org
ISSN 1751-8628
Abstract: Single carrier (SC) code division multiple access (CDMA) with block transmission has been shown to be
more effective while utilising a low-complexity equaliser to combat frequency-selective fading channels, when
compared with conventional direct sequence CDMA technology. It also has lower peak-to-average power ratio
and lower frequency sensitivity compared with multicarrier CDMA. The authors propose two blind channel
estimation methods for uplink multiple input multiple output SC-CDMA systems with block transmisssion-one
is the subspace-based method and the other is the so-called autocorrelation contribution method (ACM).
Both the methods provide close performance to the case with perfect channel knowledge at high signal-to-
noise ratio (SNR) without any training data required. It is shown that ACM yields a better performance than
the subspace method at a lower SNR, and a similar performance at a high SNR, with the advantages of
avoiding rank determination and noise power estimation as in the subspace method. In addition, the authors
integrate layered space frequency equalisation with blind channel estimation, which provides improved
performance over the conventional linear equalisation, by employing successive interference cancellation.
1 Introduction form of a cyclic prefix (CP) [4] or zero padding (ZP) [5]
are used to eliminate inter-block interference (IBI).
Code division multiple access (CDMA) [1] is an attractive
broadband technique, which features anti-jamming Multicarrier (MC)-CDMA [6] combines the advantages
capability, robustness to multipath interference, of CDMA and OFDM [4]. Yee et al. [7] as well as Chouly
interference rejection and data privacy and so on. The et al. [8] were among others who investigated the effects of
most popular forms of CDMA include direct sequence spreading coupled with MC schemes and the advantages
(DS)-CDMA and frequency hopped CDMA [1]. In availed as a result of this combination. However, MC-
frequency-selective channels, time-domain-based Rake CDMA schemes have a few drawbacks, namely the high
receivers [2], which utilise the matched filter technique peak-to-average power ratio (PAPR) and the high
to equalise, are severely affected by inter-finger sensitivity to carrier-frequency offset.
interference, self-interference as well as multiple access
interference (MAI), and thus results in poor performance. This has resulted in a renewed interest in single carrier
(SC) CDMA block transmissions with frequency domain
Orthogonal frequency division multiplexing (OFDM) [3] equalisation (FDE) [9], the key feature of which is that it
is another attractive broadband technique, which simplifies a utilises the block structure along with GB insertion in
frequency-selective fading channel into parallel flat fading the form of CP [10] or ZP as in MC-CDMA. Use of
channels. Guard bands (GB), adding redundancy in the linear equalisers for multiple input multiple output
& The Institution of Engineering and Technology 2008 IET Commun., 2008, Vol. 2, No. 3, pp. 493 – 503/ 493
doi: 10.1049/iet-com:20070014
Authorized licensed use limited to: VELLORE INSTITUTE OF TECHNOLOGY. Downloaded on August 2, 2009 at 04:29 from IEEE Xplore. Restrictions apply.
www.ietdl.org
(MIMO) systems is further complicated by the need for a noise-subspace and also is not sensitive to rank of the
large number of taps [2, 11]. To alleviate the high noise-subspace. It was used for DS-CDMA system in
complexity needs, FDE-based techniques have been [27]. The key advantages of this approach are that it is
proposed by Garg and Adachi [12], Burg et al. [11] for more robust at lower SNR as a result of the use of
MIMO-CDMA [13, 14] schemes, which provide not the whole correlation matrix instead of the noise-
only better performance but also lower complexity. The subspace as in subspace-based approach. However, the
absence of large amplitude gains and the PAPR as well as ACM has not been used in MIMO SC-CDMA uplink
synchronisation problems of MC-CDMA schemes make setup so far either. The CDMA systems considered in
SC-CDMA with FDE [9] an attractive alternative for [25– 28] all involved DS-CDMA with no guard-band
next-generation wireless communications. and utilised the inter-symbol interference (ISI)-affected
signal for channel estimation. The MIMO uplink SC-
MIMO systems have been shown to be effective in CDMA block transmission with CP considered here
enhancing capacity without any additional power or utilises the CP guard-band to provide ISI immunity as
bandwidth requirements. It requires a rich scattering well as to help in the blind channel estimation stage
environment as is the case in most wireless by using the redundant nature of the CP guard-band.
transmission channels. The genesis of MIMO was
proposed by Jack Winters [15]. Other pioneering In this paper we propose the application of two blind
works include those of Foschini, who introduced the channel estimation methods namely the subspace
Bell Labs Layered Space Time (BLAST) algorithms approach and the ACM for MIMO SC-CDMA systems
[16], and Foschini and Gans [17] who computed the with block transmission. Our work is different in a
capacity that can be achieved with MIMO and so on. way that we investigate MIMO Uplink SC-CDMA
Uplink system with block transmission. We also
Layered space frequency equalisation (LSFE) [18], integrate LSFE along with blind channel estimation to
which provides performance improvement over yield performance improvements compared with the
conventional linear algorithms, utilises FDE along with conventional minimum mean square error (MMSE)
successive interference cancellation to detect the scheme. The ACM proposed in [27] is applied here to
transmitted signals in stages. A low-complexity LSFE the uplink MIMO SC-CDMA system with block
scheme was also proposed for downlink MIMO-DS- transmission, which eliminates the need for the
CDMA systems [19], utilising an overlap-cut method determination of rank criterion as in subspace-based
to approximate guard-band insertion. However, LSFE approach. Both the blind channel methods that are
has not been used for the uplink SC-CDMA with applied utilise the GB structure of the transmitted
block transmission, especially in a MIMO setup. signal to enable blind channel estimation, whereas in
DS-CDMA systems, the same methods involve using
Conventional channel estimation schemes are based the ISI-corrupted signal. Thus the guard band in our
on training symbols [9]. This however leads to system serves the dual purpose of eliminating ISI and
significant bandwith loss. Therefore blind channel assisting in blind channel estimation. This setup also
estimation schemes are desirable. Blind signal differs from the MIMO MC-CDMA uplink case [29],
separation [20] and other higher-order statistic [21] where the diagonal structure of the channel matrix
approaches are difficult to apply in CDMA-based obtained after guard-band suppression and FFT
systems, because of the guassian nature of the signal as processing aids in blind channel estimation. As the
a result of the addition of the user signals on all diagonal structure of the channel matrix is lost in SC-
carriers. The second-order statistics (SOS)-based CDMA systems because of the shifting of the IFFT
methods [22, 23] however are more suitable and the stage from the transmitter side to the receiver side,
subspace approach is a key SOS-based method that has the guard-band intact received signal is used for blind
been used for DS-CDMA-based systems. Blind channel estimation here by making use of the Toeplitz
channel estimation techniques for DS-CDMA schemes matrix property.
based on the subspace idea [22, 23] have been
proposed in the downlink channel case [24, 25], Simulation results show that both the methods
whereas Xu [26] dealt with subspace approximation provide close performance to the case with perfect
method. A synchronous DS-CDMA setup was channel knowledge at high SNR. There is the added
examined in [24], whereas [25] an asynchronous DS- advantage of eliminating the need for rank estimation
CDMA downlink scenario was investigated. However, when using the ACM method. We also show the
the subspace-based method has not been used in effect of receiving diversity on performance of LSFE.
MIMO SC-CDMA uplink case so far. Another SOS- Also plots are shown that illustrate the stable nature
based approach is the autocorrelation contribution of SC-CDMA with FDE to various channel lengths.
method (ACM)-based blind channel estimation method The effect of varying loads, that is different
[27]. It avoids the need for an initial singular value numbers of users on the system performance, is also
decomposition (SVD) stage for determination of highlighted.
494 / IET Commun., 2008, Vol. 2, No. 3, pp. 493– 503 & The Institution of Engineering and Technology 2008
doi: 10.1049/iet-com:20070014
Authorized licensed use limited to: VELLORE INSTITUTE OF TECHNOLOGY. Downloaded on August 2, 2009 at 04:29 from IEEE Xplore. Restrictions apply.
www.ietdl.org
Section 2 describes the SC-CDMA setup with Section 5. Binary phase shift keying (BPSK) modulation is used
3 detailing the blind channel estimation methods used. for each user.
Section 4 describes the two main equaliser schemes
considered namely the MMSE equaliser and the LSFE Each user’s signal is first spread using the
scheme. The simulation results showing the improved corresponding spreading code. The spread signals after
performance of LSFE as against the linear MMSE blocking are then transferred to GB insertion stage,
equaliser are illustrated in Section 5, whereas Section where CP is appended to provide immunity against ISI.
6 concludes the paper.
As illustrated in Fig. 1, the user-spreaded signal is
The following notations are used throughout the paper. grouped into blocks of size N. These size N blocks are
Matrices and vectors are represented by boldface then passed through the CP insertion stage. The CP-
uppercase and lowercase respectively, [.]T, [.]H denote inserted signal of size P, where P ¼ N þ Lcp, is then
the transpose and conjugate transpose and I represents transmitted through the multipath fading MIMO channel
the N N identity matrix. under the assumption that Lcp L that is the CP is at
least as long as the channel order represented by L [30].
2 SC-CDMA At the receiver end as illustrated in Fig. 2, the
2.1 System model received signal has its guard-band removed before
being passed onto FFT stage. After FFT, the FDE
The system model for a MIMO SC-CDMA [10, 13 – 14] stage equalises the signal. The equalised signal is
uplink system is shown in Fig. 1. Here the system converted back to time domain via application of IFFT
employs just a block transmission scheme with a GB before hard decision device is used to obtain the
insertion stage, which is the CP system. The SC- estimates of the transmitted symbols.
CDMA with CP scenario [9] is presented here. The
use of multiple antennas at the receiver (base station) After passing through the GB removal and FFT stages
is feasible and practical as against the mobile user, [9], the frequency domain model of the received signal
where power and complexity issues become at frequency bin n can be written as
paramount. At the base-station (receiver) end, receive
diversity can be utilised by introducing multiple antennas.
X
Nt
Some assumptions made in the model described are xj [n] ¼ h~ j,i [n]ci [n]si [n] þ vj [n] (1)
as follows: i¼1
1. Nt and Nr are the number of transmit and receive where h~ j,i [n] is the Fourier transform coefficient of the
antennas, respectively, and Nt is equal to number of channel matrix between transmit antenna i and receive
users. Also a synchronous system model is assumed. antenna j on sub-carrier n. ci [n] represents the nth
value of the spreading code for user i, whereas si [n] is
2. L is the channel order which is assumed to be known the data bit transmitted from transmit antenna i on
at receiver as is the noise power. subcarrier n and v j [n] denotes the noise at the jth
receive antenna on subcarrier n. The above equation
3. The signals are assumed to be white with zero-mean over all subcarriers can be expressed as shown below
and unit variance. with H~ j,i denoting the N point FFT of the circulant
4. The spreading codes used are Walsh-Hadamard channel between jth receive and ith receive antenna,
codes of size N where N is the number of subcarriers.
Here the spreading gain is chosen to be equal to the
number of subcarriers.
& The Institution of Engineering and Technology 2008 IET Commun., 2008, Vol. 2, No. 3, pp. 493 – 503/ 495
doi: 10.1049/iet-com:20070014
Authorized licensed use limited to: VELLORE INSTITUTE OF TECHNOLOGY. Downloaded on August 2, 2009 at 04:29 from IEEE Xplore. Restrictions apply.
www.ietdl.org
where the (k, l )th entry of the circulant matrix is given Toeplitz nature of the model to obtain the subspace
by hj,i ((k l)modN). cost function as will be explained below.
Let X j ¼ [xTj [1] . . . xTj [N]]T represent the received Making use of the Toeplitz matrix property [31], we
signal at antenna j, and ci ¼ [ci [1] . . . ci [N]]T denotes the can exchange the channel and code vectors to obtain the
spreading code vector of user i , a column vector of size equivalent user-transmitted signal denoted by gi as
N 1 and si ¼ [si [1]si [2] . . . si [N]] represent the data
transmitted by user i. V j ¼ [vTj [1] . . . v Tj [N]]T is the gi ¼ C tj,i hj,i si (6)
noise at the jth receive antenna. Thus, we obtain
where hj,i denotes the channel impulse response
X
Nt between receive antenna j and user i. Ctj,i , the
Xj ¼ ~ j,i ci si þ V j
H (2) Toeplitz spreading code matrix for user i, of size
i¼1 P L is expressed as
2 3
Allowing Di ¼ ci si denote the N N matrix of user i’s bi1 0 0 . . . 0
spread signal, (2) can be written as 6 bi2 bi1 0 07
6 7
6 : : bi1 07
X
Nt 6 7
6 7
Xj ¼ ~ j,i Di þ V j
H (3) C tj,i ¼ 6 biN . . . . . . bi1 7 (7)
6 bi1 0 . . . 0 7
i¼1 6 7
6 . . 7
4 .. bi1 .. 05
The equivalent matrix vector form of the received signal
model is of the form as expressed in (4) with X biL . . . . . . bi1
denoting the net received signal over all received
antenna given by X ¼ [X T1 X T2 . . . X TNr ]T where bin (n ¼ 1, 2, . . . , N) is the nth bit of the time
domain spreading code sequence for user i.
X
Nt
X¼ ~ i Di þ V
H (4) The autocorrelation matrix of the received signal is
i¼1 written as
~ i is defined as
where H Rj ¼ E[X j X Hj ] (8)
The SVD of Rj yields
~ T1,i H
~ i ¼ [H
H ~ TN ,i ]T (5)
" #
r
L 0 U Tsj
and V represents the noise denoted as V ¼ Rj ¼ [U sj U nj ] s (9)
[V T1 V T2 . . . V TNr ]T
0 Ln U Tnj
3.1 Subspace method The key equation in the subspace method is the
minimisation of the following quadratic cost function
The subspace-based method utilises the guard-band to
estimate the signature waveform, which is the
convolution of the channel and users spreading code h^ j,i ¼ arg min khHj,i Q j,i hj,i k2 (11)
khj,i k¼1
[24, 25]. The system models used in [24, 25, 28]
were that of DS-CDMA system with no guard-band where the matrix Qj,i for user i is given by
structure and using the whole signal with ISI. Our
model is similar but with the significant difference of
having a guard-band to avoid IBI, while utilising the Q j,i ¼ C Htj,i U nj U Hnj C tj,i (12)
496 / IET Commun., 2008, Vol. 2, No. 3, pp. 493– 503 & The Institution of Engineering and Technology 2008
doi: 10.1049/iet-com:20070014
Authorized licensed use limited to: VELLORE INSTITUTE OF TECHNOLOGY. Downloaded on August 2, 2009 at 04:29 from IEEE Xplore. Restrictions apply.
www.ietdl.org
The solution is obtained by taking the eigenvector Utilising the equivalence of (13) and (15), we obtain the
corresponding to the lowest eigenvalue. following equation
& The Institution of Engineering and Technology 2008 IET Commun., 2008, Vol. 2, No. 3, pp. 493 – 503/ 497
doi: 10.1049/iet-com:20070014
Authorized licensed use limited to: VELLORE INSTITUTE OF TECHNOLOGY. Downloaded on August 2, 2009 at 04:29 from IEEE Xplore. Restrictions apply.
www.ietdl.org
w i [n] ¼ R1 ^
i [n]H i [n] (25)
498 / IET Commun., 2008, Vol. 2, No. 3, pp. 493– 503 & The Institution of Engineering and Technology 2008
doi: 10.1049/iet-com:20070014
Authorized licensed use limited to: VELLORE INSTITUTE OF TECHNOLOGY. Downloaded on August 2, 2009 at 04:29 from IEEE Xplore. Restrictions apply.
www.ietdl.org
given by
1XN 1
MSEin ¼ 1 ^ Hi [n]R1
H ^
i [n]H i [n] (27)
N n¼0
& The Institution of Engineering and Technology 2008 IET Commun., 2008, Vol. 2, No. 3, pp. 493 – 503/ 499
doi: 10.1049/iet-com:20070014
Authorized licensed use limited to: VELLORE INSTITUTE OF TECHNOLOGY. Downloaded on August 2, 2009 at 04:29 from IEEE Xplore. Restrictions apply.
www.ietdl.org
Figure 7 Performance of subspace þ LSFE and ACM þ LSFE Figure 9 Performance of LSFE and MMSE for MIMO SC-
for two transmit antenna and two, four receive antennas CDMA uplink system with N ¼ 32, L ¼ 4 and varying
MIMO SC-CDMA uplink with N ¼ 32, L ¼ 4 number of users (Nu)
Fig. 8 shows the performance of LSFE and MMSE and detection. In this analysis, only matrix
equalisers using the blind estimates as channel length multiplications and matrix inverse operations are
increases at the SNR of 20 dB. It is seen that the considered. The tabulated results are presented in
performance does not vary much as channel length Tables 1 and 2.
varies. Again the improved performance of LSFE is
illustrated as compared with MMSE equaliser.
6.1 Subspace algorithm
Fig. 9 illustrates the effect of MAI on a two transmit,
two receive antenna systems at the base-station. As the The subspace channel estimation method involves:
number of users increases, the performance gets worse,
but it is clear that the ACM-based estimates perform 1. Correlation at each receive antenna
slightly better overall as compared with the subspace-
based estimates.
The complexity of which is N 2 Ns , where Ns is the
number of symbols and N is the number of carriers.
Thus the net computations required will be
6 Complexity analysis
The numerical complexity is presented below for the
two stages involved, namely the channel estimation CostR ¼ Nr N 2 Ns (29)
Costls,i ¼ N 6 L3 (30)
Figure 8 Performance of LSFE and MMSE for 2 2 MIMO Nr, no. of receive antenna; Nt , no. of transmit antennas; P,
SC-CDMA uplink system with N ¼ 32 and varying L Block length; L, channel order; Ns , no. of symbols
500 / IET Commun., 2008, Vol. 2, No. 3, pp. 493– 503 & The Institution of Engineering and Technology 2008
doi: 10.1049/iet-com:20070014
Authorized licensed use limited to: VELLORE INSTITUTE OF TECHNOLOGY. Downloaded on August 2, 2009 at 04:29 from IEEE Xplore. Restrictions apply.
www.ietdl.org
which is futher simplified as 2. Detection of ordered stream using MMSE for each
user/stream. This would be the same value as above
Costsub ¼ Nr N 2 Ns þ Nt N 6 L3 (33) but multipleid by Nt times for each user
1. Auto-correlation for each receive antenna Costlsfe3 ¼ (Nt 1)Nr NNs (43)
& The Institution of Engineering and Technology 2008 IET Commun., 2008, Vol. 2, No. 3, pp. 493 – 503/ 501
doi: 10.1049/iet-com:20070014
Authorized licensed use limited to: VELLORE INSTITUTE OF TECHNOLOGY. Downloaded on August 2, 2009 at 04:29 from IEEE Xplore. Restrictions apply.
www.ietdl.org
anonymous reviewers for their valuable comments and Proc. Vehicular Technology Conf., September 2004, vol. 1,
suggestions on this paper. pp. 689– 693
[13] SILVA P., DINIS R.: ‘Frequency domain multiuser detection for
CP-Assisted DS-CDMA signals’. Proc. IEEE Vehicular Technology
9 References Conf., Melbourne, May 2006, vol. 5, pp. 2103–2108
[1] PRASAD R. : ‘CDMA for wireless personal [14] SILVA P., DINIS R.: ‘Multiuser detection for the uplink of
communications’ Mobile communication series (Artech prefix-assisted DS-CDMA systems employing multiple
House Publishers, 1996) transmit and receive antennas’. Proc. IEEE Vehicular
Technology Conf., Montreal, September 2006, pp. 1 – 5
[2] ADACHI F., GARG D., TAKAOKA S., TAKEDA K.: ‘Broadband CDMA
techniques’, IEEE Commun. Mag., 2005, pp. 8 – 18 [15] WINTERS J.: ‘On the capacity of radio communication
systems with diversity in a Rayleigh fading environment’,
[3] WANG Z., GIANNAKIS G.B.: ‘Wireless multi-carrier IEEE J. Sel. Areas Coomun., 1987, 5, pp. 871– 878
communications – where Fourier meets Shannon’, IEEE
Signal Process. Mag., 2000, 17, (3), pp. 29– 48 [16] FOSCHINI G.J.: ‘Layered space time architecture for wireless
communication in a fading environment when using mulitple
[4] WANG Z., MA X., GIANNAKIS G.B.: ‘OFDM or single-carrier antennas’, Bell Labs Tech. J., 1996, 1, pp. 41–59
block transmissions’, IEEE Trans. Commun., 2004, 52, (3),
pp. 380– 394 [17] FOSCHINI G.J., GANS M.J. : ‘On limits of wireless
communications in a fading environment when using
[5] MUQUET B., WANG Z., GIANNAKIS G.B., COURVILLE M., DUHAMEL P.: multiple antennas’, Wirel. Pers. Commun., 1998, 6, (3),
‘Cyclic prefixing or zero-padding for wireless multicarrier pp. 311– 335
transmissions’, IEEE Trans. Commun., 2002, 50, (12),
pp. 2136 – 2148 [18] ZHU X., MURCH R.D.: ‘Layered space frequency equalization in
a single carrier MIMO system for frequency selective channels’,
[6] HARA S. , PRASAD R.: ‘Overview of multicarrier CDMA’, IEEE Trans. Wirel. Commun., 2004, 5, (5), pp. 701–708
Commun. Mag., 1997, 35, (12), pp. 126 – 133
[19] WANG Q., LI B., YANG D.: ‘Low complexity layered space-
[7] YEE N., LINNARTZ J.P. , FETTWEIS G.: ‘Multi-carrier CDMA frequency scheme for MIMO-CDMA system’. Proc. Int.
in indoor wireless radio networks’. Proc. 4th IEEE Int. Symp. Conf. Wireless Communications, Networking and Mobile
Personal, Indoor and Mobile Radio Communications, Computing, September 2005, vol. 1, pp. 139– 142
Yokohama, Japan, September 1993, pp. 109 – 113
[20] CARDOSO J.F.: ‘Blind signal separation: statistical
[8] CHOULY A., BRAJAL A., JOURDAN S.: ‘Orthogonal multicarrier principles’, Proc. IEEE, 1998, 9, (10), pp. 2009 – 2025
techniques applied to direct sequence spread spectrum
CDMA systems’. Proc. IEEE Global Telecommunications Conf., [21] MENDEL J.M.: ‘Tutorial on higher-order statistics (spectra)
Houston, Texas, USA, November 1993, vol. 3, pp. 1723–1728 in signal processing and system theory: theoretical results
and some applications’, Proc. IEEE, 1991, 79, (3),
[9] MADHUKUMAR A.S., CHIN F., LIANG Y.C., YANG K.: ‘Single-carrier pp. 278– 305
cyclic prefix-assisted CDMA system with frequency
domain equalisation for high data rate transmission’, [22] TONG L., XU G., KAILATH T.: ‘Blind identification and
EURASIP J. Wirel. Commun. Netw., 2004, 1, pp 149– 160 equalization based on second-order statistics: a time
domain approach’, IEEE Trans. Inf. Theory, 1994, 40, (2),
[10] BAUM K., THOMAS T., VOOK F., NANGIA N.: ‘Cyclic-prefix CDMA: pp. 340– 349
an improved transmission method for broadband DS-CDMA
cellular systems’. Proc. IEEE Wireless Communications and [23] TONG L., PERREAU S.: ‘Multichannel blind identification:
Networking Conf., Orlando, March 2002, vol. 1, from suspace to maximum likelihood methods’, Proc.
pp. 183– 188 IEEE, 1998, 86, (10), pp. 1951 – 1968
[11] BURG A., RUPP M., HAENE S., PERELS D., FLEBER N., FICHTNER W.: ‘Low- [24] LIU H. , XU G.: ‘A subspace method for signature
complexity frequency domain equalization of MIMO channels waveform estimation in synchronous CDMA systems’, IEEE
with applications to MIMO-CDMA systems’. Proc. IEEE Trans. Commun., 1996, 44, (10), pp. 1346 – 1354
Vehicular Technology Conf., October 2003, vol. 1, pp. 468–472
[25] DOUKOPOULOS X.G. , MOUSTAKIDES G.V. : ‘Blind channel
[12] GARG D., ADACHI F.: ‘DS-CDMA with frequency domain estimation for downlink CDMA systems’. IEEE Int. Conf.
equalization for high-speed downlink packet access’. IEEE Communications, May 2003, vol. 4, pp. 2416 – 2420
502 / IET Commun., 2008, Vol. 2, No. 3, pp. 493– 503 & The Institution of Engineering and Technology 2008
doi: 10.1049/iet-com:20070014
Authorized licensed use limited to: VELLORE INSTITUTE OF TECHNOLOGY. Downloaded on August 2, 2009 at 04:29 from IEEE Xplore. Restrictions apply.
www.ietdl.org
[26] XU Z. : ‘Blind channel estimation via subspace Communications, Networking and Mobile Computing,
approximation’. Asilomar Conf. Signals Syst. Comput., Wuhan, September 2006, vol. 1, pp. 1 – 4
2003, vol. 52, pp. 1653 – 1657
[30] YANG K., MADHUKUMAR A.S., CHIN F.: ‘Novel two-dimensional
[27] CHEN S., CONSTANTINIDES A.G.: ‘Optimum SINR receiver in multi-stage interference cancellation scheme for uplink
dispersive CDMA channels’. Personal, Indoor and Mobile transmission of single carrier cyclic prefix-assisted CDMA
Radio Communications, 2004, PIMRC 15th IEEE Int. system’, IEE Proc. Commun., 2003, 150, (4), pp. 287 – 292
Symp., 5 – 8 September 2004, vol. 4, pp. 2519 – 2523
[31] GOLUB G.H. : ‘Matrix computations’ (John Hopkins
[28] KANG W., CHAMPAGNE B.: ‘Generalized blind subspace University Press, 1996, 3rd edn.), ISBN:0801854148
channel estimation’. Proc. Vehicular Technology Conf.,
October 2003, vol. 2, pp. 1209 – 1213 [32] PUNNOOSE S., ZHU X., NANDI A.K.: ‘Layered space frequency
equalisation for MIMO-MC-CDMA systems in frequency
[29] PUNNOOSE S., ZHU X., NANDI A.K.: ‘Blind channel estimation selective fading channels’. Proc. 5th Int. Conf. Independent
for MIMO uplink MC-CDMA systems with layered space Component Analysis and Blind Signal Separation, Granada,
frequency equalisation’. Proc. Second Int. Conf. Wireless Spain, September 2004, pp. 1181–1188
& The Institution of Engineering and Technology 2008 IET Commun., 2008, Vol. 2, No. 3, pp. 493 – 503/ 503
doi: 10.1049/iet-com:20070014
Authorized licensed use limited to: VELLORE INSTITUTE OF TECHNOLOGY. Downloaded on August 2, 2009 at 04:29 from IEEE Xplore. Restrictions apply.