Performance of Discrete Wavelet Transform – Artificial Neural Network Based Signal Detector/Equalizer for Digital Pulse Interval Modulation in Practical.

Slides:



Advertisements
Similar presentations
Feedback Reliability Calculation for an Iterative Block Decision Feedback Equalizer (IB-DFE) Gillian Huang, Andrew Nix and Simon Armour Centre for Communications.
Advertisements

OFDM Transmission over Wideband Channel
Z. Ghassemlooy, S Rajbhandari and M Angelova
MIC-CPE2010, Jordan Optimizing the Performance of Digital Pulse Interval Modulation with Guard Slots for Diffuse Indoor Optical Wireless Links Z. Ghassemlooy.
Chapter : Digital Modulation 4.2 : Digital Transmission
1 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen Data Communication, Lecture6 Digital Baseband Transmission.
The Impact of Channel Estimation Errors on Space-Time Block Codes Presentation for Virginia Tech Symposium on Wireless Personal Communications M. C. Valenti.
Wireless Transmission Fundamentals (Physical Layer) Professor Honggang Wang
S. Rajbhandari, Prof. Z. Ghassemlooy, Prof. M. Angelova School of Computing, Engineering & Information Sciences, University of Northumbria, Newcastle upon.
4.2 Digital Transmission Pulse Modulation (Part 2.1)
Optical Wireless Communications
Diversity techniques for flat fading channels BER vs. SNR in a flat fading channel Different kinds of diversity techniques Selection diversity performance.
ICTON 2007, Rome, Italy The Performance of PPM using Neural Network and Symbol Decoding for Diffused Indoor Optical Wireless Links 1 S. Rajbhandari, Z.
Sujan Rajbhandari PGNET Performance of Convolutional Coded Dual Header Pulse Interval Modulation in Infrared Links S. Rajbhandari, Z. Ghassemlooy,
Prof. Z. Ghassemlooy ICEE 2006, Iran 1 DH-PIM Employing LMSE Equalisation For Indoor Optical Wireless Communications Z. Ghassemlooy, W. O. Popoola, and.
RAKE Receiver Marcel Bautista February 12, Propagation of Tx Signal.
Z. Ghassemlooy & S Rajbhandari
Department of Electronic Engineering City University of Hong Kong EE3900 Computer Networks Data Transmission Slide 1 Continuous & Discrete Signals.
Applications of Wavelet Transform and Artificial Neural Network in Digital Signal Detection for Indoor Optical Wireless Communication Sujan Rajbhandari.
Communication Technology Laboratory Wireless Communication Group Partial Channel State Information and Intersymbol Interference in Low Complexity UWB PPM.
Dual Header Pulse Interval Modulation (DH-PIM) Dr. Nawras Aldibbiat Professor Z Ghassemlooy Optical Communications Research Group School of Computing,
IASTED- WOC- Canada 07 1 CONVOLUTIONAL CODED DPIM FOR INDOOR NON-DIFFUSE OPTICAL WIRELESS LINK S. Rajbhandari, Z. Ghassemlooy, N. M. Adibbiat, M. Amiri.
Wireless communication channel
Muhammad Imadur Rahman1, Klaus Witrisal2,
Done by Sarah Hussein 10\05\2012. Trends in modern communication systems place high demands on low power consumption, high-speed transmission, and anti-
Wireless Transmission Fundamentals (Physical Layer) Professor Honggang Wang
Normalization of the Speech Modulation Spectra for Robust Speech Recognition Xiong Xiao, Eng Siong Chng, and Haizhou Li Wen-Yi Chu Department of Computer.
Formatting and Baseband Modulation
ORTHOGONAL FREQUENCY DIVISION MULTIPLEXING(OFDM)
1 Secure Cooperative MIMO Communications Under Active Compromised Nodes Liang Hong, McKenzie McNeal III, Wei Chen College of Engineering, Technology, and.
CE 4228 Data Communications and Networking
EE 3220: Digital Communication Dr. Hassan Yousif Ahmed Department of Electrical Engineering College of Engineering at Wadi Aldwasser Slman bin Abdulaziz.
The Wavelet Tutorial: Part3 The Discrete Wavelet Transform
1 Techniques to control noise and fading l Noise and fading are the primary sources of distortion in communication channels l Techniques to reduce noise.
CHAPTER 6 PASS-BAND DATA TRANSMISSION
Signal Propagation Propagation: How the Signal are spreading from the receiver to sender. Transmitted to the Receiver in the spherical shape. sender When.
Coding No. 1  Seattle Pacific University Modulation Kevin Bolding Electrical Engineering Seattle Pacific University.
Wireless Communication Technologies 1 Outline Introduction OFDM Basics Performance sensitivity for imperfect circuit Timing and.
EELE 5490, Fall, 2009 Wireless Communications Ali S. Afana Department of Electrical Engineering Class 5 Dec. 4 th, 2009.
Numerical Evaluation SINR of an IR system Jinhui Wang.
Multiuser Detection (MUD) Combined with array signal processing in current wireless communication environments Wed. 박사 3학기 구 정 회.
1/ , Graz, Austria Power Spectral Density of Convolutional Coded Pulse Interval Modulation Z. Ghassemlooy, S. K. Hashemi and M. Amiri Optical Communications.
EE 6331, Spring, 2009 Advanced Telecommunication Zhu Han Department of Electrical and Computer Engineering Class 7 Feb. 10 th, 2009.
Digital Communications
Space-Time and Space-Frequency Coded Orthogonal Frequency Division Multiplexing Transmitter Diversity Techniques King F. Lee.
Performance analysis of channel estimation and adaptive equalization in slow fading channel Chen Zhifeng Electrical and Computer Engineering University.
Adaphed from Rappaport’s Chapter 5
1 Blind Channel Identification and Equalization in Dense Wireless Sensor Networks with Distributed Transmissions Xiaohua (Edward) Li Department of Electrical.
X. Li, W. LiuICC May 11, 2003A Joint Layer Design Smart Contention Resolution Random Access Wireless Networks With Unknown Multiple Users: A Joint.
Decision Feedback Equalization in OFDM with Long Delay Spreads
Mobile Computing and Wireless Networking Lec 02
Digital Communications Chapeter 3. Baseband Demodulation/Detection Signal Processing Lab.
Equalization Techniques By: Mohamed Osman Ahmed Mahgoub.
A Simple Transmit Diversity Technique for Wireless Communications -M
Bandpass Modulation & Demodulation Detection
APPLICATION OF A WAVELET-BASED RECEIVER FOR THE COHERENT DETECTION OF FSK SIGNALS Dr. Robert Barsanti, Charles Lehman SSST March 2008, University of New.
Sujan Rajbhandari LCS Convolutional Coded DPIM for Indoor Optical Wireless Links S. Rajbhandari, N. M. Aldibbiat and Z. Ghassemlooy Optical Communications.
doc.: IEEE /183r0 Submission March 2002 David Beberman, Corporate Wave Net, Inc.Slide 1 Single Burst Contention Resolution “Wireless Collision.
Performance of Digital Communications System
Presenter : r 余芝融 1 EE lab.530. Overview  Introduction to image compression  Wavelet transform concepts  Subband Coding  Haar Wavelet  Embedded.
WAVELET NOISE REMOVAL FROM BASEBAND DIGITAL SIGNALS IN BANDLIMITED CHANNELS Dr. Robert Barsanti SSST March 2010, University of Texas At Tyler.
An Experimental Receiver Design For Diffuse IR Channels Based on Wavelet Analysis & Artificial Intelligence R J Dickenson and Z Ghassemlooy O ptical C.
PERFORMANCE OF A WAVELET-BASED RECEIVER FOR BPSK AND QPSK SIGNALS IN ADDITIVE WHITE GAUSSIAN NOISE CHANNELS Dr. Robert Barsanti, Timothy Smith, Robert.
Digital transmission over a fading channel
Channel Equalization Techniques
Space-Time and Space-Frequency Coded Orthogonal Frequency Division Multiplexing Transmitter Diversity Techniques King F. Lee.
Advanced Wireless Networks
Source: [Yafei Tian, Chenyang Yang, Liang Li ]
Date Submitted: [March, 2007 ]
On the Design of RAKE Receivers with Non-uniform Tap Spacing
Presentation transcript:

Performance of Discrete Wavelet Transform – Artificial Neural Network Based Signal Detector/Equalizer for Digital Pulse Interval Modulation in Practical Indoor Optical Wireless Links Z. Ghassemlooy, S. Rajbhandari and M. Angelova School of Computing, Engineering & Information Sciences, University of Northumbria, Newcastle upon Tyne, UK

Outline  Optical wireless – introduction  Modulation Techniques- Overview  Mutipath induces ISI  Unequalized power penalty  Wavelet-ANN receiver  Final comments

What Optical Wireless Offers ?  Abundance bandwidth  Free from electromagnetic interference  High data rate  No multipath fading  High Directivity.  Secure data transmission  Spatial confinement.  Low cost of deployment  License free operation  Quick to deploy  Compatible with optical fibre  Simple transceiver design.  Small size, low cost component and low power consumptions. 3

Modulation Techniques  On-off keying (OOK): the most basic, simple to implement but requires a high average optical power.  Pulse position modulation (PPM): The most power efficient but require high bandwidth, susceptible to the multipath induced intersymbol interference (ISI).  Differential PPM (DPPM) and digital pulse interval modulation (DPIM): Variable symbol length, built-in symbol synchronization; improved throughputs and efficient utilization of the available bandwidth compared to PPM.  Dual header pulse interval modulation (DH-PIM): Variable symbol length, built-in symbol synchronization; the most efficient utilization of channel capacity compared to OOK, PPM and DPIM. 4

Baseband Modulation Techniques

Normalized Power and Bandwidth Requirement  PPM the most power efficient while requires the largest bandwidth.  DH-PIM 2 is the most bandwidth efficient.  DH-PIM and DPIM shows almost identical bandwidth requirement and power requirement.  There is always a trade-off between power and bandwidth Bit resolution, M Normalized bandwidth requirement PPM DH-PIM 1 DPIM DH-PIM 2 OOK Bit Resolution, M Normalized Power Requirement (dB) DH-PIM 2 PPM DH-PIM 1 DPIM

Indoor Optical Wireless Links  The key issues are: -The eye safety -shift to a higher wavelength of 1550 nm where the eye retina is less sensitive to optical radiation -power efficient modulation techniques. -Mobility and blocking -Use diffuse configuration instead of line of sight, but at cost of -reduced data rate -increased path loss -multipath induced inter-symbol-interference (ISI) -High noise at receiver due to artificial light.

Effect of Artificial Light  Dominant noise source at low data rate.  Interference produce by fluorescent lamp driven by electronic ballasts can cause serious performance degradation at low data rate.  The effect of artificial light is minimised at the receiver using combination of the optical band pass filter and electrical low pass filter.  At the high data rate, the ISI is the limiting factor in the performance of the system instead of artificial light. 2 1 Figure : Optical power spectra of common ambient infrared sources. Spectra have been scaled to have the same maximum value. 1 J. M. Kahn and J. R. Barry, Proceedings of IEEE, vol. 85, pp , A. J. C. Moreira, R. T. Valadas, and A. M. d. O. Duarte, IEE Proceedings -Optoelectronics, vol. 143, pp , 1996.

Intersymbol Interference (ISI)  Limiting factor in achieving high data rate in diffuse links.  ISI is due to broadening of pulse.  Diffuse links are characterised by RMS delay spread.  The impulse response in Ceiling bounce model is given by 1 : 9 LOS Diffuse Diffuse shadowed LOS shadowed where u(t) is the unit step function 1- J. B. Carruthers and J. M. Kahn, IEEE Transaction on Communication, vol. 45, pp , Normalized Time Amplitude Received signal for non-LOS Links

Unequalized Performance  The discrete-time impulse response of the cascaded system is  In non-LOS links, c k contains a zero tap, a single precursor tap (with the largest magnitude) and possibly multiple postcursor taps.  The optimum sampling point is at the end of each slot period T s for LOS link.  On dispersive channels, the optimum sampling point changes as the severity of ISI changes.

Unequalized Performance  For the LOS channel, the slot error probability P se of DPIM is given by: where R is the photodetector responsivity, η is the noise spectral density, P avg is the average transmitted optical signal power, R b is the bit rate, M is bit resolution and L = 2 M.  In a multipath channel, the P se is calculated by summing the error probabilities in all possible sequences. where b i is the m-slot DPIM(NGB) sequence and where  opt is the optimum threshold level, set to the midway value of RP ave (T b ) 0.5.

Unequalized Power Penalty  There is exponential growth in power penalty with increasing delay spread for all orders of DPIM.  The average optical power required to achieve a desirable error performance is impractical for normalized delay spread grater than 0.1.  To mitigate the ISI the solution is to incorporate an equalizer at the receiver. 12

Equalization  Maximum likelihood sequence detector : Though the optimum solution, not suitable for variable symbol length modulation schemes like DPIM since symbol boundaries are not known.  Hence sub-optimum solutions based on finite impulse response filters would be the preferred option.  But equalization based on the finite impulse response (FIR) filter suffers from severe performance degradation in time varying and non-linear channels. 1  The equalization problem can be formulated as classification problem and hence artificial neural network can be used to reduce the effect of ISI. 2,3 1- A. Hussain, J. J. Soraghan, and T. S. Durrani, IEEE Transactions on Communications, vol. 45, pp , J. C. Patra and N. R. N. Pal, Signal Processing, vol. 43, pp , L. Hanzo, C. H. Wong, and M. S. Yee, Adaptive wireless transceivers: Wiley-IEEE Press, 2002, pp

Equalization: A Classification Problem  Classification capability of FIR filter equalizer is limited to a linear decision boundary, which is a non-optimum classification strategy 1.  FIR base equalizers suffer from severe performance degradation in time varying and non-linear channel 2.  The optimum strategy would be to have a nonlinear decision boundary for classification.  ANN is employed for equalization because of its capability to form complex nonlinear decision regions. - In fact both the linear and DFE are a class of ANN 3.  Wavelet based equalization L.Hanzo, et al, Adaptive wireless transceivers: Wiley-IEEE Press, 2002, pp C. Ching-Haur, et al, Signal Processing,vol. 47, no. 2, pp S. Haykin, Communications Magazine, IEEE, vol.38, no.12, pp , Dec D. Cariolaro et al, IEEE Intern. Conf. on Communications, New York, NY, USA, pp , 2000.

Wavelet Transform Neural Network Block Diagram of Receiver Based on Classification Optical Receiver Feature Extraction Pattern Classification Post- Processing Optical Signal  For efficient classification, feature extraction tools are incorporated in the receiver.  The receiver is made modular by having separate block for : (a) Feature extraction (wavelet transform) and (b) pattern classification (ANN).  WT-ANN based receiver outperforms the traditional equalizers R. J. Dickenson and Z. Ghassemlooy, International Journal of Communications Systems, Vol. 18, No. 3, pp , 2005.

Feature Extraction Tools Time-Frequencies Mapping Fourier Transform Short-Time Fourier Transform Wavelet Transform No time- frequency Localization Fixed time-frequency resolution: Uncertainty problem No resolution problem :Ultimate Transform

CWT vs. DWT  Infinite scale in CWT, having highly redundant coefficients.  Redundancy in CWT can be removed by utilizing the DWT.  The DWT is easier to implement using filter bank of high pass and low pass filters.  Reduced computational time compared CWT.  Possibility of denoising of signal by thresholding the wavelet coefficient in DWT. 17

Discrete Wavelet Transform x[n] h[n] 2 g[n] 2 cD 1 cA 1 h[n] 2 g[n] 2 cD 2 cA 2 Level 1 DWT coefficients Level 2 DWT coefficients... Signal Filtering Down- sampling  DWT coefficient can efficiently be obtained by successive filtering and down sampling.  Signal is decomposed using high pass h[n] and a low pass g[n] filters and down sampled by 2.  The two filter are related to each other and are known as a quadrature mirror filter.

Denoising Signal using DWT  Denoising is performed by hard/soft thresholding of the detail coefficients. - Hard thresholding - Soft thresholding - The threshold level  for universal threshold scheme : : the variance of the wavelet coefficient.  Denoised signal where is the inverse WT.

WT-ANN Based Receiver Model  The receiver incorporates a feature extractor (DWT) and a pattern classifier (ANN).  16-samples per bit.  Signal is decimated into W-bits discrete sliding window. (i.e. each window contains a total of 16W discrete samples ).  Information content of the window is changed by one bit.  3-level DWT of each window is calculated.  DWT coefficients are denoised by: a) Thresholding : A threshold is set and ‘soft’ or ‘hard’ thresholding are used for detail coefficients. b) Discarding coefficients: detail coefficients are completely discarded.  The denoised coefficient are fed to ANN.  ANN is trained to classify signal into two binary classed based on the DWT coefficients.

Simulation Parameters 21 ParametersValue Data rate R b 200 Mbps Channel RMS delay spread D rms 1-10 ns No. of samples per bit16 Mother waveletDiscrete Meyer ANN typeFeedforward back propagation No. of neural layers2 No. of neurons in 1 st layer4 No. of neurons in 2 nd layer1 ANN activation functionlog-sigmoid, tan-sigmoid ANN training algorithmScaled conjugate gradient algorithm ANN training sequence100 symbols Minimum error1 -30 Minimum gradient1 -30 DWT levels3

Simulation Flowchart 22

Results  Unequalized DPIM- worst error performance.  The unequalized error performance is not practically acceptable for highly diffuse channel like channels with D rms > 5ns.  Both linear and DWT-ANN equalizers show improve error performance compared to unequalized cases.  The DWT-ANN based receiver showed a significant improvement in SER performance compared to linear equalizer.  The SNR gain with DWT-ANN at the SER of is ~ 8.6 dB compared to linear equalizer.  Performance of DWT-ANN also depends on selection of mother wavelet, with discrete Meyer wavelet showing the best performance.  Further improvement in SER performance can be achieved by using error control coding. 23 Figure : The SER performance against the SNR for unequalized, Linearly equalized and a DWT-ANN based receiver at data rate of 200 Mbps for diffuse links with D rms of 1, 5 and 10 ns.

Conclusions  The traditional tool for signal detection and equalization is inadequate in time-varying non-linear channel.  Digital signal detection can be reformulated as feature extraction and pattern classification.  Both discrete and continuous wavelet transform is used for feature extraction.  ANN is trained for classify received signal into binary classes.  DWT-ANN equalizers performance offers an SNR gain of almost 8 dB at SER of at data rate of 200 Mbps for all values of channel delay spread.  The rapid increase in the processing time of electronic devices can make the system practically feasible.  Practical implementation of the proposed system in the process of being carried out at the photonics Lab, Northumbria University.

Acknowledgement  Northumbria University for supporting the research.  OCRG and IML lab for providing require software for simulation. 25

Questions/Suggestions/Comments Thank you!