Professional Documents
Culture Documents
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/258817673
CITATIONS READS
4 151
2 authors:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Kandarpa Kumar Sarma on 31 May 2015.
294
2. BASIC CONSIDERATIONS OF network learns while performing signal processing, which do not
stop during learning.
RECURRENT NEURAL NETWORKS
(RNN)s 3. SYSTEM MODEL FOR CHANNEL
ANNs are mathematical model or computational model based on
ESTIMATION TECHNIQUE USING RNNS.
biological neural networks [11]. It consists of an interconnected
group of artificial neurons and processes information using a In this paper, a channel estimation technique for STBC coded
connectionist approach to computation. In this section, a brief MIMO transmission has been proposed over Rayleigh faded
description of the RNN architecture used for the channel wireless channel with BPSK modulation, using high degree of
estimation problem has been described. In general, a Neural adaptability of RNNs. Diversity in data transmission is based on
Network (NN) is a machine that is designed to model the way in the idea that the probability that multiple statistically independent
fading channels simultaneously experience deep fading is very
low. MIMO is thus defined as the way of providing multiple
antennas at the transmitter and at both link ends of a
communication system respectively. The basic principle behind
295
STBC, blocks of data are transmitted from different transmitter at 4. EXPERIMENTAL DETAILS AND
different time instants for which a specific coding scheme is used
[3]-[5]. There is a special version of STBC called Alamouti code RESULTS
which uses two transmit antennas and N receive antennas and can
accomplish a maximum diversity order of 2N. It has the coding In wireless communications systems, each of the multipath
components have different relative propagation delays and
c1 c2
attenuations which results in filtering type of effect on the
matrix , where * denotes complex conjugate [3].
c2
*
c1* received signal. The mobile radio channel can be modeled as a
linear time varying channel, where the channel changes with time
Channel estimation is regarded as a classification task in which and distance. The received signal Y can be expressed as a
RNNs can form decision regions in the space of received symbol convolution of the transmitted signal X with channel impulse
sequences. Figure 3 shows the block diagram of the RNN assisted response h, as Y=X*H +N, where N is the Additive White
channel estimator block comprising of two independent neural Gaussian Noise and H is the channel matrix. If the RNN estimates
networks NN1 and NN2 respectively. These two NNs work the signal as XE, an error matrix can be generated as, e= X-XE,
such that XE=X*HE+N is the signal generated by the RNN.
296
The RNNs with specific configuration are shown in Table 1. The Figure 8: Performance of RNN assisted MIMO system with
training part is completed initially and tested extensively for all STBC for BPSK modulation
possible combinations of 64-bit data blocks. Training is carried
out till mean square error (MSE) approaches the desired goal of Table 4 gives knowledge about the behavior of RNN with change
10-3. Several configurations for the RNN are utilized for training. in activation functions in three layers with trainbfg algorithm.
The length of the hidden layers has been fixed by trial and error Hyperbolic tangent sigmoid transfer function in all layers seems
method. Table 2 shows the performance variation during training to give a better result than the other combination. This
of the RNN with variation in size of the hidden layer for MSE configuration has been selected for further training and testing of
goal 10-3. The size of the hidden layer has been fixed to be same the RNNs as channel estimator in STBC-MIMO systems.
with the input layer, as it gives an efficient result in terms of
convergence time and number of epochs required to reach the
goal. Table 3 shows the performance variation of three training
algorithms, namely Gradient descent with adaptive learning rule
back propagation (traingda), Gradient descent with momentum
and adaptive learning rule (traingdx, commonly known as
Levenburg-marquardt (LM)) and BFGS quasi-Newton back
propagation (trainbfg) respectively. The last algorithm shows a
way to attain better performance than the other two. Hence the
RNNs are trained with this memory less quasi-Newton back
propagation (trainbfg) algorithm as a measure of training
standardization.
5. CONCLUSION
297
channel. This technique is found to be more bandwidth efficient performance comparison with standard architecture. Proceedings
compared to pilot-based channel estimation techniques. Simulated of the Institution of Electrical Engineers, 137(pt. 1), 221225.
results in terms of Bit Error Rates (BER) vs. Signal to Noise
(SNR) ratio also depict the effectiveness of the learning capability [8] Zhang, L., & Zhang, X. 2007. MIMO channel estimation and
of RNNs for the task of channel estimation over wireless fading equalization using three-layer neural network with feedback.
channel. Tsinghua Science and Technology, 12(6), 658662.
298