You are on page 1of 13

This article was downloaded by: [National Institute of Technology - Warangal]

On: 16 August 2014, At: 04:02


Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

International Journal of Electronics


Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/tetn20

Improving performance of DS-CDMA


systems using chaotic complex
Bernoulli spreading codes
a b
Mohammad Farzan Sabahi & Ali Dehghanfard
a
Department of Electrical Engineering, University of Isfahan,
Isfahan, Iran
b
Department of Electrical Engineering, Azad University, Bushehr
Branch, Bushehr, Iran
Accepted author version posted online: 31 Jan 2014.Published
online: 19 Feb 2014.

To cite this article: Mohammad Farzan Sabahi & Ali Dehghanfard (2014): Improving performance
of DS-CDMA systems using chaotic complex Bernoulli spreading codes, International Journal of
Electronics, DOI: 10.1080/00207217.2014.888781

To link to this article: http://dx.doi.org/10.1080/00207217.2014.888781

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
Downloaded by [National Institute of Technology - Warangal] at 04:02 16 August 2014
International Journal of Electronics, 2014
http://dx.doi.org/10.1080/00207217.2014.888781

Improving performance of DS-CDMA systems using chaotic complex


Bernoulli spreading codes
Mohammad Farzan Sabahia* and Ali Dehghanfardb
Downloaded by [National Institute of Technology - Warangal] at 04:02 16 August 2014

a
Department of Electrical Engineering, University of Isfahan, Isfahan, Iran; bDepartment of
Electrical Engineering, Azad University, Bushehr Branch, Bushehr, Iran
(Received 19 May 2013; accepted 12 August 2013)

The most important goal of spreading spectrum communication system is to protect


communication signals against interference and exploitation of information by unin-
tended listeners. In fact, low probability of detection and low probability of intercept
are two important parameters to increase the performance of the system. In Direct
Sequence Code Division Multiple Access (DS-CDMA) systems, these properties are
achieved by multiplying the data information in spreading sequences. Chaotic
sequences, with their particular properties, have numerous applications in constructing
spreading codes. Using one-dimensional Bernoulli chaotic sequence as spreading code
is proposed in literature previously. The main feature of this sequence is its negative
auto-correlation at lag of 1, which with proper design, leads to increase in efficiency of
the communication system based on these codes. On the other hand, employing the
complex chaotic sequences as spreading sequence also has been discussed in several
papers. In this paper, use of two-dimensional Bernoulli chaotic sequences is proposed
as spreading codes. The performance of a multi-user synchronous and asynchronous
DS-CDMA system will be evaluated by applying these sequences under Additive
White Gaussian Noise (AWGN) and fading channel. Simulation results indicate
improvement of the performance in comparison with conventional spreading codes
like Gold codes as well as similar complex chaotic spreading sequences. Similar to
one-dimensional Bernoulli chaotic sequences, the proposed sequences also have nega-
tive auto-correlation. Besides, construction of complex sequences with lower average
cross-correlation is possible with the proposed method.
Keywords: DS-CDMA system; spread spectrum; chaotic communication; complex
chaotic sequence

1. Introduction
In spreading spectrum systems, main information is spread in a wide bandwidth. One of
the main methods to spread the spectrum is Direct Sequence Code Division Multiple
Access (DS-CDMA) in which information of all users will be sent simultaneously at the
same frequency band (Proakis & Salehi, 2008; Rovatti, Mazzini & Setti, 2004;
Stavroulakis, 2005; Torrieri, 2011). The spectrum will be spread by using the sequences
with specialised features which are called spreading sequences. Indeed, for each user, a
particular sequence will be sent instead of each information bit. The receiver uses the
same spreading sequence to detect the received data. As a result of spreading, the
transmitted signal occupies more bandwidth than the original message. Properties of
spreading sequences are mainly determined by their auto-correlation and crosscorrelation.

*Corresponding author. Email: sabahi@eng.ui.ac.ir

© 2014 Taylor & Francis


2 M.F. Sabahi and A. Dehghanfard

The code should be orthogonal or quasi-orthogonal to each other in different delays, to


use in multipath channels ideally. The most commonly used sequences in DS-CDMA are
maximal length (m-sequence), Gold and Kasami (Proakis & Salehi, 2008; Stavroulakis,
2005; Torrieri, 2011), which are implementable by Linear Feedback Shift Registers
(LFSRs) (Torrieri, 2011). In recent decades, using chaotic sequences as spreading codes
has been of great interest. Chaotic sequences, despite the deterministic nature, show the
irregular and pseudo-random behaviour. These sequences can be produced in arbitrary
length and they are entirely non-periodic. Due to extreme sensitivity of chaotic sequences
in initial conditions, numerous chaotic sequences with desired properties can be created.
Generated chaotic sequences are almost orthogonal with the low cross-correlation value
Downloaded by [National Institute of Technology - Warangal] at 04:02 16 August 2014

that made them desirable to be used in spread spectrum systems (Stavroulakis, 2005).
Unlike the conventional pseudo-noise sequences, chaotic sequences can be made by
simple recursive equations. These sequences with different characteristics and parameters
have been used in several references (Rahnama & Talebi, 2013; Rovatti & Mazzini, 1998;
Stavroulakis, 2005; Tsuneda, 2005). In Tsuneda (2005) and Rovatti and Mazzini (1998), it
has been shown that chaotic sequences with exponentially vanishing auto-correlation
function can improve the performance of spreads spectrum systems. In Tsuneda and
Miyazaki (2010) and Cong and Shaoqian (2000), chaotic sequences with negative auto-
correlation at lag of 1 have been studied as a near optimal spreading sequence. Complex
chaotic sequences are introduced in Kurian, Puthusserypady, and Htut (2005) and
Mazzini, Setti, and Rovatti (1997), and performance of the DS-CDMA system based on
them has been analysed. The spread spectrum system can have better performance in
terms of cross-correlation and the average Bit Error Rate (BER), using two-dimensional or
complex sequences. This is due to more flexibility to select real and imaginary parts of
each sequence. In Kurian et al. (2005), the complex IKEDA map is used to generate two-
dimensional chaotic sequences.
In a multi-user asynchronous DS-CDMA systems (such as the reverse link of cellular
systems), each user is affected by interference of other users. The average interference can
be described by the auto-correlation function of codes (Pursley, 1997). So by designing
the sequences with appropriate auto-correlation functions, the proper spreading codes
sequences can be produced. This idea has been used to proper design of one-dimensional
chaos spreading sequences in Tsuneda and Miyazaki (2010). In this paper, the idea of
utilising two Bernoulli chaotic sequences to form two-dimensional (complex) sequences is
presented. This paper shows that the auto-correlation function of such two-dimensional
sequence has similar behaviour with the one-dimensional sequence in average.
Additionally, according to the new complex sequences, it is expected that spreading
codes with a better cross-correlation than the one-dimensional codes can be selected.
Different sections of the paper will be arranged as follows. In the second section, at
first, the DS-CDMA systems using two-dimensional spreading code is reviewed. In
Section 3, Bernoulli chaotic sequence is studied, and in Section 4, the proposed two-
dimensional Bernoulli chaotic sequence is introduced. In Section 5, analysis and computer
simulations of the DS-CDMA system with complex Bernoulli chaotic spreading code are
presented. The simulations are carried out for Additive White Gaussian Noise (AWGN)
channel and Rayleigh fading channel. Section 6 is devoted to conclusions.

2. DS-CDMA system with two-dimensional spreading code


Figure 1 illustrates thePblock diagram of a DS-CDMA transmitter in complex mode (for
the kth user). bk ðtÞ ¼ 1 n¼1 bk pðt  nTbÞ is the data waveform for the kth user, where
n
International Journal of Electronics 3

Figure 1. Block diagram of DS-CDMA transmitter with complex spreading sequence.


Downloaded by [National Institute of Technology - Warangal] at 04:02 16 August 2014

P
bnk 2 f1g and Tb is the bit time duration. ck ðtÞ ¼ 1 n¼1 ck pðt  nTc Þ is the spreading
n

code for the kth user and ck 2 f1  jg. Tc is called the chip time. pðÞ is the rectangular
n

pulse and ratio of TTbc is called spreading factor.


In the receiver end, to detect the data bits, the received waveform is firstly down-
converted to baseband. Despreading is performed by taking the dot product of the base-
band signal with the hermitian transpose of the spreading sequence (assuming perfect
synchronisation at code level). After despreading, most of the transmitted bit energy
concentrated in the real parts of the signal will provide the complex spreading sequences
which have proper auto-correlation properties. Hence, the real part alone is considered for
the detection purpose.
In synchronous DS-CDMA systems (e.g. in the forward link of the cellular system)
with K active users, the received signal for kth user after despreading and sampling can be
shown as follows:

XK
y½n ¼ 2zk bnk þ z ðρ
l¼1; lÞk l R;l
þ ρI;l Þbnl þ ν½n (1)

where zk is the amplitude of the transmitted signal for the kth user. ν[n] is the noise term
due to despreading. From (1), it can be seen that the interference term has two cross-
correlation terms ρR,l and ρI,l corresponding to the real and imaginary part of the spreading
sequence. ρR,l and ρI,l are the cross-correlation values between the real and imaginary parts
of spreading sequences of the kth user, and lth user, respectively.
In asynchronous systems (e.g. in the reverse link in cellular systems), signals from
different users receive with random different time shifts to the kth user receiver. So, for
modelling the interference, the partial correlation functions should be used. Pursley (1997)
showed that it is still possible to express the average interference at the receiver output of
asynchronous DS-CDMA system for the kth user as a function of the cross-correlation
parameters ρR,l and ρI,l between kth user and lth user ðl ¼ 1; 2; ::: K; lÞkÞ (Pursley, 1997).
If the real and imaginary parts in (1) are opposite in sign, the sum can be made much
smaller than the individual values. This property of the complex spreading sequences can
be utilised for interference cancellation. In other words, there are more degrees of freedom
to select a number of sequences with the above-mentioned property among the two-
dimensional spreading codes. It is notable that the conventional spreading codes (like
Gold codes and Kasami codes) introduce only a limited number of codes with desirable
auto-correlation and cross-correlation values for certain spreading factor. On the other
hand, chaotic sequences with different initial conditions can produce numerous proper
sequences, and selecting good spreading sequences is much easier.
4 M.F. Sabahi and A. Dehghanfard

3. Chaotic Bernoulli sequence with negative auto-correlation


In a DS-CDMA system, the average interference seen by user k from user i is obtained as
follows (Pursley, 1997; Tsuneda, 2005):
XN 1 XN 1
rk;i ¼ 2N 2 þ 4 l¼1
Ak ðlÞAi ðlÞþ l¼1N
Ak ðlÞAi ðl þ 1Þ (2)

where N is the spreading factor and Ak ð:Þ is the auto-correlation function of real spreading
n oN1
ðnÞ
sequence ck used for kth user and is defined as follows:
n¼0
Downloaded by [National Institute of Technology - Warangal] at 04:02 16 August 2014

8 PN 1l ðnÞ ðnþ1Þ


>
< n¼0 ck ck ð0  l  N  1Þ
Ak ðlÞ ¼ P
ck ck ð1  N  l  0Þ
N 1þl ðnlÞ ðnÞ
>
: n¼0 (3)
0 ðj l j > N Þ

Using Equation (2), the average Signal to Noise ratio (SNR) at the output of correlation
receiver for ith user in AWGN channel is as follows (Pursley, 1997):
 1
1 XK N0
SNRi ¼ r þ
k¼1; kÞi k;i
(4)
6N 3 2Eb

Where K is the number of active users, N20 is the power spectral density of white noise and
Eb is signal energy of ith user for each bit. In other words, we can say that system quality
is directly related to the auto-correlation functions. Different chaotic sequences with
proper auto-correlation features for DS-CDMA systems are presented. Generally, chaotic
sequences are generated by recursive maps and have particular properties. One of these
properties is extreme sensitivity to initial values, as for the two initial values that are very
close together, after a few steps, the output values will have a large relative distance from
each other. One of the maps used to generate chaotic sequence is the Bernoulli map,
which is defined as follows:

2xn 0  x < 0:5
τðxnþ1 Þ ¼ (5)
2xn  1 0:5  x < 1

By choosing the initial point x0 ð0; 1Þ, a real valued chaotic sequence fxn ¼ τ n ðx0 Þg is
obtained, in which τ n ð:Þ means that the function τð:Þ is applied for n times. To construct a
spreading code, a binary converter function is used for converting the continuous values
ðxn Þ to  1 values. Binary converter function is defined as follows (Tsuneda & Miyazaki,
2010):

Bð xÞ ¼ Θ18 ð xÞ  Θ38 ð xÞ þ Θ34 ð xÞ (6)

where Θt ð xÞ is the Threshold function as follows:



1 ðx < t Þ
Θ t ð xÞ ¼ (7)
1 ðx  t Þ
International Journal of Electronics 5

It is also possible to use other binary functions (Tsuneda & Miyazaki, 2009a). Average
auto-correlation function for the binary sequence is defined as follows (Kohda & Tsuneda,
1997):

Z
 
CðlÞ ¼ E BðτðxÞÞBðτ l ðxÞÞ ¼ BðτðxÞÞBðτ l ðxÞÞf ðxÞdx (8)

where f ð:Þ is the Invariant Density Function for τð:Þ and the integral is calculated over
interval ½0; 1. In Tsuneda et al. (2004), it is shown for one-dimensional Bernoulli
 binary
Downloaded by [National Institute of Technology - Warangal] at 04:02 16 August 2014

sequences, constructed by the binary function Bð:Þ, that f ðxÞ ¼ 1 and E Bðτ l ðxÞÞ ¼ 0. If
all active users use the Bernoulli sequences generated by this way, and the initial values of
sequences are statistically independent of each other, by averaging the Equation (2), the
average interference parameter (AIP) for each user can be written as follows:

XN1 XN1
r ¼ 2N 2 þ 4 l¼1
ðN  lÞ2 CðlÞ þ 2 l¼1
ðN  lÞðN  l þ 1ÞCðlÞCðl  1Þ (9)

Normalised AIP is defined as R ¼ limN !1 2Nr 2 and can be supposed as a criterion for
spreading code performance. It is clear that for a completely random sequence, C(l) is in
form of δð:Þ and R = 1. In Rovatti and Mazzini (1998), chaotic sequences with exponen-
tially vanishing auto-correlation are introduced and the better performance than the
pseudo-noise code (that have auto-correlation function in form of δð:ÞÞ is shown. For
these sequences, CðlÞ ¼ λl ð0 < jλj < 1Þ and can be easily seen from (9) that R ¼ λ 1λ þλþ1
2
2 .
pffiffiffi pffiffi
The value of R is minimised at λ ¼ 2 þ 3 and has the value R ¼ 23 ¼ 0:866, which is
less than 1 for completely random sequences. Also in Rovatti and Mazzini (1998), the
method to create a chaotic sequence with exponentially vanishing auto-correlation func-
tion is presented.
If a sequence has an auto-correlation function with negative value at lag one and zero
at the other points, it can be shown that its properties is very similar to the sequence with
exponentially vanishing auto-correlation function (Tsuneda & Miyazaki, 2010; Tsuneda
et al., 2004). In fact in these kinds of sequences we have:

8
< 1 ðl ¼ 0Þ
CðlÞ ¼ " ðl ¼ 1Þ (10)
:
0 ðl  2Þ

and normalised AIP from (9) is calculated as R ¼ 2"2 þ " þ 1. Minimum value of R is
achieved at " ¼  14 and equal to R ¼ 78 ¼ 0:875 which is very close to optimum value of
spreading sequences that have exponentially vanishing auto-correlation (R = 0.866). In
Tsuneda and Miyazaki (2009b) using Bernoulli map a sequence has been designed that
have average auto-correlation function in form of Equation (10) with " ¼  14 . The auto-
correlation function of Bernoulli sequence of lengths 63 is shown in Figure 2 (Tsuneda
et al., 2004). Performance of DS-CDMA system utilising these codes is evaluated in
Tsuneda and Miyazaki (2010) and their proper features are presented.
6 M.F. Sabahi and A. Dehghanfard

0.5

Auto-correlation
0

–0.5

–1
0 10 20 30 40 50 60
Downloaded by [National Institute of Technology - Warangal] at 04:02 16 August 2014

Time delay

Figure 2. Average auto-correlation function of Bernoulli map in length of 63 (Tsuneda et al.,


2004).

4. Two-dimensional Bernoulli spreading sequence


As mentioned before, in case of using two-dimensional or complex sequences, there is
more freedom of choice of sequences with appropriate cross-correlation properties. In this
paper, two different Bernoulli sequences (with different initial conditions), as real and
imaginary parts of a two-dimensional sequence is used. At first, we show that the complex
sequence generated with this approach also has negative auto-correlation similar to real
Bernoulli sequences. Two generated Bernoulli sequences by Bernoulli map with different
initial conditions x; y 2 ½0; 1 and binary function B(.) according to (6) are considered.
Assume that the initial conditions x and y are statistically independent. The complex
binary sequence is defined as follows:
 N 1
1 
pffiffiffi Bðτ n ðxÞÞ þ jBðτ n ðyÞÞ (11)
2 n¼0

Factor of p1ffiffi2 has been considered to have the same power as real sequence. Average auto-
correlation function is obtained as follows:

1 h i
CðlÞ ¼ E BðτðxÞÞ þ jBðτðyÞÞÞðBðτ l ðxÞÞ  jBðτ l ðyÞÞ
2Z
(12)
1
¼ ðBðτðxÞÞ þ jBðτðyÞÞÞðBðτ l ðxÞÞ  jBðτ l ðyÞÞÞf ðxÞf ðyÞdx dy
2

where f(.) is invariant density function for τð:Þ and the integralis calculated over interval
[0,1]. As said before, it is shown that f(x) = 1 and E Bðτ l ðxÞÞ ¼ 0 for one-dimensional
Bernoulli binary sequence constructed by B(.). Because of the deterministic nature of the
Bernoulli map and the independence of x and y, it can be written:

1    
CðlÞ ¼ ðE BðτðxÞÞBðτ l ðxÞÞ þ E BðτðyÞÞBðτ l ðyÞÞ Þ (13)
2

Following (10) for one-dimensional Bernoulli sequences, it can be seen that the same
relationship is true for the complex sequence too. So, in the average, the similar behaviour
to one-dimensional Bernoulli sequences is expected. It should be noted that proper
behaviour of the one-dimensional Bernoulli sequence is obtained in the average. In the
International Journal of Electronics 7

practical systems, a limited number of spreading sequences must be selected and used. In
the suggested two-dimensional sequence, it is easier to select sequences with low cross-
correlation values. To construct a set of suitable sequences for using in multi-user system,
the same method as Kurian et al. (2005) is used. Applying different initial conditions (or
using different binary functions), we will produce a large number of binary Bernoulli
sequences. From this large pool of sequences, two sequences are selected randomly,
which form the real and imaginary parts of first complex spreading sequence. Cross-
correlation between these two sequences and rest of the sequences are calculated.
Sequences with negative and positive correlation values are grouped and then sorted.
The sequences with lowest cross-correlation values from each group are selected as real
Downloaded by [National Institute of Technology - Warangal] at 04:02 16 August 2014

and imaginary parts of the next complex sequence. These steps are repeated until the
required numbers of sequences are obtained. In the subsequent steps, the average cross-
correlation with previous sequences will be calculated for selecting the new sequence. In
each step, the new sequence has low average cross-correlation values with the previous
sequences in both real and imaginary parts. Furthermore, because of having the opposite
sign, sum of these values will be smaller.

5. Simulation results
5.1. Synchronous system
Computer simulations are carried out to evaluate the performance of synchronous DS-
CDMA system using the 31-chip complex Bernoulli chaotic sequences under different
channel conditions. The BER performance of the new system is compared with the system
based on the IKEDA complex chaotic code of length 31 chips (introduced in Kurian et al.
(2005)) as well as the 63-chip Gold sequence. The Binary Shift Phase Keying (BPSK)
modulation scheme is assumed. Since the proposed scheme uses same bandwidth as of the
63-chip Gold sequence based BPSK system, comparison between the 31 chip complex
Bernoulli spreading system and 63-chip Gold-sequence-based BPSK system is justified.
The BER curves versus SNR of three mentioned methods under AWGN channel are
presented in Figure 3. As can be seen, complex Bernoulli chaotic sequences outperform
Gold and IKEDA sequences in the same conditions.

10–1
IKEDA MAP
BERNOULLI MAP
GOLD CODES

10–2
BER

10–3

10–4

10–5
1 2 3 4 5 6 7 8 9
SNR

Figure 3. BER of synchronous DS-CDMA system under AWGN channel.


8 M.F. Sabahi and A. Dehghanfard

100
BERNOULLI MAP
GOLD CODE
IKEDA MAP

10–1
Bit error rate
Downloaded by [National Institute of Technology - Warangal] at 04:02 16 August 2014

10–2

10–3
0 2 4 6 8 10 12 14 16 18
Eb/No, dB

Figure 4. BER of synchronous DS-CDMA system under Rayleigh fading channel.

Figure 4 illustrates the BER curves under multipath Rayleigh fading channel. A
channel with three independent paths is considered. A tapped delay line model for channel
is assumed. The taps are zero mean complex Gaussian variables with variances equal to
0.7, 0.2 and 0.1. Other simulation conditions are assumed as previous. As can be seen, the
complex Bernoulli spreading sequence has better performance than the Gold codes and
slightly better BER than IKEDA sequences.
In Figure 5, the average BER versus the number of active users are shown. To
examine the interference of users, the channel is considered without noise. As be seen,
by using complex Bernoulli sequences as spreading codes, the system capacity is
improved.

100

10–1

10–2
BER

10–3

GOLD CODES
BERNOULLI MAP
IKEDA MAP
10–4
5 10 15 20 25 30 35
Number of users

Figure 5. Average BER of synchronous DS-CDMA system versus the number of active users.
International Journal of Electronics 9

5.2. Asynchronous system


The uplink of a wireless communication system is essentially asynchronous. Hence, the
performances of the proposed system have to be evaluated under asynchronous channel
conditions. In a multi-user asynchronous DS-CDMA system, the average interference
seen by any user can be computed with only the auto-correlation functions of correspond-
ing spreading codes (Pursley, 1997). So, the appropriate codes can be produced by careful
designing the auto-correlation function. As mentioned before, this idea has been used in
Tsuneda and Miyazaki (2010) to design an appropriate one-dimensional chaotic sequence.
Two-dimensional Bernoulli chaotic sequence, which is introduced in this paper, also has
the same properties of one-dimensional sequence (Tsuneda and Miyazaki, 2010). So it is
Downloaded by [National Institute of Technology - Warangal] at 04:02 16 August 2014

expected that by using the introduced method in Section 4, the sequences with desired
cross-correlation values can be selected. Utilising these sequences leads to lower BER and
increased capacity under asynchronous channel conditions.
Asynchronous system is simulated in presence of 10 active users. According to Figure 1,
transmitted signal for the kth user can be written as sk ðtÞ ¼ zk bk ðtÞcosðωc t þ ’k ðtÞ þ θk Þ.
The BPSK modulation is assumed and ’k ðtÞ represents the effect of complex spreading code
ck ðtÞ. Received signal can also be expressed as follows:

XK
rðtÞ ¼ s ðt
k¼1 k
 τ k Þ þ vðtÞ (14)

where τ k is the sequence delay and θk is transmitter oscillator phase for the kth user. V(t) is
additive Gaussian noise. The optimum receiver for the ith user employs matched filter on
si ðtÞ. Without loss of generality, we can assume that τ i ¼ 0 and θi ¼ 0 and other τ k and θk
(for kÞi) have uniform distribution in the intervals ½0; Tb  and ½0; 2π , respectively.
Similar to synchronous channel, AWGN and multipath Rayleigh fading channels (with
the same parameters) are considered for simulation. Three kinds of spreading sequence are
utilised under asynchronous channel conditions. Figure 6 demonstrates BER perfor-
mances in asynchronous DS-CDMA system. The improvement in system performance
using complex chaotic Bernoulli sequences can be seen.

100
GOLD CODES
BERNOULLI MAP
IKEDA MAP

10–1
Bit error rate

10–2

10–3
0 2 4 6 8 10 12 14
Eb/No, dB

Figure 6. BER in asynchronous DS-CDMA system under AWGN channel.


10 M.F. Sabahi and A. Dehghanfard

100
GOLD CODE
IKEDA MAP
BERNOULLI MAP

10–1
Bit error rate
Downloaded by [National Institute of Technology - Warangal] at 04:02 16 August 2014

10–2

10–3
0 2 4 6 8 10 12 14
Eb/No, dB

Figure 7. BER in asynchronous DS-CDMA system in Rayleigh fading channel.

In Figure 7, BER curves of asynchronous DS-CDMA system under Rayleigh fading


channel are shown. BER value in simulated curves is rather high and, as expected, an
irreducible amount of error is remained due to Inter Symbol Interference (ISI). Obviously,
channel coding can help to resolve both problems. In Figure 7 it can be seen that the
performance of the systems based on the complex Bernoulli sequence and the IKEDA
sequence in asynchronous channel is almost the same and superior to Gold sequence.

6. Conclusion
In this paper, two-dimensional Bernoulli chaotic sequences are introduced. They can be
used to generate complex spreading sequences for utilising in DS-CDMA system.
Amongst the complex sequences it is much easier to choose a proper set of sequences
for multi-user DS-CDMA systems. One-dimensional Bernoulli chaotic sequence (gener-
ated by applying an appropriate binary function to output of the Bernoulli map) has been
utilised before to design near-optimum spreading codes. The considerable properties of
Bernoulli chaotic sequence are due to the negative auto-correlation at lag one. Combining
of two one-dimensional Bernoulli sequences with different initial conditions, to form a
complex two-dimensional sequence, also has the mentioned properties. Employing the
new sequences can improve the performance of DS-CDMA systems compared with the
conventional spreading codes as well as similar complex spreading codes, such as the
codes generated by IKEDA map. The performance improvement is due to the following
reasons: the proper auto-correlation function and more freedom of choice of appropriate
codes among complex codes. Results, both in synchronous and asynchronous systems,
confirm decrease in BERs and increase in cellular capacity.

References
Cong, L., & Shaoqian, L. (2000). Chaotic spreading sequences with multiple access performance
better than random sequences. IEEE Transactions on Circuits and Systems: Fundamental
Theory and Applications, 47, 3.
International Journal of Electronics 11

Kohda, T., & Tsuneda, A. (1997). Statistics of chaotic binary sequences. IEEE Transactions on
Information Theory, 43(1), 104–112.
Kurian, A. P., Puthusserypady, S., & Htut, S. M. (2005). Performance enhancement of DS/CDMA
system using chaotic complex spreading sequence. IEEE Transactions on Wireless
Communications, 4, 3.
Mazzini, G., Setti, G., & Rovatti, R. (1997). Chaotic complex spreading sequences for asynchronous
DS-CDMA—Part I: System modelling and results. IEEE Transactions on Circuits and Systems
Part I, 44, 937–947.
Proakis, J., & Salehi, M. (2008). Digital communications (5th ed.). New York, NY: McGraw Hill.
Pursley, M. B. (1977). Performance evaluation for phase-coded spread-spectrum multiple-access
communication. Part I: System analysis. IEEE Transactions on Communications, 25, 795–799.
Rovatti, R., & Mazzini, G. (1998). Interference in DS-CDMA systems with exponentially vanishing
Downloaded by [National Institute of Technology - Warangal] at 04:02 16 August 2014

autocorrelations: Chaos-based spreading is optimal. Electronics Letters, 34(20), 1911–1913.


Rahnama, N., & Talebi, S. (2013). Performance comparison of chaotic spreading sequences
generated by two different classes of chaotic systems in a chaos-based direct sequence-code
division multiple access system. IET Communications, 7(10), 1024–1031.
Rovatti, R., Mazzini, G., & Setti, G. (2004). On the ultimate limits of chaos-based asynchronous
DS-CDMA-I: Basic definitions and results. IEEE Transactions on Circuits and Systems, 51,
1336–1347.
Stavroulakis, P. (2005). Chaos applications in telecommunications. Hoboken, NJ: CRC Press.
ISBN-13: 978-0849338328.
Tsuneda, A. (2005). Design of binary sequences with tunable exponential autocorrelations and run
statistics based on one-dimensional chaotic maps. IEEE Transactions on Circuits and systems-I,
52(2), 454–462.
Tsuneda, A, & Miyazaki, Y. (2009a). Performance evaluation of LFSR-based spreading sequences
with negative auto-correlation designed by chaos theory of modulo-2 added sequences.
European Conference on Circuit Theory and Design, ECCTD 2009, Antalya (pp. 141–144).
doi:10.1109/ECCTD.2009.5274956
Tsuneda, A., & Miyazaki, Y. (2009b). Performance evaluation of spreading sequences with negative
auto-correlation based on chaos theory and gold sequences. Fourth International Workshop on
Signal Design and Its Applications in Communications 2009, IWSDA '09, Fukuoka (pp. 169–
172). doi:10.1109/IWSDA.2009.5346422
Tsuneda, A., & Miyazaki, Y. (2010). Binary spreading sequences with negative auto-correlation
based on chaos theory and gold sequences for application to asynchronous DS/CDMA com-
munications. IEICE Transactions on Fundamentals of Electronics, Communications and
Computer Sciences, E93.A(11), 2307–2311.
Tsuneda, A., Yoshioka, D., & Hadate, T. (2004, August 30). Design of spreading sequences with
negative auto-correlations realizable by nonlinear feedback shift registers. IEEE Eighth
International Symposium on Spread Spectrum Techniques and Applications, ISSSTA 2004,
Sydney (pp. 330–334). doi:10.1109/ISSSTA.2004.1371716
Torrieri, D. (2011). Principles of spread-spectrum communication systems (2nd ed.). Berlin:
Springer. ISBN-13: 978-1441995940.

You might also like