Spectral Efficient Blind Channel Estimation Technique for MIMO-OFDM Communications

International Journal of Advances in Applied Sciences (IJAAS)
Vol. 7, No. 3, September 2018, pp. 286~297
ISSN: 2252-8814, DOI: 10.11591/ijaas.v7.i3.pp286-297  286
Journal homepage: http://iaescore.com/online/index.php/IJAAS
Spectral Efficient Blind Channel Estimation Technique for
MIMO-OFDM Communications
Renuka Bhandari, Sangeeta Jadhav
Department Of (E&TC), Dr. D.Y. Patil Institute of Engineering & Technology, Pimpri, Pune-411018
Army Institute of Technology Pune-411015, India
Article Info ABSTRACT
Article history:
Received Jun 9, 2017
Revised Feb 20, 2018
Accepted Mar 11, 2018
With emerge of increasing research in the domain of future wireless
communications, massive MIMO (multiple inputs multiple outputs) attracted
most of researchers interests. Massive MIMO is high-speed wireless
communication standards. A channel estimation technology plays the
essential role in the MIMO systems. Efficient channel estimation leads to
spectral efficient wireless communications. The critics of Inter-Symbol
Interference (ISI) are the challenging tasks while designing the channel
estimation methods. To mitigate the challenges of ISI, we proposed the novel
blind channel estimation method which based on Independent component
analysis (ICA) in this paper. Proposed channel estimation it works for both
blind interference cancellation and ISI cancellation. The proposed Hybrid
ICA (HICA) method depends on pulse shape filtering and ambiguity removal
to improve the spectral efficiency and reliability for MIMO communications.
The Kurtosis operation is used to measure the complex data at first to
estimate the common signals. Then we exploited the advantages of 3rd and
4th order Higher Order Statistics (HOS) to priorities the common signals
during the channel estimation. In this paper, we present the detailed design
and evaluation of HICA blind channel estimation method. We showed the
simulation results of HICA against the state-of-art techniques for channel
estimation using BER, MSE, and PAPR.
Keyword:
Blind channel estimation
Error rates
Independent component
analysis
Interference
MIMO-OFDM
Copyright © 2018 Institute of Advanced Engineering and Science.
All rights reserved.
Corresponding Author:
Renuka Bhandari,
Department Of (E&TC), Dr. D.Y. Patil Institute of Engineering & Technology,
Pimpri, Pune-411018, Army Institute of Technology Pune-411015, India.
Email: bhandarirenuka33@gmail.com
1. INTRODUCTION
Since from last decade, continues growth in the requirements for higher data rates on constrained
resources and available bandwidth. This demand for higher data rate is resulting in the significant attention of
researchers to initiate the working towards future wireless communications [1]. In future wireless
communications, the resource utilization using MIMO (Multiple Input Multiple Output) is better than other
methods [2] [3]. Hence massive MIMO is the crucial technology for future communication systems like 5G.
The MIMO combined with transmission method like OFDM (orthogonal frequency division multiplexing)
also the CDMA (code division multiple access). The critical part for MIMO systems is the design of efficient
channel estimation to improve the overall communication performance and minimize the error rates.
Therefore, several channel estimation methods designed in literature. The methods reported in [4-7] depend
on ICA and wavelet-based channel estimation for MIMO wireless systems.
The channel estimation techniques of MIMO-OFDM divided into the three types such as semi-blind,
blind channel, and training based evaluation methods [4]. In training based techniques, plan known training

IJAAS ISSN: 2252-8814 
Spectral Efficient Blind Channel Estimation Technique for MIMO-OFDM… (Renuka Bhandari)
287
samples used to perform the extensive channel estimations. The least square (LS) and MMSE are the cases of
training based channel assessments techniques. In blind channel estimation techniques, SOS (second-order
stationary statistics) or HOS techniques used for transporting maximum spectral efficiency. In the semi-blind
channel estimation, the main order statistics utilized. The semi-blind channel estimation technique
consolidated attributes of training based and blind based channel estimation are interconnected with MIMO-
OFDM respectively. The semi-blind strategy sets aside longer time for the channel estimation and
subsequently increases the overall communication costs. The blind channel estimation technique is superior
to other two types of channel estimation methods. Thus in this paper we focused on designing the novel blind
channel estimation method.
In OFDM systems, block-based symbol transmission method used; hence for 4G communications,
the technique of block-based and iterative channel estimation introduced. For future wireless communication
systems, same methods will be applicable. Therefore, it motivates to use blind channel estimation to obtain
the initial symbol estimation and furthermore it is utilized the initial symbol evaluation to increases the more
considerable dedication channel estimations. This channel estimation process repeated iteratively with soft
information exchange to enhance the performance of channel estimation and data symbol estimations [8].
Further, consequently prompts the fast increment in low mobility based application and hence influence to
design the blind channel estimation which requires a large number of samples based on quasi-static
channel conditions.
The current investigations exhibited the SOS and HOS based blind channel estimation techniques.
However, none of such process equipped for interference signals cancelation for MIMO-OFDM [11]. An
interference signal generates the blind channel or scalable client in MIMO-OFDM wireless systems. The
HOS relies upon ICA strategy for interference cancelation in MIMO systems, yet not expressly outlined and
tended to for blind channel estimation. Moreover, ISI (inter-symbol interference) likewise have the critical
effect on the performance of blind channel estimation and spectral efficiency. In this paper, we proposed
distinctive spectral efficient channel estimation based on modified ICA approach called HICA. The proposed
HICA method designed to mitigate the challenges of blind interference, ISI cancellations, and error
minimizing efficiently using the pulse shape filtering, Kurtosis, HOS functions, and ambiguity removal. We
present the extensive simulation analysis for proposed blind channel estimation method using BER, MSE,
and PAPR performance metrics. The PAPR evaluation for MIMO systems recently introduced in [13]-[18],
hence we conducted the PAPR performance investigation in this paper. In part II, the brief literature of past
works exhibited, In part III, the proposed method for blind channel estimation discussed. In part IV, the
simulation outcomes illustrated. The conclusion based on the results presented in section V.
2. RELATED WORK
The channel estimation is the crucial phase of wireless communications; designing channel
estimation for MIMO gains the significant attention since from last decade. We discussed some of the recent
methods presented for channel estimation as well as PAPR analysis.
The blind channel estimations by using the repeated index approach reported in [8]. They proposed
subspace blind channel assessments design relies upon the rehashed list with the same yield from separates to
ultimately strategies with a few quantities of symbols. Our work is not the same as this work as we in light of
altered ICA way to deal with blind interference cancellations and error minimization.
In [9], another blind channel estimation technique in the MIMO-OFDM system with the equilateral space-
time piece code investigated. This approach composed the new weighted covariance matrix of data recognize
in the direction to obtain the errors in the code. All the non-scalar ambiguities are resolved using method
reported in [9], but failed to address the challenges of common signal estimation and prioritizing the
estimated common signals.
Further, the SOS based blind channel estimation method reported in [10] based on the novel blind
recursive approach in MIMO-OFDM systems. The subspace based tracking designed for the fast varying
wireless channels. To get SOSs update fast, data used from time and frequency domain due to frequency
correlation of this wireless channels.
In [11], the blind channel estimation technique using the SOS and subspace models for the MIMO-
OFDM system reported. Authors efficiently utilized the null space produced by OSTBC. The method
designed with the goal of minimum received blocks. The experimental results demonstrated on the single
input and single output only. Our approach depends on multiple inputs and multiple output antennas for the
channel estimation.
Another recent approach for channel estimation reported in [12]. The blind channel estimation
techniques utilized for the LTE (Long-Term Evaluation) networks. The channel estimation performed using
the wavelet transform with ICA. The wavelet transforms used preceding ICA channel estimation to diminish

 ISSN: 2252-8814
IJAAS Vol. 7, No. 3, September 2018: 286 – 297
288
the errors in approaching signals. The denoising approach was intended to deal with the blind interference
cancelation. It is exploiting the ICA. The ISI cancelation techniques are not utilized in this technique.
In this paper, we proposed the novel channel estimation strategy which differs from previous works
of channel estimation discussed above. The proposed blind channel estimation method based on ICA aims to
ISI and blind interference cancelations. At first, to mitigate the noise interference and ISI effect, we designed
the pulse shape filtering to apply on modulated signals. At channel estimation phase, the source of signals
estimated using the kurtosis values, then common signals are detected and then prioritize common estimation
signals using 3rd and 4th order HOS. The lightweight method used for the ambiguity removal.
Additionally, we presented the PAPR evaluation of proposed blind channel estimation technique for
MIMO-OFDM systems in this paper. The PAPR analysis of space-time block-coded (STBC) MIMO-OFDM
method for the 4G wireless networks evaluated in [13]. The numbers of techniques introduced to decreases
PAPR in the (STBC) MIMO-OFDM system such as SLM, clipping and filtering and partial transmit
sequence. Their simulation outcomes claim that clipping and filtering delivered effective PAPR reduction
than the others methods and only SLM technique conserves the PAPR reduction in reception part of the
signal. We studied some recent works reported for PAPR performance evaluation in [14]-[18].
3. SYSTEM MODEL
The system model for proposed HICA channel estimation is showing in Figure 1 for
MIMO-OFDM systems.
Figure 1. Proposed MIMO-OFDM System Model
In proposed system model, at transmitter data X from users is randomly generated and then
forwarded to modulation process on each symbol. IFFT and CP operations performed at each transmitter. To
mitigate the effect of ISI cancellation, we applied the pulse shape filtering on IFFT data. Once symbols are
ready to transmit over the Rayleigh channel through transmitter’s antennas towards receiver’s antennas, the
measurement of PAPR performs on each symbol. At the receiver side, after the pulse reshaping, CP removal
and FFT operations, proposed HICA method for efficient channel estimation and signal detection performed.
Figure 2 showing the complete functionality of HICA block. After estimation, the reverse operations applied
at the receiver side to get the original data at the receiver side. The modulation and demodulation are
performed by using QPSK technique. Let’s consider S is the sample of signal, M is a transmitted signal
which include the many numbers of samples. An is an attenuation for nth channel. The Factor of An is
complex number. The authors investigates the pulse shaping for the deleting the ISI impacts to reduces the
error rates and optimize the spectral efficiency. The HICA method for blind channel estimation consists of
steps like source estimation, common signal estimation; prioritize the signals, ambiguity removal, and final
signal detection at the receiver side. The functionality of complete architecture presented in the form of
algorithms below.

IJAAS ISSN: 2252-8814 
289
Figure 2. HICA channel estimation phases
Algorithm 1: MIMO Transmitter
Input:
X; random data
n; number of transmitter and receivers
Output:
T1…Tn; transmitted data
1. Generate random data X
2. Representation of X in different symbols as per the number of
transmitter and receiver antennas X1…Xn
3. Apply the modulation on X1…Xn.
4. IFFT on X1…Xn
5. Apply the Pulse shaping on input symbols using Eq. (1)
X1…Xn = Pf
* (X1…Xn) (where Pf
is pulse shaping filter)
6. Cyclic Prefix on X1…Xn
7. Measure PAPR X1…Xn using Eq. (11).
8. Transmit data X1…Xn through Rayleigh channel
Algorithm 2 is showing the process at receiver side.
Algorithm 2: MIMO Receiver
Input:
Yr = [T1…Tn]; set of transmitted signals
Output:
X` = [X`1…X`n];set of received signals
1. Pulse Reshaping
T1…Tn =(1/Pf
* (T1…Tn)) (where Pf
is pulse shaping filter)
2. FFT on estimated signals T1…Tn
3. CP Removal T1…Tn
4. Channel estimation using algorithm 3
T`1…T`n = HICA (T1…Tn)
5. X`= Demodulation T`1…T`n
6. Return X`
The algorithm 3 shows the process for proposed channel estimation method called HICA. The
proposed channel estimation approach is the iterative process to optimize the channel estimation performance
and minimize the errors of estimation.

 ISSN: 2252-8814
290
Algorithm 3: Channel Estimation
Input:
Yr = [T1…Tn]; set of received signals
Output:
ŜiF = [T`1…T`n]; set of estimated signals
1. Initialization HICA iterations in IT and set it = 0;
2. Random W initialization
3. Defining the objective function Jold ← J (W).
4. Gradient computation of objective function using Eq. (4).
5. W updating according to negative gradient direction, W ←W
− µJw.
6. W normalization according to unitary constraint, W ←W / W
.
7. If (Jold − J (W) < ε )
8. then go back to step 2
9. end if
10. Set of signals estimation ŝ [n] = WH
x Yr (where, Yr is the
set of received signal)
11. Form every estimated signal ŝ [n] as vector which
represented by Vit[n]
12. If it <1 goes to step 2 else, continue.
13. Find the set of the common vectors for all runs of algorithm
up to itth
run.
14. If (no common vectors)
15. Goto step 2
16. Else
17. break
18. end if
19. Apply Eq. (5) to prioritize the common estimated signals
with J (ŝit).
20. Selection of desired signals (m) with largest J (ŝit) in order
performs the blind interference cancellation.
21. Ambiguity Elimination on estimation common signals using
Eq. (9).
22. ŜiF = [T`1…T`n];
23. Return estimated signals ŜiF
Algorithm 3 HICA
In the proposed method, we first enable the pulse shaping techniques in each user symbol in the
order to minimize the internal symbol interference at the transmitter side. Pulse shape is simple ISI
cancellationtechniques. We utilized the square root increased cosine filter in sampling region to perceive the
pulse shaping filter. Table 1 shows the symbols used by proposed channel estimation method.
Table 1. List of Symbols
Symbol Meaning
gt (t) Transmitter side pulse shape filter for each symbol
gr (t) Receiver side pulse shape filter for each symbol
T Number of transmitters
R Number of Receivers
L Number of taps
iR ith
receiver
iT ith
transmitter
W Separation Matrix
v non-gaussianity for random variable
K Kurtosis

IJAAS ISSN: 2252-8814 
291
The composite channel is represented as T * R with matrix H (t). The (iR, iT) channel using pulse
phase filtering is represented by:
ℎ𝑖 𝑅′ 𝑖 𝑇(t) = ℎ𝑖 𝑅′ 𝑖 𝑇′ 𝑐(𝑡) ∗ 𝑔𝑡(𝑡) ∗ 𝑔𝑟(𝑡) (1)
Where hi R' iT'c(t)is the (iR, iT) element of H (t). Here the channel can be represented as the L tap
FIR filters array for blind channel estimation [19]. After the designing of pulse shaping filter which we
applied to time domain (i.e., after IFFT operation) and perform the subsequent tasks of transmitter side. The
data is transmitted from the transmitter’s antennas towards the receiving antennas via Rayleigh channel.
At receiver end, the channel estimation is performed after the pulse re-shaping, FFT and cyclic
prefix removal operations (Algorithm 2). For the channel estimation, we proposed the algorithm 3
called HICA.
As observed in algorithm 3, initially signals sources from the receiver analysis mixture are
determine by using the partition of matrix W. The proposed channel estimation algorithm is based on the
iterative process for excellent channel estimation, thus at each iteration, the order of initial signal estimation
is varying due to the random initialization of separation matrix. But, if the important information discovered
in estimated signals, then it will be utilized for the iteration. The identified important information called as
common estimated signals. The HOSs 3rd and 4th order derivatives applied to the common estimated signals
to prioritize them. The outcome of HOS is the selected desired signals (m) by canceling the interference
components from
the signals.
The objective function based on W introduced based on the principle of non-gaussianity maximizing
observation signals. The outcome of non-gaussianity for random data (v) containing the complex data which
is measured by the operator Kurtosis K[s]:
𝐾 [𝑠] = 𝐸[|𝑣|4] − 2(𝐸[|𝑣|2])2
− 𝐸[𝑣. 𝑣]𝐸[𝑣∗
𝑣] (2)
Where (.)*
is represents the complex conjugate. Before prioritising the common estimation signals,
the Kurtosis function is applied to discover the common estimation signals. The signal estimation using W
based on the minimization of objective function J (W) within the unitary constraint
The estimation of W is performed by the minimization of J (W) objective function within the WW H
= IR unitary constraint. The unitary constraint used to alleviate the negative outcomes of kurtosis value for
various modulation methods. Thus, the objective function using the estimation signals ŝ[𝑛] kurtosis values
represented by:
𝑊 = {min 𝑊 𝐽(𝑊) = ∑ 𝐾[𝑀
𝑗=1 ŝ[𝑛]] (3)
The minimization of objective function J (W) satisfied using the gradient computation
represented by:
Jw =
∂J(W)
∂W
= K(WH
ŝ[n])[E{ŝ[n])3}] (4)
Where (w) is one of the vectors from W separation matrix. According to the constraint for the
objective function (Jw), the object function gradient complemented projecting W over the interval after each
step performed by dividing the W by its norm.
The iterative execution using different random initialization matrix W performed to evaluates the
general signals at the receiver side. The two estimated signals of various computations are examine as
different if the spectral angle mapper (SAM) of their vectors vi[n] and vj[n] is beyond the threshold value ε.
The general signal estimation results used to prioritize the signals based on the 3rd
(Eq. 5) and 4th
(Equation
5) higher order statics.
𝐽�ŝ 𝑞� = �
1
12
� [𝑄 𝑞
3
]2
+ (
1
48
)[𝑄 𝑞
4
− 3]2
(5)
Where, 𝑄 𝑞
3
= 𝐸�ŝ 𝑞
3
� = �
1
𝑇
� ∑ (ŝ 𝑞
𝑇
𝑛=1 [𝑛])3
is the 3rd
order of statistics and
𝑄 𝑞
4
= 𝐸�ŝ 𝑞
4
� = �
1
𝑇
� ∑ (ŝ 𝑞
𝑇
𝑛=1 [𝑛])4
is the 4th
order of statistics of estimated signals, and q is the
execution index.

 ISSN: 2252-8814
292
Once the signal estimated common signals prioritise, the next step of ambiguity removal applied to
suppress the unwanted components from the estimated signals. Scientifically the estimated ŝi [n] is not
matching to the original transmitted s[n] due to presence of ambiguity matrix A in ŝi [n]. This can be
represented as:
ši [n] =A Ⅹ ŝi [n] (6)
Two indeterminacies forming the A as:
A = P×D (7)
Where P is the permutation ambiguity matrix and (D) is phase rotation ambiguity matrix.
The removal of such ambiguities performs to reduce the error rates in proposed channel estimation
techniques. The ambiguities suppressed by multiplying the estimated signals and remove from the original
signals in first step represented by𝐿 𝐹:
𝐿 𝐹 = arg min 𝐿ℰ𝐺 ∥ 𝑠[𝑛] − ŝ[𝑛] 𝑥 A ∥2
(8)
After the 𝐿 𝐹 computation, the multiplying it with the estimated signals as:
ŝiF [n] = LF x ŝ[n] (9)
For the performance evaluation along with BER and MSE, we computed the PAPR for MIMO
systems. The computation of PAPR is done as per given below:
The PAPR of the signal, x (t), is then provided the peak prompt power to the average power,
following are shown the formula,
PAPR =𝑙𝑙𝑙10 ∗
𝑚𝑚𝑚0≤𝑡≤𝑇|𝑥(𝑡)|2
𝐸[|𝑥(𝑡)2|]
(10)
Where E [.] is the predicated operator. From the central limit formulas, for most values of N, the X
(t) is a real and imaginary values. The PAPR computation showing in Eq. (10) is applicable for Single Input
Single Output (SISO) systems, therefore in MIMO systems; same formula is utilized to compute the PAPR
for each transmitter given by:
𝑃𝑃𝑃𝑃 𝑀𝐼𝐼𝐼−𝑂𝑂𝑂𝑂=𝑚𝑚𝑚𝑖≤𝑖≤𝑛 𝑡
𝑃𝑃𝑃𝑃𝑖 (11)
Where PAPRi denotes the PAPR of transmit ith
antenna.
4. SIMULATION RESULTS AND DISCUSSION
The performance analysis of the innovated channel estimation technique for MIMO systems has
been shown in this section against the state-of-art techniques. The simulation performed for this study under
different network conditions such as varying number of sender antennas and accepted antennas. Table 2
shows the simulation parameters used to evaluate the proposed method. For comparative study, we used the
conventional ICA technique [20] and recent wavelet denoising of ICA (WD-ICA) [12] method.
Table 2. MIMO-OFDM Simulation Parameters
FFT Size 64, 128, 256
Transmit Antennas 2, 4
Receive Antennas 2, 4
Block size 8
Sub band size 20
SNR level 0:5:30
Number of iterations 10
Blind Estimation Method ICA [20], WD-ICA [12] and Proposed HICA
Modulation Technique QPSK
Number of subcarriers 256
Filter type Pulse shaping filter
Oversampling factor 4
Channel Type Rayleigh

IJAAS ISSN: 2252-8814 
293
To check the reliability of proposed channel estimation method, we performed the simulations based
on (1) changing the number of transmitter and receiver antennas such as 2x2 and 4x4 and (2) varying FFT
sizes such as 64, 128, and 256. The performance measured in three parameters such as BER, MSE, and
PAPR to check the trade-off between spectral and PAPR efficiency for each investigated method.
4.1. 2x2 Transmit/Receive Antennas
This section presents the simulation results for 2x2 (2-transmit and 2-receiving antennas) for MSE,
BER and PAPR performance metrics for each varying FFT size.
The performance of MSE and BER is significantly improved by proposed HICA channel estimation
method due to iterative error minimization approach using the Kurtosis and HOS functions for the common
channel estimations. However, the performance PAPR is not promising for all the methods. In an average of
64 FFT size, WDICA method showed minimum PAPR as compared to ICA and proposed HICA method.
The Figures 6 to 8 shows the results using the FFT size 128. The PAPR ratio increased as the FFT
size increases. The Figures 9 to 11 shows the BER, MSE and PAPR results using the FFT size 256. The
proposed channel estimation method was demonstrating the superior performance against the state-of-art ICA
based methods for the channel estimation using varying FFT size. However, the PAPR results show the
proposed channel estimation method more PAPR rate as compared to conventional ICA and WD-ICA
methods. Figure 11 shows the FFT size having the impact of PAPR performance, as the FFT size increases
the PAPR is also growing. The result of BER and MER proves the proposed channel estimation techniques
are the dependable techniques MIMO- OFDM systems. The impacts of FFT size on BER and MSE is
showing in Tables 1 and 2 respectively.
Increasing FFT size leads to minimizing the BER and MSE values for all the methods as the number
of subcarriers increases to transmit each symbol. However, the PAPR performance has the reverse impact as
shown in table 3. Increased FFT size is showing the worst PAPR performance for all the methods. For each
FFT size, the average PAPR result of WD_ICA having minimum PAPR rate. In this sub-section, we
evaluated the impact of varying FFT size on BER, MSE, and PAPR performances using 2x2 transmit/receive
antennas.
Figure 3. 2x2 BER analysis using FFT size 64 Figure 4. 2x2 MSE analysis using FFT size 64
Figure 5. 2x2 PAPR analysis using FFT size 64 Figure 6. 2x2 BER analysis using FFT size 128

 ISSN: 2252-8814
294
Figure 7. 2x2 MSE analysis using FFT size 128 Figure 8. 2x2 PAPR analysis using FFT size 128
Figure 11. 2x2 PAPR analysis using FFT size 256
Table 3. Average BER Performance Analysis (2x2)
FFT Size ICA WDICA HICA
64 0.0445 0.0068 0.0032
128 0.0408 0.0062 0.003
256 0.0368 0.0065 0.0027
Table 4. Average MSE Performance Analysis (2x2)
64 0.0779 0.0529 0.0064
128 0.0385 0.0261 0.0032
256 0.0198 0.013 0.0016
Table 5. Average PAPR Performance Analysis (2x2)
64 13.4883 13.4443 13.4823
128 16.2325 16.1961 16.2552
256 19.0180 19.010 19.0375

IJAAS ISSN: 2252-8814 
295
4.2. 4x4 Transmit/Receive Antennas
After the investigation of FFT size on channel estimation methods, to claim the reliability and
scalability of proposed channel estimation, we used 4x4 transmit/receive antennas for MIMO-OFDM
communications. This section presents the simulation results for 4x4 (4-transmit and 4-receiving antennas)
results for MSE, BER and PAPR performance metrics for each varying FFT size.
The varying transmitting and receiving antennas not having the significant impact on HICA
performance. However, the HICA performance is significantly improved as compared to state-of-art methods
using 64 FFT size (Figure 12 and Figure 13), using 128 FFT size (Figure 15 and Figure 16) and using 256
FFT size (Figure 18 and Figure 19). The channel estimation method should be lightweight and spectral
efficiency. The results of MSE and BER show that HICA the spectral efficient way for MIMO-OFDM
systems. The PAPR ratio is increased as the FFT size increases using 4x4 antennas scenario as well.
The effect of increasing the antennas on MSE, BER and PAPR performances are review in Tables 6,
7 and 8 accordingly.
The increasing antennas from 2x2 to 4x4 does not show the significant impact of performances,
except little increase in MSE and BER rates as the compared 2x2 scenario. The purpose of evaluating the
channel estimation methods using different antenna sizes is to check the reliability of the proposed method.
The results show that for all the cases of varying FFT size and antenna scenarios, suggested HICA method
delivered efficient channel estimation performance as compared to state-of-art methods. There is little
decrease in PAPR rate using 4x4 situations as compared to 2x2 scenarios. From all the results, the
performance of MSE and BER are the contrast to PAPR performance for all the channel estimation methods.
Achieving the trade-off between PAPR and error rates is one of the leading research problems for MIMO-
OFDM systems. Finally, Table 9 shows the average processing time performance for each channel estimation
method. The results of processing time demonstrate the proposed method has very less computation overhead
as differentiates to both state-of-art channel estimation techniques. The simulation conducted on the I5
processor with 4GB RAM.
Figure 14. 4x4 PAPR analysis using FFT size 64 Figure 15. 4x4 BER analysis using FFT size 128

 ISSN: 2252-8814
296
Figure 16. 4x4 MSE analysis using FFT size 128 Figure 17. 4x4 PAPR analysis using FFT size 128
Figure 20. 4x4 PAPR analysis using FFT size 256
Table 6. Average BER performance analysis (4x4)
64 0.0525 0.0081 0.0037
128 0.045 0.0071 0.0033
256 0.04 0.0062 0.0028
Table 7. Average MSE performance analysis (4x4)
64 0.0788 0.0528 0.0066
128 0.0394 0.0261 0.0033
256 0.0195 0.013 0.0016
Table 8. Average PAPR performance analysis (4x4)
64 13.2 13.18 13.21
128 16.17 16.16 16.17
256 18.99 18.98 18.97
Table 9. Average processing time performance
Method ICA WD-ICA HICA
Time (Seconds) 49.55 47.2 36.23

IJAAS ISSN: 2252-8814 
297
5. CONCLUSION AND FUTURE WORK
In this research paper, we investigate the novel hybrid blind channel estimation technique based on
modified ICA technique using pulse shape filtering approach. The HICA effectively utilized the Kurtosis and
HOS features to estimate and prioritize the common signals respectively. The idea of ambiguity removal
helps to reduce the error rates further. The proposed HICA method designed and evaluated with conventional
ICA and recent WD-ICA regarding BER, MSE and PAPR results. HICA method is showing the BER, and
MSE performances significantly improved as compared state-of-art methods with the different antenna and
FFT sizes using the QPSK modulation methods. However, the PAPR performance is not efficient using the
HICA technique. HICA method failed to reduce the PAPR rate against the state-of-art methods. For future
work, it will be interesting to try and check PAPR reduction methods with HICA to optimize the PAPR
performance as well.
REFERENCES
[1] C. Shin, Jr., R. W. Heath, and E. J. Powers, “Blind channel estimation for MIMO-OFDM systems,” IEEE Trans.
Veh. Technol., vol. 56, no. 2, pp. 670–685, Mar. 2007
[2] Yufei.jiang, xuzhu, enggeelim, yihuang. Orthogonal Sequences Based Multi-CFO Estimation and Semi-Blind ICA
Based Equalization for Multiuser Comp Systems. Computer Science &Information Systems. 2012; 9(4): 1385-1406.
[3] J Gao, X Zhu, AK Nandi. Independent Component Analysis for Multiple-Input Multiple-Output Wireless
Communication Systems. Signal Processing. 2011; 91(4): 607-623.
[4] Chiu Shun Wong, Dragan Obradovic. Independent Component Analysis (ICA) for Blind Equalization of Frequency
Selective Channels. 14thIEEE workshop Neural Networks for Signal Processing. 2003; 419-428.
[5] Obradovic D, Madhu N, Szabo A, Wong CS. Independent Component Analysis for Semi-Blind Signal Separation
in MIMO Mobile Frequency Selective Communication Channels. IEEE International Joint Conference on Neural
Networks (IJCNN). 2004; 1: 53–58.
[6] M Sifuzzaman, MR Islam, MZ Ali. Application of Wavelet Transform and its Advantages Compared to Fourier
Transform. Journal of Physical Sciences. 2009; 13: 121-134.
[7] Karanpreet Kaur. Disceret Wavelet Transform based OFDM System using Convolutional Encoding. Master thesis.
Patiala: Thapar university; 2014
[8] Xiangli Shao, Jian Chen and Yonghong Kuo, "Blind Channel Estimation for MIMO-OFDM Systems Based on
Repetition Index", 2011 International Conference on Internet Computing and Information Ser-vices, IEEE, 2011
[9] Nima Sarmadi, Marius Pesavento, "Closed-Form Blind Channel Estimation In Orthogonally Coded Mimo-Ofdm
Systems: A Simple Strategy To Resolve Non-Scalar Ambiguities", 2011 IEEE 12th International Workshop on
Signal Processing Advances in Wireless Communications.
[10] Chao-Cheng Tu and Benoît Champagne, "Blind Recursive Sub-space-Based Identification of Time-Varying
Wideband MIMO Channels", IEEE Transactions On Vehicular Technology, VOL. 61, NO. 2, FEBRUARY 2012
[11] Jian-Da Jiang, Tzu-Chiao Lin, and See-May Phoong, "New Sub-space-Based Blind Channel Estimation for
Orthogonally Coded MIMO-OFDM Systems", 2014 IEEE International Conference on Acoustic, Speech and Signal
Processing (ICASSP).
[12] Gamal Mabrouk Abdel-Hamid, Reham S. Saad, "Blind Channel Estimation Using Wavelet Denoising of
Independent Component Analysis for LTE", Indonesian Journal of Electrical Engineering and Computer Science
Vol. 1, No. 1, January 2016.
[13] Leila Sahraoui, Djmail Messadeg, Nouredinne Doghmane, "Analyses And Performance Of Techniques Papr
Reduction For Stbc Mimo-Ofdm System In (4g) Wireless Communication", International Journal of Wireless &
Mobile Networks (IJWMN) Vol. 5, No. 5, October 2013
[14] P.Sunil Kumar, M.G.Sumithra, E.Praveen Kumar, "Performance Analysis of P APR reduction in STBC MIMO-
OFDM System", 2013 Fifth International Conference on Advanced Computing (ICoAC)
[15] Ying-Che Hung and Shang-Ho (Lawrence) Tsai, Senior Member, "PAPR Analysis and Mitigation Algorithms for
Beam forming MIMO OFDM Systems", IEEE Transactions On Wireless Communications, Accepted For
Publication, 2014
[16] S. Sujitha, R. Ramachandran, "Performance Analysis of PAPR Reduction in MIMO OFDM System Using
Modified Constant Modulus Algorithm", IJAREEIE, Vol. 3, Special Issue 2, April 2014
[17] Sadhana Singha, Arvind Kumarb, "Performance Analysis of Adaptive Clipping Technique for Reduction of PAPR
in Alamouti Coded MIMO-OFDM Systems", 6th International Conference on Advances in Computing &
Communications, ICACC 2016, 6-8 September 2016, Cochin, India
[18] Gourav Misra, Arun Agarwal, "A Technological Analysis and Survey on Peak-to-Average Power Reduction
(PAPR) in MIMO-OFDM Wireless System", International Conference on Electrical, Electronics, and Optimization
Techniques (ICEEOT) – 2016
[19] Feng Wan, Wei-Ping Zhu and M.N.S. Swamy, “Semi-Blind Channel Estimation of MIMO-OFDM Systems with
Pulse Shaping”, 2008 IEEE International Symposium on Circuits and Systems
[20] LIU Haiyuan, SUN Jiancheng, “Blind MIMO-OFDM Channel Estimation Based on ICA and KRLS Algorithm,
“2009 5th International Conference on Wireless Communications, Networking and Mobile Computing.

Spectral Efficient Blind Channel Estimation Technique for MIMO-OFDM Communications

More Related Content

What's hot (20)

Similar to Spectral Efficient Blind Channel Estimation Technique for MIMO-OFDM Communications (20)

More from IJAAS Team (20)

Recently uploaded (20)

Spectral Efficient Blind Channel Estimation Technique for MIMO-OFDM Communications