:
Ntaps 1

r(t) =
n= l=0

l ejl s[n]d(tnT l ; n)+w(t) (1)

III. W IRELESS C HANNEL Both conventional and proposed TPC bit estimation techniques are tested in a multi-path wireless channel environment. In a wireless medium, the transmitted signal is reected by many obstacles between a transmitter and a receiver, thus creating a multi-path channel. If the reecting objects are close to each other, superposition of the reected signals causes fast

where T is the symbol interval, s[n] S is the nth symbol, l is the average amplitude and l is the channel phase of the lth channel tap at delay l and d(t; n) is the spreading waveform at the nth symbol. Here, S is the symbol alphabet for QPSK modulation (i.e. 2E (1j), E being transmitted T signal energy per symbol) and w(t) is white Gaussian noise with zero mean and double sided spectral density N0 /2. The spreading waveform is given by:
NSF 1

d(t; n) =
l=0

c[l; n](t lTc )

(2)

TABLE II ITU M ULTI -PATH C HANNEL P ROFILES

Tap 1 2 3 4 5 6

ITU Pedestrian A,v =3 km/h Relative Average Delay [ns] Power [dB] 0 0.0 110 -9.7 190 -19.2 410 -22.8 NA NA NA NA

ITU Pedestrian B, v =7 km/h Relative Average Delay [ns] Power [dB] 0 0.0 200 -0.9 800 -4.9 1200 -8.0 2300 -7.8 3700 -23.9

ITU Vehicular A, v =30 km/h Relative Average Delay [ns] Power [dB] 0 0.0 310 -1.0 710 -9.0 1090 -10.0 1730 -15.0 2510 -20.0

ITU Vehicular B, v =120 km/h Relative Average Delay [ns] Power [dB] 0 -2.5 300 0 8900 -12.8 12900 -10.0 17100 -25.2 20000 -16.0

where Tc is the chip interval, NSF = T /Tc is the spreading factor, c[l; n] S is the lth element of the spreading sequence for the nth symbol, and (t) is the chip waveform. IV. UMTS D OWNLINK R ECEIVER S TRUCTURES A. Conventional Coherent Rake Receiver In the conventional UMTS receiver, coherent demodulation is utilized for symbol detection. To coherently combine multipath components, the RAKE receiver is employed. Each RAKE nger is allocated to a multi-path component which carries a signicant amount of power. The time delays of the multi-path components are assumed to be perfectly tracked. Let y[n] denote chip match ltered and sampled sequence of y(t) at the receiver. The despread pilot symbols on a RAKE nger can be written as: P [n] = 1 NSF
NSF 1 l=0

Fig. 2.

Channel estimation by pilot symbols at a RAKE nger.

Ntaps

s(n) =
l=1

h rl [n] l

(6)

where s(n) is the nth estimated symbol and r[n] is the despread trafc transmission. B. Proposed Receiver The proposed receiver utilizes a non-coherent detection algorithm where power increase and decrease decisions are represented by two different vector notations. The vector is made up of pilot and TPC bits. Lets consider slot format 12, where we have 8 pilot and 4 TPC bits. Therefore power increase and decrease vectors of length 12 can be represented by two orthogonal vectors as illustrated in eq. 7. The robustness of the power control algorithm increases as the total number of pilot and TPC bits in a slot increases. One concern in this method is the possible trade off between TPC and data bit reliability. In the proposed algorithm, after deciding the TPC bit by non-coherent detection (e.g. power increase), the whole vector made up of pilot and TPC bits (e.g. pow+ vector in 7) is used for channel estimation which is required to demodulate the data bits. Therefore, data BER depends on TPC BER in the proposed algorithm.
pow pow
+

y[nNSF + l]c [l; n]

(3)

where denotes the complex conjugate. From 1, the pilot signal on the lth RAKE nger can be calculated as: P l [n] = l ejl s + I[n] (4)

where s is the known pilot symbol and I[n] is the interferenceplus-noise. For coherent symbol detection, the channel attenuation parameter hl = l ejl should be estimated by using despread pilot signals P l [n]. By using the dedicated pilot symbols, the channel parameters on the lth RAKE nger can be extracted as: hl = [l ejl ] = 1 Npi |s|2
Npi 1

P l [n]s n
n=0

(5)

where Npi in 5 is the number of dedicated pilot symbols per slot. Channel estimation algorithm for a single RAKE nger is illustrated in Figure 2. Channel estimation must be done individually in each temporal RAKE nger. Then the received symbols are rotated back, so as to undo the phase rotation caused by the channel. Such channel-compensated symbols can then be simply summed together to recover the energy spread over all delay positions [11]. This processing is called maximal ratio combining, and is expressed as

= [+1 + 1 + 1 + 1 + 1 + 1 + 1 + 1 + 1 + 1 + 1 + 1] = [+1 + 1 + 1 + 1 + 1 + 1 1 1 1 1 1 1]

(7)

V. S IMULATION R ESULTS In this section, BER performances of conventional and proposed TPC bit detection techniques are compared in different channel conditions and for different slot formats. At the receiver side, RAKE receiver with perfectly placed ngers are utilized for multi-path channels. A simulator of UMTS

downlink transmitter and receiver is developed to evaluate the BER performance. The simulator is implemented in C++ computer language. It utilizes IT++ signal processing and communications library [12]. The performance of the UMTS system with different detection algorithms could be compared by various parametric simulations. Different slot formats (effecting TPC and pilot bit numbers), different channel multipath proles (i.e. ITU pedestrian or vehicular channels) or user speeds all affect the algorithms performance. The positions and the number of the data, TPC and pilot bits are kept the same in the simulations as in the standard. In the simulations, TFCI bits are assumed as data bits. The modulation technique in WCDMA downlink is QPSK [13]. The complex symbols are spread to a chip rate of 3.84 Mcps. The spreading codes used in UMTS are Orthogonal Variable Spreading Factor (OVSF) codes. The spreading codes of length 16, 32, 64 and 128 are used in simulations. The main objective in this work is to demonstrate the superiority of the proposed technique in all the conditions. Therefore, in the rst set of simulations, we kept the slot format constant and analyzed the algorithm performances with different multipath and user speeds. In the second set of simulations, the channel condition and user speed are kept constant and the performance of the algorithms are compared with different slot formats. Figure 3 illustrates the TPC bit detection performance of conventional and proposed algorithms with a certain slot format and in different propagation environments. Slot number 12 with spreading factor (SF) 64, channel bit rate of 60 ksps, 4 TPC and 8 pilot bits is chosen as the simulated slot. The rst thing we notice is that, with the 4 multi-path ITU Pedestrian A Channel with strong rst tap gain, small multi-path spread and low mobility, both algorithms perform almost the same. But, as the multi-path spread and user mobility increase the proposed algorithm outperforms the conventional algorithm by 2 dB and 6 dB at BER of 102 on the ITU Pedestrian B Channel with user speed 7 km/h and ITU Vehicular A Channel with user speed 30 km/h sequentially. And additionally, the performance of the conventional detection has an error oor for ITU Vehicular Channel with user speed of 120 km/h whereas the proposed algorithm satises BER value of 102 with reasonable Eb/N0 value. Figure 4 presents the second set of simulation results illustrating the performances of both conventional coherent and proposed non-coherent algorithms with the same multipath radio channel with constant user velocity, but different slot formats. The simulated slot formats represents data rates ranging from 60 kbps to 960 kbps. Therefore, in order to accommodate all bit rates and possible channel conditions for the indicated channel rates, a reasonable multi-path radio channel with acceptable user speed should be chosen. ITU Vehicular A Channel with user speed of 30 km/h is chosen for evaluation. In all the slot formats chosen, the proposed non-coherent TPC bit detection technique gives a better result than the coherent detection. On average 5 dB performance gain is observed with the proposed algorithm at a BER of 102 .

VI. C ONCLUSION AND F UTURE WORK The conventional coherent TPC bit detection algorithm of WCDMA assumes that the TPC bit reliability is only the function of channel estimation success and the relative power of TPC bits. In this paper, a new TPC estimation algorithm using non-coherent detection is proposed which utilizes both pilot and TPC bits to predict power control commands. It has been demonstrated that the proposed algorithm provides an average of 5 dB gain in various wireless communication channels and slot formats at a practical TPC BER of 102 , compared to the conventional coherent algorithm. In the simulations, we just compared the TPC error rate performances of both algorithms. In order to have a clear opinion and fair comparison of both algorithms,the data BER of the proposed algorithm should also be estimated. Our preliminary studies show that data BER performance is affected slightly by using the proposed TPC estimation technique and it depends on TPC detection performance. R EFERENCES
[1] 3GPP-TSG-RAN Working Group 1 meeting 3. Ber on power control bits, March 1999. [2] Fuyun Ling. Pilot assisted coherent ds-cdma reverse-link communications with optimal robust channel estimation. IEEE ICASSP, pages 263266, November 1997. [3] F Adachi H Andoh, M Sawahashi. Channel estimation using time multiplexed pilot symbols for coherent rake combining for ds-cdma mobile radio. Proc. PIMRC97, pages 954958, 1997. [4] J. Choi. Multipath cdma channel estimation by jointly utilizing pilot and trafc channels. IEEE Proc.-Comm., 146(5):312318, October 1999. [5] Werner Weichselberger Thomas Baumgartner, Alberto Gil Ferras. Optimum power ratio between pilot and information bits if user specic beamforming is used in UMTS FDD. ISWPMC 2003, 2:10251030, 22 September 2003. [6] P.M. Grant G. J. R. Povey and R. D. Pringle. A decision directed spread spectrum rake receiver for fast fading mobile channels. IEEE Trans. Veh. Tech., 45:491502, 1996. [7] 3GPP-TSG-RAN Working Group 4 meeting 17. Proposal for userspecic beamforming for utra fdd, May 2001. [8] Third Generation Partnership Project. Universal mobile telecommunications system (umts); physical channels and mapping of transport channels onto physical channels (fdd). technical report etsi ts125 211 v5.3.0. Technical Report ETSI TS125 211, European Telecommunications Standards Institute, December 2002. [9] TIA/EIA/IS-95-A:. Mobile station-BTS compatibility standard for dualmode wideband spectrum cellular system. Technical report, 1993. [10] Recommendation ITU-R M.1225. Guidelines for evaluation of radio transmission technologies for IMT-2000. Technical report, Internationla Telecommunications Union (ITU), 1997. [11] H. Holma and A. Toskala. WCDMA for umts. John Wiley and Sons, third edition edition, 2004. [12] IT++ Library. http://itpp.sourceforge.net/., March 2005. Simulation Environment. [13] Third Generation Partnership Project. Universal mobile telecommunications system (umts); spreading and modulation(fdd). technical report etsi ts125 211 v5.3.0. Technical Report ETSI TS125 213, European Telecommunications Standards Institute, September 2002.

10

ITU Ped. A Ch. with user V = 3 km/h Coh Det. Noncoh Det.

10

ITU Ped. B Ch. with user V = 7 km/h Coh Det. Noncoh Det.

10 BER

10 BER

10

10

10

10

10 15 EbN0 (dB) ITU Vec. A Ch. with user V = 30 km/h Coh Det. Noncoh Det.

20

10

10

10 15 EbN0 (dB) ITU Vec. B Ch. with user V = 120 km/h Coh Det. Noncoh Det.

20

10

10 BER 10
2

BER 10
2

10

10

10 EbN0 (dB)

15

20

10

10 EbN0 (dB)

15

20

Fig. 3.

Performance of conventional coherent detection and non-coherent detection in different ITU multi-path channels with xed slot format (12)

10

Slot # 8, SF = 128, NTPC = 2, NPilot = 4 Coh Det. Noncoh Det.

10

Slot # 12, SF = 64, NTPC = 4, NPilot = 8 Coh Det. Noncoh Det.

10 10 BER
1

BER 10
2

10

10

10

10

10 15 EbN0 (dB) Slot # 13, SF = 32, NTPC = 4, NPilot = 8 Coh Det. Noncoh Det.

20

10

10 EbN0 (dB) Slot # 14, SF = 16, N = 8, N


TPC

15 = 16 Coh Det. Noncoh Det.

20

10

Pilot

10 BER 10
2

BER 10
3

10

10 EbN0 (dB)

15

20

10

10 EbN0 (dB)

15

20

Fig. 4.

Performance of conventional coherent detection and non-coherent detection in xed multi-path channel with different slot formats

S-ar putea să vă placă și