Design and performance of multicarrier CDMA system in ... - IEEE Xplore

3 downloads 19339 Views 359KB Size Report
Design and Performance of Multicarrier. CDMA System in Frequency-Selective. Rayleigh Fading Channels. Shinsuke Hara, Member, IEEE, and Ramjee Prasad, ...
1584

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 48, NO. 5, SEPTEMBER 1999

Design and Performance of Multicarrier CDMA System in Frequency-Selective Rayleigh Fading Channels Shinsuke Hara, Member, IEEE, and Ramjee Prasad, Senior Member, IEEE

Abstract— This paper presents the advantages and disadvantages of a multicarrier code-division multiple-access (MC-CDMA) system. The transmitter/receiver structure and the bandwidth of transmitted signal spectrum are compared with those of a conventional direct-sequence (DS) CDMA system, and an MC-CDMA design method, how to determine the number of subcarriers and the length of guard interval is discussed. The bit error rate (BER) lower bounds for DS-CDMA and MA-CDMA systems are derived and their equivalence is theoretically demonstrated. Finally, the BER performance in downlink and uplink channels with frequency-selective Rayleigh fading is shown by computer simulation. Index Terms—Code division multiaccess, fading channels, radio communication.

I. INTRODUCTION

C

ODE-DIVISION multiple-access (CDMA) system has been considered to be a candidate to support multimedia services in mobile communications because it has its own capabilities to cope with asynchronous nature of multimedia data traffic, to provide higher capacity over conventional access schemes such as time-division multiple-access (TDMA) and frequency-division multiple-access (FDMA), and to combat hostile channel frequency selectivity. Direct-sequence (DS) and frequency-hopping (FH) CDMA systems have been subject to extensive research [1]. Recently, a new CDMA system based on a combination of CDMA scheme and orthogonal frequency division multiplexing (OFDM) signaling, which is called “multicarrier (MC) CDMA system,” has been reported in [2]–[4]. Much attention has been paid to it, because the signal can be easily transmitted and received using the fast Fourier transform (FFT) device without increasing the transmitter and receiver complexities and it is potentially robust to channel frequency selectivity with a good frequency utilization efficiency. So far, a lot of reports have been dedicated for the bit error rate (BER) analysis of MC-CDMA system and the BER comparison between MC-CDMA and DS-CDMA sysManuscript received March 12, 1997; revised December 10, 1997. This paper was presented in part at VTC’96, Atlanta, GA, and ISSSTA’96, Mainz, Germany. S. Hara is with the Department of Electronic, Information and Energy Engineering, Graduate School of Engineering, Osaka University, Osaka, Japan. R. Prasad is with the Department of Electrical Engineering, Delft University of Technology, Delft, The Netherlands. Publisher Item Identifier S 0018-9545(99)07368-5.

tems in frequency-selective Rayleigh fading channels [5]–[11]. In these works, “independent fading characteristic at each received path” has been often assumed for the BER analysis of DS-CDMA system, whereas “independent fading characteristic at each received subcarrier” for the BER analysis of MC-CDMA system. However, in general, fading characteristics among subcarriers are highly correlated, and the subcarrier correlation is uniquely determined by the multipath delay profile of the channel. Therefore, when we discuss the BER performance of MC-CDMA system and compare it with that of other multiple-access systems such as DSCDMA, it is essential to make a fair assumption for all the systems compared, such as the same channel frequency selectivity and channel time selectivity as well as the same modulation/demodulation format, transmission rate and processing gain. Furthermore, when we design a MC-CDMA system and discuss the BER performance, it is essential to carefully determine two transmission parameters; the length of guard interval and the number of subcarriers, because they significantly affect the BER performance. In this paper, we discuss the advantages and disadvantages of MC-CDMA system. Here, in order to focus much attention on the MC-CDMA concept, a conventional DS-CDMA system is introduced for a comparison purpose. MC-CDMA system inevitably requires linear amplification, because it is very sensitive to nonlinear amplification. This requirement could be realizable for base stations, so in this sense, we can say that MC-CDMA system is well suited for downlink channel. Therefore, in the BER investigation, we discuss a bit more the downlink performance, although the uplink performance is shown as well. We show the BER performance with four different combining strategies such as orthogonality restoring combining (ORC), equal gain combining (EGC), maximum ratio combining (MRC), and minimum mean square error combining (MMSEC). They are all categorized into single-user detection scheme applicable for downlink and uplink channels, however, as shown later, the MC-CDMA uplink performance is still poor even with quasi-synchronous scenario. The paper is organized as follows. Section II explains a frequency-selective fast Rayleigh fading channel to carry out the MC-CDMA system design and the BER evaluation. Section III shows the DS-CDMA and MC-CDMA systems and outlines the four different combining strategies for the MCCDMA system. Section IV discusses a MC-CDMA design

0018–9545/99$10.00  1999 IEEE

HARA AND PRASAD: DESIGN AND PERFORMANCE OF MULTICARRIER CDMA SYSTEM

1585

where and are the zero-order Bessel function of the first kind and the maximum Doppler frequency, respectively. are indepenNote that, for different the path gains dent, identically distributed (i.i.d.) in an uplink channel and identically distributed in a downlink channel. III. DS-CDMA Fig. 1. Multipath delay profile.

method, namely, how to determine the number of subcarriers and the length of guard interval to minimize the BER for a given channel condition. Section V shows the theoretical BER lower bounds for both systems and proves their equivalence. Section VI demonstrates the BER performance of MC-CDMA and DS-CDMA schemes in (synchronous) downlink and quasisynchronous uplink channels and discusses the advantages and disadvantages in terms of “bandwidth of transmitted signal spectrum” and “attainable BER performance.” Finally, Section VII draws the conclusions. II. CHANNEL MODEL As a frequency-selective fast Rayleigh fading channel, we assume a wide-sense stationary uncorrelated scattering (WSreceived paths in the complex SUS) channel [12] with equivalent low-pass time-variant impulse response (1) where is the user index, and are the time and the delay, is the Dirac delta function, is the th respectively, path gain which is a mutually independent complex Gaussian for different random process with zero mean and variance , and is the propagation delay for the th path. Fig. 1 shows the corresponding multipath delay profile given by (we assume that there is no signal whose propagation delay exceeds the symbol duration)

AND

MC-CDMA SYSTEMS

OFDM scheme is robust to frequency-selective fading, however, it has severe disadvantages such as difficulty in subcarrier synchronization and sensitivity to frequency offset and nonlinear amplification, which result from the fact that it is composed of a lot of subcarriers with their overlapping power spectra and exhibits a nonconstant nature in its envelope. One may not imagine the introduction of OFDM signaling in CDMA scheme, because CDMA scheme itself is robust to frequency-selective fading. The OFDM drawback may be only a burden to the CDMA scheme. However, the combination of OFDM signaling and CDMA scheme has one major advantage that it can lower the symbol rate in each subcarrier so that a longer symbol duration makes it easier to quasi-synchronize the transmissions [15]. For instance, in [16], a multicarrier-based DS-CDMA scheme is proposed for a quasi-synchronous system. Therefore, in this paper, we assume a quasi-synchronous uplink channel, in addition to a (synchronous) downlink channel, and then we discuss the BER performance of MC-CDMA and DS-CDMA systems in the channels. In order to focus much attention on the BER variations by different combining strategies, we assume a perfect subcarrier synchronization with no frequency offset and no nonlinear distortion and perfect subcarrier amplitude/phase estimation for MC-CDMA system. On the other hand, for DSCDMA system, we assume a perfect carrier synchronization and perfect path gain estimation. A. DS-CDMA System Fig. 2(a) shows the DS-CDMA transmitter for the th user with binary PSK modulation/coherent (CBPSK) format. The complex equivalent low-pass transmitted signal is written as

(2) (5) is the expectation and is the complex conjugate. where can With (2), the root mean square (rms) delay spread be calculated [13]. On the other hand, the channel time selectivity is characterized by the normalized time autocorrelation function [12]

and are the th information bit and the th where and chip duration chip of the spreading code with length respectively, is the symbol duration is is the chip pulse waveform. For the symbol rate), and is given by instance, when a rectangular pulse is used,

(3) Assuming that an omnidirectional monopole antenna is used at the receiver and the angular distribution of wave arrival on is given by [14] each path is uniformly distributed,

(4)

otherwise.

(6)

The bandwidth (main lobe) of the transmitted signal spectrum for rectangular pulse waveform is given by (7)

1586

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 48, NO. 5, SEPTEMBER 1999

a fraction of the symbol corresponding to a chip of the spreading code is transmitted through a different subcarrier. For MC transmission, it is essential to have frequency nonselective fading over each subcarrier. Therefore, if the original symbol rate is high enough to become subject to frequency-selective fading, the signal needs to be first serial-to-parallel converted before spreading over the frequency domain. The basic transmitter structure of MC-CDMA scheme is similar to that of OFDM scheme used in digital audio broadcasting (DAB) system [17]. The main difference is that the MC-CDMA scheme transmits the same symbol in parallel through a lot of subcarriers whereas the OFDM scheme transmits different symbols. Fig. 3(a) shows the MC-CDMA transmitter for the th user with CBPSK format. The input information sequence is first converted into parallel data sequences and then each Serial/Parallel converter output is All the multiplied with the spreading code with length (corresponding to the total number data in total of subcarriers) are modulated in baseband by the inverse discrete fourier transform (IDFT) and converted back into serial data. The guard interval is inserted between symbols to avoid intersymbol interference caused by multipath fading, and finally the signal is transmitted after radio frequency (RF) upconversion. The complex equivalent low-pass transmitted signal is written as

(a)

(b)

(c) Fig. 2. DS-CDMA system: (a) transmitter, (b) power spectrum of its transmitted signal, and (c) Rake receiver.

(11) (12) (13)

and for the Nyquist pulse waveform with rolloff factor of [see Fig. 2(b)] (8) Fig. 2(c) shows the -finger DS-CDMA Rake receiver for the th user. The received signal through the channel given by (1) is written as

is the spreading code where is the symbol duration at subcarrier, with length is the minimum subcarrier separation, and is the rectangular symbol pulse waveform defined as otherwise.

(14)

The bandwidth of the transmitted signal spectrum is written as [see Fig. 3(b)] (9) is the convowhere is the number of total active users, is the complex additive Gaussian lution operation, and The decision variable noise with zero mean and variance is written as at

(10) B. MC-CDMA System MC-CDMA transmitter spreads the original signal using a given spreading code in the frequency domain. In other words,

(15) (16) where is the bandwidth expansion factor associated with the guard interval insertion. Note that, in (11), no spreading operation is done in the time domain. Equation (12) shows that the symbol duration at subcarrier level is times as long as the original symbol duration due to serial/parallel conversion. Although the minimum subcarrier separation is given by (13), the subcarrier separation is [see the hatched subcarrier for to power spectra in Fig. 3(b)]. Therefore, when setting

HARA AND PRASAD: DESIGN AND PERFORMANCE OF MULTICARRIER CDMA SYSTEM

1587

(a)

(b)

(c) Fig. 3. MC-CDMA system: (a) transmitter, (b) power spectrum of its transmitted signal, and (c) receiver.

one, the transmitted waveform given by (11) becomes all the subcarriers. same as an OFDM waveform with On the other hand, the received signal is written as

of the received signal scattered in the frequency domain. The decision variable is the sum of the weighted baseband components given by (we can omit the subscription without loss of generality)

(18)

(17) is the received complex envelope at the where th subcarrier of the th user. MC-CDMA receiver requires coherent detection for successful despreading operation. This could give us an impression that the structure of MC-CDMA receiver is very complicated, as compared with that of normal OFDM receiver which employs differential detection to avoid complicated subcarrier recovery. Fig. 3(c) shows the MC-CDMA receiver for the th user. After downconversion, the -subcarrier corresponding to the components is first coherently detected with DFT and received data to combine the energy then multiplied with the gain

(19) and are the complex baseband comwhere ponent of the received signal after downconversion and the complex additive Gaussian noise at the th subcarrier at respectively. Now, we discuss the following four combining strategies. 1) Orthogonality Restoring Combining (ORC): Choosing the gain in the downlink channel as (20)

1588

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 48, NO. 5, SEPTEMBER 1999

the receiver can eliminate multiple-access interference perfectly [4]

The normalized autocorrelation function of the th subcarand for the th user is rier between written as (see Appendix)

(21) However, in (21), low-level subcarriers tend to be multiplied by the high gains, and the noise components are amplified at weaker subcarriers. This noise amplification effect degrades the BER performance. Note that the ORC is applicable only for the downlink channel. 2) Equal Gain Combining (EGC): The gain for the equal gain combining is given by [2]

(29)

(22) 3) Maximum Ratio Combining (MRC): The gain for the maximum ratio combining is given by [2] (23)

(30)

In the case of a single user, the MRC can minimize the BER. 4) Minimum Mean Square Error Combining (MMSEC): Based on the minimum mean square estimation (MMSE) criterion, the error must be orthogonal to all the baseband components of the received subcarriers (24) is given by [4] (31) (25) the Note that, in the downlink application, for small gain becomes small to avoid the excessive noise amplification, it becomes in proportion to the whereas for large in order to recover inverse of the subcarrier envelop orthogonality among users [4]. IV. MC-CDMA SYSTEM DESIGN In order to determine the number of subcarriers and the length of guard interval, we derive the autocorrelation function of the received signal. The received signal for the th user is given by (26) where follows:

is given by (1) and

can be classified as

(27) The Fourier coefficient of the is given by [see Fig. 3(b)]

th subcarrier at

(28)

HARA AND PRASAD: DESIGN AND PERFORMANCE OF MULTICARRIER CDMA SYSTEM

1589

(32) is the chip rate. where In OFDM scheme, generally, when the transmission rate in the case of MC-CDMA scheme) is given, the transmission performance becomes more sensitive to the time selectivity as the number of subcarriers ( ) increases because the wider symbol duration is less robust to the random FM noise, decreases because the wider whereas it becomes poor as power spectrum of each subcarrier is less robust to the frequency selectivity (see Fig. 4). On the other hand, the transmission performance becomes poor as the length of guard increases because the signal transmission in the interval guard duration introduces the power loss, whereas it becomes decreases more sensitive to the frequency-selectivity as because the shorter guard duration is less robust to the delay and spread (see Fig. 5). Therefore, for given there exists an optimal value to minimize the BER in both and [18]. and must maximize In the MC-CDMA system, the autocorrelation function given by (29)–(32), because it means a measure to show how much the received signal can be distorted in the time-frequency-selective fading channel (how we can place the signal on the time-frequency plane, suffering from less distortion)

Fig. 4. Optimum in the number of subcarriers.

.. . .. .

..

(35)

.

where is the transpose. In the above equation, we assume a perfect autocorrelation characteristic for the spreading codes. The BER of time domain -finger DS-CDMA Rake receiver for the case of a single user is uniquely determined by the (in this case, the eigenvalues are clearly eigenvalues of [19]. For example, when are different each other, the BER is given by BER

(33) Therefore, with (33), we can determine two transmission and parameters:

(36) (37)

V. BER LOWER BOUND (38)

A. DS-CDMA System as the received signal vector, the time domain Defining is given by covariance matrix

where

is the total power of the received signal. Also, when are all the same [12]

BER (34)

(39)

1590

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 48, NO. 5, SEPTEMBER 1999

other BER

(44) (45)

are all the same

Also, when BER

(46) C. BER Lower Bound Equivalence When there are paths in the symbol duration at subcarrier in the multipath delay profile shown in Fig. 1, we obtain time domain covariance matrix with the following time resolution of

.. . .. . .. .

Fig. 5. Optimum in the length of guard interval.

(40) Note that the -finger Rake receiver can achieve the minimum BER (the BER lower bound) [12]. B. MC-CDMA System For the case of a single user, the frequency domain MCCDMA Rake receiver based on the maximum ratio combining can achieve the minimum BER (the BER lower bound) [12]. as the received signal vector, the frequency Defining is given by domain covariance matrix (41)

.. . .. . .. . ..

(47)

. ..

.

are where the nonzero eigenvalues of The corresponding frequency domain covariance is given by matrix with frequency resolution of (48) where

is the

DFT matrix given by (49)

We define value

as the eigenvector corresponding to the eigen(50)

(42) is the matrix with element and is the spaced frequency correlation function which is defined as the Fourier transform of the multipath delay profile

where

Also, we define

as (51)

Now, we can theoretically prove that the frequency domain covariance matrix has all the same eigenvalues as the time domain covariance matrix as follows:

(43) as the nonzero eigenvalues of Defining the BER is given by a form similar to (36) or (40) [19]. For example, when are different each

(52)

HARA AND PRASAD: DESIGN AND PERFORMANCE OF MULTICARRIER CDMA SYSTEM

1591

Fig. 6. Optimum number of subcarriers and optimum length of guard interval.

The above equation clearly shows that the nonzero eigenvalues are Therefore, as long as we use of the same frequency-selective fading channel, the BER lower bound of the MC-CDMA system is all the same as that of the DS-CDMA system. Also, the assumption of independent fading characteristic at each subcarrier implies a frequencyselective fading at each subcarrier, because it requires indepaths uniformly scattered in the symbol duration pendent at subcarrier. VI. NUMERICAL RESULTS In order to demonstrate the numerical results, we assume: ns; • rms delay spread • Doppler power spectrum with maximum Doppler freHz; quency Msymbols/s (BPSK format); • transmission rate for the MC• Walsh–Hadamard codes with CDMA system; for the DS-CDMA system • Gold codes with First, in order to design the MC-CDMA system and to select one best-suited combining strategy in the MC-CDMA system, we assume a simple two-path multipath delay profile often encountered in urban and hilly areas [20], [21], where the first and second paths have the same power (two-path i.i.d. delay profile) [22]. A. MC-CDMA System Design Fig. 6 shows the optimal values in the number of subcarriers and the length of guard interval versus the normalized as a function of the Doppler frequency rms delay spread where and are normalized by This figure is obtained from the maximization of (33). For given and both and increase as increases. and For the above parameters, we obtain so we select as an optimal set. It means that the original data sequence and is first converted into eight parallel sequences then each sequence is mapped onto 32 subcarriers, and that the length of the guard interval is negligibly short, as compared with the symbol duration at subcarrier.

From (7) and (16), and are calculated as 97.8 and 186 MHz, respectively, so the bandwidth of the DS-CDMA signal is 1.9 times as wide as that of the MC-CDMA signal as long as the same rectangular pulse format is employed. However, from (8), if a Nyquist pulse is employed in the DS-CDMA system, the difference in the signal bandwidth diminishes as the rolloff factor becomes small (there is no Therefore, it can be concluded difference when that, MC-CDMA system has no major advantage in terms of signal bandwidth, as compared with DS-CDMA system. However, note that, when Nyquist filters are introduced in the transmitter and receiver for baseband pulse shaping in DSCDMA system, the Rake receiver may wrongly combine paths. This is because noise causing distortion in autocorrelation characteristic often results in wrong correlation [23]. How many users the system can accommodate depends on the attainable BER performance, in other words, the combining strategy employed in the MC-CDMA system and the number of fingers in the DS-CDMA system. We discuss the BER performance in the following two sections. B. Downlink BER Performance Figs. 7–9 show the downlink BER performance of MCCDMA scheme with EGC, MRC, and MMSEC for the twopath i.i.d. delay profile, respectively. Here, the theoretical BER In these figures, lower bound is given by (40) with the BER for the ORC is shown, and furthermore, the BER of a MC-FDMA scheme is also shown, which supports 32 users at most, assigning a different set of 8 subcarriers to each user. This scheme obtains no frequency diversity effect, so the theoretical BER is given by [12] BER

-

(53)

The BER performance of the ORC is independent of the number of users, however, it is worse than that of the MCFDMA scheme. Therefore, we do not have to employ the ORC even when we can estimate the channel condition perfectly. The MRC can perform better when the number of users is less than eight. For the case of the more users, however, the performance abruptly becomes worse, because the interference

1592

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 48, NO. 5, SEPTEMBER 1999

Fig. 7. Downlink BER of MC-CDMA system with EGC.

Fig. 8. Downlink BER of MC-CDMA system with MRC.

Fig. 9. Downlink BER of MC-CDMA system with MMSEC.

resulting from distorted code orthogonality is multiplied in the combining process. On the other hand, as the interference is not multiplied in the EGC, it can perform better than the MRC for the cases of 16 and 32 users. The MMSEC can perform best among the four combining strategies, although it requires information on the number of total active users and the noise power, in addition to the channel condition. In the following downlink BER comparison, we select the MMSEC as the best combining strategy. Fig. 10 shows the BER comparison between DS-CDMA and MC-CDMA schemes for the two-path i.i.d. delay profile. For

Fig. 10. BER comparison in a downlink channel with two-path i.i.d. multipath delay profile.

Fig. 11. BER comparison in a downlink channel with seven-path exponential multipath delay profile.

the DS-CDMA scheme, the BER of two (full)-finger Rake combiner is a little worse than the lower bound even for the case of a single user because of the self-interference resulting from the imperfect autocorrelation characteristic of the Gold codes. Also, the BER of one-finger Rake combiner, which selects a larger path, is worse than that of two-finger Rake combiner, because it always misses a part of the received signal energy scattered in the time domain. On the other hand, for the MC-CDMA scheme, the MMSEC outperforms the full-finger DS-CDMA Rake combiner. This is because the MMSEC-based MC-CDMA scheme can effectively combine all the received signal energy scattered in the frequency domain. Fig. 11 shows the BER comparison for a seven-path exponential delay profile often encountered in urban areas[13], ns where the power decays exponentially with We obtain as an optimal set for the delay profile, and the theoretical BER lower bound is given by (38) with For the DS-CDMA scheme, the BER performance depends on how many fingers the Rake receiver employs. Usually, a one-, two-, three, or four-finger Rake receiver is used depending on hardware limitation. Therefore, if the received signal is composed of more paths than the number of Rake fingers, the receiver misses a part of its energy. In these figures,

HARA AND PRASAD: DESIGN AND PERFORMANCE OF MULTICARRIER CDMA SYSTEM

Fig. 12.

Uplink BER of MC-CDMA system with EGC.

the BER’s of one-, two-, and three-finger Rake combiners are worse than that of seven (full)-finger Rake combiner, because they always miss a larger part of the received signal energy scattered in the time domain. On the other hand, for the MCCDMA scheme, the MMSEC outperforms the three-finger DS-CDMA Rake combiner, and for the case of eight users and more, it can better perform than the full-finger DS-CDMA Rake combiner. From all the results obtained in this section, it can be concluded that, it could be difficult for DS-CDMA receiver to employ all the received signal energy scattered in the time domain, whereas MC-CDMA receiver can effectively combine all the received signal energy scattered in the frequency domain. DS-CDMA receiver needs to make efforts to select larger paths, on the other hand, MC-CDMA receiver does not care about where the received signal energy is, because the energy always exists where it should be. We believe that it is a major advantage of MC-CDMA scheme over DS-CDMA scheme, although it requires much complexity in the receiver structure such as subcarrier synchronization. The MMSECbased MC-CDMA must be a promising access scheme in a downlink channel.

1593

Fig. 13. Uplink BER of MC-CDMA system with MRC.

Fig. 14. Uplink BER of MC-CDMA system with MMSEC.

C. Uplink BER Performance Figs. 12–14 show the uplink BER performance of MCCDMA system with EGC, MRC, and MMSEC for the twopath i.i.d. delay profile, respectively. In these figures, the BER of MC-FDMA scheme and the BER lower bound are also shown. The MMSEC can perform best among the three combining strategies, although there is no large difference in the attainable BER. Therefore, we select it as the best combining strategy. Figs. 15 and 16 show the BER comparison between DSCDMA and MC-CDMA schemes for the two-path i.i.d. and seven-path exponential delay profiles, respectively. As compared with the DS-CDMA scheme, the MMSEC performs well only for the case of a single user and otherwise performs poor. This is because the code orthogonality among users is totally distorted by the instantaneous frequency response. Therefore, in the uplink application, a multiuser detection scheme is required, which jointly detects the signals in order to mitigate the nonorthogonal properties [24].

Fig. 15. BER comparison in an uplink channel with two-path i.i.d. multipath delay profile.

VII. CONCLUSIONS In this paper, we have discussed the advantages and disadvantages of MC-CDMA system, and have shown the BER performance by computer simulation. MC-CDMA system has no major advantage over DSCDMA system in terms of required bandwidth, because the bandwidth of MC-CDMA signal spectrum is almost the same as that of DS-CDMA signal spectrum. Also, in terms of transmission performance, the BER lower bound of MCCDMA system is all the same as that of DS-CDMA system.

1594

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 48, NO. 5, SEPTEMBER 1999

Fig. 16. BER comparison in an up channel with seven-path exponential multipath delay profile.

Therefore, If we make every effort to improve the BER in each system, there is no difference in the attainable BER as long as the same channel is used. DS-CDMA system cannot always employ all the received signal energy scattered in the time domain, whereas MCCDMA system can effectively combine all the received signal energy scattered in the frequency domain, although it requires much complexity in the receiver structure such as subcarrier synchronization. The MMSEC-based MC-CDMA must be a promising scheme in a downlink channel, although estimation of the noise power as well as subcarrier condition is required. On other hand, in the uplink application, a multiuser detection is required because the code orthogonality among users is totally distorted by the channel frequency selectivity.

(57) where the first term of (57) represents the desired signal component, the second and third terms represent the intersubcarrier interference (ICI) and the intersymbol interference (ISI) is the Gaussian random components, respectively, and With (57), noise component with zero mean and variance and in (29) are written as

APPENDIX Equation (11) is rewritten as (58) (54) (55) where

has the following property: (56)

Substituting (1), (26), and (54) into (28) leads

(59) Taking account of and (59) leads (30)–(32).

substituting (4) into (58)

HARA AND PRASAD: DESIGN AND PERFORMANCE OF MULTICARRIER CDMA SYSTEM

ACKNOWLEDGMENT The authors wish to thank Prof. Y. Bar-Ness of the New Jersey Institute of Technology, Dr. J-P. Linnartz of Philips Research, and Assistant Prof. M. Okada of Osaka University for their helpful comments and fruitful discussions. REFERENCES [1] R. Prasad, CDMA for Wireless Personal Communications. Norwood, MA: Artech, 1996. [2] N. Yee, J.-P. Linnartz, and G. Fettweis, “Multi-carrier CDMA in indoor wireless radio networks,” in Proc. IEEE PIMRC’93, Sept. 1993, pp. 109–113. [3] K. Fazel and L. Papke, “On the performance of convolutionally-coded CDMA/OFDM for mobile communication system,” in Proc. IEEE PIMRC’93, Sept. 1993, pp. 468–472. [4] A. Chouly, A. Brajal, and S. Jourdan, “Orthogonal multicarrier techniques applied to direct sequence spread spectrum CDMA systems,” in Proc. IEEE GLOBECOM’93, Nov. 1993, pp. 1723–1728. [5] K. Fazel, S. Kaiser, and M. Schnell, “A flexible and high performance cellular mobile communications systems based on orthogonal multicarrier SSMA,” Wireless Personal Communications, vol. 2, nos. 1–2. Norwell, MA: Kluwer, Nov. 1995, pp. 121–144. [6] T. M¨uller, H. Rohling, and R. Gr¨unheid, “Comparison of different detection algorithms for OFDM-CDMA in broadband Rayleigh fading,” in Proc. IEEE VTC’95, July 1995, pp. 835–838. [7] S. Kaiser, “OFDM-CDMA versus DS-CDMA: Performance evaluation for fading channels,” in Proc. IEEE ICC’95, June 1995, pp. 1722–1726. , “On the performance of different detection techniques for [8] OFDM-CDMA in fading channels,” in Proc. IEEE GLOBECOM’95, Nov. 1995, pp. 2059–1063. [9] S. Hara, T.-H. Lee, and R. Prasad, “BER comparison of DS-CDMA and MC-CDMA for frequency selective fading channels,” in Proc. 7th Tyrrhenian Int. Workshop on Digital Communications, Sept. 1995, pp. 3–14. [10] S. Hara and R. Prasad, “DS-CDMA, MC-CDMA, and MT-CDMA for mobile multi-media communications,” in Proc. IEEE VTC’96, Apr. 1996, pp. 1106–1110. [11] R. Prasad and S. Hara, “An overview of multi-carrier CDMA,” in Proc. 4th IEEE Int. Symp. Spread Spectrum Techniques and Applications (ISSSTA’96), Sept. 1996, pp. 107–114. [12] J. G. Proakis, Digital Communications, 3rd ed. New York: McGrawHill, 1995, pp. 758–785. [13] W. C. Y. Lee, Mobile Communications Engineering. New York: McGraw-Hill, 1995, pp. 40–44. [14] W. C. Jakes, Jr., Microwave Mobile Communications. New York: Wiley, 1974, pp. 19–26. [15] Q. Chen, E. S. Sousa, and S. Pasupathy, “Performance of a coded multi-carrier DS-CDMA system in multi-path fading channels,” Wireless Personal Commun., vol. 2, nos. 1–2, pp. 167–187, 1995. [16] V. M. DaSilva and E. S. Sousa, “Multicarrier orthogonal CDMA signals for quasi-synchronous communication systems,” IEEE J. Select. Areas Commun., vol. JSAC-12, no. 5, June 1994. [17] P. Dambacher, Digital Broadcasting. New York: IEEE Press, 1996, pp. 97–114. [18] S. Hara, M. Mouri, M. Okada, and N. Morinaga, “Transmission performance analysis of multi-carrier modulation in frequency selective fast Rayleigh fading channel,” Wireless Personal Commun., vol. 2, no. 4, pp. 335–356, 1995/1996. [19] P. Monsen, “Digital transmission performance on fading dispersive diversity channels,” IEEE Trans. Commun., vol. COM-21, pp. 33–39, Jan. 1973. [20] T. S. Rappaport, Wireless Communications, Principles, and Practice. Upper Saddle River, NJ: Prentice-Hall, 1996, pp. 188–189. [21] R. Steele, Mobile Radio Communications. London, U.K.: Pentech, 1992, pp. 727–729. [22] L. J. Cimini, “Analysis and simulation of a digital mobile channel using orthogonal frequency division multiplexing,” IEEE Trans. Commun., vol. COM-33, pp. 665–675, June 1985. [23] S. Sampei, Applications of Digital Wireless Technologies to Global Wireless Communications. Upper Saddle River, NJ: Prentice-Hall, 1997, pp. 315–332. [24] A. Duel-Hallen, J. Holtzman, and Z. Zvonar, “Multiuser detection for CDMA system,” IEEE Personal Commun., vol. 2, pp. 46–58, Apr. 1995.

1595

Shinsuke Hara (S’87–M’90) received the B.Eng., M.Eng., and Ph.D. degrees in communication engineering from Osaka University, Osaka, Japan, in 1985, 1987, and 1990, respectively. From April 1990 to March 1996, he was an Assistant Professor in the Department of Communication Engineering, Osaka University. Since April 1996, he has been with the Department of Electronic, Information and Energy Engineering, Graduate School of Engineering, Osaka University, and now, he is an Associate Professor. Also, from April 1995 to March 1996, he was a Visiting Scientist at Telecommunications and Traffic Control Systems Group, Delft University of Technology, Delft, The Netherlands. His research interests include satellite, mobile and indoor wireless communications systems, and digital signal processing. Dr. Hara is a member of IEICE of Japan.

Ramjee Prasad (M’88–SM’90) was born in Babhnaur (Gaya), Bihar, India, on July 1, 1946. He is now a Dutch Citizen. He received the B.Sc. (Eng) degree from the Bihar Institute of Technology, Sindri, India, and the M.Sc. (Eng.) and Ph.D. degrees from the Birla Institute of Technology (BIT), Ranchi, India, in 1968, 1970, and 1979, respectively. He joined BIT as a Senior Research Fellow in 1970 and became Associate Professor in 1980. While he was with BIT, he supervised a number of research projects in the areas of microwave communications and plasma engineering. From 1983 to 1988, he was with the University of Dar es Salaam (UDSM), Tanzania, where he became Professor of Telecommunications at the Department of Electrical Engineering in 1986. At USDM, he was responsible for the collaborative project “Satellite Communications for Rural Zones” at Eindhoven University of Technology, The Netherlands. Since February 1988, he has been with the Telecommunications and Traffic-Control Systems Group, Delft University of Technology (DUT), The Netherlands, where he is actively involved in the area of wireless personal and multimedia communications (WPMC). He is Head of the Transmission Research Section of IRCTR (International Research Centre for Telecommunications-transmission and Radar) and also Program Director of a newly established Center for Wireless Personal Communications (CEWPC). He is currently involved in the European ACTS project FRAMES (future radio wideband multiple access system) as a Project Leader of DUT. He is Project Leader of several international industrial-funded projects. He has published over 300 technical papers and authored and coedited three books: CDMA for Wireless Personal Communications, Universal Wireless Personal Communications, and Wideband CDMA for Third Generation Communications (Boston, MA: Artech). He is Coordinating Editor and Editor-in-Chief of the Kluwer international journal of Wireless Personal Communications. His current research interest lies in wireless networks, packet communications, multiple-access protocols, adaptive equalizers, spread-spectrum CDMA systems, and multimedia communications. Dr. Prasad has served as a member of advisory and program committees of several IEEE international conferences. He has also presented keynote speeches, invited papers, and tutorials on WPMC at various universities, technical institutions, and IEEE conferences. He was Organizer and Interim Chairman of the IEEE Vehicular Technology/Communications Society Joint Chapter, Benelux Section. He is now the Elected Chairman of the joint chapter. He is also founder of the IEEE Symposium on Communications and Vehicular Technology (SCVT) in the Benelux and was the Symposium Chairman of SCVT’93. He is a member of the editorial board of other international journals, including the IEEE COMMUNICATIONS MAGAZINE and the IEE ELECTRONICS COMMUNICATION ENGINEERING JOURNAL. He was the Technical Program Chairman of the PIMRC’94 International Symposium held in The Hague, The Netherlands, during September 19–23, 1994, and also of the Third Communications Theory Mini-Conference in conjunction with the GLOBECOM’94 held in San Francisco, CA, November 27–30, 1994. He is the Conference Chairman of IEEE Vehicular Technology Conference, VTC’99 (Fall), Amsterdam, The Netherlands, to be held on September 26–29, 1999. He is listed in the U.S. Who’s Who in the World. He is a Fellow of the IEE and the Institution of Electronics and Telecommunication Engineers and a member of NERG (The Netherlands Electronics and Radio Society).