PAPER Radio Resource Management and Power Control for W-CDMA Uplink with High Data Rate Packet Transmission

Transcription

1 202 IEICE TRANS. COMMUN., VOL.E88 B, NO.5 MAY 2005 PAPER Radio Resource Management and Power Control for W-CDMA Uplin with High Data Rate Pacet Transmission Yoshitaa HARA a), Kuniyui SUZUKI, Koji KANEKO, and Taashi SEKIGUCHI, Members SUMMARY In wideband code division multiaccess (W-CDMA) uplin, immediate accommodation of high data rate pacet causes power control error and maes active users signal quality deteriorate in a beginning of a frame. To avoid the deterioration, we propose a new radio resource management (RRM) which accommodates high data rate traffic gradually in several frames. The proposed RRM reduces the signal quality deterioration in the beginning of the frame. We also propose an effective power control scheme, where a power increase command is sent to all users before a new high data rate pacet is transmitted. Simulation results show that joint utilization of the proposed two methods is effective to eep signal quality good for all users. ey words: code division multiaccess, radio resource management, power control, pacet transmission. Introduction Wideband code division multiaccess (W-CDMA) systems are expected to provide flexible data rate services such as voice, data, and internet access in wireless communications []. The specifications have been developed in Third Generation Partnership Project (3GPP) and, recently, the specifications for high data rate pacet transmission has become a ey issue for enhanced CDMA [2]. In W-CDMA Release99, that is the first standardized specification, the system was optimized for voice telephony. The base station (BS) manages each user s maximum data rate and a user can change his data rates autonomously in each frame based on his maximum rate constraint. Since there are many voice users, the variance of total interference power is reduced by statistical averaging effect. Apart from voice telephony, in high data rate pacet transmission, a few users transmit pacets with much larger power. Since statistical averaging effect is not sufficient, it is difficult to apply the same management with voice telephony. In such a situation, it is more desirable to manage the users data pacets frame by frame to reduce the interference variance. In addition, the frame-by-frame management has the advantage of using scheduling, based on delay insensitive characteristics of data traffic [3] [6]. In practical W-CDMA systems, the high data rate pacet services will be mixed with voice services. Therefore, the radio resource Manuscript received June 2, Manuscript revised November 4, The author is with Mitsubishi Electric Information Technology Centre Europe B.V. (ITE),, allee de Beaulieu, CS 0806, Rennes Cedex 7, France. The authors are with Information Technology R &D Center, Mitsubishi Electric Corporation, Kamaura-shi, Japan. a) hara@tcl.ite.mee.com DOI: 0.093/ietcom/e88 b management (RRM) for high data rate pacet is required to harmonize with voice telephony. One of the problems in an integrated voice and data system is that immediate accommodation of high data rate pacet impairs active users signal quality in a beginning of a frame. Normally, on W-CDMA uplin, closed-loop power control is used so that the received signal-to-interferenceplus-noise ratio (SINR) corresponds to the target SINR. However, immediate increase of much larger power by the data pacet prevents active users from maintaining their target SINRs. Then, active voice users signal quality will deteriorate instantaneously and it taes a certain duration to recover the signal quality. To avoid the deterioration of the signal quality, it is important to consider a feasible method to accommodate voice and data. Although there are many papers on CDMA radio resource management [3] [5], the previous papers have not dealt with this problem because most of them assume a perfect power control. In this paper, we propose a new RRM and a transmit power control () scheme for W-CDMA uplin, to avoid the deterioration of active users signal quality. In our RRM, high data rate traffic is accommodated gradually in several frames. Active users do not experience drastic increase of interference power and can eep their signal quality good. In addition, our scheme sends power increase commands to all users before a new high data rate pacet is transmitted. Our scheme compensates the increased interference due to the high data rate accomodation by increasing the power of desired signal. Using these two methods, we will show that all users can maintain good signal quality under high data rate pacet transmission. 2. Radio Resource Management and Transmit Power Control for W-CDMA Uplin In this section, we describe conventional RRM, new RRM, and new for W-CDMA uplin. 2. Conventional Radio Resource Management In W-CDMA uplin, RRM schemes based on interference power have been widely investigated [7] [9]. In these schemes, the BS manages traffic to satisfy the constraint of KE b SF + N 0 + I other I 0,max, () where K is the number of active users, I 0,max is the maximum allowable total interference, I other is the interference Copyright c 2005 The Institute of Electronics, Information and Communication Engineers

2 HARA et al.: RADIO RESOURCE MANAGEMENT AND POWER CONTROL FOR W-CDMA UPLINK 203 from other cells, E b is the bit energy, N 0 is the noise power density, and SF is the spreading factor. Normally, the allowable interference-to-noise-ratio (INR) η max = I 0,max /N 0 is set to 6 to 0 [db] [7]. In case of multimedia traffic with various transmission rates, the BS is required to manage traffic to satisfy [7] i= E b,i SF i + N 0 + I other I 0,max (2) where E b, and SF are the bit energy and the spreading factor for the -th user, respectively. To achieve the constraint of (), the literature [8] uses a threshold T bloc, where a new call is bloced if the observed interference level is above T bloc. This access control strategy will be useful for voice services, whereas there are some problems to apply it for multimedia communications [3] [6], [0] [5]. One problem is that new access requests with different transmission rates cannot be managed by the same threshold T bloc. Consider to accommodate a new user with data rate 2.2 bps (60 bps after channel coding) and SF = 64 or another new user with data rate 384 bps (960 bps after channel coding) and SF = 4. Assuming the same target SINR after despreading, the new user requires 6 times larger transmission power in 384 bps than in 2.2 bps. Since higher data rate user causes larger interference, admission of the new user depends on his data rate. Therefore, a feasible management scheme applicable to different rates is required. Another problem is how to accommodate the high data rate pacet without causing power control errors for active users. Immediate accommodation of the high data rate pacet causes drastic increase of interference and maes the signal quality of active users deteriorate due to power control errors. Therefore, a feasible method to avoid the power control errors is required. Figure (a) shows an example of traffic load in the conventional RRM scheme. The active users signal quality deteriorates in a beginning of a frame due to power control errors. Thus, in multimedia communications, () or (2) is not enough to eep all active users signal quality good and we propose new RRM and schemes to solve these problems. 2.2 New Radio Resource Management We propose a new RRM scheme which restricts the interference variation between frames within D [db]. Figure 2 shows the flowchart of our RRM, which manages the traffic as follows: [RRM-] The BS measures total interference power I c before despreading process in the current frame. [RRM-2] The BS determines power threshold T=min{I c 0 D/0, I 0,max },wherei 0,max is the BS s maximum allowable interference level. [RRM-3] The BS assigns new traffic on the constraint that the total interference power after convergence is less Fig. Traffic load in CDMA frames. (a) Conventional method. (b) Proposed method. Fig. 2 Flowchart of proposed radio resource management. than T in the next frame. The details of [RRM-3] is addressed in Sect. 3. The conventional RRM corresponds to the case with D. Under our RRM, all users experience smaller increase of interference and less degradation of signal quality. Figure (b) shows an example of traffic load in our RRM scheme. Although the transmission rate admitted by our RRM is not enough to accomodate a high data rate in one frame, the high data rate can be admitted gradually using several frames. For example, in case of D = 2 [db], the total interference power can be increased by 0 [db] using 5 frames, which corresponds to the maximum allowable interference level under η max = 0 [db]. Therefore, the maximum delay for a new high data rate user can be ept within 5 frames. 2.3 New Transmit Power Control In W-CDMA uplin, active user s transmit power is adjusted to meet his target SINR by a closed-loop power control scheme. In the normal power control scheme, the BS estimates the output SINR Γ for the -th user. Comparing the

3 204 IEICE TRANS. COMMUN., VOL.E88 B, NO.5 MAY 2005 Fig. 3 Flowchart of proposed power control method. estimated SINR Γ with the target SINR γ, the BS sends a command cmdtothe-thuserineveryslotin downlin. Then, cmd is given by { 0 Γ cmd = >γ (3) Γ γ The active user increases or decreases his transmit power of the next slot iteratively by a power step () [db] in case of cmd=0 or, respectively. command in the last slot is reflected on the first slot in the next frame. When a high data rate user is newly accommodated, the power control scheme tries to increase active users power. However, the received SINR does not meet the target SINR immediately, because it taes several slots to converge the transmit power. Therefore, quality degradation will occur in the beginning of the frame. To compensate the degradation, our scheme increases active users signal power using traffic nowledgecompleted by RRM. Figure 3 shows the flowchart of our scheme, which controls transmit power as follows: [-] The BS completes RRM for the next frame. [-2] If traffic increases in the next frame, the BS sends power increase commands ( cmd = ) to all active users and a new admitted user in the last slot of the current frame; otherwise the BS uses the normal scheme. Note that our scheme deals with power increase between frames and that the other parts are the same with the normal scheme. Here, the frame-by-frame power step [db] from the slotby-slot power step () [db]. The frame-by-frame power step can be achieved by broadcasting to all active users in downlin. Using the power step [db], we can change the transmit power more dynamically between [db] is the same with the slot-by-slot power step () [db]. In our scheme, all users have one step ( () [db]) higher transmit power and compensate increase of interference in the beginning of the next frame. Therefore, active users can eep his signal quality when a new high data rate user is accommodated. For further study, we also consider another case of different frame-by-frame power step frames. If [db] is larger than D [db], the interference variation can be perfectly compensated by our scheme and the quality deterioration will not occur in the beginning of the frame. Our scheme enables fast power convergence using an appropriate step. In our study, a constant step [db] is initially studied and a method to decide an appropriate value [db] is presented in Sect In the process of the power control scheme, a user who has the maximum power level cannot increase his transmit power and suffers from increased interference. The user will be forcedly terminated when his received SINR is below a certain threshold. Since the received SINR converges to the same value in the proposed and normal schemes, both schemes have the similar performance of the forced termination. More exactly, our may have a little more forced terminations in case of too large step [db], where the total received power temporally exceeds the power level to be converged. However, this effect can be dissolved by using an appropriate step presented in Sect Therefore, employment of our scheme with the appropriate step has little effect on the forced termination and we hereafter study the performance of active users who don t reach the maximum power level. 3. Traffic Management Strategy In this section, we explain details of [RRM-3], or a traffic management method on the constraint of total received power. For simple description, we assume a perfect SINRbased power control and no multipath environment in our analysis. The effects of a practical power control and multipath environments are evaluated by simulations. 3. Total Received Power in Base Station Suppose K active users in the current frame. Under a perfect SINR-based power control, we have SF P = γ, =,...,K (4) P i + P IN i=,i where γ is the target SINR for the -th user, P is the received power for the -th user, and P IN is the sum of the outer-cell interference and noise power. Using (4), the received power for the -th user can be expressed as [] P IN P = SF /γ + { ψ(k)} (5) ψ(k) = SF /γ + (6) = where ψ(k) < from P > 0. Therefore, the total received The current W-CDMA systems correspond to the case with = (). Some modifications of specifications will be required to use () (2).

4 HARA et al.: RADIO RESOURCE MANAGEMENT AND POWER CONTROL FOR W-CDMA UPLINK 205 power in the BS is given by ψ(k) P + P IN = P IN ψ(k) + P IN = P IN = (7) ψ(k) The total received power increases rapidly as ψ(k) approaches to. 3.2 RRM Strategy Suppose K active users in the current frame and K + K active users in the next frame. To eep the interference variation between frames within D = 0 log 0 (T/I c ) [db], it is required to satisfy /0 ψ(k + K) 0D ψ(k). (8) Using (6), (8) is rewritten as SF /γ D /0 = K+ K =K+ <. (9) SF /γ + Traditionally, the criterion to accommodate active users under an ideal power control is presented by [], [2] K+ K = <. (0) SF /γ + The difference of (9) from (0) is the existence of ( 0 D /0 ), which expresses the constraint of interference variation. As a special case of D, (9) is identical to (0). Considering users which finish sending pacets in the current frame, the constraint of (9) is rewritten as K+K SF = /γ D /0 SF =K+ /γ + K+K +K 2 + <, () SF /γ + =K+K + where =,...,K is the continuous active users, = K +,...,K + K is the terminated active users in the current frame, and = K + K +,...,K + K + K 2 is the new users in the next frame. Equation () includes both low and high rate users. In an actual W-CDMA system, the BS does not manage the voice users transmission rate frame by frame. The BS manages only voice user s maximum data rate and the voice user changes his data rates autonomously in each frame. Since there are many voice users, the interference variance is reduced by statistical averaging effect and the mean interference power is decreased by a voice activity factor. Therefore, the BS regards all voice users as the continuous active users, the number of whom is reduced by the voice activity factor. A voice user with a silent period is not regarded as a terminated user, because the BS is holding his admission. On the contrary, the data rates of the high rate users are managed by a frame-by-frame RRM since one high rate user has large effect on the system performance. According to the RRM s admission, the high rate users are classified into the continuous active users, the terminated active users, and the new users. In practice, there are many types of RRM algorithms to determine the terminated users and the new users in (). For example, a scheduling algorithm determines the active users data rate considering propagation condition, fairness, and quality of service (QoS). Although the conventional schedulers [3] [6] enhance the throughput, they accommodate the high data rate immediately on the constraint of a constant maximum received power and cause the quality degradation due to power control errors. Our RRM scheme can dissolve this drawbac and the scheduling algorithm on the constraint of () effectively enhances the throughput maintaining good signal quality. 4. Numerical Results Let us evaluate the performance of the proposed RRM and schemes by simulations. 4. Simulation Setup and Parameters Table lists the simulation parameters. In the simulations, the cellular system with a central cell and neighboring cells is assumed as shown in Fig. 4. Active users are randomly distributed in the central cell and the base station in the central cell is assumed to receive the outer-cell interference which power is the same with the noise. Assume that each user transmits a signal using a 0 ms frame with 5 slots in asynchronous W-CDMA uplin. The -th user has the spreading factor SF, and the transmit power W (n) in the n-th slot. Considering characteristics of long codes, an output SINR of the BS s RAKE receiver for Table Simulation parameters. Chip rate 3.84Mchip/s Channel 4-path Rayleigh channel Pass loss exponent 3.5 Modulation (Data) BPSK Modulation (Spreading) QPSK SINR estimation error σ = [db] step () = db (slot by slot) Amplitude ratio of I:Q Number of BS antennas Maximum allowable INR Number of users Pacet transmission time interval(tti) =, 2, 3, 4dB (frame by frame) β d :β c Single antenna η max = 0 [db] Traffic#:20 users Traffic#2: user frame(0 ms)

5 206 IEICE TRANS. COMMUN., VOL.E88 B, NO.5 MAY 2005 Table 2 Traffic types. TrafficType Data rate SF γ β c : β d # 2.2 bps 64.0 db :5 # 2-0(control ch. only) 2.4 bps db 5:0 #2-44 bps 8.0 db 5:5 # bps 4.0 db 5:5 # 2-3(2 codes) 768 bps db 5:5 Fig. 4 Cell configuration. Table 3 Pacet transmission rate admitted by RRM with D [db]. (Request: 768 bps) st frame 2nd frame 3rd frame RRM (D = [db]) 44 bps 384 bps 528 bps RRM (D = 2[dB]) 384 bps 768 bps 768 bps Conventional 768 bps 768 bps 768 bps the -th user is given by [6] Γ (n) = i= SF W (n) L W (n) i α i, + P IN l= ξ g,l 2 W (n) ξ (2) min{l,l l} 2 α i, = g,l L+ l L h i,l+ l (3) l=max{, l} L 2 ξ = g,l h,l, (4) l= where L is the number of paths and g,l and h,l are the RAKE combining weight and the propagation coefficient for the -th user and the l-th path, respectively. We assume 4- path Rayleigh channel with the same average power for each path and ideal RAKE combining weight g,l (= h,l ). The average power of each path is determined by the distance between the user and the central base station according to the pass loss exponent of 3.5. In actual CDMA uplin, I and Q channels have different signal power with the ratio of β 2 d : β 2 c [7]. An output SINR Γ (n) includes the signal power averaged over I and Q channels. In the simulation, the estimated SINR Γ (n) is given by Γ (n) = ɛγ (n), (5) where ɛ is the estimation error of Gaussian variable with standard deviation σ = [db]. In the transmit power control scheme, the BS compares the estimated SINR Γ with the target SINR γ and sends a command to the -th user according to (3). Table 2 lists the types of users in the simulation. Traffic type # is the low rate voice user with 2.2 bps, SF = 64, and γ = [db]. Traffic type #2 is the high rate data user, whose transmission rate adaptively changes within the maximum rate of 768 bps. A single spreading code is used for 384 bps with SF = 4andγ = [db], whereas two spreading codes are multiplexed for 768 bps with SF = 4 and γ = 2 [db]. When a high rate data user pauses a certain duration, he only transmits a control channel (traffic #2-0) to maintain power control with 2.4 bps, SF = 256, and γ = 3[dB]. In the beginning of each simulation trial, 20 active voice users and a high rate user only transmitting a control channel are considered. This situation is made by updating all users transmit power 60 times according to (3), iteratively. Then, a high rate user (traffic #2) requests a new transmission of 768 bps and the BS determines his transmission rate in the next frame based on our RRM scheme. According to the determined transmission rate from the BS, the high rate user starts to send a pacet in the next frame, increasing the transmit power by (γ /SF )/(γ 0 /256) (see Appendix), where γ 0 is the target SINR for a control channel. In the following, we evaluate the performance of our RRM and schemes. 4.2 Admitted Transmission Rate by Our RRM Table 3 lists transmission rates of traffic #2 admitted by our RRM. The admitted transmission rate depends on D and becomes higher as the number of frames increases. Under a small D, it taes many frames to accommodate the full data rate. On the other hand, under a large D, the immediate accommodation of high rate pacets maes active users signal quality deteriorate. Therefore, an appropriate D is important to achieve reasonable data rate accommodation without drastic increase of interference. From Table 3, it is seen that D = 2 [db] is an appropriate solution. 4.3 Characteristics of Received Signals Figure 5 shows the average SINR of active voice users when a high data rate pacet with 384 bps or a pacet with bps is newly accommodated in the first frame. Average output SINR Γ (n) is calculated by averaging output SINR over simulation trials and multiple active users. In the figure, the signal quality deteriorates under the normal scheme, whereas our scheme lessens the deterioration of the signal quality. As increases, the signal quality becomes better. When is larger than D, the deterioration of the signal quality can be perfectly dissolved in the beginning of the frame. Figure 6 shows the average SINR of newly assigned high rate data user. The signal quality of the high rate user is also improved by the

6 HARA et al.: RADIO RESOURCE MANAGEMENT AND POWER CONTROL FOR W-CDMA UPLINK 207 Fig. 5 Average SINR of active voice users. (a) 384 bps pacet admitted. (b) bps pacet admitted. Fig. 6 Average SINR of active data users. (a) 384 bps pacet admitted. (b) bps pacet admitted. proposed. Under = 4 [db], we can eep all users signal quality good, even when a high data rate pacet with bps is accommodated in the first frame. As becomes smaller, immediate high data rate accommodation maes active users signal quality worse and joint utilization of our and RRM schemes become more important to eep the signal quality good. 4.4 Received Power in Base Station Figure 7 shows the transient of average interference to noise power ratio (INR) E[η](η = I c /N 0 ) for the case of accommodating a pacet with 384 bps or a pacet with 768 bps. In the figure, the interference variation can be ept within 2 [db], when accommodating a pacet with 384 bps. This result agrees with the theoretical consideration, which suggests that a transmission rate up to 384 bps can be accommodated within the interference variation of D = 2[dB]. In the simulation, SINR estimation errors, quantized power steps for, and multipath environments are considered. The similar results between theory and simulation show that theoretical expression in () becomes a good approximation in an actual W-CDMA uplin. It is also found that the interference variation is very large when accommodating a pacet with 768 bps. 4.5 Method to Decide Appropriate Power Step In practice, it is desirable to use an appropriate power step for fast power convergence and for small excess signal power. Here, we present a method to decide an appropriate value. From theoretical consideration in (7), the total received power increases by 0 log 0 ( ψ )/( ψ 2 )[db] between frames, where ψ = (6) SF /γ + ψ 2 = = K+K SF /γ + + = K+K +K 2 + =K+K + =K+ SF /γ + SF /γ +. (7)

7 208 IEICE TRANS. COMMUN., VOL.E88 B, NO.5 MAY 2005 maintaining good signal quality for all users. There are many types of algorithms to satisfy the constraint of our RRM scheme. Our method can be jointly used with other algorithms, such as scheduler considering the propagation characteristics in W-CDMA uplin. Considering easy implementation and low cost, our RRM and schemes will be practical in W-CDMA systems with high data rate pacet transmission. References Fig. 7 Total interference to noise power ratio. (a) 384 bps pacet admitted. (b) bps pacet admitted. To compensate the interference increase and to minimize the excess signal power, a minimum integer [db] satisfying 0 log 0 ( ψ )/( ψ 2 ) is suitable for the power step. Based on this method, is given as 2 and 3 [db] to accommodate 384 bps and bps, respectively, in our simulation conditions. From Figs. 5 to 7, it is confirmed that = 2 and 3 [db] eeps the required signal quality minimizing the excess signal power in accommodating 384 bps and bps. Also, the total received power within a frame is maintained as almost constant. Therefore, an appropriate value provided by the above method is useful for fast power convergence and for small excess signal power. 5. Conclusion This paper presented new RRM and schemes to avoid the deterioration of active users signal quality caused by a newly admitted high rate pacet. Numerical results show that the joint utilization of our RRM and is effective in [] F. Adachi, M. Sawahashi, and H. Suda, Wideband DS-CDMA for next generation mobile communications system, IEEE Commun. Mag., vol.36, pp.56 69, Sept [2] [3] J. Damnjanovic, A. Jain, T. Chen, and S. Sarar Scheduling the cdma2000 reverse lin, Proc. VTC 02 Fall, pp , Sept [4] S. Ramarishna and J.M. Holtzman, A scheme for throughput maximization in a dual-class CDMA system, IEEE J. Sel. Areas Commun., vol.6, no.6, pp , Aug [5] K. Dimou and P. Godlewsi, MAC scheduling for uplin transmission in WCDMA UMTS, Proc. VTC 0 Spring, vol.4, pp , April 200. [6] K. Kumaran and L. Qian, Uplin scheduling in CDMA pacet-data systems, Proc. INFOCOM, pp , April [7] A.J. Viterbi and A.M. Viterbi, Earlang capacity of a power controlled CDMA system, IEEE J. Sel. Areas Commun., vol., no.6, pp , Aug [8] Y. Ishiawa and N. Umeda, Capacity design and performance of call admission control in cellular CDMA systems, IEEE J. Sel. Areas Commun., vol.5, no.8, pp , Oct [9] S. Shin, C.H. Cho, and D.K. Sung, Interference based channel assignment for DS-CDMA cellular systems, IEEE Trans. Veh. Technol., vol.48, no., pp , Jan [0] N. Dimitriou and R. Tafazolli, Quality of service for multimedia CDMA, IEEE Commun. Mag., vol.38, pp.88 94, July [] A. Sampath, P.S. Kumar, and J.M. Holtzman, Power control and resource management for a multimedia wireless CDMA system, Proc. PIMRC 95, vol., pp.2 25, Sept [2] A. Sampath and J.M. Holtzman, Access control of data in integrated voice/data CDMA systems: Benefits and tradeoffs, IEEE J. Sel. Areas Commun., vol.5, no.8, pp.5 526, Oct [3] S.-J. Oh, T.L. Olsen, and K.M. Wasserman, Distributed power control and spreading gain allocation in CDMA data networs, Proc. INFOCOM, pp , March [4] J.W. Chang, J.H. Chung, and D.K. Sung, Admission control scheme for soft handoff in DS-CDMA cellular systems supporting voice and stream-type data services, IEEE Trans. Veh. Technol., vol.5, no.6, pp , Nov [5] V.A. Siris, Resource control for elastic trafficincdmanetwors, Proc. MOBICOM, pp , Sept [6] Y. Hara and T. Seiguchi, Exact analysis of received signal in CDMA uplin, IEICE Technical Report, RCS , March [7] 3GPP RAN, 3G TS , V3..0, Sept Appendix Let us consider a user transmitting the control channel with the SINR γ 0, the spreading factor SF 0 = 256, the propagation loss U, and the transmit power W 0. Also, consider another case of the same user transmitting a high data rate

8 HARA et al.: RADIO RESOURCE MANAGEMENT AND POWER CONTROL FOR W-CDMA UPLINK 209 pacet with the SINR γ, the spreading factor SF,andthe transmit power W. Assuming no multipath environment and the same interference power I 0 for simplicity, we have SF 0 UW 0 I 0 = γ 0, SF UW I 0 = γ (A ) Therefore, a user changing the data rate from the control channel to the high data rate pacet needs to increase the power by W = γ /SF. (A 2) W 0 γ 0 /SF 0 In practice, it is necessary to consider the effects of multipath environments and interference variation due to accommodation of thehigh datarate pacet. Theseeffects are evaluated in our simulations. Taashi Seiguchi received the B.E., M.E., and D.Eng. degrees in Electrical Engineering from Keio University, Toyo, Japan, in 985, 987, and 990, respectively. In 990, he joined Mitsubishi Electric Corporation. In 994, he moved to ATR Optical and Radio Communications Research Laboratories. From 996 to 997, he was with ATR Adaptive Communications Research Laboratories. Currently, he is a manager at Information Technology R & D Center, Mitsubishi Electric Corporation, and has been engaged in research and development of signal processing for sensor systems. He is a member of IEEE. antenna arrays. Yoshitaa Hara received the B.E., M.E., and Dr. Eng. degrees from the University of Toyo, Toyo, Japan, in 993, 995, and 2003, respectively. In 996, he joined Mitsubishi Electric Corporation. From 999 to 200, he was also a senior research engineer at YRP Mobile Telecommunications Key Technology Research Laboratories Co., Ltd. Since 2003, he has been with the Mitsubishi Electric Information Technology Centre Europe B.V. (ITE). His research interests include CDMA systems and adaptive Kuniyui Suzui received the B.S. degree from Kyoto University, Kyoto, Japan in 990. In 990 he jointed Mitsubishi Electric Corporation. He has engaged in research and development of radio communication networ system. Koji Kaneo received the B.E. degree from Niigata University, Niigata, Japan in 984. In 984 he joined Mitsubishi Electric Corporation. He has engaged in research and development of radio communication system and equipment.