© 2012, IJARCSSE All Rights Reserved Page | 356

Volume 2, Issue 7, July 2012 ISSN: 2277 128X

International Journal of Advanced Research in Computer Science and Software Engineering Research Paper Available online at: www.ijarcsse.com

Reduction of ICI in OFDM Based WLAN System using

Neural Network Miss Shruti M. kallurwar

Assistant Prof. Rahila Patel

Computer Science and Engineering, Computer Science and Engineering,

Rajiv Gandhi College of Engineering Research Rajiv Gandhi College of Engineering Research

and Technology, Chandrapur, India and Technology, Chandrapur, India

Abstract:— Orthogonal frequency division multiplexing (OFDM), because of its resistance to multipath fading, has attracted

increasing interest in recent years as a suitable modulation scheme for commercial high-speed broadband wireless

communication systems. OFDM can provide large data rates with sufficient robustness to radio channel impairments.

Orthogonal frequency division multiplexing (OFDM) is one of the multi-carrier modulation (MCM) techniques that transmit

signals through multiple carriers. These carriers (subcarriers) have different frequencies and they are orthogonal to each

other. A major problem in OFDM is its vulnerability to frequency offset errors between the transmitted and received signals.

In such situations, the orthogonality of the carriers is no longer maintained, which results in Intercarrier Interference (ICI). In

this paper interference is reduced by neural network.

Keywords – Intercarrier Interference, OFDM, Neural Network.

---

I. INTRODUCTION

Orthogonal Frequency Division Multiplexing is a special

case of multi-carrier modulation and widely used in

wireless communication system like wireless local area

network and digital audio broadcasting (DAB). However

one of the problems in OFDM systems is its sensitivity to

phase offset and frequency offset caused by Doppler

frequency drift and multipath fading [1]. In such

situations, the orthogonality of the carriers is no longer

maintained, which results in Intercarrier Interference

(ICI). ICI results from the other sub-channels in the same

data block of the same user. ICI problem would become

more complicated when the multipath fading is present. If

ICI is not properly compensated it results in power

leakage among the subcarriers, thus degrading the system

performance [2].

Some techniques are previously developed for

reducing the effect of ICI: Frequency domain equalization

but it only reduce the ICI caused by fading distortion

which is not the major source of ICI [3]. Time Domain

Windowing only reduce the ICI caused by band limited

channel which is also not the major source of ICI. The

major source of ICI in OFDM is its vulnerability to

frequency offset errors between the transmitted and

received signals, which may be caused by Doppler shift in

the channel or by the difference between the transmitter

and receiver local oscillator frequencies.

In this paper the Neural Network schemes is

applied to the system to reduce the effect of ICI on

receiver side in OFDM system and compare their result.

The paper is organized as follows: Section II illustrates

the OFDM modulation techniques; Section III explains

basic block diagram which we used for the

implementation; Section IV explain how the effect is

reduced; Section V analyses the result; Section VI makes

the concluding remarks.

II. OFDM SYSTEM DESCRIPTION

A basic OFDM system contains modulation scheme,

serial to parallel transmission, parallel to serial

transmission and IFFT/FFT [4]. Fig 1 illustrates the block

diagram of OFDM system. The input data stream is

converted into parallel data stream and mapped with

modulation scheme. Then the symbols are mapped with

inverse fast Fourier transform (IFFT) and converted to

serial stream. The complete OFDM symbol is then

transmitted through the channel.

Fig 1: Block diagram of FFT based OFDM system

Therefore OFDM symbol can be expressed as

(1)

Where denotes the sample of the OFDM signal,

denotes the modulated symbol within subcarrier and

N is the number of subcarriers.

Input

data

Modulat

ion scheme

Serial

To Parallel

IFFT

FFT

Parallel

To

serial

Channel

Serial

To Parallel

Demod

ulation scheme

Output

data

Parallel

To Serial

Volume 2, Issue 7, July 2012 www.ijarcsse.com

© 2012, IJARCSSE All Rights Reserved Page | 357

On receiver side this symbols are converted back

to parallel stream and mapped with FFT then with

demodulation scheme and converted to serial data as

output data.

The demodulated symbol stream is given by:

(2)

Where w (m) corresponds to the FFT of the

samples of the w (n).

III. HIPERLAN2

For continuous data transformation and receiving, here we

are using Hiperlan2 architecture. OFDM is the modulation

used in the physical layer of HiperLAN2. For the

subcarrier modulation we have choice between BPSK,

QPSK, and 16-64 QAM, but in this paper we worked on

BPSK and 16QAM; the symbol period used is 3.6μs with

a guard interval of 0.8μs (optionally 0.4μs). The

demodulation is coherent. OFDM obviously provides

intentionally wide frequency band and a potential bit-rate.

A basic block structure of Hiperlan2 shows below.

Following are the various components Hiperlan2

architecture-

Binary generator – It generate binary bit signal from

the input data.

Fig 2: Block diagram of Hiperlan2

Convolution encoder- It takes the bit sequence and

constraint length as input and it returns the encoded

sequence.

Matrix interleaver - This function simply interleaves

two vectors. The vectors can be of different lengths.

If one vector is longer, the leftover elements are just

appended to the output vector. This can be done in a

couple of lines if the vector lengths are known, but

this function handles most of the possibilities

automatically, and should save a few minutes.

OFDM Transmitter- An inverse FFT is computed on

each set of symbols, giving a set of complex time-

domain samples. These samples are then quadrature-

mixed to pass band in the standard way. The real

and imaginary components are first converted to the

analogue domain using digital-to-analogue

converters (DACs); the analogue signals are then

used to modulate cosine and sine waves at

the carrier frequency, respectively. These signals are

then summed to give the transmission signals.

OFDM receiver- The receiver picks up the signal,

which is then quadrature-mixed down to baseband

using cosine and sine waves at the carrier frequency.

This also creates signals centered on, so low-pass

filters are used to reject these. The baseband signals

are then sampled and digitized using analogue-to-

digital converters (ADCs), and a forward FFT is used

to convert back to the frequency domain.

Modulator Baseband- This block modulates a signal

using modulation technique, example- BPSK or 16

QAM

Demodulator Baseband- This block demodulates a

signal using modulation technique , example- BPSK

or 16 QAM

General Block Deinterleaver: - The General Block

Deinterleaver block rearranges the elements of its

input vector without repeating or omitting any

elements. If the input contains N elements, then

the Elements parameter is a column vector of

length N. The column vector indicates the indices, in

order, of the output elements that came from the input

vector.

Viterbi decoder- It is most often used for decoding a

bit stream that has been encoded using convolution

code

IV. NEURAL NETWORK

The term neural network was traditionally used to refer

to a network or circuit of biological neurons. The modern

usage of the term often refers to artificial neural

networks, which are composed of artificial neurons or

nodes. Thus the term has two distinct usages:

1. Biological neural networks are made up of real

biological neurons that are connected or

functionally related in a nervous system. In the

field of neuroscience, they are often identified as

Binary

generator

Viterbi

decoder

Convolutional

encoder

Matrix

interleaver

Demodu

lator Baseban

d

Normalize Modulator

Baseband

OFDM

receiver

OFDM

transmitter

Channel

Denormalize General

block

Deinterleaver

Unipolar

to Bipolar

Converter

Matrix

deinterleaver

Volume 2, Issue 7, July 2012 www.ijarcsse.com

© 2012, IJARCSSE All Rights Reserved Page | 358

groups of neurons that perform a specific

physiological function in laboratory analysis.

2. Artificial neural networks are composed of

interconnecting artificial neurons. Artificial

neural networks may either be used to gain an

understanding of biological neural networks, or

for solving artificial intelligence problems

without necessarily creating a model of a real

biological system. The real, biological nervous

system is highly complex: artificial neural

network algorithms attempt to abstract this

complexity and focus on what may

hypothetically matter most from an information

processing point of view.

3.

V. PROPOSED MODEL

In this Paper the effect of ICI is reduced using artificial

neural network in which we used Kohennes unsupervised

learning method. With the help of Kohennes unsupervised

learning method we generate a simulink block in Matlab

and insert this block into fig 2 in receiving side with

OFDM receiver to reduce the effect of ICI.

The principal goal of a Kohenens Unsupervised Learning

Map is to transform an incoming signal pattern of

arbitrary dimension into a one or two dimensional discrete

map, and to

Perform this transformation adaptively in a topologically

ordered fashion.

We therefore set up our Kohenens Unsupervised Learning

Map by placing neurons at the nodes of a one or two

dimensional lattice. Higher dimensional maps are also

possible, but not so common. The neurons become

selectively tuned to various input patterns (stimuli) or

classes of input patterns during the course of the

competitive learning. The locations of the neurons so

tuned (i.e. the winning neurons) become ordered and a

meaningful coordinate system for the input features is

created on the lattice. The Kohenens Unsupervised

Learning Map thus forms the required topographic map of

the input patterns.

Algorithm of Kohenens Unsupervised Learning Map –

1) Initialize each node’s weights.

2) Choose a random vector from training data and

present it to the SOM.

3) Every node is examined to find the Best

Matching Unit (BMU).

4) The radius of the neighborhood around the BMU

is calculated. The size of the neighborhood

decreases with each iteration.

5) Each node in the BMU’s neighborhood has its

weights adjusted to become more like the BMU.

Nodes closest to the BMU are altered more than

the nodes furthest away in the neighborhood.

6) Repeat from step 2 for enough iterations for

convergence.

In this paper we are showing various results on various

channels (AWGN, Rayleigh, Rician) and with different

modulation scheme (BPSK, 16QAM). When we are

passing signals through the channel, signals are affected

by many types of interferences present in the channel but

we are working on only ICI so we are introducing ICI by

own using matlab simulink. ICI occurs because of

Frequency offset, Phase offset and Non linearly distorted

signals, frequency offset and phase offset blocks are

available in matlab and non linearly distorted signals are

generated when we attach High Power Amplifier (HPA)

block. All this blocks are attached at the transmitting end

(after OFDM transmitter) one by one to generate ICI on

the signals and in receiving side Neural is applied on this

signal to reduce the effect.

VI. SIMULATION RESULT

Modulation schemes BPSK and 16 QAM are used in

simulation as they are used in many OFDM standards on

different channels (AWGN, Raylegh, Rician). In this

paper we are comparing total 18 modules, 18 modules are

generated with different combination of channels,

Modulation scheme and three effects which introduce ICI.

Other simulation parameters are given in table 1.

Table 1: Basic simulation parameters.

Number of subcarriers (N) 48

IFFT size 64

Number of OFDM symbols 80

Signal to noise ratio 1,3,5,7,8….40dB

In this paper we are showing result in the form of

table, table2 consists of Signal to Noise (SNR) values,

Simulated BER values for schemes without Neural

Network and with Neural Network. Table 2 shows how

much effect of ICI is reduced using Neural Network and

comparing their result.

Table 2: Result of Simulation

Table 2 a: Shows the result of ICI caused by Non Linear

distortion with BPSK Modulation on all three channels

Sr. No

. Channel

SNR Values

3 6 9 12

1

Conventional

Hiperlan2 receiver

with AWGN channel

model

0.4749 0.2667

0.0086 0.15

2

Proposed Neural based receiver with

AWGN channel

model, 0.4711 0.2317 0.0079

0.00002

3

Conventional

Hiperlan2 receiver with Racian channel

model 0.6212 0.6113 0.3284 0.0254

4

Proposed Neural based receiver with

Racian channel

model, 0.6209 0.6101 0.2284 0.0254

Volume 2, Issue 7, July 2012 www.ijarcsse.com

© 2012, IJARCSSE All Rights Reserved Page | 359

5

Conventional Hiperlan2 receiver

with Rayleigh channel model 0.6619 0.6455 0.5837 0.4473

6

Proposed Neural

based receiver with

Rayleigh channel model 0.6475 0.6435 0.5819 0.4473

Table 2b: Shows the result of ICI caused by Phase Offset with

BPSK Modulation on all three channels

Sr. No.

Channel SNR Values

3 6 9 12

1

Conventional Hiperlan2 receiver

with AWGN channel

model 0.4643 0.2562 0.0081 0.0125

2

Proposed Neural based receiver with

AWGN channel

model, 0.4598 0.2201 0.0068

0.00009

3

Conventional Hiperlan2 receiver

with Racian channel

model 0.6034 0.5864 0.3192 0.0211

4

Proposed Neural

based receiver with

Racian channel model, 0.5902 0.5762 0.3145 0.0209

5

Conventional

Hiperlan2 receiver

with Rayleigh channel model 0.6523 0.641 0.5785 0.4398

6

Proposed Neural

based receiver with

Rayleigh channel model 0.64 0.6386 0.5764 0.4276

Table 2 c: Shows the result of ICI caused by Frequency Offset

with BPSK Modulation on all three channels

Sr. No.

Channel SNR Values

3 6 9 12

1

Conventional

Hiperlan2 receiver

with AWGN channel model 0.4569 0.243 0.0062 0.0089

2

Proposed Neural

based receiver with

AWGN channel

model, 0.451 0.2242 0.0059

0.00014

3

Conventional

Hiperlan2 receiver with Racian channel

model 0.5345 0.5173 0.3072 0.013

4

Proposed Neural

based receiver with

Racian channel model, 0.5209 0.4998 0.2143 0.0101

5

Conventional Hiperlan2 receiver

with Rayleigh

channel model 0.6328 0.6356 0.5733 0.4365

6

Proposed Neural

based receiver with

Rayleigh channel model 0.6253 0.62 0.5694 0.4213

Table 2d: Shows the result of ICI caused by Non Linear distortion with 16 QAM Modulation on all three channels

Sr. No.

Channel SNR Values

3 6 9 12

1

Conventional

Hiperlan2 receiver with AWGN channel

model 0.4069 0.0988 0.0008

2

Proposed Neural

based receiver with AWGN channel

model 0.3912 0.0933 0.0001 0.000002

3

Conventional

Hiperlan2 receiver

with Racian channel model 0.5247 0.5154 0.5225 0.3223

4

Proposed Neural

based receiver with Racian channel

model 0.5244 0.5143 0.51 0.2322

5

Conventional

Hiperlan2 receiver with Rayleigh

channel model 0.5487 0.5297 0.5341 0.3145

6

Proposed Neural

based receiver with Rayleigh channel

model 0.5378 0.5198 0.5101 0.2098

Table 2e: Shows the result of ICI caused by Phase Offset with 16 QAM Modulation on all three channels

Sr. No.

Channel SNR Values

3 6 9 12

1

Conventional

Hiperlan2 receiver

with AWGN channel model 0.4064 0.0976 0.0007

2

Proposed Neural

based receiver with AWGN channel

model, 0.3899 0.0786 0.0006 0.000001

3

Conventional

Hiperlan2 receiver

with Racian channel model 0.5107 0.5109 0.5223 0.3199

4

Proposed Neural

based receiver with Racian channel

model, 0.4984 0.5103 0.501 0.2124

5

Conventional

Hiperlan2 receiver

with Rayleigh channel model 0.5412 0.5203 0.5291 0.3112

6

Proposed Neural

based receiver with Rayleigh channel

model 0.5309 0.5113 0.5201 0.2065

510

510

Volume 2, Issue 7, July 2012 www.ijarcsse.com

© 2012, IJARCSSE All Rights Reserved Page | 360

Table 2f: Shows the result of ICI caused by Frequency offset with 16 QAM Modulation on all three channels

Sr. No

. Channel

SNR Values

3 6 9 12

1

Conventional Hiperlan2 receiver

with AWGN

channel model

0.406

2 0.0967 0.0006

2

Proposed Neural based receiver with

AWGN channel

model

0.386

7 0.0623

0.0005

4

0.00000

3

3

Conventional Hiperlan2 receiver

with Racian channel

model

0.439

7 0.51 0.5219 0.3195

4

Proposed Neural based receiver with

Racian channel

model 0.42 0.4991 0.5009 0.2052

5

Conventional Hiperlan2 receiver

with Rayleigh

channel model

0.536

7 0.5188 0.52 0.3089

6

Proposed Neural based receiver with

Rayleigh channel

model

0.522

5 0.5078 0.5188 0.2012

VII. CONCLUSION

Orthogonal frequency division multiplexing

(OFDM) is a very important modulation technique in

wideband wireless communication and multimedia

communication systems. The paper concentrates on

reducing the effect of ICI using Neural Network. This

paper shows effect of ICI is much more reduced, table 2a

to 2f shows the result of simulation and it also shows the

difference between values of systems with Neural

Network and without Neural Network. From the table 2a

to 2f we have seen that with BPSK modulation we get

better result than the 16QAM with all channels. Such a

technique will improve the performance of the existing

OFDM systems.

ACKNOWLEDGMENT

This research paper is made possible through the help and

support from everyone, including: parents, teachers and in

essence, all sentient beings. Especially, please allow me to

dedicate my acknowledgment of gratitude toward the

following significant advisors and contributors:

First and foremost, I would like to thank Prof. Rahila

Patel for her’s most support and encouragement. She

kindly read my paper and offered invaluable detailed

advices on grammar, organization, and the theme of the

paper.

Second, I would like to thank Prof. Shubhangi

Rathkanthiwar to read my paper and to provide valuable

advices.

The product of this research paper would not be possible

without this two person.

REFERENCES

[1]“OFDM for wireless multimedia communications” by

Richard van Nee, Ramjee Prasad 2000

[2]” Performance Analysis of a New ICI-Self-

Cancellation-Scheme in OFDM Systems “Yi-Hao Peng,

Ying-Chih Kuo, Gwo-Ruey Lee, Student Member, IEEE

and Jyh-Horng Wen, Member, IEEE, 2007

[3] “Frequency-Domain ICI Estimation, Shortening, and

cancellation in OFDM Receivers” Charles K. Sestok

Texas Instruments, Inc. Dallas, TX, USA and Predrag

Radosavljevic Dept. of ECE, Rice University Houston,

TX, USA

[4] Mark Engels, “Wireless OFDM Systems: How to

Make Them Work?” Kluwer Academic Publishers, first

edition, 2002.

[5] “Performance Analysis of a New ICI-Self-

Cancellation-Scheme in OFDM Systems” Yi-Hao Peng,

Ying-Chih Kuo, Gwo-Ruey Lee, Student Member, IEEE

and Jyh-Horng Wen, Member, IEEE 2007

[6] Zhao, Y. and HAggman, G., ”Intercarrier Interference

Self Cancellation Scheme for Mobile Communication

Systems.” IEEE Transactions on Communications, Vol.

49, No. 7, 2001

[7] Brahmaji T.A.R.K, ―”An Efficient ICI Cancellation

Technique for OFDM Communications Systems”, M.S.

Thesis, Dept. of Electrical Engineering National Institute

of Technology Rourkela, 2009.

[8]” An Experimental Study of ICI Cancellation in OFDM

Utilizing GNU Radio System” Miftahur Rahman, Sajjadul

Latif, and Tasnia Ahad North South University Dhaka,

Bangladesh. 2011

[9] Peiris, B. J., “A technique to reduce inefficiencies in

self ICI cancellation based OFDM system.” IEEE, 2002

[10] K. Sathananthan and Rajtheva, R. M. A. P.,

”Analysis of OFDM in the presence of frequency Offset

and a method to reduce performance Degradation.” IEEE,

Globecom 2000, San Francisco, California, 2000

[11] “Joint Phase/Amplitude Estimation and Symbol

Detection for Wireless ICI Self-Cancellation Coded

OFDM Systems” Hsiao-Chun Wu and Xiaozhou Huang

2004

[12] S.V. Rathkanthiwar, C.G. Dethe, K.D. Kulat, ‘A

Neural Approach for Compensation of Effects of PAPR

Causing BER Degradation in Hiperlan/2’, IEEE

International Conference on Modeling and Simulation

(ICCMS- 2011),

[13] S.V. Rathkanthiwar, C.G.Dethe, K. D. Kulat, ‘Self

organizing map approach for improvement in BER

performance of OFDM receiver’, International conference

on signal acquisition and processing ICSAP 2011

[14] S.V. Rathkanthiwar, C.G.Dethe, K. D. Kulat, ‘A

neural approach for overall improvement of OFDM

system’, IET conference SEISCON 2011, 20-22 July2011.

510