Al-Qadisiya Journal For Engineering Sciences

Vol. 4 No. 3

Year 2011

COMPUTER MODELING OF SMART ANTENNA SYSTEM

Muhammed salah sadiq
Asst. Lec.
Department of technical electrical power
Technical college /al-musayab

ABSTRACT
This research studied the adaptive smart antenna system using in cellular phone applications using
MATLAB simulation.
In this work, the proposed novel least mean square algorithm (No-LMS) build and studied the
results as comparative work with results of two another algorithms called Standard LMS algorithm
(S-LMS), and normalized LMS algorithm (N-LMS). The computer simulation work results based
No-LMS algorithm have better performance refer to obtain the optimum convergence factor (CF)
and shows that robustness smart system, good tracking capability, and high adaptation accuracy
than the other algorithms.
The smart antenna system that is based on LMS-algorithms for all types (S-LMS),(N-LMS),and
(No-LMS) show that this system is affected by the convergence factor, the computer simulation
results shows the minimum mean square error (MSE = 0.007) is obtained for the value of
convergence factor equal to (0.2). also the smart system is affected by the number of the antennas
are used in smart array and number of samples interval, at least three antennas gives the poor
steering of array with large MSE value (0.15).
The test of the three algorithms for different values of signal to noise ratio (SNR) show that the
(No-LMS) algorithm gives the minimum MSE values compared with the two those obtain from
other algorithms.
KEYWORDS: smart antenna system, computer modeling, noise ratio.

(N-LMS) (S-LMS)

(No-LMS)

(CF)

.
(No-LMS)

(No-LMS)
(LMS)

.
(CF=0.2 )

(MSE = 0.007)

.(.)

Al-Qadisiya Journal For Engineering Sciences

Vol. 4 No. 3

Year 2011

(No-LMS)
.
1. INTRODUCTION
A smart antenna system combines multiple antenna elements with a signal processing capability to
optimize its radiation and/or reception pattern automatically in response to the signal environment
(Trees2002). Over the last few years the demand for wireless services has risen dramatically. This
fact introduces a major technological challenge to the design engineer: that is to increase the overall
performance and efficiency of the wireless system with an increased number of users under the
constraints of spectrum efficiency, power usage and cost. Most of the research on this topic, until
very recently, has been largely focused on the development of modulation and coding techniques as
well as communication protocols, very little attention has been paid to the overall transceiver
structure and antenna technology. Recently developed smart antenna technology may be the
solution to satisfying the requirements of next generation wireless networks (Chris, et.al, 2003).
Smart antenna for mobile communication has received enormous interests worldwide in recent
years. In the last decade, wireless cellular communication has experienced rapid growth in the
demand for provision of new wireless multimedia services such as internet access, multimedia data
transfer and video conferencing. Smart antennas involve processing of signal induced on an array of
antennas. They have application in the areas of radar, sonar, medical imaging location based
application and cellular phone applications (Shaukat 2009).
This paper represents the study of performance and design of adaptive smart antenna algorithm used
in cellular phone applications. This algorithm is studied for different levels of intelligence. The
performance of novel least mean square (No-LMS) algorithm with standard-LMS algorithm and
normalized LMS (N-LMS) algorithm has been compared.
Smart antennas shown in Figure (1) refer to a group of antenna technologies that increase the
system capacity by reducing the co-channel interference and increase the quality by reducing the
fading effects (Trees, 2002).
A smart antenna array containing M identical elements can steer a directional beam to maximize the
signal from desired users, signals of interest (SOI), while nullifying the signals from other
directions, signals not of interest (SNOI) (Chris2003).
There are many adaptive algorithms that can be used to adjust the weight vector, the beam former
must be implemented subject to a number of contradictory demands, regarding to the best choice of
the algorithm.
The analog-to-digital converters (ADCs) in such systems will be located as close to the antennas as
possible in order to achieve almost complete digital processing. In order to realize this, ADCs
capable of digitizing a high-frequency wideband signal at very high sampling rates will be required
along with wideband or multi-band antennas and RF analog devices. However, direct analog-todigital conversion at over sampling rates of very high RF or IF signals, typically ranging between
hundreds of MHz to several GHz, may not yet be practical because the reasonably price ADCs and
sufficiently high-speed digital devices, such as current signal processors and buffer memories,
cannot be used. The under sampling technique is always useful by performing frequency down
conversions and quantization at the same time (Minseok, 2004).
The most appropriate criterions include:(1) Computational complexity, defined as the number of snapshots required to converge to the
optimum solution,
(2) Robustness, which is an ability of the algorithm to behave satisfactorily under finite word
precision numerical operation, and
(3) Implementation issue. One important class of beamforming algorithms are the non blind
algorithms in which training signal is used to adjust the array weight vector (Shaukat, 2009).

Al-Qadisiya Journal For Engineering Sciences

Vol. 4 No. 3

Year 2011

2- TYPES OF SMART ANTENNAS (RAPPAPORT 98)

There are basically two approaches to implement antennas that dynamically change their pattern to
mitigate interference and multi path affects while increasing coverage and range. They are: Switched beam
Adaptive Arrays
The Switched beam approach is simpler compared to the fully adaptive approach. It provides a
considerable increase in network capacity when compared to traditional omni directional antenna
systems or sector-based systems. In this approach, an antenna array generates overlapping beams
that cover the surrounding area as shown in Figure (2a).
When an incoming signal is detected, the base station determines the beam that is best aligned in the
signal-of-interest direction and then switches to that beam to communicate with the user. The
Adaptive array system is the smarter of the two approaches. This system tracks the mobile user
continuously by steering the main beam towards the user and at the same time forming nulls in the
directions of the interfering signal as shown in Figure (2b). Like switched beam systems, they also
incorporate arrays. Typically, the received signal from each of the spatially distributed antenna
elements is multiplied by a weight. The weights are complex in nature and adjust the amplitude and
phase. These signals are combined to yield the array output. These complex weights are computed
by a complicated adaptive algorithm, which is pre-programmed into the digital signal-processing
unit that manages the signal radiated by the base station.
3. BENEFITS OF SMART ANTENNA TECHNOLOGY (AHMED2005):Smart antennas have several advantages over other antenna systems. For example, smart antennas
are able to increase the data transfer rate of a wireless signal as well as reduce the number of errors
or obstructed pieces of data. Smart antennas are also able to calculate the direction of arrival of a
wireless signal and effectively alert the user to where the signal is the strongest. Smart antennas are
easy to use and depend on plug-and-play technology. Many advantages that realize by smart system
are:3.1 REDUCTION IN CO-CHANNEL INTERFERENCE
Smart antennas have a property of spatial filtering to focus radiated energy in the form of narrow
beams only in the direction of the desired mobile user and no other direction.
3.2 RANGE IMPROVEMENT
Since smart antennas employs collection of individual elements in the form of an array they give
rise to narrow beam with increased gain when compared to conventional antennas using the same
power. The increase in gain leads to increase in range and the coverage of the system. Therefore
fewer base stations are required to cover a given area.
3.3 INCREASES IN CAPACITY
Smart antennas allow more users to use the same frequency spectrum at the same time bringing
about tremendous increase in capacity.
3.4 MITIGATION OF MULTI PATH EFFECTS
Smart antennas can either reject multi path components as interference, thus mitigating its effects in
terms of fading or it can use the multi path components and add them constructively to enhance
system performance.
3.5 COMPATIBILITY
Smart antenna technology can be applied to various multiple access techniques such as TDMA,
FDMA, and CDMA. It is compatible with almost any modulation method and bandwidth or
frequency band.

Al-Qadisiya Journal For Engineering Sciences

Vol. 4 No. 3

Year 2011

3.6 SIGNAL GAIN

Inputs from multiple antennas are combined to optimize available power required to establish given
level of coverage.
4. THE LMS AND N-LMS ALGORITHMS:4.1 LMS-ALGORITHM (HYKIN 96)
The algorithm uses a gradient descent to estimate a time varying signal. The gradient descent
method finds a minimum, if it exists, by taking steps in the direction negative of the gradient. It
does so by adjusting the filter coefficients so as to minimize the error.
The gradient is the Del operator (partial derivative) and is applied to find the divergence of a
function, which is the error with respect to the nth coefficient in this case. The LMS algorithm
approaches the minimum of a function to minimize error by taking the negative gradient of the
function. LMS algorithm can be implemented as shown in Figure (3)
The desired signal d (n) is tracked by adjusting the filter coefficients c (n). The input reference
signal x (n) is a known signal that is fed to the FIR filter. The difference between d (n) and y (n) is
the error e (n). The error e (n) is then fed to the LMS algorithm to compute the filter coefficients c
(n+1) to iteratively minimize the error.
4.2 N-LMS-ALGORITHM (RALLAPALL2007)
Most parameters of the NLMS algorithm are the same as the LMS algorithm, except that the step
size or the coverage factor is now bounded between 0 and 2. The normalization term, makes the
convergence rate independent of signal power. The normalized least mean square algorithm
(N-LMS) is an extension of the LMS algorithm, which bypasses the step size issue by selecting a
different step size value, , for each iteration of the algorithm. This step size is proportional to the
inverse of the total expected energy of the instantaneous values of the input vector. The N-LMS
algorithm shows far greater stability with unknown signals. This combined with good convergence
speed and relative computational simplicity makes the NLMS algorithm ideal for the real time
adaptive echo cancellation system. As the NLMS is an extension of the standard LMS algorithm,
the N-LMS algorithms practical implementation is very similar to that of the LMS algorithm.
5. ADAPTIVE SMART ANTENNA MODEL (ASA-MODEL)
Figure (4) demonstrates the adaptive smart antenna model; the ASA-model is proposed and
discussed as comparative work between (No-LMS) algorithm with S-LMS algorithm and (N-LMS)
algorithms by use the Matlab simulation program.
The system model shown can be used to test the different algorithms performance and scope the
novel and optimal one in different environments.
Where:u (r) : represents the input signal from uniform linear array.
m (r) : the output signal from the unknown channel.
I (r): the interference signal.
y (r): the received signal is obtained by summing the interference signal with m (r).
The signal y (r) is applied to the adaptive equalizer to reduce the noise that is generated by the
effect of the channel. The output of the equalizer gives the signal x (r). This signal is compared with
the original transmitted signal u (r) to obtain the error signal e (r).
Let the unknown channel is represented mathematically as a function (1/Jc), and the adaptive
equalizer as (E). In the case of absence the interference effect (ideal channel), that mean the output
of equalizer x (r) is the same input signal u(r) and we can obtain (Shannon8):Jc = E
That produces

(1)

Al-Qadisiya Journal For Engineering Sciences

Vol. 4 No. 3

Year 2011

E
(2)
1
Jc
From figure(4) the unknown channel is assumed the linear channel with response same as the
response of the finite impulse response filter with time (t):- adaptive

J c [ j c 0 , j c1 , j c 2 ,........... j cn 1 ]

(3)

The parameters of the equalizer can written as:E (r ) [e0 , e1 , e 2 ,........., e n 1 ]

(4)

For the nth LMS-filter parameters:- Ei (0) 0, fori 0,1,2,3,................n 1

The output of unknown channel:M (r ) [m(r 1), m(r 2), m(r 3),..........., m(r n 1)]

(5)

The output is calculated as:M (r )

1
(u ( r ) J c ( 2 : n ) M T ( r )
J c (1)

(6)

By adding the interference signal the received signal become and can show in figure (5):y ( r ) m( r ) I ( r )

(7)

Y (r ) [ y (r ), y (r 1), y (r 2),.........., y (r n 1)]

(8)

The equation (8) represents the received signal in matrix form, and the output of the equalizer is:-

The resultant error signal:-

x(r ) E (r 1)Y T (r )

(9)

e( r ) u ( r ) x ( r )

(10)

If the error signal is equal to the interference signal that refer to the adaptive smart antenna system
based the LMS-algorithms is estimated the unknown channel successfully and the standard
LMS-algorithm can represented:E (r ) E (r 1) Y (r ).e(r )

(11)

where () is the convergence factor (Cowan85).

6. COMPUTER SIMULATION TEST:6.1 CHOICE OF CONVERGENCE FACTOR
The convergence factor () controls how far we move along the error function surface at each update step.
Convergence factor certainly has to be chosen > 0 (otherwise we would move the coefficient
vector in a direction towards larger squared error). Also this parameter must not become too large.
Furthermore, too large a convergence factor causes the LMS algorithm to be instable, i.e., the
coefficients do not converge to fixed values but oscillate. Closer analysis (Hykin96) reveals, that the
upper bound for for stable behavior of the LMS algorithm depends on the largest Eigen value

Al-Qadisiya Journal For Engineering Sciences

Vol. 4 No. 3

Year 2011

(max) of the tap-input auto-correlation matrix and thus on the input signal. For stable adaptation
behavior the coverage factor has to be:-

max

(12)

The convergence time of the LMS algorithm depends on the convergence factor (). If is small,
then it may take a long convergence time and this may defeat the purpose of using an LMS filter.
However if is too large, the algorithm may never converge (Amrita2010).
6.2 NOVEL ALGORITHM
There is proportion between the stability of the adaptive algorithm and the convergence ratio for the
smart system. If the -value is large that produces the fast convergence ratio but this fast
convergence ratio led to low stability and low accuracy for smart system. In other hand, the low
values give good stability the accuracy.
The No-LMS algorithm is suggested to give the suitable convergence factor, this factor is
dependent upon the equation:-

E[(u (r )) 2 ]
.
2.E[(e(r )) 2 ]

(13)

Where:(u(r))2: input signal power.

(e(r))2: error signal power.
E[ ]: the expected value.
: the standard deviation. The graph of the standard deviation shown in Figure (7)

A
2 10 0.1SNR

(14)

Equation (14) graph can be shown in figure (6)

Where:A: the input signal amplitude.
SNR: the signal to noise ratio.
Now the error signal equation can be written:
e(r ) w H (r ).u (r )

(15)

Where wH(r) the estimated of the tap weight of LMS-filter.

7. RESULTS AND DISCUSSION
The computer simulation test using matlab simulation program for ASA-model based three
algorithms called S-LMS, N-LMS, and No-LMS show that:From the Figure (7) the No-LMS algorithm has the optimum performance compared with the other
two algorithms (S-LMS, and N-LMS) for small values of the Mean Square Error (MSE) and the
sample interval, whereas the two other algorithms need the large values of mean square error (MSE)
and the sample interval at the same value of convergence factor (CF =0.2) to be stable.
By varying the value of the convergence factor from the value 0.04 and below, the MSE values is
increased and that produces unstable system as shown In Figure (8), but from this values the NoLMS algorithm gives the good system stability but for the large numbers of sample interval, the
N-LMS algorithm produces low stability for large number of sample interval, and the S-LMS
algorithm fail.

Al-Qadisiya Journal For Engineering Sciences

Vol. 4 No. 3

Year 2011

The relationship between the mean square error and the number of the convergence factor for
increase the values of CF from 0.26 and above, the system tends to be unstable that can be shown in
Figure (9).
Figure (10) represents the inverse proportion between the number of antennas and the MSE values
when the number of antennas is increased that led to reduce the value of the MSE and vice versa;
that means the ASA system is affected by the number of the antennas in the array.
The other test that appears the good performance of (No-LMS) algorithm over the other two
algorithms can be shown in Figure (11), when varying the values of SNR the (No-LMS) algorithm
still gives conceivable errors than that obtained from the other algorithms (S-LMS, and N-LMS).
8. CONCLUSIONS
The optimum convergence factor for the minimum mean square error is obtained to No-LMS
algorithm whereas S-LMS and N-LMS algorithms gives mean square error values larger than that
obtain from No-LMS for same value of convergence factor; that means this algorithm (No-LMS)
gives best performance also in bad conditions.
The smart system used in cellular phone applications is affected by the fine varying for convergence
factor, and the stability of the whole system depends on these fine variations of different types of
algorithms, these variations of same as slow tuning for small steps, also this small value variations
of convergence factor appear the considered effect on system performance.
The computer simulation test shows that number of antenna in smart array also has large effect on
the stability of the system based to this array.
The S-LMS gives bad performance and fails in worst environments.
9. REFERENCES
1. Ahmed El Zooghby "Smart Antenna Engineering", Artech house, Second edition, p-7, 2005.
2. Amrita Rai, Amit Kumar "Analysis and Simulation of Adaptive Filter with LMS Algorithm",
International Journal of Electronic Engineering", Vol (2), pp121-123, 2010.
3. C.F.C Cowan, P.M. Grant "Adaptive Filter, Prentice Hall, 1985.
4. Chris Loadman, Zhizhang Chen, and Dylan Jorgen "An overview of Adaptive Antenna
Technologies for Wireless Communications" , Communication Networks and Series Research
Conference, Session A3, Moneton, New Brunswick Canada, 2003.
5. Hykin, Simon "Adaptive Filter Theory, Third Edition, Prentice Hall Inc.NJ, 1996.
6. Minseok Kim Hardware Implementation of Signal Processing in Smart Antenna Systems for
High Speed Wireless Communication ,PhD. Thesis, Yokohama National University , 2004.
7. N.P. Rallapall, S. Sharma, and A. Jain Simulation of Adaptive Filter for Hybrid Echo
Cancellation, IEI Journal-ET, Sept, 2007.
8. Rappaportm T. S. " Smart Antennas: Adaptive Array Algorithms and Wireless Position
Location, New York, IEEE Press, 1998.
9. S.F.Shaukat, Mukhtar Ul Hassan, and R. Farooq "Sequential Studies of Beam Forming
Algorithms for Smart Antenna Systems, World Applied Sciences Journal, Vol-6, pp 754-758,
2009.

Al-Qadisiya Journal For Engineering Sciences

Vol. 4 No. 3

Year 2011

10. Shannon Liew "Adaptive Equalizers and Smart Antenna System, PhD.Thesis, 2002,University
of Queensland.
11. Trees H.V. "Detection, Estimation, and Modulation Theory, PartIV, Optimum Array
Processing, John Wiely&Sons, 2002.

Figure (1) Smart antenna block diagram

Figure (2a) Switched beam system

Figure (2b) Adaptive array system

Al-Qadisiya Journal For Engineering Sciences

Vol. 4 No. 3

Figure (3) The LMS-algorithm

Figure (4) The ASA-Model

Year 2011

Al-Qadisiya Journal For Engineering Sciences

Vol. 4 No. 3

Year 2011

y(r) signal
2.5
2
1.5
1
0.5
0
-0.5
-1
-1.5
-2
-2.5

100

200

300

400

500

600

700

800

900

1000

Figure (5) The received signal y (r)

1
0.9
0.8

g a u s s ia n

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0

5
t im e

Figure (6) The standard deviation

10

Al-Qadisiya Journal For Engineering Sciences

Vol. 4 No. 3

Year 2011

0
S-LMS
N-LMS
No-LMS

-5
S-LMS

-15

-20
N-LMS
-25
-30

-35

No-LMS

500

1000

1500

2000
2500
Sample interval

3000

3500

4000

Figure (7) MSE versus the sample interval

for CF = 0.2

0
S-LMS
No-LMS
N-LMS

-5
-10

mean square error

mean square error

-10

S-LMS

-15

-20

N-LMS

-25
No-LMS

-30

-35

500

1000

1500

2000
2500
Sample interval

3000

Figure (8) MSE versus the sample interval

for CF 0.04

3500

4000

Al-Qadisiya Journal For Engineering Sciences

Vol. 4 No. 3

Year 2011

Figure (9) MSE versus the convergence factor

0.11
0.1
0.09
0.08

MSE

0.07
0.06
0.05
0.04
0.03
0.02
0.01

10

15

20
25
No. of antenna

30

35

Figure (10) MSE versus the number of antennas

40

Al-Qadisiya Journal For Engineering Sciences

Vol. 4 No. 3

Year 2011

0.09
S-LMS
N-LMS
No-LMS

0.08
0.07
0.06

MSE

0.05
0.04
0.03
0.02
0.01
0
2

5
SNR dB

Figure (11) Error versus the signal to noise ratio