ANN-assisted Booth Multiplier for Design of Convolutional Encoder for Application in Wireless Communication

8/13/2019 ANN-assisted Booth Multiplier for Design of Convolutional Encoder for Application in Wireless Communication

1/25

ANN-assisted Booth Multiplier for Design of

Convolutional Encoder for Application in

Wireless CommunicationIICAI 2011, SIT Tumkur

Parismita Gogoi and Kandarpa Kumar Sarma

Deptt. of Electronics & CommunicationTechnology

Gauhati University, Guwahati-781014, AssamPresented by: Parismita Gogoi


2/25

Plan of Talk

Introduction

Theoretical Background

System ModelExperimental Details

Results Derived

Conclusion

References

Wednesday, December 14,2011

2IICAI-2011, 14th-16th December, Tumkur


3/25

Block Diagram Of A Digital CommunicationSystem




4/25

Requirement Of Channel Coding In

Digital Communication

Error performance vs. Bandwidth

Transmitted signal power vs. BandwidthData rate vs. Bandwidth

Channel Capacity vs. Bandwidth




5/25

Convolutional encoder and decoder in a

Communication Link




6/25

Basic Definitions Of Convolutional Codes




7/25

Booth Algorithm For Multiplication

Booths algorithm is a popular method for2s complement multiplication.

designed by Andrew D. Booth in 1951




8/25

Artificial Neural Network (ANN)




9/25

System model


IICAI-2011, 14th-16th December, Tumkur 9


10/25

System Model

Figure: Booth multiplier designed using ANNWednesday, December 14,2011



11/25

ANN specification




12/25

Performance variation after 1000 epochsduring training with variation of size of

the hidden layer




13/25

Effect on average MSE convergence

with variation of activation

functions




14/25

Experimental Details




15/25

BER plot for QAM

Thursday, December 15, 2011 IICAI-2011, 14th-16th December, Tumkur 15


16/25

BER plot for QPSK



17/25

Improvement in BER in % shown by the

ANN-assisted Booth multiplier (16-QAM and

QPSK AWGN channel)


I i BER i % h b h


18/25

Improvement in BER in % shown by theANN-assisted Booth multiplier (16-QAM

and QPSK Rayleigh Faded channel)


I i BER i % h b h


19/25

Improvement in BER in % shown by theANN-assisted Booth multiplier (16-QAM

and QPSK Rician Faded channel)



20/25

Results

For QAM in AWGN channel the improvement in BER

due to coding is between 3138% , For QPSK, this

improvement is between 23-39% .

The corresponding margins for QAM and QPSK in

Rayleigh channel is 15

33% and 21-54% respectively.

A similar set of values for QAM and QPSK in Rician

channel is found to be between 5-15.6% and 5-23% .

This shows that the best performance derived from

coding is with QPSK in Rayleigh faded channel which isover 50%. This is significant and justifies the necessity

of using Booth Algorithm aided Convolutional coding.



21/25

Comparison of time taken and average

accuracy in % for different bit-sizes



22/25

Conclusion

The work here describes a novel approach of designing

a Booth multiplier used as part of a Convolutional

encoder.

The Booth multiplier constituted by several ANNs

provides 32 to 60% improvement in processing speedcompared to conventional approach of the design.

Similarly, the accuracy also improves by 2.1 to 4.2%which is also significant. These two aspects combined

with less dependence on hardware makes it suitablefor a range of applications including wirelesscommunication.



23/25

Contd.

The drawback is with regards to the initial

complexity and tedious work related to the

fixing of the ANN configuration. Moreover, for

cases greater that 16-bit encoding, the

present work will require a bit expansion and

re-orientation of the configuration. The entire

approach, however, can be made adaptive tosuit specific encoding-transmit cases.



24/25

References Kalita S. , Gogoi P., Sarma K.K.: Convolutional Coding Using Booth Algorithm for

Application in Wireless Communication,International Journal of Electronics Signals andSystems (IJESS), 2011, vol1 (Iss1), pg 37 41.

Sklar, B.,: Digital Communication - Fundamental and Application, 2nded., PearsonPublication, 2005.

Shannon, C. E.: À Mathematical Theory of Communication, The Bell SystemTechnical Journal, pp- 379 - 427, 623-656, 1948.

Viterbi, A. J.: `Èrror Bounds for Convolutional Codes and an Asymptotically OptimumDecoding Algorithm, IEEE Transactions on Information Theory, Vol.IT-13, pp. 260-269,

April, 1967. Meng W., Wei-Ping Z., Nakamura, S: `À hybrid fuzzy neural decoder for convolutional

codes, Proceedings of the 1998 IEEE International Symposium on Circuits and Systems,vol. 3, pp. 235 238, 1998

Berber, S.M. Kecman, V.: ``Convolutional decoders based on artificial neuralnetworks, Proceedings. 2004 IEEE International Joint Conference on Neural Networks,

vol. 2, pp. 1551 1556, 2004. Mason K. and SodhaJ.: `À Neural Network Coded Modulation Decoder, Caribbean

Journal of Mathematics and Computer Science, Vol. 13, pp. 35-41, 2006.

Haykin S.: Neural Networks- A Comprehensive Foundation. 2nded., Pearson Education,New Delhi, 2003.

Proakis. J. G.: Digital Communication, 4thed., McGraw-Hill Series, 2001.

Mano M. M.: Computer System Architecture, Third ed., Pearson Prentice hall, 2006.Thursday, December 15, 2011 IICAI-2011, 14th-16th December, Tumkur 24


25/25

THANK YOU



ANN-assisted Booth Multiplier for Design of Convolutional Encoder for Application in Wireless Communication

Documents