Artificial Neural Network Channel Estimation for OFDM System

Transcription

1 International Journal of Electronics and Computer Science Engineering 1686 Available Online at ISSN Artificial Neural Network Channel Estimation for OFDM System 1 Kanchan Sharma, Shweta Varshney 2 1,2 Dept. of Electronics and Communication Technology 1,2 Guru Gobind Singh Indraprastha University,Delhi, India 1 - joinkanchansharma@gmail.com 2 - shwetavarshney05@gmail.com Abstract- This paper uses Artificial Neural Network (ANN) for channel estimation based on Levenberg-Marquardt training algorithm in OFDM systems over Rayleigh fading channels. This technique utilizes the learning property of neural network. By using this feature, there is no need of any matrix computation and proposed technique is less complex. This technique is useful to achieving the high data rate, transmission capability with high bandwidth, efficiency and its robustness to multipath delay. In OFDM system, the Channel estimation is an essential problem so the Pilot-aided channel estimation has been used; a good choice of the pilot pattern should match the channel behavior both in time and frequency domains. In this arrangement, the performance of the channel estimation is analyzed with estimators based on Least Square Algorithm is carried out through MATLAB Simulation. The performance of OFDM with ANN is evaluated on the basis of Bit Error Rate (BER). The OFDM with ANN has been shown to perform much better than the OFDM without ANN. Keywords OFDM. Channel Estimation. Artificial Neural Network. Least-Square Error I. INTRODUCTION OFDM is becoming a very popular multi-carrier modulation technique for transmission of signals over wireless channels. Now OFDM is widely used for high-speed communications over frequency selective channels. OFDM divides the high data rate stream into parallel lower data rate and hence prolongs the symbol duration, thus helping to eliminate Inter Symbol Interference (ISI). It also allows the bandwidth of subcarriers to overlap without Inter Carrier Interference (ICI) as long as the modulated carriers are orthogonal. Therefore OFDM is considered as an efficient modulation technique for broadband access in a very dispersive environment. The frequency selective fading, is caused by multipath could lead to carriers used, being heavily attenuated due to destructive interference at the receiver. The result of this is the carriers being lost in the noise [1].To increase performance of OFDM system under frequency selective channels; the channel estimation is required before demodulation of OFDM signals [2]. The channel estimation is a process of characterizing the effect of the transmission medium on the input signal. In OFDM system there are several techniques for channel estimation [2-14].Among these techniques; Block type Pilot based channel estimation technique is more popular. The Block type Pilot based estimation techniques can be based on Least-Square (LS). The LS estimators have low complexity. In this paper, we propose an artificial neural network (ANN) based on channel estimation technique as an alternative to Block type pilot based channel estimation technique for OFDM systems over Rayleigh fading channels. The Simulation results show that ANN based on channel estimator gives better results as compared to Block type pilot based channel estimator for OFDM systems over the Rayleigh fading channel. The organization of this paper is as follows. In section II, description of OFDM system model is given. ANN based channel estimator is described in section III. Simulation results are offered in section IV and finally, section V Concludes the paper.

2 Artificial Neural Network Channel Estimation for OFDM System 1687 II.SYSTEM OUTLINE A. OFDM System Model A block diagram of the OFDM system is shown in Fig.1.Firstly the binary information is mapped using baseband modulation schemes such as QAM.Then the serial-to-parallel conversion is applied to baseband modulated signals. After modulation the symbol rate reduced to R=(R/log2M), where M is constellation is applied to baseband modulated signals. This reduces data rate by N times where N is number of parallel streams. Each of parallel streams constitutes tiny bandwidth in the spectrum. So these streams almost undergo flat fading in the channel. This is the greatest advantage of OFDM. After inserting pilots either to all subcarriers with a specific period of blocks or within a uniform period of frequency bins in all blocks, the serial-to-parallel converted data is modulated using Inverse Fast Fourier Transform (IFFT). After IFFT, the time domain signal is given by following equation: s(n)=ifft(s(k)),n=0,1,2..n-1 Where N is the length of FFT, s (k) is baseband data sequence. After IFFT, the guard interval called as cyclic Prefix is inserted to prevent Inter-Symbol-Interference (ISI). This interval should be chosen to be larger than expected delay spread of the multipath channel. The guard time includes the cyclically extended part of the OFDM symbol in order to eliminate the Inter-Carrier-Interference (ICI). The symbol extended with the cyclic prefix is given as follows: Where Nc is the length of the cyclic prefix. The resultant signal st(n) will be transmitted over frequency selective time varying fading channel with additive white Gaussian Noise (AWGN). The received signal is given by following equation: yt (n) = st (n) h (n) + w (n) (3) Where h(n) is the impulse response of the frequency selective channel and W(n) is AWGN. At the receiving end, firstly the cyclic prefix is removed. Then the signal y(n) without cyclic prefix is applied to FFT block in order to obtain following equation: After FFT block, assuming there is no ISI demodulated signal is given by following equation: Where H(k) is FFT[h(n)] and W(k) is FFT[w(n)].

3 IJECSE,Volume1,Number 3 Kanchan Sharma and Shweta Varshney 1688 Fig1 Block diagram of base band model of general OFDM system From Eq.5 that before demodulation, the channel estimation should be done at the receiver side, in order to compensate the effects of the channel on the received signal B. ArtificialNeuralNetworks(ANNs) Artificial Intelligence is a branch of study, which enhances the capability of computers by giving them human-like intelligence. The brain architecture has been extensively studied and attempts have been made to simulate it. Artificial Neural Networks (ANN) represent an area of artificial intelligence (AI). They are basically an attempt to simulate the brain. An artificial neuron model consists of a linear combination followed by an activation function. This network utilized the different type of activation functions; the common ones, which are the sufficient for most applications, are the sigmoidal and hyperbolic tangent functions. B.I Multi-layer Feed-Forward (FF) ANN First, consider the single layer FF network. A layer is a set of neurons or computational nodes at the same level. A set of inputs can be applied to this single layer of neurons. This would then become a single layer FF network. This single layer could be called the output layer (OL). Each node in the OL is called an output neuron. This structure can be extended to a multi-layer FF ANN by adding one or more layers to the existing network. These additional layers thenbecome hidden layers (HL). Each node in the HL is called a hidden neuron. They appear as intermediate neurons between the input and the output layers. In Figure 2, the three layers of an ANN are shown. The first layer is the input layer where the input datavector is passed into the network.the input is a two dimensional vector. Following that is the hidden layer containing 3 hidden neurons. Figure 2 Multilayer feed forward network B.II. ANN Based OFDM

4 Artificial Neural Network Channel Estimation for OFDM System 1689 In this paper the ANN based channel estimation is designed using MATLAB.We have simulated BER in 16-QAM modulated OFDM signal. The system model comprises of the blocks as shown in Figure 3. Figure 3 OFDM with ANN For ANN the symbols after parallel to serial converter block, which is QAM modulated are taken as the training data. The data after passing through the channel are taken as target. With these training data and target data we have trained the network for varying the SNR. After that we have tested the network with different OFDM signals. Now these tested data are demodulated and with these we calculate the BER. The parameters of the proposed technique are given in table1. Table -1Parameters of artificial neural networks parameter value Number of inputs 2 Number of hidden layers 2 Number of neurons 10,10 Epoch number 500 Training Function Levenberg-Marquart Transfer Function Tansig Performance Mean Aquare Error III. EXPERIMENT AND RESULT The BER performance has been observed for the OFDM signal for 16-QAM modulated signals in Rayleigh faded channel in fig4. In AWGN as well as faded channel OFDM with LS estimation 10-1 BER SNR Figure4. BER for LS estimators to OFDM Systems.

5 IJECSE,Volume1,Number 3 Kanchan Sharma and Shweta Varshney 1690 The ANN is trained with signal input from the transmitter side. For 16-QAM, the training sample set is a (1x8640) complex type matrix. The target sample set is presented to the ANN in the form of a matrix of size (1x8640) complex type matrix. The learning of the ANN is done in the training phase during which the ANN adjusts its weights according to the specific coding logic applied at the transmitter end. The ANN is trained for 1000 epochs and it performance is measured by Mean Square Error. During this phase, on an average, the ANN reaches this MSE goal with an accuracy of nearly 100%.this is confirmed by over twenty trials. During the simulation, severely faded data, mixed with AWGN is decoded by trained ANN to test its effectiveness and confirm its feasibility in that role. This test also assesses its accuracy of performance. Fig 5 shows the BER versus SNR of the LS with ANN channel estimation algorithms of OFDM system. It is seen that ANN based OFDM exhibits better performance in terms of lower Bit Error Rate (BER) ANN OFDM BER SNR db Fig5 BER plot for LS estimator with ANN to OFDM systems ANN OFDM Theoretical OFDM OFDM with LS estimation 10-1 B E R SNR db Fig6 Bit Error Rate of the OFDM system with and without the neural network Finally we can compare the BER plots of OFDM with LS estimation, theoretical OFDM and ANN OFDM.From the figure we can say that the best performance is obtained with the proposed ANN based technique for QAM. As shown in Figs.6, the best results are obtained with ANN based estimation technique. These results are very promising and show that neural networks can be efficiently used in an OFDM system.

6 Artificial Neural Network Channel Estimation for OFDM System 1691 IV.CONCLUSION In this paper an artificial neural network channel estimation technique based on levenbergmarquardttraining algorithm has been proposed for OFDM systems over Rayleigh fading channel. BER analysis of the ANN based estimator is obtained and compared with the LS estimation techniques. The application of ANN for Rayleigh multipath fading channel in modulated environment may come to an effective way to improve the BER probability in OFDM system. It shows the ANN transforms to be a suitable tool which makes channel estimation better as well as received data performance is significantly good in wireless communication. ANN specially feed-forward networks. With better configuration of the ANN and optimization conditions of training and testing, the ANN can be used as the efficient method for recovery of symbols at different fading condition. V. REFERENCE [1] Cebrail Çiflikli A. Tuncay Öz sahin A. Ça grı Yapici, Artificial Neural Network Channel Estimation Based on Levenberg-Marquardt for OFDM Systems, Wireless Pers Commun (2009) 51: DOI /s [2] Colieri, S., Ergen, M., Puri, A., & Bahai, A. (2002). A study of channel estimation in OFDM systems. In Proceedings of the IEEE 56th Vehicular Technology Conference, September 2002 (Vol. 2,pp ). [3] Edfors, O., Sandell, M.,Wilson, S. K., & Borjesson, P. O. (1998). OFDM channel estimation by singular value decomposition. IEEE Transactions of Communications, 46, [4] Tolochko, I., & Faulkner, M. (2002). Real time LMMSE channel estimation for wireless OFDM systems with transmitter diversity. In Proceedings of the 56th IEEE VTC, Vancouver, Canada (pp ). [5] Gacanin, H., Takaoka, S., & Adachi, F. (2005). Pilot-assisted channel estimation for OFDM/TDM with frequency-domain equalization. In Vehicular Technology Conference, September 2005, USA. [6] Colieri, S., Ergen, M., Puri, A.,&Bahai, A. (2002). Channel estimation techniques based on pilot arrangement in OFDM systems. IEEE Transactions on Broadcasting, 48(3), [7] Doukopoulos, X. G., & Moustakides, G. V. (2004). Adaptive algorithms for blind channel estimation in OFDM systems. In 2004 IEEE International Conference on Communications, June 2004 (Vol. 4,pp ). [8] Roy, S., & Li, C. (2002). A subspace blind channel estimation method for OFDM systems without cyclic prefix. IEEE Transactions on Wireless Communications, 1, [9] Patra, J. C., Pal, R. N., Baliarsingh, R., & Panda, G. (1999). Nonlinear channel equalization for QAM signal constellation using artificial neural networks. IEEE Transactions on Systems, Man, and Cybernetics,29(2), [10] Siu, S., Gibson, G. J., & Cowan, C. F. N. (1990). Decision feedback equalization using neural network structures and performance comparison with standard architecture. Proceedings of the Institution of Electrical Engineers, 137(pt. 1), [11] Naveed, A., Qureshi, I. M.,Cheema, T. A.,&Jalil, A. (2004).Blind equalization and estimation of channel using artificial neural network. In 8th International Multitopic Conference, INMIC 2004 (pp ). [12] Zhang, L., & Zhang, X. (2007). MIMO channel estimation and equalization using three-layer neural network with feedback. Tsinghua Science and Technology, 12(6), [13] Sun, J., & Yuan, D. F. (2006). Neural network channel estimation based on least mean error algorithm in the OFDM systems. Advances in Neural Networks. Berlin: Springer. van de Beek, J. J., Edfors, O., Sandell, M.,Wilson, S. K.,&Borjesson, P. O. (1996).On channel estimation in OFDM systems. In Proceedings of IEEE VTC 96, November 1996 (pp ).