Zigbee is one of the major sources of interference in 2.4GHz band for WLANs. It is seen whenever any Zigbee system is operating near to the WLAN system and transmitting signal at same frequency, time as of WLAN’s, the later ones performance detoriate severely. So in this paper an algorithm is proposed to estimate Zigbee interference component present in all OFDM based WLANs sub-carriers and cancel out the Zigbee interference from the received signal of WLANs receiver for multipath fading channels in frequency domain. Simulation results shows for high SNR values full cancellation of Zigbee interference or zero BER is possible.

1. ACEEE Int. J. on Network Security, Vol. 02, No. 01, Jan 2011
Cancellation of Zigbee interference in OFDM
based WLAN for multipath channel
Minakshmi Roy1, H.S. Jamadagni2
1
Indian Institute of Science, CEDT, Bangalore, India
Email: rmina@cedt.iisc.ernet.in
2
Indian Institute of Science, CEDT, Bangalore, India
Abstract— Zigbee is one of the major sources of interference II. OVERVIEW OF THE WLAN AND ZIGBEE
in 2.4GHz band for WLANs. It is seen whenever any Zigbee
system is operating near to the WLAN system and
transmitting signal at same frequency, time as of WLAN’s, the
later ones performance detoriate severely. So in this paper an
algorithm is proposed to estimate Zigbee interference
component present in all OFDM based WLANs sub-carriers
and cancel out the Zigbee interference from the received
signal of WLANs receiver for multipath fading channels in
frequency domain. Simulation results shows for high SNR
values full cancellation of Zigbee interference or zero BER is
possible.
Index Terms— WLAN, Zigbee, OFDM, OQPSK, narrowband
I. INTRODUCTION
Recent development in wireless technology shows that next
generation wireless system will provide users with variety Figure 1: Zigbee interference in OFDM based WLAN
of wireless services assuming coexistence of different
wireless technology at the same time and same place. At IEEE802.11g[7] standard based WLAN uses OFDM
present IEEE802.11g standard based WLAN which is modulation technique. In OFDM modulation multiple low
intended for wireless network, is one of the most popular data rate carriers are combined by transmitter to form a
device adopted in home, office, institution etc. It can composite high data rate transmission. IEEE802.11g
operate within range of 100 meter distance in 2.4GHz ISM supports different data like 6, 9, 18, 36, 48,54Mbps. Use of
band. cyclic prefix makes OFDM signal immune to multipath
Zigbee is another wireless technology intended for personal effect, inter carrier interference (ICI), intersymbol
area network. It transmit within a range of 10meters in interference (ISI) etc.
same 2.4 GHz unlicensed ISM band. As both the system Zigbee[5] system is designed for mainly low cost, low
operates in same ISM band (i.e 2.405-2.480GHz) whenever power applications. In the 2.4GHz range bit rate of Zigbee
they operate at the same time results in collision. is 250kbs/s. Zigbee uses direct sequence spread spectrum
From the literature search some techniques, to mitigate (DSSS) to make the signal robust against interference.
narrowband interference by estimating interference Zigbee system has 16 different 32-chip sequences. So to
component with OFDM null carriers using pseudo inverse make the signal DSSS 4bit information sequence is mapped
of narrowband signal’s transfer function [3] or erasing the into any of the 32 chip sequence which results in 2Mchip/s
sub carriers[1] can be found. But all this techniques having chip rate. Then chips are modulated by Offset-QPSK.
there own limitation. In this paper an algorithm is
proposed to cancel Zigbee interference in OFDM based III. INTERFERENCE MODEL AND CANCELLATION
WLANs. In the proposed cancellation algorithm
interference component in all the sub carriers are estimated In this section first a mathematical model of Zigbee
and cancelled. It is assumed that WLAN is monitoring the interference is given, and then an algorithm is proposed to
channel before transmission using spectrogram [2]. From estimate and cancel Zigbee interference from OFDM based
this it can detect any Zigbee signal interference. WLAN receiver.
The paper is organized as follows: In section 2 The received signal in the OFDM receiver for the kth
Overview of OFDM system and Zigbee interference subcarrier is given by[9] [5][6] :
modeling is given. In section 3 Cancellation algorithm,
estimation of frequency, power of Zigbee and channel tap Rk = H kU k + I k + N k (1)
value is given In section 4 simulation results are given
Section 5 conclude the paper summarizing of the Where U k is the complex MQAM signal and H k denotes
simulation results. the kth channel component corresponds to equivalent
channel transform matrix and I k is the interference
35
© 2011 ACEEE
DOI: 01.IJNS.02.01.194

2. ACEEE Int. J. on Network Security, Vol. 02, No. 01, Jan 2011
component corresssponds to kth sub-carrier of OFDM or S Zn,l = BZn,l • Yn 2 (6)
system and N k is the AWGN.
Zigb where Y n2 = [ D 0 D 1 ]T , n is any positive integer
The idea is to model interference I k as function of
transmitted unknown binary data to model the interference ⎛ πt ⎞ (7)
as given below: P (t ) = cos ⎜ ⎟ c o s 2π f h t
⎝ 2TZ ⎠
p −1
⎛ πt ⎞
Ik = ∑a
n=0
n,k Dn + Ck (2) Q (t ) = sin ⎜ ⎟ sin 2 π f h t
⎝ 2TZ ⎠
(8)
The value of p depends on the number of unknown Where subscript l denotes the lth Zigbee channel and
transmitted binary data of Zigbee system transmitted within subscript n is given to denote nth time instant. Matrix BZ
one symbol duration of the OFDM system. The values of n,l
is an N×2 matrix consisting of 2 columns of Pl (t), Ql (t)
an , k , Ck can be found from the fixed parameters of both multiplied by the square root of the power P. In equation
system like modulation type, pulse shape used in the (6) the superscript “2” denotes that matrix, Yn
2
is
Zigbee system , tap values of the receiver low pass filter,
cyclic prefix remove matrix, FFT matrix in OFDM generated from 2 unknown binary data. Similarly for p
receiver. (where p is even) unknown binary data can be written as:
A. Interference Model of Zigbee signal in OFDM receiver S Zn,l = BZpn,l • Yn p
p
(9)
This subsection show how to model the Zigbee
p
interference for each of the OFDM subcarrier as a function Where S Zn,l , BZpn,l Yn p is of dimension N1×1, N1×p,
of unknown binary data transmitted within a symbol ,
duration of OFDM signal. p×1 respectively. The superscript “p” denotes the number
of binary Zigbee data transmitted during one the OFDM
T
The OQPSK modulated signal can be written as[5]: symbol period and Ynp = ⎡D0 D D2.... Dp−1 ⎤ where D0,D1…Dp-
1 ⎣ ⎦
1are binary data transmitted by Zigbee system. So vector of
sz (t ) = PD0 p (t ) cos 2π f ht Zigbee samples
p
S Zn,l depend on the particular
(3)
+ PD1 p (t − TZ ) sin 2π f h t combination binary data vector x j = { D 0 D1 D 2 ....D p −1 } .
Where D0 , D1 ∈ {+1, −1} denotes binary data for in Therefore, to denote the dependency on a particular
combination of data input vector the equation(9) is written
phase , quadrature phase component of OQPSK signal and as:
p(t ) is the shaping pulse which is given by[5]:
S Zn,l, j = BZn,l • Yn,pj
p p
(10)
⎛ πt ⎞
p (t ) = cos ⎜ ⎟ 0 ≤ t ≤ 2TZ (4)
p
⎝ 2TZ ⎠ The received Zigbee data vector S Z n,l in OFDM
=0 elsewhere
receiver is passed through a low pass filter of impulse
And f h is the centre frequency among any of the 16 Zigbee
and Ts
OFDM
response hlpf(t). Let Ts be the sampling
channels of the Zigbee and P is the signal power. Let there
are N samples for the TZ symbol duration, then the time of the OFDM and BT signal respectively. Then the
equation(3) can be written in matrix form as given: filtered Zigbee samples in OFDM system will be[3]:
N1 −1
(
⎡P n2T
⎢
l ) Z
Q n2TZ
l ( ) ⎤
⎥ i ( n) = ∑ hlpf ( nTsOFDM − mTsZigbee )SZn,l,j ( m)
p
(11)
(5)
⎢P ( n2TZ +TS )
⎢l
(
Q n2TZ +TS
l ) ⎥
⎥
m=0
⎢ ⎥ ⎡D ⎤
SZn,l = P ⎢. 0
⎥ ⎢D ⎥ After passing through the cyclic prefix removal block
⎢. ⎥⎣ 1 ⎦ the above equation can be written in matrix form as:
(
⎢l Z
) (
⎢P n2T + N −1 T Q n2T + N −1 T
( )S l Z ( )S ) ⎥
⎥
i NFFT = Rcp • hlpf • SZn,l,j = h1 • Bn,l •Yn,j
p
⎢ ⎥ p p
⎣ ⎦ (12)
Rcp is cyclic prefix removal matrix of dimension NFFT
×(Ncp +NFFT ), Ncp is the length of cyclic prefix. hlpf is
generated from tap value of receiver low pass filter.
Τ
i N FFT = ⎡ i0 i1 .. iΝ FFΤ ⎤ , h1 = Rcp • hlpf ,
⎣ -1 ⎦
36
© 2011 ACEEE
DOI: 01.IJNS.02.01.194

3. ACEEE Int. J. on Network Security, Vol. 02, No. 01, Jan 2011
After FFT operation in the OFDM receiver the vector
consist of interference component for the OFDM sub 2MHz Wide
carriers is given by : Zigbee Spectrum
p p
I j = F • h1 • BZ n,l • Yn, j (13) Sub carriers
Where F is 64×64 FFT matrix. And
fstart ffinish
1 2…. k …………. NFFT
I = [I 0, j , I 1, j , I 2, j ....I k , j ....I 63, j ]T .
Let Ik denotes the interference corresponding to the kth
22 MHz wide WLAN Spectrum
subcarrier and Fk denotes the kth row of the FFT matrix
Figure 2: Zigbee inside WLAN frequency band
F, then Ik is expressed as:
C. Channel estimation and Interference power estimation:
p p
I k, j = Fk • h1 • BZ n,l • Yn, j (14) Training signal are used to estimate both channel taps
and the instantaneous power of the interference signal. Let
According IEEE802.11g standard [4] one OFDM symbol uTrain is the set of training signal after IFFT in the
period TOFDM=4µs and one Zigbee[5] symbol period
transmitter of an OFDM system.
TZ=.5µs. As Zigbee is using O-QPSK data per cos or sin
pulse will remain unchanged for 2TZ=1µs. So it can be
found for 4us of OFDM signal there will be 16 unknown
uTrain = F H • U Train = ⎡u0 u1 . .. . u N FFT −1 ⎤
⎣ ⎦ (18)
data for Zigbee (combining both sin and cos pulse’s data),
basically which implies 9 combination of unknowns. As If u1Train is the received matrix at the receiver before FFT
OFDM will remove the cyclic prefix length which is one operation then it is given by:
fourth of the symbol duration, that is 0.8 µs, so for the rest
3.2 µs there will 8 unknown Zigbee data. Now Yn, j is a
p u1Train = Rcp • H 0 • Tcp • uTrain (19)
8×1 matrix consist of 8 unknown binary data and matrix %
= H •u (20)
Train
p
BZ n,l is of N1×8.Let
%
where H is NFFT× NFFT circulant matrix which is given
p below:
B Z n,l = [C1 C2 C3 C4 C5 C6 C7 C8] ( 15)
⎡h0 0 . . . . . . . . 0 hL-1 . . h2 h1 ⎤
So the eq(14) can be rewritten as: ⎢h h 0 . . . . . . . . 0 h . h3 h2 ⎥
⎢ 1 0 L-1 ⎥
7
⎢. . . . . . . . . ⎥ (21)
⎢ ⎥
I k,l, j = ∑a D
n=0
n n (16) ⎢ hL-2 hL-3 . h1 h0 0 . . . . . . 0 hL-1
H = ⎢hL-1 . . . . h1 h0 0 . . . . . 0. . . . . . .0
%
⎥
⎥
⎢ ⎥
⎢0 hL-1 . . h1 h0 0. . . . . . . .0 ⎥
⎢. . . ⎥
Then a0 = Fk • h1 • C1 ; a1 = Fk • h1 • C2 ; ⎢ ⎥
⎢. . . ⎥
a2 = Fk • h1 • C3; a3 = Fk • h1 • C4 ; a4 = Fk • h1 • C5 ; ⎢ ⎥
⎣ 0 . . . . 0 hL-1 . . . . . h1 h0 ⎦
a5 = Fk • h1 • C6 ; a6 = Fk • h1 • C7 ; a7 = Fk • h1 • C8
Now equation (20) is rewritten as:
B. Interference frequency estimation u1Train = uTrain • h
% (22)
Let Δf be the sub carrier spacing of the OFDM sub ⎡u0 u N FFT −1 ....u N FFT − ( L −1) ⎤
carriers and kth sub carrier have maximum energy. If the ⎢ ⎥
⎢ u1 u 0 ...........u N FFT − ( L − 2 ) ⎥
OFDM frequency band ranges from fstart to ffinish then
Zigbee’s frequency is given by Where u
% Train = ⎢. . . ⎥ (23)
⎢ ⎥
f h = round ( f start + ( Δf × k ) ) (17)
⎢.
⎢
. . ⎥
⎥
⎢u u u
⎣ N FFT −1 N FFT − 2 N FFT − L ⎥
⎦
Where, ffinish- fstart = BMHz , B is the estimation band of
WLAN. and h = (hL −1 .......h1 h0 ) T and uTrain is a matrix of
%
dimension NFFT×L consisting training signals.
37
© 2011 ACEEE
DOI: 01.IJNS.02.01.194

4. ACEEE Int. J. on Network Security, Vol. 02, No. 01, Jan 2011
Replacing h1 • BZn,l = h2 in equation(12), the Zigbee
p
filter ,FFT matrix of the OFDM receiver and Vk can be
interference in OFDM receiver before IFFT can be written expressed as:
as: 8
Vk = Fk • h1 • BZn,l (34)
= h2 • Yn, j
p
N FFT
i (24) And
8
x j = Yn, j (35)
Now in presence of AWGN noise and Zigbee interference
the received signal before FFT operation is written in Where Fk is subset of FFT matrix corresponds to the
matrix form as: chosen set of p=8 null carriers and is of cdimension
1×NFFT,. h1 is generated from low pass filter response of
u1Train = u1Train + i N FFT + n1
ˆ (25)
8
OFDM receiver, BZ n,l is generated from transmitted
Where n1 is AWGN noise
frequency, modulation ,pulse shaping characteristics of
ˆ % (
u1Train = ( uTrain • h ) + h2 • Yn, j + n1
p
) (26)
Zigbee system. Assuming hopping frequency, power of the
interference signal has been measured perfectly using
OFDM training signals], estimation of interference in
⎡h ⎤ OFDM data signals is done as follows:
= [ uTrain h2 ] ⎢ p ⎥ + n1
% (27)
1. For each data input vector x j , where
⎢Yn, j ⎥
⎣ ⎦
= Ay + n1 (28) j ∈ {0,1,...., 255} , generate Vk and find the interference
for each of the set of p null sub-carrier using the equation
where A = [ uTrain h2
% ] is of dimension NFFT×(L+2p). below:
( )
T Vk x j = Fk • h1 • BZn,l • x j = I k, j .
8
(36)
T p
and y = h Yn, j is matrix of dimension (L+2p)×1
consist of unknown channel vector and unknown Zigbee 2. Take the Difference Ek, j = Yk − I k , j and obtain the
signal vector. From eq(29) unknown parameters of matrix y
can be estimated near to the actual value if noise is magnitude of Ek, j . Average Ek, j over all the chosen set
negligible by the equation given below: of null sub-carriers using the equation below:
y = A u1Train − A n1
ˆ -1 -1
(29) 1 p −1
Ej = ∑E
p k =0 k, j
(37)
Let be the vector of estimated vector Zigbee signal from
then estimated Zigbee samples corresponding to the p
symbol will be: 3. Choose xj that will minimize the average error E j
)p ) 4. For the above chosen x j obtain the interference
S Zn,l, j = BZn,l • Yn,pj
p
(30)
component for all the sub-carriers of one OFDM signal.
B. Interference Cancellation Algorithm
5. Subtract the above obtained estimated Zigbee
The received signal for the null sub carriers of the OFDM interference for each sub-carrier from the received signal.
receiver is given by[5]: Repeat the above procedure from step 1 to 5 for the next
OFDM signal to cancel out Zigbee interference.
Rk = I k + N k (32)
IV. SIMULATION RESULTS
It can be shown I k can be expressed as function of In all simulation results SIR is ratio of total OFDM power
transmitted binary input signal of the Zigbee system as: of all sub-carriers with interferer (BT) power. SNR is the
ratio of total OFDM power with additive white Gaussian
I k = Vk • x j (33) noise. In figure 6 technique1 refers to the narrowband
cancellation method[3] where estimation of transmitted
data of narrowband interference and reconstruction of its
Where x is matrix of unknown transmitted binary inputs waveform is done by measuring interference information
corresponds one OFDM symbol period and on certain unmodulated null sub-carriers and using
erasures[1] refers to the techniques of replacing nulls in the
{
x∈−1−1−1−1−1−1−1−1,......,11111111 ={ x0 ,x1,........x255}
} sub-carriers where the narrowband interference signal has
hopped in OFDM frequency range.
and Vk is matrix generated depending upon modulation ,
pulse shaping of the Zigbee system and equivalent low pass
38
© 2011 ACEEE
DOI: 01.IJNS.02.01.194

5. ACEEE Int. J. on Network Security, Vol. 02, No. 01, Jan 2011
Figure6. Comparison of uncoded BER of 64QAMWLAN with different
techniques in AWGN for scenario in figure2. Technique1 is the technique
Figure 3: Comparison of coded BER for 64QAM OFDM based WLAN used in ref[3].
with and without Zigbee in AWGN
V. SUMMARY
This paper proposes an algorithm for extracting out Zigbee
interference component present in all OFDM subcarriers
and cancelling those estimated Zigbee signal’s part from
received signal in OFDM receiver. Simulation result shows
algorithm.
gives significant improvement in performance particularly
when SNR value is high, almost zero BER can be obtained
in presence of Zigbee interference for OFDM based
WLANs. As initial estimation of amplitude or frequency
cannot be obtained accurately for low SNR values
performance in terms of BER will not improve much using
proposed algorithm for low SNR. So applying this
algorithm will nullify the Zigbee interference present in all
OFDM based WLAN subcarriers and give one a step
solution towards coexistence of WLAN and Zigbee system
Figure 4. Comparison of Coded BER for 64QAM OFDM based WLANs
in presence of Zigbee interference in flat Rayleigh fading channel using REFERENCES
proposed cancellation algorithm
[1] Doufexi, A. Arumugam, S. Armour and A. Nix, An
Investigation of the Impact of Bluetooth Interference on the
Performance of 802.11g Wireless Local Area Networks,
pages:680-684,Vol1.,April,2003
[2]Miller, Rob Xu, Wenyuan Kamat, Pandurang Trappe,
Wade , Service discovery and Device Indentification in cognitive
Radio Networks, Workshop on Networking Technologies for
Software Defined Radios (SDR)Networks, San Diego ,
CA,pges40-47, June 2007.
[3] Dan Zhang; Pingyi Fan; Zhigang Cao, A novel narrowband
interference canceller for OFDM systems Wireless
Communications and Networking Conference, 2004. WCNC.
2004 IEEEVolume 3, Issue , 21-25 March 2004 Page(s): 1426 -
1430 Vol.3
[4] IEEE802.11g standard.
[5] IEEE Std 802.15.4-2006
[6]ZhengdaoWang and Geogios B. Giannakis, Wireless
Multicarrier Communication: Where Fourier Meets Shannon ,
Transaction on IEEE Signal Processing,May 2000
Figure 5. Comparison of Coded BER for 16QAM OFDM, BPSK-OFDM [7]Peter Klenner, The OFDM multicarrier system,
based WLANs in presence of Zigbee interference in flat Rayleigh fading Commmunication Technology Laboratory,-SS 2006
channel using proposed cancellation algorithm [8]Henik Schulze, Christian Luders, Theory and Applications of
OFDM and CDMA, Wideband Wireless Communications, John-
Wiley and Sons, Ltd, 2005.
[9]Ramjee Prasad, OFDM for Wireless Communication
39
© 2011 ACEEE
DOI: 01.IJNS.02.01.194