Communications and Network, 2013, 5, 31-35
http://dx.doi.org/10.4236/cn.2013.53B2007 Published Online September 2013 (http://www.scirp.org/journal/cn)
Interference Cancellation Algorithm for 2 × 2 MIMO
System without Pilot in LTE
Otgonbayar Bataa1, Erdenebayar Lamjav1, Uuganbayar Purevdorj1, Young-il Kim2,
Khishigjargal Gonchigsumlaa2
1School of Information and Communication Technology (SICT) of Mongolian University
of Science and Technology, Ulaaanbaatar, Mongolia
2Electronics and Telecommunications Research Institute (ETRI), 138 Gaejeongno, Yuseong-gu, Daejeon, Korea
[e-mail], [e-mail]
Received May, 2013
ABSTRACT
Interference cancellation system (ICS) for 3GPP/LTE system is the broadband cancellation system, which receives
forward signal through the donor antenna. We proposed new algorithm of received signal with pilot and non-pilot de-
sign. Although repeater design needs our project, so in this paper we discuss about interference cancellation algorithm
for 2x2 MIMO systems without pilot in LTE. First explain the general principle structure of 3GPP/LTE, next determine
our new design and algorithm. Finally, we simulated our mathematic extraction of proposed new algorithm on MAT-
LAB.
Keywords: MMSE; Lookup-table; Threshold; Cost Function; Viterbi Algorithm
1. Introduction
Interference cancellation system (ICS) for the donor and
service antenna amplifies the received signals and sends
them to base station and mobile to get stronger signal.
During this process, it cancels the interference signal
between the donor and receiving antenna. There are us-
ing types of cancellation algorithms, some existing adap-
tive cancellation algorithms work in the frequency do-
main, using the reference tones carried in the OFDM
signal. Other algorithms use the time domain methods
like the LMS. In general these algorithms are classified
by interference cancellation algorithm with and without
pilot signals. We described interference cancellation al-
gorithm without pilot signals in this paper. It is feasible
to implement ICS in OFDMA systems such as LTE. Al-
though IC techniques can be applied to both downlink
and uplink of LTE, due to complexity considerations, IC
is considered mainly as a technique for the UL and im-
plemented in the base station receiver. ICS techniques
can be used to cancel both intra-cell and inter-cell inter-
ference.
2. Proposed Channel Estimation Method and
Equalization
2.1. Non Pilot Channel Estimation
The MMSE estimator minimizes the MSE of the channel
estimates, but the complexity is high compared to the
ML or LS estimators. The LS and MMSE method were
compared in and for OFDM systems and the MMSE was
found to outperform the ML in low SNRs. The calcula-
tion of the MMSE estimate requires a large matrix inver-
sion. The complexity of the MMSE estimator can also be
reduced by considering only the high energy channel taps.
Transform domain techniques may also be used for
obtaining the channel estimates for the whole bandwidth.
The inverse FFT transforms the channel frequency re-
sponse into time domain where the low power taps can
be eliminated and the noise reduced channel can be
transformed back to frequency domain with the FFT.
MMSE filtering can also be used to predict the channel
of the current OFDM symbol based on channel estimates
from previous symbols, i.e. time and frequency domain
correlation of the channel frequency response can be
exploited in the channel estimation. For improved per-
formance in MIMO-OFDM systems, the spatial correla-
tion can be included in the MMSE channel estimation.
2.2. Equalization Techniques
In general, there are three categories of equalization
techniques. Our technique is the time–frequency domain
equalization with channel shortening. A time-domain
equalizer is inserted to reduce the MIMO channels to the
ones with the channel length shorter than or equal to the
C
opyright © 2013 SciRes. CN

O. BATAA ET AL.
32
CP length, and then, a one-tap frequency-domain equal-
izer is applied to each subcarrier. When the MIMO channels
are shortened by the time-domain equalizer, residual ICI
and ISI are introduced. They cannot be eliminated by the
subsequent frequency-domain equalizer and, thus, limit
the performance.
This has resulted in new receiver concepts using dif-
ferent equalization techniques. These techniques are
briefly explained below.
 Time-domain equalizer (TEQ)
The time domain equalization (TEQ) is a short FIR
filter at the receiver input that is designed to shorten the
duration of the channel impulse response (L). Thus it
allows a reduction in the guard interval length. Using a
filter with up to 20 coefficients, the effective channel
impulse response of a typical AWGN channel can easily
be reduced by a factor of 10. Different cost functions
such as minimum mean squared error, maximum short-
ening signal-to-noise ratio (SNR), and minimum in-
ter-symbol interference (ISI), and maximum bit rate have
been proposed to design the time domain equalizer
(TEQ).
2.3. Iterative Algorithm
In OFDM, channel estimation can be performed with a
blind or a non-blind technique. The blind channel esti-
mation method does not require the use of training se-
quences or pilot symbols and enables a more efficient use
of the available bandwidth. The channel estimates are
obtained using the statistical properties of the received
data which is collected over a certain time period.
ML equalizer, and can be used to compute the coeffi-
cients of suboptimal but lower-complexity equalizers such
as the minimum mean-squared error (MMSE) linear equa-
lizer (LE). Even though the MMSE-LE can be estimated
directly, having the channel estimates allows us to choose
which equalizer is more appropriate for the channel.
3. Mathematic Extraction for Non Pilot
Channel Estimation
We assume a single UE receiving desired signal from the
serving BS as well as inter-cell interference signals from
neighboring BSs. The BSs from different cells are as-
sumed to be synchronized in time and frequency. The UE
has NR receive antennas which are used for performing
inter-cell interference suppression. Each BS has one an-
tenna to transmit one stream of information.
We define s as the vector that stands for the d pilot
symbols, which are multiplexed with s to form a block of
N=Nd+Np transmitted symbols s. For simplicity, we
X

Figure 1. The block diagram of an OFDM system.
Figure 2. Proposed Iterative model for MIMO receiver.

O. BATAA ET AL. 33
Figure 3. Proposed system model of ICS Repeater without Pilot signal for 3GPP/LTE.
consider unit-energy QPSK with the symbol alphabet αk,
k=1,…,4, which are used for desired signal as well as all
other interference signals. In this work, the receiver is
assumed to have perfect CSI between the serving BS and
the UE. If SIC is applied, the UE also requires knowl-
edge about the interference link, e.g., the channel be-
tween the UE and the interfering BS and the modulation
scheme of the interference signals. Notice that the pilot
symbols here are only used to calculate the beam forming
weights, not for performing channel estimation.
3.1. MMSE Estimator
This effectively equalizes the frequency-selective chan-
nel. First, consider the infinite length filter case: The
output of the equalizer is
ˆ[][][][ ]
nk nkk
kkk
x
ntkqxkn 


 


k  
nj
Is where the equalized channel IR is
nj
j
qf 




The difference between the Tx.ed data and the equal-
izer output is:
ˆ [][] []nxnxn


and the MMSE cost function is:
22
ˆ {| []|}{| [][]|}JEnE xnxn  
Principle of orthogonality:
*
{[][ ]}0,Etnknk  
We can calculate the MMSE equalizer by either mini-
mizing J over w:
22
2
12
0
ˆ {|[] |}{[][] |}
{|[][]|}
|[]([()][])|}
k
k
L
kl
kl
JE nExnxn
Exn tnk
Exnfxl nkvnk






 

 


 





Then Jmin is,
2* **
ˆ {|[ ]|}{[ ][ ]}{[ ]([ ][ ])}
J
En EnnEnxnxn   
Due to the principle of orthogonality,
*
*
min
2*
0
ˆ
{[][]}0 then
{[][]}
{|[] |}{[][]}
1
k
k
jk
k
Enxn
JEnxn
Exn Etnkxn
fb











 
 


3.2. MMSE Calculation
0
ˆ[][] [] []
K
nk k
knk K
x
nqxn qxknk 






nj
where the WMF output/equalizer input is
nj
j
qf 




and the convolution of the equalizer and the equivalent
channel IRs is


O. BATAA ET AL.
34
[][][]
k
k
tnfxnkn





Obviously, the variance of noise is
22
0
k
nk
kk
Nc 



The ISI terms are
[]
nk
kn
Dqxk



For a fixed sequence of information symbols
{[ ]},
j
j
x
xkD D
Then, the probability of error for this sequence is
0
2
0
2
1
()2{() }
()
1
2()
[]
Mj j
j
n
K
k
kK
M
PDPDNq
M
qD
MQ
M
Nnk











Average probability is found by averaging over all
J
D
{} {}
M
Mj j
xj
PPDPx
{}
M
J
PD is dominated by the sequence yielding high-
est
J
D which occurs when x[n]= ±(M-1) and the signs
of x[n]’s match the corresponding {qn}.
*
0
(1)||
j
k
k
DM q



3.3. Error Model
MMSE equalizer aims at minimizing





 





2
2
2
ˆ
ˆ
1
2
T
J EnE xnxn
E exnxn
xn
xn
xn
xnNL

 












Expanding the cost function
 


 



 





 
22
**
ˆ
TT
TT T
HHH
JEexnxnEexnw n
EewFxnwn
xne Fwnw







 

T
Optimum equalizer coefficients are:



*0
1
12 22
0
20
TH
w
THTH H
xx
JwRp
wpReF FFI





 
Substituting back to the MSE term

212
min
1
2222 2
1
22
11
H
xx
TH H
xx xx
TH
JpRp
eF FFIFe
eIFFe











 

 


4. Flowchart of Proposed Signal Processing
Technique without Pilot Signal for ICS
In result section, the simulations of the algorithms de-
veloped two cases. Our proposed design implemented
both cases, which are pilot and non-pilot receiver simu-
lated on MATLAB. Figure 5 shows non-pilot design for
2-2 Tx-Rx MIMO system. This figure included TDI and
FDI based matrix interpolation that requires modulation
Figure 4. Proposed combination signal processing technique
of ICS for 3GPP/LTE.

O. BATAA ET AL.
35
QPSK, 16QAM, 64QAM. The first iterative step ended
to check maximum threshold is larger than Jmin value. If
true no more iterative step, otherwise next iterative step
continue. Proposed methods based two cases of ICS re-
ceiver structure. So we define only one general algorithm,
which is used to any system with ICS block. If system
determined input signal with pilot, then signal go to
matched filtering block, otherwise select to adaptive filter.
5. Conclusions and Future Works
In finally, linear MMSE time-domain equalization tech-
nique has been proposed for general MIMO-OFDM sys-
tems. The added CP at the end of each MIMO-OFDM
symbol converts the linear convolution in the channel
into circular convolution. Simulations have demonstrated
that the proposed MMSE time domain equalizer tech-
nique is effective in suppressing ICI and ISI and robust
against the number of shifts in excess of the CP length.
Finally, in this project the MMSE equalization used to
non-pilot system, LS equalization used to pilot system,
that are mainly considered as proved by any works. It
would be very interesting to extend the ideas of the po-
lynomial approach and transceiver/repeater designs to
new practical system based channel interference cancel-
lation methods.
Figure 5. The symbol error rate comparison of the pro-
posed non-pilot technique with that of a 2Tx−2Rx at the
MMSE receiver.
6. Acknowledgements
This work was supported by the IT R&D program of
KCC/MKE [2010-10035206, The Development of IMT-
Advanced Mobile IPTV core technology], Electronics
and Telecommunications Research Institute of Korea and
Mobicom corporation of Mongolia.
Figure 6. The symbol error rate comparison of the pro-
posed non-pilot and pilot technique with any modulation
order at the receiver.
Table 1. Simulation parameters.
REFERENCES
Parameter Value
Modulation schemes QPSK,16QAM,64QAM
Cyclic prefix Normal
FFT/IFFT block size 1024
Number of iterations 6<
Threshold 1.2
Packet size in symbols 10^3
Channel tape 6
Channel estimator with pilot MLSE using Viterbi algorithm
Channel estimator without pilot MMSE
Channel equalizer with pilot Least Square equalizer
Channel equalizer without pilot Linear MMSE equalizer
Receiver antenna MMSE receiver
[1] S. Mahboob, Interference Cancellation System, Simon
Frazer University, 2009.
[2] J. Ketonen, Equalization and Channel Estimation Algo-
rithms and Implementations for Cellular MIMO-OFDM
downlink; University of Oulu Graduate School; Juvenes
Print, Tampere 2012.
[3] A. Mehmood and W. A. Cheema, Channel Estimation for
LTE Downlink, Blekinge Institute of Technology, 2009.
[4] J. C. Henao, Y. J. Chen, K. Fang, M. Grieger, P. Helbing
and R. Legouable, Advanced Receiver Signal Processing
Techniques: Evaluation and Characterization.
[5] B. M. Donlan, Ultra-wideband Narrowband Interference
Cancellation and Channel Modeling for Communications.
[6] Mr. C. Pirak “Spatial Interference Cancellation and
Channel Estimation for Multiple-input Multiple-output
Wireless Communication Systems,” Doctor Thesis.