Page 1

An Effective Hybrid GACSA -based Multi-user Detection for Ultra-WideBand

Communications Systems

Jyh-Horng Wen

Department of Electrical Engineering,

Tunghai University

No. 181, Sec. 3, Taichung Harbor Rd.,

Taichung, Taiwan, R.O.C.

e-mail :jhwen@thu.edu.tw

Ho-Lung Hung, Chien-Chi Chao

Department of Electrical Engineering,

Chienkuo Technology University,

Changhua, Taiwan.

e-mail: hlh@cc.ctu.edu.tw

Chia-Hsin Cheng

Department of Electrical Engineering,

National Formosa University

No. 64, Wunhua Rd., Huwie, Yunlin

632, Taiwan,

e-mail:chcheng@nfu.edu.tw

Abstract—In this paper, we investigate the performance

of various interference cancellation techniques in direct-

sequence ultra-wideband (DS-UWB) communication

systems. Multiple access interference (MAI) causes the

performance of the conventional single user detector in

DS-UWB systems to degrade. Due to high complexity of

the optimum multiuser detector, suboptimal multiuser

detectorswith less complexity

performance have received considerable attention. A

hybrid approach that employs a genetic algorithm (GA)

and chaos algorithm (CSA) for the MUD problem in

UWB communication systems is proposed. By taking

advantageof heuristic values and the

intelligence of GACSA, the proposed detector offers

almost the same bit error rate (BER) performance as the

full-search-based optimum multiuser detector does,

while greatly reducing the computational complexity.

The near–far resistance of the GACSA-based multiuser

detector is also examined. The good behavior of the

proposed approach is demonstrated by means of

comparisons in term of bit error rate (BER)

performance and implementation complexity with the

classical Rake receiver and different multiuser receivers

previously proposed in the literature on this subject.

and reasonable

collective

Keywords-Ultra-wideband, genetic algorithm, chaos

alhorithm, multiuser detection

I.

INTRODUCTION

Ultra- wideband (UWB) technology is currently being

investigated as a promising solution for short range, high-

capacitywireless communications

conventional radio systems, UWB systems have a number

of advantages that make them attractive for consumer

communication applications, such as low cost, low power

consumption, low complexity, and high data rate

transmission, etc [1-2]. To realize the multiple accesses

technique in UWB systems, two commonly used approaches

is time hopping (TH) and direct sequence (DS) techniques

[3-4]. When using DS technique, pseudo-random code is

applied to spread the data bit into multiple chips, just as in

systems.Unlike

conventional DS code division multiple assess (DS-CDMA)

systems, and the users are separated by independent spread.

In the DS-UWB communication system, multiple access

interference (MAI) is the main source of interference.

Additionally, it is well known that MAI limits DS-UWB

system capacity. Multiuser detection (MUD) is a powerful

technique to combat MAI and to improve the performance

of UWB systems. Considering the large complexity

involved in optimal multiuser detection (OMUD), which is

exponential in the number of active users, most of the

current work is centered around investigating suboptimal

approaches.

Some sub-optimal approaches, such as differential

receive and adaptive receivers were proposed [5-6]. These

techniques do not require channel estimation and allow

capturing a large amount of the received energy. To reduce

the complexity of this optimum

suboptimum multiuser have been developed in the past

several years [4], [5-12]. For examples, the minimum-mean-

squared error (MMSE) MUD has been described in [5],

while an interference cancellation (IC) based MUD has been

proposed in [6] [9]. The traditional receiver for such a UWB

system is a simple matched filter [1-2] and the performance

is degraded due to the MAI and ISI. Moreover, using

intelligent computation techniques seems to be a feasible

approach to achieve a bit error rate (BER) performance

close to that of the OMUD when reducing computational

complexity. Hybrids MUD have been proposed [7-8] as

have other distinct approaches [9]. While each has its own

merits and drawbacks, we will focus on a multistage

approach which performs parallel interference cancellation

(PIC) at each stage.

In resent years, there are many new intelligence MUD

techniques which utilize some genetic algorithms (GA) [10-

14], particle swarm optimization (PSO) [15], and neural

networks (NN) [16] have lower computational complexity.

In all kinds of techniques, the evolutionary computation

algorithm has proven to be an effective way to design the

sub-optimum multiuser detectors. In [9-10], it is shown that

GA based MUD approaches the single-user performance

bound at a lower complexity as compared with optimal

detector various

Digital Object Identifier: 10.4108/ICST.WICON2010.8624

http://dx.doi.org/10.4108/ICST.WICON2010.8624

Page 2

maximum likelihood (ML) optimum detector. GA can get

the optimum solution for multidimensional engineering

problems. Furthermore, GA algorithm has suffered from

some deficiencies. It is well known that premature

convergence degrades the performance of GA and reduces

the search ability [11]. In addition, a change in the genetic

population through generation results in the destruction of

previous knowledge of the problem [12].

The GA is a stochastic search algorithm whose

procedures are based on the Darwinian models of natural

selection and evolution [4]. Given some arbitrary initial

solutions, the GA will generate the better solution through a

series of genetic operations including selection, crossover,

and mutation. Furthermore, the GA searches the solution

space in parallel, that is, a set of possible solutions are

manipulated in the same generation, so multiple local

optimum can be reached simultaneously and thereby the

likelihood of finding the global optimum is increased. The

genetic algorithm based multiuser detector (GA-MUD) [11-

13] is one of the sub-optimum multiuser detectors that

evolves from the OMUD by replacing the exhaustive search

scheme with the GA [11], and has attracted much attention

in recent years. By means of GA’s powerful search ability,

the GA-MUD can attain sub-optimum performance with

less complexity compared to the OMUD [12]. One of the

proposed GA-MUDs uses the GA to provide a good initial

point for the successive stage of the multistage detectors. A

modified GA that adjusts the operations of the GA to

improve the BER performance is proposed in [13]. However,

the performance of the GA-MUD still can not approach a

sub-optimal, which is far from the single user bound, at high

system load. Thus, after a low-complexity GA detector, the

parallel interference cancellation (PIC) scheme [10] is used

to simultaneously subtract the interference from each user’s

received signal. However, at heavy system load, the

multistage conventional PIC (CPIC) approach suffers

performance degradation due to a poor cancellation, which

is brought about by the relatively high error rate of bit

decisions in the preceding stage [12]. Thus, the partial

cancellation contrarily is a better policy than the complete

cancellation [13]. Consequently, in this paper, we proposed

a low-complexity iterative MUD approach for DS-UWB

communication systems which can effectively alleviate the

harmful effects of MAI. The proposed method is based upon

the use of iterative processing techniques, which have

already been successfully applied to

communication system and, in particular, to the MUD case.

Using this approach, the CSA is embedded into the GA to

improve further the fitness of the population at each

generation. Such a hybridization of the GA the CSA reduces

its computation complexity by providing faster convergence.

The remainder of this correspondence is organized as

follows. Section II describes the signal model of the DS-

UWB system and Section III introduces the proposed

particle swarm optimization technique, and multiuser

detection based on the GA-CSA technique. Section IV,

much wireless

simulation results

performance of the proposed detector. Conclusion and

discussion are given in Section V.

areprovided to demonstrate the

II.SYETEM AND CHANNEL MODEL

A. Transmitter Model

In this section describes a simple model for a DS-UWB

communication system employing multiuser detector which

will be employed for the proposed of analysis in this paper.

We assume a K-users DS-UWB system over the UWB

indoor multipath fading channels, where each user employs

unique DS spreading code. The transmitted signal qk(t) for

the kth user is obtained by spreading a set of binary phase-

shift keying (BPSK) data symbol {bk[i]} onto a spreading

waveform sk(t), which is written as follows:

P

q tE b i s t

?

where Ek is the symbol energy of the kth user, P is the

packet size,

? ?

[ ]1

is the ith data symbol of the kth

user, and Tb is the symbol interval duration. The spreading

waveform sk(t) is also written as follows:

1

1

( )(

k k n

n

G

?

N

Gc

?

of the kth user, Nc is the chip numbers, Tc is the chip

interval duration, and

( ) w t

duration Tc= Tb / Nc.

1

( )[ ] (),

kkkkb

i

iT

??

?

(1)

kb i ? ?

,

0

),

c

N

?

c

s tc w tnT

?

??

(2)

where

2

k n

,

1

c

n

??

, k=1,2,…,K,

? ?

? ?

,

1

k n

c

is the nth chip

is the chip waveform of

B. Multipath Channel Model

In this paper, we use the IEEE 802.15.3a indoor channel

model, which is based on a modified Saleh-Valenzuela (S-V)

model where multipath components arrive in clusters, each

of which could contain several components namely rays

[17]. For the UWB indoor transmission environment, the

channel impulse response of UWB indoor channel model is

modeled as

k L

h tt

? ??

?

?

,,

1

( )()

k k lk l

l

??

?

??

,

1

(1),

k L

k lc

l

tlT

? ?

?

???

(3)

where Lk denotes the total number of propagation paths of

the kth user,

, k l

?

is the channel coefficient of the lth path of

?

is the multipath delay of the lth path of

the kth user. In this thesis, we suppose that the multipath

delay

, k l

?

is an integral multiple of Tc, L1 = L2= … = LK =

L, and the system is assumed to be synchronous. In system

performanceanalysis, the commonly adopted UWB

the kth user and

, k l

Digital Object Identifier: 10.4108/ICST.WICON2010.8624

http://dx.doi.org/10.4108/ICST.WICON2010.8624

Page 3

multipath channel models are those standard statistical

models established by IEEE 802,15.3a task group, i.e.,

CM1~CM4. These standard models are based on modified

S-V model, of which the clustered multipath components

obey a lognormal amplitude distribution and an equi-

probable polarity distribution.

C.

communication Systems

This section

conventional multiuser detector and optimum MUD for

UWB communication systems. When passing the signal

through the indoor environment, the obstacles in the

transmitted path will cause the multipath transmission.

Therefore, the total received signal can be formulated as

follows:

Review of Multiuser Detectors in DS-UWB

describes the construction of the

1

( )r t ( ) ( )( )

K

kk

k

q t h tn t

?

???

?

11

[ ] ( b i v t)( ), n t

KP

kkkb

ki

E iT

??

???

??

(4)

where ? is linear convolution, n(t) is zero-mean additive

white Gaussian noise and

( )

k

v t

as template signal of the kth user, which is a convolution

between the kth user’s spreading code and channel

coefficient.

( )( )

kk

s th t

??

is defined

D. Conventional Detector

The template signal vk(t) that is transmitted over a

channel is corrupted by channel noise. Hence, the function

of the receiver must detect the template signal vk(t) for each

user. According to [5], we note that a filter which is

matched to a template signal vk(t) of duration (Nc+L-1)Tc is

characterized by an impulse response. The channel response

for kth user can be written as follows:

*

,( )( ).

opt kk

htvt

??

(5)

So, the output of the filter which is matched to a template

signal vk(t) can be written as follows:

( ) ( )( )

kopt k

y tr tht

??

( )()

k

r tvt

???

KP

E b i v tiT

??

and the discrete-time impulse response sampling at t = iTb is

represented as follows:

[ ]().

kkb

y iy iT

?

Then the discrete-time received signal after sampling (iTb) is

written as follows:

KP

y iEbj v ij

??

,

*

*

k

*

k

11

[ ]()() ( )(),

mmmb

mi

vtn tvt

??? ? ???

??

(6)

(7)

*

k

*

k

11

[ ] [ ] [] [ ]

? ?

[ ]n i [ ]

?

kmmm

mj

vivi

????

??

,

11

[ ]j R[ , ]i j[ ]

KP

mmm kk

mj

Ebn i

?

??

??

??

,,

1

desired signal

ISI

[ ][ , ]

i i

[ ] [ , ]

i j

P

kk k kkkk k

j

j i

?

E b i R

??? ??? ?

Eb i R

?

??

?

??? ? ???? ?

,

11

MAI

[ ]j R [ , ]i j[ ],

KP

mmm kk

m

m k

?

???? ? ????? ?

j

Eb n i

?

??

??

??

(8)

where

*

k

,[ , ]

m k

R

[ ]

k

n i

?

[][ ],

?

m

i j

?

v i

?

jvi

??

?

?

*

k

[ ]n i[ ].vi

Hence, the signal that received by a CD can be detected:

ˆ

[ ] sgn

k

bi

?

where a CD structure is shown in Fig. 1.

??

[ ] ,

y i

CD

k

(9)

E. Maximum Likelihood Detector

The optimum MUD performs maximum-likelihood

sequence detection jointly across all users’ sequences [5-6].

According to [4-6], the optimal multiuser detector can be

achieved by maximum a posteriori (MAP) estimation.

Because the probability of bk[i] = +1 is equal to the that of

bk[i] = -1, the maximum likelihood (ML) estimation can be

generalized by the MAP estimation. As a result, the optimal

multiuser detector that fulfils ML sequence estimation [5]

gives the best performance. However, its computational

complexity which grows exponentially with the number of

the users forbids application in real system. The search for b

is a combinatorial optimization problem detailed in [5]

whose complexity grows exponentially with K. Though

impractical given large K, its performance establishes a

benchmark for multiuser design. In UWB systems, the

number of zero cross-correlation entries in R can very be

large and can help decouple the ML sequence detection

problem into much smaller, independent ML sequence

detection problems.

ML

ˆ

arg max2

KP

? ? ?

?

b

??

[ 1, 1]

.

TT

??

?

??

b b Ay b ARA b

(10)

III.GENETIC ALGORITHM AND CHAOS ALGORITHM

1.Solution Representation

For the subsequent genetic operations, the trial solution to

the addressed problem must be encoded into the string form

first. An encoded solution is referred to as a chromosome

and its elements are referred to as the genes. The multiuser

detection can be regarded as an optimization problem that

finds the most likely combination of the binary transmitted

bits

pOMD,

. Since the configuration of the trial solution

ˆ

,,

ˆ

,

ˆ

[

,2,1,pKpp

bbb

?

is already an antipodal binary string of

ˆb

]

Digital Object Identifier: 10.4108/ICST.WICON2010.8624

http://dx.doi.org/10.4108/ICST.WICON2010.8624

Page 4

length K, the encoding process is unnecessary.

2.Initialization

For each time the GA being carried out, a chromosome

set with Pc members named the chromosome population is

created in order to produce the better solutions by applying

the genetic operations, where Pc is known as the population

size. Generally, the larger the population size, the faster the

convergencerate but the higher the computational

complexity. In this paper, the seed chromosome in the initial

population is created by the Rake receicer. The Rake

receiver first makes

]

ˆ

,,

ˆ

,

ˆ

[

ˆ

2,1 K,pp ,pp

bbb

?

?

b

according

equivalent baseband signal r(i) in (10) and then passes its

output to the GA for further processes.

a rough

to

decision

received the

3.

In the cost evaluation process the GA employs the

problem-dependent objective function to evaluate the cost or

fitness of a solution, which represents how closely the

chromosome fits the addressed problem. A high fitness or

low cost reflects the excellence of the chromosome. The

objective of our system is to find the bˆ that has the

minimum cost. Consequently, we define the cost function of

a chromosome for the ith bit duration as

.

[ 1, 1]

? ? ?

?

b

Cost Evaluation

??

ML

ˆ

b

arg max2.

KP

TT

??

?

??

b Ay bARA b

(11)

4.Selection

Selection is the operation that chooses the chromosomes

from the parent population to constitute a selected

population for crossover. Since the parent chromosomes

with advantageous genes are more likely to produce the

better offspring chromosomes, only certain parent

chromosomes will have the chance to produce the offspring

chromosomes. The selected population consists of Pc

chromosomes chosen from the parent population with

probabilities that are inversely proportional to their costs by

using the roulette wheel selection scheme [11]. Then, the

selection rate of the kth chromosome is

?

?

j

?

?

?

c P

j

k

S,k

C

C

P

1

1

1

)

ˆ(

b

)

ˆ(

b

, (12)

where

)

ˆ(

j

C b

is the cost of the jth chromosome.

5.Crossover

The crossover is the operation that exchanges the genes

of two chromosomes to generate a new pair of offspring

chromosomes. Through

chromosomes are expected to be superior to their producer

because they inherit the merits of both parents. Three

commonly used crossover schemes are one-point, multi-

point, and uniform crossover [4]. The simplest crossover

crossover, the offspring

scheme, the one-point crossover, is utilized in this study.

The probability that the crossover operation is applied to the

selected chromosomes, which is known as the crossover rate,

is set to 1.

6.Offspring Mutation

The mutation is the operation that randomly alters the

genes of the crossover results for increasing the diversity of

genes. Sometimes the members of the selected population

are not diverse enough to find any better solution even if the

global optimum is not reached yet. This situation is known

as the premature convergence and may happen when one of

the parents has a relatively low cost compared to the others',

in which this outstanding chromosome predominates the

selected population, resulting in a considerable amount of

identical offspring chromosomes. Therefore, the mutation

operation is introduced to prevent the premature

convergence. The GA with a high mutation rate Pm is more

likely to escape from the local optimum but has the slower

convergence. Conventionally, the mutation rate Pm is

usually smaller than the crossover rate. In this paper, the Pm

is set by 0.1 to compromise between the optimization and

convergence.

7.Elitist Replacement

The mutation and crossover are the operations that

modify genes within chromosomes. To avoid destroying the

good solutions during the mutation and crossover, we

replace a small portion of offspring population with the

good chromosomes in the parent population. This is called

the elitist replacement. In

replacement is implemented by selecting the best Pc

chromosomes from the combination of the parent and

offspring population [11].

this paper, the elitist

8.Iteration

Unlike the other sub-optimum multiuser detectors, the

GA is an iteration scheme. We can repeat the procedures

mentioned in parts 3~7 to refine our solution. In GA, each

iteration is called a generation. It is known that when the

pre-defined generation number Gn is reached the iteration

will be terminated and the best chromosome in the last

generation will be taken as the result of the detection. With

the aid of the elitist replacement, the best chromosome of

the offspring population is never worse than that of the

parent population. This ensures the discovery of better

solution after certain generations.

The standard adaptive genetic algorithm (AGA) is

proposed by Srinvas [11]. its main idea is that when the

fitness values of population tend to convergence, the

probability of the occurrence of the genetic operators will be

increased so as to avoid the premature convergence, where

when the fitness values of population tend to divergence, the

probability of the occurrence of the genetic operator will be

decreased so as to converge to the optimum. Figure 1 the flow

chart of the GA-based detection scheme for DS-UWB systems.

Digital Object Identifier: 10.4108/ICST.WICON2010.8624

http://dx.doi.org/10.4108/ICST.WICON2010.8624

Page 5

The probabilities of crossover and mutation are defined as

follows:

??

?

??

?

??

?

??

?

?

?

?

?

?

?

?

?

avge

?

f

avge

avge

c

ffe

ff

ff

ff

e

p

,

,

3

max

max

1

( 13 )

??

??

??

??

?

?

?

?

avge

avge

avge

m

ffe

ff

f

ff

e

p

,

,

4

max

max

2

(14 )

where

average fitness value of the population, respectively. f ? is

the larger one of the fitness values of two individuals to be

crossed, and f it is the fitness value of the individual to be

mutated.

321

,,eee

and

the range [0,1].

The basic idea of AGA is that , during the genetic search

process, when the individual fitness value is less than the

average fitness values of the population, larger rates of

crossover and mutation operators should be adopted, and

vice versa. This scheme can prevent the genetic search

process from prematurely converging to local optimal

solutions by regulating the balance between exploration and

exploitation in the solution space. The procedure of AGA

can be depicted as follows:

Step 1: Initialize the population and various parameters.

Step 2: Calculate the fitness values including individual

fitness values and average fitness values of the population.

Step 3: Compute the crossover and mutation rates

according to formulas (13 ) and (14 ).

Step 4: execute genetic operation including selection,

crossover and mutation.

Step 5: when the current generations is less than the

maximal generation, turn to step 2; otherwise, terminate the

program and return the optimal searching. Adaptive Genetic

algorithm combined with chaos searching [17] (GACSA):

The proposed approach GACSA is developed on the

basis of the standard GA by introducing chaos searching and

the other set of crossover and mutation rates so as to guide

to the whole population to evolve in the solution space. In

GACSA, the chaos searching algorithm is used to local

exploration for obtaining the local optimum while the AGA

with two sets of crossover and mutation rates is responsible

for global exploitation. In order to fulfill chaos searching,

here, a premature decision identifier ? is employed: assume

max

f

and

avge

f

are the maximal fitness value and

4e are constants predetermined in

that

if

is the average fitness values at generation i , which

1

, where

is computed by ?

?

P

n

i

nf

P

1

i

nf

represents the fitness

value of the nth individual in the ith generation, P is the

population size. At the same time, suppose the best

individual’s fitness is

ifmax, f ? is the average of all the

individuals whose fitness values are greater than

ff

?

??

max

?

crossover and mutation operators have a significant effect

on the convergence of GA during the search process. Thus,

the other set of crossover and mutation rates adopted in this

paper is devised with ? as follows:

?

??

exp(1

e

if . And

then let

i

. It is known that the rates of

??

?

??

?

??

???

??

?

??

???

)exp(1

1

03 . 003. 0

)

1

3 . 01

2

1

?

?

e

p

p

mut

c

(15 )

where

The detailed procedure of GACSA is described as follows:

Step 1: Initialize the population and various parameters.

Step 2: Calculate the fitness values of each individual and

the premature decision identifier? .

Step 3: When ? is greater or equal to

generation (GEN) is greater than maximal generation (M),

M/2, carry out the following steps for chaos searching.

1). Let d=1.

2). Execute chaos search from step 2 to 5 as shown in

procedure chaos optimization algorithm, and the rest (D-1)

individuals keep invariant.

3).

1

?? dd

. If d reaches D, terminate the chaos search;

otherwise, turn to 2) to optimize the next variable. When ?

? and GEN is less or equal to M/2,

calculate the rates of crossover and mutation operators by

formulas 4) and 5), and carry out the corresponding genetic

operations. Otherwise, turn to Step 4.

Step 4: Calculate the crossover and mutation rates according

to equations (13 ) and (14).

Step 5: Execute genetic operations including selection,

crossover and mutation.

Step 6: When the GEN is smaller than the maximal

generation (M), turn to Step 2; otherwise, terminate the

program and return the optimal solutions.

1e and

2e are the positive numbers predetermined.

*

? and current

is greater or equal to

*

IV.

SIMULATION RESULTS

In this section, the simulations of multiuser transmission for

DS-UWB radio systems under the modified S-V channel

that GACSA based MUD algorithm is adopted are shown in

Figs. 2-5. The UWB CM 1-4 which are discussed in paper

indicate the different transmission distance for indoor

environment, and all Rake receivers is adopted for CM 1-4.

We assume that the packet size is 4 bits and the number of

users is 10 on DS-UWB systems

Digital Object Identifier: 10.4108/ICST.WICON2010.8624

http://dx.doi.org/10.4108/ICST.WICON2010.8624

Page 6

Generate Initial Population

p

bˆ

Evaluate Cost

Start with g = 0

Chromosomes Selection

No

Yes

Crossover to Generate New Chromosomes

g = g+1

Mutate the New Chromosomes

End

g > Gn

Fig. 1. The flow chart of the GA-based detection scheme for DS-

UWB systems.

. Fig. 2 shows the performance of GACSA based MUD,

the optimum detector, and several suboptimum detectors,

for K=10. The performance of HNN detector for DS-UWB

systems is better with the increase of iteration. However, the

performance improvement is no longer obvious when it

achieves about 100 iterations. That is because the HNN

detector has many local minimum, but it is unable to

determine which minimum is the global minimum. Since

the information of R is known for HNN detector, the

performance of HNN detector always is better than CD

when it achieves about 50 iterations. Unfortunately, the

performance of HNN detector is poorer than original GA,

original PSO, GACSA and OMUD detectors for DS-UWB

in CM 1. On the other hand, as for the BER of the original

GA-based MUD, an error floor is observed foe the results

show in the figure. This is because of the limitations of the

GA associated with the particular set of individual and

generation values. In contrast, the BER of the proposed

GACSA based MUD demonstrated a perfect approach

similarto that of OMUD

complexity compared with that of the original PSO- and

GA-based MUD.

Further, in Figs.3-5 the performance of GACSA based

detector is depicted for CM 2, CM 3 and CM 4, respectively.

The performance of HNN detector is approximated other

suboptimum detectors at SNR=0-6dB for other DS-UWB

channel models. But, its performance is worse at SNR=8-

12dB for others. The neuron output of HNN detector with

sign activation is either +1 or -1, but the neuron output of

HNNdetector is distributed form -1 to +1. As can be

observed in Figs. 2-4, the performance of our algorithm is

better than that of all existing suboptimum schemes with

same level of complexity. Moreover, the performance of

GACSA detector is approximated OMUD at SNR=6-12dB

for other UWB channel models.

with less computational

V.

CONCLUSION

To reduce computational complexity of the optimal

multi-user detector, a novel hybrid algorithm that employs

GA and CSA is present. In this paper, we proposed a new

suboptimum multiuser detector for DS-UWB systems,

which utilizes a hybrid algorithm to decide on the

transmitted bits. Using this approach, the CSA is embedded

into GA to improve further the fitness of the population at

each generation. Such a hybridization of the GA with the

CSA based MUD reduces its computational complexity by

providing faster convergence. The complexity of the

proposed is approximately

(O KS

that its performance is significantly better and more robust

comparedto other examined

Simulations results are provide to show that the proposed

detector has significant performance improvement over the

detectors based on HNN, PSO and CD in term of MAI.

())

Tpi

PP

?

, and it is evident

suboptimum schemes.

Fig.. 2 The simulation of BER for DS-UWB systems that employs GACSA,

CD, GA, PSO, ML and HNN detectors with UWB CM 1when K=10

Fig. 3 The simulation of BER for DS-UWB systems that employs GACSA,

CD, GA, PSO, ML and HNN detectors with UWB CM 2

Digital Object Identifier: 10.4108/ICST.WICON2010.8624

http://dx.doi.org/10.4108/ICST.WICON2010.8624

Page 7

Fig. 4 The simulation of BER for DS-UWB systems that employs GACSA,

CD, GA, PSO, ML and HNN detectors with UWB CM 3

Fig.5 The simulation of BER for DS-UWB systems that employs GACSA,

CD, GA, PSO,ML and HNN detectors with UWB CM 4

REFERENCES

[1].

Win M Z, Scholtz RA. Impulse radio: How it works. IEEE

Commun. Lett; vol.2, pp. 36-38, 1998.

Linging Yang, Georgios B. Giannakis. Ultra-Wideband

communications an idea whose time has come. IEEE Signal

Processing Magazine ; pp.26-54, 2004.

Ismail Guvenc, Huseyin Arslan. A review on multiple

access interference cancellation and avoidance for IR-UWB.

ELSEVIER Singal Processing ;vol.87, pp.623-563, 2007.

[2].

[3].

[4].

Verdú S. Multiuser Detection. Cambridge, U.K.: Cambridge

Univ. Press, 1998.

Li Q, Rusch L. A. Multiuser detection for DS-CDMA UWB

in the home environment, IEEE J. Select Areas Commun.

vol.20, pp.1701-1711. Dece.2002.

Nejib Boubaker and Khaled Ben Letaief, Combined

multiuser successive interference cancellation and partial

rake reception for Ultra-wideband wireless communication,

2004, pp. 1209-1212.

J. R. Foerster, The performance of a direct-sequence spread

ultra wide-band system in the preseence of multipath,

narrowband interference, and multiuser interference, IEEE

Conf. Ultra wideband Syst. Technologies, pp. 87-91, May

2002.

Yoon YC, Kohno R. Optimum multiuser detection in Ultra

wide-band multiple-access communication systems, In Proc

IEEE International confer on Commun. New York City,

NY, pp.812-816, 2002.

Lili Lin and Anding Wang, New hybrid multiuser receiver

for DS-UWB system, IEEE Wireless Communications,

Networking and Mobile Computing, WiCOM , pp.1-5, Oct.

2008.

Tan-Hsu Tan, Yung-Fa Huang, Chung-Weng Lin, and Ray-

Hsiang Fu, Performance improvement of

detection using a genetic algorithm in DS-CDMA UWB

systems over an extreme NLOS multipath channel. In Proc.

of IEEE SMC 2006, vol. 3, pp. 1945-1950, Oct. 2006.

D. B. Fogle, Evolutionary Computation: Toward a New

Philosophy of Machine Intelligence, 2nd ed. Piscatewy, NJ:

IEEE Press, 2000.

C. Ergun and K. Hacioglu, “Multiuser detection using a

genetic algorithm in CDMA communication systems,” IEEE

Trans. Commun., vol. 48, no. 8, pp. 1374-1383, Aug. 2000.

K. Yen and L. Hanzo, “Genetic-algorithm-assisted multiuser

detection in asynchronous CDMA communications,” IEEE

Trans. Veh. Technol., vol. 53, Issue 5, pp. 1413 - 1422, Sept.

2004.

S. Abedi and R. Tafazolli, “Genetically modified multiuser

detection for code division multiple access systems,” IEEE J.

Select. Areas in Commun., vol. 20, pp. 463-473, Feb. 2002.

H.-L. Hung, and J.-H. Wen, An Adaptive Multistage

Multiuser Detector for MC-CDMA

Systems Using Evolutionary Computation Technique,” to

appear in Springer: Wireless Personal Communications,

2009.

Sitao Wu and Tommy W, S. Chow, Self-Organizing and

self-evolving neurons: a

optimization, IEEE Trans. on Neural Networks, vol. 18, pp.

385-396, March 2007.

Silva C. P. Survey of Chaos and its applications. Proceeding

of the 1996 IEEE MTT-S Internation microwave

symposium digest, San Francisco, USA, pp. 1871-1874.

Saleh AA, Valenzuela RA. A statical model for indoor

multipath propagation. IEEE J. Select. Areas Commun.;

vol.5, pp. 128-137, 1987.

[5].

[6].

[7].

[8].

[9].

[10].

multiuser

[11].

[12].

[13].

[14].

[15].

Communication

[16].

new neural network for

[17].

[18].

Digital Object Identifier: 10.4108/ICST.WICON2010.8624

http://dx.doi.org/10.4108/ICST.WICON2010.8624