MULTICARRIER TECHNIQUES: OWSS AND MC-CDMA

Ing. Tomáš KAŠPAREC, Doctoral Degreee Programme (1)

Dept. of Radio Electronics, FEEC, BUT
E-mail: [email protected]

Supervised by: Prof. Vladimír Šebesta

ABSTRACT
This article deals with comparing of two multicarrier systems in AWGN channel.
Comparative parameter is BER (Bit Error Rate). MC-CDMA (Multicarrier Code Division
Multiple Access) uses orthogonality of cosine functions. While OWSS (Orthogonal wavelet
spread spectrum) utilizes so called OWSS pulses. These techniques change frequency
selective fading channel into more flat fading channels. Multicarrier techniques are proposed
to next generation of mobile communications, e.g. 4th generation.

1 INTRODUCTION

Multicarrier CDMA (Code Division Multiple Access) communication is a combination

of multicarrier modulation scheme and CDMA concept. Multicarrier communication (MC)
schemes divide fast data stream into more streams, which are not so fast. Each the data stream
modulates other subcarrier. Subcarriers should be orthogonal. Complex exponentials are good
subcarriers for example. The purpose to use multicarrier transmission is to extend the symbol
duration so that a frequency selective fading channel is divided into a number of flat fading
channels, because subcarriers have much narrower bandwidths. CDMA is a multiple access
technic using some properties of spreading sequences for separation different users.
Three types of multiple access schemes based on a combination of code division and
multicarrier techniques were proposed, such as “multicarrier (MC-) CDMA,” “multicarrier
DS-CDMA,” and “multitone (MT-) CDMA” [1]. Those three techniques use the cosine
waveforms as subcarriers. Further MC systems are known which use another shape of
subcarriers. In [2] a new signaling system is proposed which using a new class of signaling
pulses. Called OWSS (Orthogonal Wavelet Spread Spectrum), they are generated through a
combination of OWDM (Orthogonal Wavelength-Division Multiplexing) and spread-
spectrum concepts. It is known techniques with wavelet packet transformation [3,4].
This article reviews the multicarrier CDMA schemes, specially MC-CDMA scheme and
a little modified OWSS concept. In this paper a downlink wireless communication AWGN
channel is assumed, in which the signals from different users are assumed synchronized. The
article shows their properties in AWGN channel. In the following, MC-CDMA scheme is
introduced. Next section is devoted OWSS signaling system. Systems simulation results are
given in the last section.
2 MC-CDMA

The multicarrier CDMA schemes can be categorized mainly into two groups. One
spreads the user symbols in the frequency domain, the other spreads user symbol in the time
domain. The MC-CDMA transmitter spreads the original data stream over different
subcarriers using a given spreading code in the frequency domain. A fraction of the user
symbol corresponding to a chip of the spreading code is transmitted through a different
subcarrier. In this article only downlink channel is considered. We do not pay attention to the
auto-correlation characteristic of the spreading code. So we can use the spreading code, which
is orthogonal. The Walsh-Hadamard codes are used [4].
cos(2πf1t)
c1

c2 cos(2πf2t)

Σ
Data stream cos(2πfMt)
cM
Time

Data stream

Time
Fig. 1: MC-CDMA transmitter scheme.

Figure 1 shows the MC-CDMA transmitter scheme, where M denotes the number of
subcarrier and c = [c1 c2 ... cM] is a spreading Walsh-Hadamard code with processing gain M.
Then we consider processing gain and number of subcarriers are the same.

cos(2πf1t) q1

LPF
cos(2πf2t) q2
Received
signal LPF
Σ
cos(2πfMt)
qM

LPF

Fig. 2: MC-CDMA receiver.

Figure 2 shows the MC-CDMA receiver, where after the serial-to-parallel conversion,
the m-th subcarrier is multiplied by the gain qm to combine the received signal energy
scattered in the frequency domain. The decision variable is given by equation (1).
 M J
D j = ∑ qm ∑ z mj a j cmj + nm (1)
m =1 j =1

J
Where, y m = ∑ z mj a j c mj + nm and nm are the complex baseband component of the
j =1

received signal after down-conversion with subcarrier frequency synchronization and the
complex additive Gaussian noise at the m-th subcarrier, respectively, zmj and aj are the
complex envelope of the m-th subcarrier and the transmitted symbol for the j-th user,
respectively, and J is the number of active users. If we use ORC (Orthogonality Restoring
Combining) then the gain is qmj=cmjzm*|zm|2. Other detection strategies are in [1].

3 OWSS

OWSS is a proposed signaling system for 100-150 MB/s wireless LANs. OWSS pulses
are generated through a combination of orthogonal wavelength-division multiplexing and
spread-spectrum concepts. The system offers several advantages including robustness to
frequency-selective fading, effective equalization due, in part, in to the wide spectrum and
wide time-support of the pulses used, high data rate [2]. The concept uses the pulses φm(t),
m = 0, 1, ...,M-1 from an orthonormal set over certain interval of time instead cosine functions
in MC-CDMA. Then output signal over time T is
M −1
s 0 (t ) = ∑a m φ m (t ), (2)
m =0
where M is the number of various pulses in orthonormal set, and the length of pulses is
T = MTS, where TS is the basic symbol interval. Each basis pulse φm(t) serves to create a
“virtual” channel over which the symbol am is carried. A supersymbol is vector of symbol
A = [a0 a1 ... aM-1]T. The basic sets are generated from wavelets. All OWDM basis functions
can be generated through the process of scaling or shifts from a single function, called the
mother wavelet. And it is reason why they are indexed by two indices, one signifying scale,
the other shift. Thus, if ψ(t) is the mother wavelet, the basis functions are of the form:
ψ m ,n = 2 − m / 2 ψ(2 - m t − n), (3)
where m is scale index and n shift index, both indexes are integer.
The orthonomal basis sets are generated through a tree structure. To illustrate, let us
define g0 = [2-1/2 2-1/2] and g1 = [2-1/2 -2-1/2], g0 is a scaling function and g1 constitutes the well-
known Haar wavelet. Then we show in fig. 3 two full trees, which can generate basis function
over support intervals of 2 and 4, respectively.
3 2 G1(z)
High-pass filter
∑ 2 G1(z)
1 2 G1(z) h(t)
h(t) 2 2 G0(z)
∑ ∑
0 2 G0(z) 1 2 G1(z)
Low-pass filter ∑ 2 G0(z)
0 2 G0(z)
(a) (b)

Fig. 3: Synthesis tree, (a) one-stage,(b) two-stages.

 In the fig. 3 there are circles, with up arrow, which denote upsampling device, and the
digit 2 signifies that the upsampling ratio is two. G0(z) and G1(z) is transfer function of low-
pass filter and high-pass filter, respectively. They are gained by z-transform of g0 and g1. In
each of the two cases shown in this figure, a member of the basis set is generated by applying
a unit pulse at any one, and only one, input node while all other nodes receive a zero input.
The transmitted signal for the ith user is
M −1
s ( i ) (t ) = ∑ a n( i ) ∑ c m( i ) φ m (t − nT ) (4)
n m =0
where an is data symbol, c(i) = [c1 c2 ... cM] is spreading code for the i-th user. Spreading
codes are Walsh-Hadamard codes, because these codes are orthogonal (see section 2). In [2]
are proposed receiver techniques based on DFE (Decision Feedback Equalizer) and
possibility combination of OWSS and STC (Space-Time Coding).

4 SIMULATION

Simulation was done in Matlab. Parameters of simulation are in table (1). At first data
are generated for each user. These data are bipolar (1 or -1) and the probability of ones is the
same as probability of minus ones. Data are input of OWSS or MC-CDMA transmitter, where
data are sampled and modulated. AWGN (Additive White Gaussian Noise) is added to output
from transmitter, Matlab function awgn is used. In this simulation we consider downlink
channel, and good chip synchronization. Output of channel is filtered by low-pass filter. RRC
(Root Raised Cosine) filter is used. The parameters of filter are: bandwidth is M∆f, in this
case bandwidth is 8, roll-off factor is 0.5. In receiver the averaging on symbol duration is
consider instead of low-pass filtering (see fig.2 block LPF). The simulation outputs are two
graphs, which both show comparing of OWSS and MC-CDMA fig.4. On the left picture there
is BER performance as function of SNR (Signal to Noise Ration) in dB and on the right one
there is dependence BER on number of user.

OWSS MC-CDMA
The data symbol duration Tc = 1 Tc = 1
Number of subcarriers M=8 M=8
Symbol duration at subcarrier Ts = MTc Ts = MTc
Processing gain (length of code) M M
Sample period Tvz = 0.0125 Tvz = 0.0125
Subcarrier separation ∆f = 1/ Tc
Wavelet haar
Length of data vector 20 000 20 000
Tab. 1: Simulation parameters
5 SIMULATION RESULT AND DISCUSSION

Simulation results may be seen on the fig. 4. The picture on the left side gives
comparison of BER performance between OWSS based on wavelet and MC-CDMA. System
OWSS has better BER performance than MC-CDMA. The similar situation shows the picture
on the right side, OWSS has better performance than MC-CDMA.

Fig. 4: Simulation result, BER performance

ACKNOWLEDGMENTS
This research has been supported by the research grant GACR (Grant Agency of Czech
Republic) No. 102/04/0557 "Development of the digital wireless communication resources".

REFERENCES
[1] Hara, S., Prasad, R.: Overview of Multicarrier CDMA. IEEE Communications Magazine,
35, Dec. 1997, p. 126-133.
[2] Jain, V., K., Myers, B., A.: OWSS: A New Signaling System For 100-150 MB/s Wireless
LANs. IEEE Wireless Communications. Aug. 2003, p. 16-24.
[3] Zhang, H., Fan, H., H., Lindsey, A.: Wavelet Packet Waveforms For Multicarrier CDMA
Communications. Proceedings of IEEE ICASSP 2002, Florida,USA, Vol. 3, pp. 2557-
2560.
[4] Shi, Q., Latva-aho, M.: Spreading Sequences for Asynchronous MC-CDMA Revised:
Accurate Bit Error Rate Analysis. IEEE Transaction On Communications, 51, Jan. 2003,
p. 8-11.