Geneti Algorithm Optimization Tool For Channel Estimation and Symbol Detection in Mimo-OFDM Systems
Geneti Algorithm Optimization Tool For Channel Estimation and Symbol Detection in Mimo-OFDM Systems
Geneti Algorithm Optimization Tool For Channel Estimation and Symbol Detection in Mimo-OFDM Systems
K.I.T.S. Ramtek
Apoorvaagrawal4271@gmail.com
Abstract
The quality of wireless media is described by three parameters. These parameters are
its transmission range, transmission rate and reliability. In the conventional OFDM
systems one parameter can be increased on the cost of decreasing other two parameters.
However by combining MIMO with OFDM systems, all the three parameters can be
improved simultaneously. Symbol detection and channel estimation are the two essential
tasks of MIMO-OFDM system. These tasks can be excellently achieved by various other
recently developed algorithms such as maximum likelihood (ML) detector, LMS, RLS etc.
All these algorithms face a common problem of robustness. Also the complexity of these
algorithms is very high in the system with large number of transmitters and receivers and
having large constellation size. Therefore, we are using the NLMS estimator. But it
doesnt provide the optimal solution. Genetic algorithm has the advantages of
significantly less computational complexity, greater robustness and is closer to the
optimal solution. Hence in this paper we are using Genetic algorithm based NLMS
estimator to accomplish these tasks and to achieve results near to optimal solution.
Comparisons between the results obtain from GA optimized NLMS estimator and plane
NLMS estimator has been shown for better understanding.
Keywords: Genetic algorithm (GA), MIMO-OFDM systems, symbol detection, channel
estimation
1. Introduction
Nowadays, wireless communication system has received a great attention because of
rapidly increasing demand for high data communication. Therefore various wireless
digital communication techniques have been developed to fulfill our day to day
necessities. Also the services which are using wireless technology are being common. So
the bandwidth of wireless link is decreasing rapidly. To fulfill these demands, high-speed
wireless communication and higher network capacity are required. The MIMO-OFDM
system which is a combination of Pros of MIMO and OFDM systems is currently the
best solution to meet these requirements [1]. Also, a significant capacity increase has
been provided for OFDM systems by combining them with multiple-input, multipleoutput (MIMO) technology in many communication systems, such as WLAN,
HIPERMAN, and 4G wireless cellular systems [2].
There are various existing techniques for symbol detection and channel estimation of
MIMO-OFDM systems. A detector with parallel partial candidate-search algorithm is
described for estimation but it has greater computational complexity [3]. Recently, in a
Turkish research Differential evolution algorithm is used for symbol detection in MIMOOFDM system. It has less computational complexity but it lacks robustness [4]. A
recently researched NLMS algorithm for channel estimation and symbol detection in
MIMO-OFDM system is better than the above two algorithms as it is robust and the
complexity is low. In NLMS algorithm, step size is used to repetitively update the
channel. Hence its a robust technique. But it doesnt provide the optimal solution.
Genetic algorithm has the advantages that it is closer to the optimal solution [4]. Hence a
genetic algorithm based NLMS estimator is being developed as an optimization tool for
channel estimation and symbol detection.
The structure of this paper is as follows: Section II describes the MIMO-OFDM
system and its parameters. Section III discusses the techniques and algorithm used for
symbol detection and channel estimation. Section IV consists of simulation result.
Finally, conclusion in Section V.
46
*S/P: Serial to parallel convertor *FFT: Fast Fourier transform *MIMO: Multiple input
multiple output *P/S: parallel to serial convertor
Specifications
BPSK, QPSK, 8QAM, 16QAM
STBC
AWGN, Rayleigh, Racian
64
Zero level guard period
47
48
49
complex multimodal problems with a frequent change in nature the genetic algorithm is
the best choice for optimization.
If the stopping criteria are not met, multiple weights are selected from current
population in order to form new weights. The selection process chooses weights based
upon their fitness score where higher scoring weights have a better chance to be selected.
Once the selected weight have been chosen, they are modified by mutation, a random bit
flip or crossover, which combines two chromosomes into one. The new weight is used in
the next iteration of the algorithm.
In this paper we are developing a tool based on genetic algorithm which will provide
optimal solution of the results from NLMS detector for channel estimation.
4. Simulation Results
MIMO-OFDM system is implemented under Matlab environment. Channel estimation
has been done first by using plane NLMS algorithm and then by GA based NLMS
estimator. The simulation parameters of MIMO-OFDM systems and channel estimation
are given in Table 1 and 2 respectively. We have assumed to do perfect channel
estimation. IFFT size is set to 64 in the MIMO-OFDM system with QPSK modulation.
The output of NLMS method is then applied to GA optimization tool. The parameters
which have been considered is shown in Table 2. Figure 4 shows the performance of
NLMS estimator and Figure 5 shows the BER versus the SNR of the NLMS detector for
a MIMO-OFDM system.
While, Figure 6 shows the performance of GA based NLMS estimator and Figure 7
shows the BER versus the SNR of the GA based NLMS detector for a MIMO-OFDM
system.
Following are the parametric specification which has been considered during
simulation:
Table 2. Basic Simulation Parameters Which Has Been Used
PARAMETER
No. of bits transmitted
Frequency
Amplitude
Channel used
SNR
Modulation type
50
VALUE
110 bpm
8000
50
AWGN
10
BPSK
2000
Binary Data
4000
6000
1
0.5
0
4
4
x 10
Channel Data, Symbols:589413
100
0
-100
10
10
15
5
x 10
Received Data (over channel)
0.1
0
-0.1
x 10
Received Bits, SNR:15
1
0.5
0
15
x 10
Tx PSNR:5.60 dB, Error:0.0182
400
200
0
2000
x 10
Est PSNR :11.81 dB, Error:0.0000
1000
0
-1000
6000
0
2000
4000
6000
4000
-6
10
BER
X: 10
Y: 1.762e-007
-7
10
-8
10
10
11
12
13
14
SNR
Figure 5. The BER versus the SNR of the NLMS Detector for a MIMO-OFDM
System
From the above graph, the BER for SNR= 10 is near to 10^-7.
51
Now, the simulation results by apply GA based NLMS estimator are as follows:
Input Data
400
200
0
2000
Binary Data
4000
6000
1
0.5
0
4
4
x 10
Channel Data, Symbols:589413
100
0
-100
0.1
0
-0.1
10
10
15
5
x 10
Received Data (over channel)
x 10
Received Bits, SNR:15
15
5
x 10
Rx PSNR:5.53 dB, Error:0.01812335
400
200
0
0
2000
4000
6000
1
0.5
0
4
4
x 10
Est PSNR :13.15 dB, Error:0.00000010
500
0
-500
0
2000
4000
6000
-6
BER
10
-7
X: 10
Y: 6.993e-008
10
-8
10
10
11
12
13
14
SNR
Figure 7. The BER versus the SNR of the GA based NLMS Detector for a
MIMO-OFDM System
From the above graph, the BER for SNR= 10 is near to 10^-8.
From the above simulation results, it can be seen that both PSNR and BER has been
comparatively improved by using GA optimized NLMS estimator.
5. Conclusion
We have presented implementation of plane NLMS estimator and GA optimized
NLMS estimator on MIMO-OFDM systems. Here the comparison between both the
estimator is described by PSNR improvement. The PSNR improvement of later over the
previous one is shown in the Table 3. According to the table GA optimized NLMS
52
estimator has improved PSNR over plane NLMS as it provides the optimal solution. Also
from the above simulation results we can conclude that the GA optimized estimator
provides better PSNR then simple NLMS estimator also the BER is reduced by GA
optimized NLMS estimator It is also simple to implement and is robust technique. Hence
we can conclude that Genetic algorithm is an optimization tool for symbol detection and
channel estimation in MIMO-OFDM systems.
Table 3. Comparison Table
Value
PSNR
of Value of BER
1.145 * 10^-3
5.53 db
9.06 db
13.15 db
Without
Estimation
References
G. Foschini, Layered space-time architecture for wireless communication in a fading environment
using multi-element antennas, Bell Labs Tech Journal, vol. 1, no .2, (1996), pp. 41-59.
[2] V. Zelst and T. Schenk, Implementation of a MIMO-OFDM based wireless LAN system, (2004) in
IEEE Transaction on signal Processing, vol. 52, (2004), pp. 483-494.
[3] P. Radosavljevic, K. Jin Kim, H. Shen and J. R. Cavallaro, Parallel Searching-Based Sphere Detector
for MIMO Downlink OFDM Systems, IEEE transactions on signal processing, vol. 60, no. 6, (2012).
[4] M. Jiang and L. Hanzo, Multiuser MIMO-OFDM for Next-Generation Wireless Systems, vol. 95, no.
7, (2007).
[5] Samanta, Genetic Algorithm: An Approach for Optimization (Using MATLAB), International
Journal of Latest Trends in Engineering and Technology (IJLTET).
[6] Z. Kuder and S. Hanus, Different Approaches to Achieving Higher Capacity in MIMO-OFDM
Systems, (2013) elektrorevue, vol. 4, no. 3, (2013) September.
[7] R. Yazdil and T. Kwasniewski, Low Complexity Sphere Decoding Algorithms, 978-1-4244-24894/08/$20.00 2008 IEEE, (2008).
[8] J. Dhiman, S. Ahmad and K. Gulia, Comparison between Adaptive filter Algorithms (LMS, NLMS
and RLS), International Journal of Science, Engineering and Technology Research (IJSETR), vol. 2,
no. 5, (2013).
[9] E. Saberinia, J. Tang, A. Tewfik and K. Pardhi, Pulsed-OFDM Modulation for Ultrawideband
Communications, IEEE transactions on vehicular technology, vol. 58, no. 2, (2009).
[10] M. Ghogho, D. McLernon, A. Swami and E. Alameda-Hernandez, Channel Estimation and Symbol
Detection for Block Transmission Using Data-Dependent Superimposed Training, IEEE signal
processing letters, vol. 12, no. 3, (2005).
[1]
Author
Apoorva S. Agrawal
M.Tech Final semester student
Research Scholar
53
54