Performance of Multicarrier CDMA With ... - Semantic Scholar

2 downloads 221 Views 460KB Size Report
(CDMA) systems as an engineer at Qualcomm from 1995 to 1997, and has served as a ... leave from Stanford and founded Atheros Com- munications, Inc.
IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 52, NO. 5, MAY 2004

811

Performance of Multicarrier CDMA With Successive Interference Cancellation in a Multipath Fading Channel Jeffrey G. Andrews, Member, IEEE, and Teresa H. Y. Meng, Fellow, IEEE

Abstract—A high capacity, low complexity, and robust system design for a successive interference cancellation (SIC) system is developed and analyzed. Multicarrier code-division multiple access (MC-CDMA) is used to suppress multipath and to overcome the multipath channel estimation problem in single-carrier SIC systems. In addition, an optimal power control algorithm for MC-CDMA with SIC is derived, allowing analytical bit-error rate expressions to be found for an uncoded system. Low-rate forward error-correcting codes are then added to the system to achieve robustness. It is found that the capacity of the coded system approaches the additive white Gaussian noise capacity for SIC, even in a fading multipath channel with channel estimation error. This indicates that MC-CDMA is very attractive for systems employing SIC. Index Terms—Channel estimation error, interference cancellation (IC), multicarrier code-division multiple access (MC-CDMA), multiuser detection (MUD), power control (PC).

I. INTRODUCTION

C

ODE-DIVISION multiple access (CDMA) is the multiple-access technology used in the third-generation cellular systems. It is estimated that by 2007, there will be about two billion cellular subscribers worldwide, the majority of them using CDMA technology [1]. Techniques that increase the capacity of CDMA networks will be vital in facilitating this growth. CDMA systems have the interesting property that their capacity is typically limited by multiple-access interference (MAI), rather than noise. Due primarily to interference reduction using variable-rate transmission and frequency reuse, commercial CDMA systems currently have a higher capacity than systems using other multiple-access technologies. However, the currently achieved capacities lie far below theoretical bounds. The potential for dramatically increasing the capacity of CDMA systems using advanced signal processing at the receiver was first shown over 15 years ago [2], and has been a Paper approved by S. N. Batalama, the Editor for Spread Spectrum and Estimation of the IEEE Communications Society. Manuscript received March 5, 2002; revised September 21, 2002; April 4, 2003; and August 26, 2003. This paper was presented in part at the IEEE International Symposium on Spread Spectrum Technology and Applications, Prague, Czech Republic, September 2002. J. G. Andrews is with the Wireless Networking and Communications Group, Department of Electrical and Computer Engineering, University of Texas at Austin, Austin TX 78701 USA (e-mail: [email protected]). T. H. Meng is with the Department of Electrical Engineering, Stanford University, Stanford, CA 94305-4070 USA (e-mail: [email protected], [email protected]). Digital Object Identifier 10.1109/TCOMM.2004.826240

topic of intensive research ever since. Despite the theoretical promise of these techniques, known broadly as multiuser detection (MUD), industry has not yet adopted multiuser receivers to increase capacity. The primary reasons for this skepticism fall into two categories, complexity and robustness. A large amount of MUD research has focused on the complexity problem. Two quite different, suboptimal approaches for MUD emerged, which had much lower complexity than the optimum multiuser detector: interference cancellation (IC) [3]–[5] and adaptive filtering [6], [7]. The IC techniques can be broadly broken into serial (successive) and parallel schemes for cancelling MAI. Successive interference cancellation (SIC) is considered in this paper, due to its potential compatibility with current commercial systems, allowance of strong error-correcting codes, and its robustness in an asynchronous environment. Despite its apparent simplicity, SIC has the potential to achieve impressive spectral efficiency, as it has been shown to approach the Shannon capacity of an additive white Gaussian noise (AWGN) channel under ideal conditions [3]. SIC systems have historically suffered from four primary problems. Each user’s signal must be estimated and subtracted out from the composite signal before decoding the next user. If the signal estimation is sufficiently inaccurate, future users will not be decoded reliably. Second, this successive process, while saving hardware, takes more time than parallel detection. Third, a specific ordering of user powers must be enforced for the users to achieve similar performance [8]–[10]. Finally, multipath propagation poses a particular problem, as each multipath component must be cancelled. In fact, capacity in many IC systems drops off proportionally to the number of multipath components [11], [12]. In this paper, a low-complexity system design using multicarrier CDMA (MC-CDMA) with SIC (MC-SIC) is developed that addresses the above degradations and disadvantages. MC-CDMA has the additional advantage over CDMA that multipath suppression is not dependent on the spreading factor, so high data rates may be accommodated in a smaller bandwidth. The latency problem still exists, but the proposed system design minimizes the signal processing required for each iteration, and the latency is linear with the number of users. Given the exponential growth in the capabilities of integrated circuits, this is, at most, a temporary problem. Previous authors [13]–[18] have proposed the use of MC-CDMA with IC, but important system design issues such as multipath, latency, and estimation error have often been

0090-6778/04$20.00 © 2004 IEEE

812

IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 52, NO. 5, MAY 2004

neglected. In this paper, by using optimum power control (PC), capacity for a realistic system exceeds that of previous work, while also confronting these impairments. In this paper, two different models for an MC-SIC system are considered. The first is an uncoded system similar to that analyzed in [19] and [20], but with an IC block that allows successive detection at the receiver. An optimum power-control distribution for MC-SIC is derived, and used to analyze the system performance when the IC is imperfect. The analytical bit-error rate (BER) results show the dramatic improvement that SIC allows over a normal MC-CDMA system [19], [20]. The second system is a more realistic and higher performance design. The principal difference is that it uses low-rate superorthogonal codes for spreading, with each coded symbol placed on a subcarrier rather than simple replicates of each bit. The decoding, and hence, subcarrier combining, is implemented using maximum-likelihood sequence estimation (MLSE) with a Viterbi decoder, rather than an integrator and threshold check. Further, a low-complexity inverse fast Fourier transform (IFFT) is used for forming the subcarriers, and a fast Fourier transform (FFT) is used for demodulation. A BER approximation is derived empirically using coding gain for this system. II. SYSTEM MODEL In order to attain analytical results for BER performance, and to compare meaningfully with previous work, the uncoded system is used for the analysis. The analysis follows the same basic procedure as in [19], but differs significantly in three respects. First, the proposed system is a SIC system, so each user experiences different amounts of interference, and requires a different power level, as seen in Section III-C. Second, imperfect channel estimation is assumed for both the subcarrier combining and interference regeneration. Third, the binary phaseshift keying (BPSK)-modulated signal is quadrature spread over the sine and cosine channels in order to allow better suppression of other-user interference [3], so the analysis here is for a quadrature channel. A. Transmitter The transmitted signal for user is described as shown in Fig. 1 and in (1) and (2) at the bottom of the page, where is the transmit power of the th user, is the data bit of user at time , is the complex and spreading sequence for user on subcarrier , is the caris the rier frequency of subcarrier in radians per second, phase for user ’s th subcarrier and is independent and iden, tically distributed (i.i.d.) and uniformly distributed in is the symbol interval, or equivalently, the bit time, deis defined as notes the real part of a complex number, and

Fig. 1. MC-CDMA transmitter for analysis, with the spreading code and