The Efficient Denoising Artificial Light Interference using Discrete Wavelet Transform

with Application to Indoor Optical Wireless System

S. Rajbhandari, Prof. Z. Ghassemlooy, Prof. M. AngelovaSchool of Computing, Engineering & Information Sciences,

University of Northumbria, Newcastle upon Tyne, UK.sujan.rajbhandari@unn.ac.uk

http://soe.unn.ac.uk/ocr

Content

Introduction to indoor optical wireless system (OWS)

Challenges in OWS.

Artificial light interference, its effect in indoor OWS links and techniques to mitigate.

DWT based denoising.

Realization of the propose system.

Future works

Conclusion

History of Optical Communication

The very first form of wireless speech communication was achieved at optical wavelengths in 1878 by Alexander Graham Bell, more than 25 years before Reginald Fessenden did the same thing with radio1.

1 Alexander Graham BELL, American Journal of  Sciences, Third Series, vol. XX, no.118, Oct. 1880, pp. 305- 324.2 F. R. Gfeller and U. Bapst, Proceedings of the IEEE, vol. 67, pp. 1474- 1486, 1979.

Diagram of photophone from Bell paper 1

Development of LASER in 60’s, optical fibre and semiconductor has made the modern communication possible.

The modern era of indoor wireless optical communications was proposed in 1979 by F.R. Gfeller and U. Bapst 2. In fact it was the first LAN proposed using any medium.

Optical Wireless System (OWS): Overview

1 M. Kavehrad, Scientific American Magazine, July 2007, pp. 82-87.

Typical optical wireless system components

Optical wireless connectivity 1

Communication system using light beams (visible and infrared) propagated through the atmosphere or space to carry information.

Optical transmitter Light Emitting Diodes (LED) Laser Diodes (LD)

Optical receiver p-i-n Photodiodes. Avalanche Photodiodes.

Links Line-of-sight(LOS) Non-LOS Hybrid

What OWS offers

Abundance bandwidth High data rate License free operation High Directivity small cell size can support multiple devices

within a room Free from electromagnetic interference suitable for hospital and

library environment.

cannot penetrate opaque surface like wall Spatial confinement Secure data transmission

Compatible with optical fibre (last mile bottle neck?) Low cost of deployment Quick to deploy Small size, low cost component and low power consumptions. Simple transceiver design. No multipath fading

Challenges (Indoor)

Challenges Causes (Possible ) Solutions

Power limitation Eye and skin safety. Power efficient modulation techniques, holographic diffuser, transreceiver at 1500ns band

Noise Intense ambient light (artificial/ natural)

Optical and electrical band pass filters, Error control codes

Intersymbol interference (ISI)

Multipath propagation (non-LOS links)

Equalization, Multi-Beam Transmitter

No/Limited mobility

Beam confined to small area.

Wide angle optical transmitter , MIMO transceiver.

Shadowing Blocking

LOS links Diffuse links/ Cellular System/ wide angle optical transmitter

Limited data rate Large area photo-detectors

Bandwidth-efficient modulation techniques /Multiple small area photo-detector.

Strict link set-up LOS links Diffuse links/ wide angle transmitter

Common Baseband Digital Modulation Techniques

OOK Simple to implement High average power requirement Suitable for Bit Rate greater than 30Mb/s Performance detiorates at higher bit rates

PPM Complex to implement Lower average power requirement Higher transmission bandwidth Requires symbol and slot synchronisation

DPIM Higher average power requirement

compared with PPM Higher throughput Built in symbol synchronisation Performance midway between PPM and

OOK.

DH-PIM The highest symbol throughput Lower transmission bandwidth than PPM and DPIM Built in symbol synchronisation Higher average power requirement compared with PPM and DPIM. Complex decoder

Artificial Light Interference (ALI)

Dominant noise source at low data rate.

Spectral overlapping of signal and interference produce by fluorescent lamp driven by electronic ballasts

can cause serious performance degradation as the interference amplitude can be much higher than signal amplitude.

The effect of noise is minimised using combination of the optical band pass filter and electrical low pass filter.

Optical power spectra of common ambient infrared sources. Spectra have been scaled to have the same maximum value.

Wavelength (m)

No

rma

lise

d p

ow

er/

un

it w

ave

len

gth

0

0.2

0.4

0.6

0.8

1

1.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1.1

1.2

1.3

1.4

1.5

Sun Incandescent

x 10

1st window IR

Fluorescent

2nd window IR

Pave)amb-light >> Pave)signal (Typically 30 dB with no optical filtering)

Fluorescent Light Interference Model1

mhigh(t) high frequency component.

mlow(t) low frequency component.

A. J. C. Moreira, R. T. Valadas, and A. M. d. O. Duarte, IEE Proceedings -Optoelectronics, vol. 143, pp. 339-346.

Low frequency component

High frequency componentOptical power penalty due to FLI

ALI-Possible Solutions

Differential receiver1

Differential optical filtering2

Electrical high pass filter3,4

Polarisers 5

Angle diversity receiver 6,7

Discrete wavelet transform based denoising8,9 1 J. R. Barry, PhD Dissertation, University of California at Berkeley, 19922 A.J.C Moreira, R. T. Valadas, A. M. De Oliveira Duarte, Optical Free Space Communication Links, IEE Colloquium on , vol., no., pp.5/1-510, 19 Feb 1996. 3 R. Narasimhan, M. D. Audeh, and J. M. Kahn, IEE Proceedings - Optoelectronics, vol. 143, pp. 347-354, 1996.4 A. R. Hayes, Z. Ghassemlooy , N. L. Seed, and R. McLaughlin, IEE Proceedings - Optoelectronics vol. 147, pp. 295-300, 2000.5S. Lee, Microwave and Optical Technology Letters, vol. 40, pp. 228-230, 2004.6R. T. Valadas, A. M. R. Tavares, and A. M. Duarte, International Journal of Wireless Information Networks, vol. 4, pp. 275-288, 1997 .7J. M. Kahn, P. Djahani, A. G. Weisbin, K. T. Beh, A. P. Tang, and R. You, IEEE Communications Magazine, vol. 36, pp. 88-94, 1998.8 S. Rajbhandari; Z. Ghassemlooy; and M. Angelova, IJEEE, Vol. 5, no. 2 ,pp102-111. 2009.9 S. Rajbhandari; Z. Ghassemlooy; and M. Angelova, Journal of Lightwave Technology, on print.

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

Discrete Wavelet Transform

Level 1 DWT coefficients

Level 2 DWT coefficients

x[n]

h[n] 2

g[n] 2

y1h

y1l

h[n] 2

g[n] 2

y2h

y2l

Signal

FilteringDown-

sampling

Coefficient can efficiently be obtained by successive filtering and down sampling.

The two filter are related to each other and are known as a quadrature mirror filter.

Reconstruction is inversion of decomposition process filter, up sample and combine.

n

nkgnXkhy ]2[][: cD n

nkhnxkly ]2[][:cA

(14)

DWT based Denoising

Multiresolutional analysis tree

DWT is a multiresolutional analysis (MRA) tool signals are divided into half-frequency bands at each level of the decomposition.

Separate the received signal into different frequency bands.

Remove the frequency band that corresponding to interference.

Reconstruct the signal using inverse DWT.

Challenge: spectral overlap between the signal and interference (both signals have high PSD at a low frequency region).

The denoising should be carried out to ensure that information lost is minimum.

System Descriptions

Transmitter filterp(t)

Matched filter r(t)

2Pavgn(t)

Outputslots

Inputbits

sample

mfl(t)

Multipath channel

h(t)

x(t) v(t)

R

z(t)y(t)

Wavelet Denoising

dis(t)

id̂

y(n) t

DWT Processing IDWT

FLI is a low frequency band signal, the approximation coefficients need to be manipulated.

For denoising proposes, the approximation coefficients corresponding to the FLI are made equal to zero so that reconstructed signal is free from FLI.

The signal is then reconstructed using the inverse DWT .

DWT based Denoising

Received OOK signal in the presence of the FL interference,

The eye-diagram of received signal corrupted by ALI

The eye diagram of received signal with wavelet denoising.

Complete closer of eye in the eye-diagram of the signal corrupted by ALI high BER.

Wide opening of the eye with wavelet denoising.

The number of decomposition level for DWT is calculated using:

where is the ceiling function

Approximate cut-off frequency of 0.5 MHz is used as it provide near optimum performance.

65.0log2 ETk s

x

DWT based Denoising

PSD of the OOK with FLI and DWT denoising at 2 Mbps

PSD of the OOK with FLI and DWT denoising at 200 Mbps

No significant changes in PSD at frequency > 0.5 MHz.

Significant portion of the spectral content at < 0.3 MHz is removed with no DC contents.

Spectral overlap between signal and interference power penalty.

Performance of OOK with DWT

The normalized OPP to achieve a error rate of 10-6 for OOK, 8-PPM and 8-DPIM for ideal and interfering channels and with DWT denoising at data rates of 10 - 200 Mbps.

DWT based receiver reduces the optical power requirement significantly.

Above data rate of 40 Mbps, the optical power penalty for OOK-NRZ is less than 1.5 dB.

Optical power penalty is the highest for OOK due to a high DC content.

Optical power penalty for PPM and DPIM is ~0.5 dB.

Since the PPM has zero spectral component near DC value, PPM offers improved performance.

DWT vs. HPF

DWT HPF

Performance Displays similar or better performance compared to the best achieved with the HPF.

Significantly inferior performance at high data rate compared to DWT.

Optimization

Optimization is not necessary as decomposition level can only be positive integer.

Optimization is necessary to obtain best performance.

Complexity Reduced complexity compared to HPF.Example, the maximum number operation for ‘db8’ wavelet is 60n, n length of input signal.

High.Example, for a HPF of order L, the total number of floating point operations is nL/2. L=2148 at data rate of 200 Mbps .

Realization Easy as repetitive structure is used. Realization becomes difficult with increasing in order.

0 0.5 1 1.5 20

1

2

3

4

5

6

7

8

Cut-off frequency (MHz)

Opt

ical

pow

er p

enal

ty (

dB)

20 Mbps50 Mbps100 Mbps200 Mbps

5 6 7 8 9 10 110

1

2

3

4

5

6

7

8

Decomposition level

Opt

ical

pow

er p

enal

ty (

dB)

20 Mbps50 Mbps100 Mbps200 Mbps

Implementation- DWT

Implementation- TI DSP

Using TI DMS320C6713 DSP board + Matlab/Simulink

DSP Board

Conclusion

Indoor optical wireless systems will have a major role in future indoor personal communication.

A number of key challenges needs to be address before fully potential can be realized.

Artificial light interference is a dominant noise source that impair the link performance.

Artificial light interference can be reduced effectively by using the discrete wavelet transform.

Discrete wavelet transform provide improved performance with reduced complexity compared to the high pass filter.

Discrete wavelet transform based denoising can easily be realized using DSP.

Acknowledgement

Northumbria university for providing an studentship.

My supervisors: Prof. Maia Angelova and Prof. Fary Ghassemlooy.

All my colleagues.

Finally my family members.