Вы находитесь на странице: 1из 5

FSO transmission of halftoned image over DGG

turbulence channel

Stefan Pani, Hranislav Miloevi Bojan Prlinevi


Faculty of Natural Science and Mathematics Higher Technical Professional School in Zvecan
Kosovska Mitrovica, Serbia Zvecan, Serbia
Emails: stefanpnc@yahoo.com mhrane@gmail.com Email: prlincevic@hotmail.com

Vera Petrovi Olga Taseiko


The School of Electrical and Computer Engineering of Institute of Computational Technologies of SB RAS
Applied Studies Krasnoyarsk, Russia
Belgrade, Serbia Email: taseiko@gmail.com
Email: vera.petrovic@viser.edu.rs

AbstractIn this paper we will analyze free space optics by short dimension of wavelengths used for optical
(FSO) transmission of halftoned image over Double Generalized transmission. However, FSO transmission manifests some
Gamma (DGG) turbulence channel. First, we will present undesired characteristics that occur during transmission, such
halftoning method, and algorithm for FSO transmission
simulation. Further, we will observe standard performance
are: acquisition and pointing are subjected to difficulties,
criterions of reconstruceted image, such are Bit Error Rate various harmful effects arise due to influence of atmospheric
(BER) and Peak Signal-to-Noise Ratio (PSNR) versus factors (i.e. haze, fog, rain, Sun, turbulence). Major
properties of observed FSO link: link distance, shaping and performance impairment in FSO links, which degrades the link
severity parameters of irradiance and Rytov variance. Presented performance is atmospheric turbulence-induced fading [9].
analysis is very general and encompasses a large number of FSO Time varying spatial fluctuations of the optical beam appear as
transmission scenarios, since Double Gamma-Gamma turbulence
an outcome of atmosphere refractive index variations. These
channel model as general turbulence channel model can be
reduced to various turbulence channel models as it special cases. variations are caused by atmospheric turbulence, and are
displayed as irradiance fluctuations of the received signals.
Keywords Free-space optical communication, Double Such phenomena is known as FSO fading or scintillation.
Generalized Gamma Channel, Image processing, Halftone image. Mixed RF/FSO cooperative systems were observed in [20-21].
Various attempts have been made in order to obtain for
I. INTRODUCTION accurate mathematical and numerical modeling of FSO
propagation. Double Weibull distribution turbulence model is
M ultimedia content is very present in everyday
communication, since distribution of multimedia content mostly used for modeling of scintillation of signal with
different intensities of turbulences. This model fits turbulence
such as newspapers and books and is very present in all forms
of digital communication. Some of produced images are measurements within systems with large aperture on the
produced over the halftoning and they contain only two colors, receiving side [10]. Log-Normal distribution turbulence model
black and white. When this halftone image is viewed from a [11] on the other hand, corresponds to the regimes of weak
certain distance, it appears as the original image [1]. Halftoning atmospheric turbulence [3], [12]. Best fit for turbulence
methods developed in past years can be categorized into scenarios that occurs in terrestrial communication over sparsely
ordered dithering [2], error diffusion [3], [4], dot diffusion [5], populated areas and suburbs is Ricean turbulence model[13].
[6] and direct binary search [7], [8]. Gamma-Gamma turbulence model is simplified model of
Usage of high-data-rate free-space optical (FSO) scintillation that can be often used for various propagation
transmission systems has grown recently, since FSO scenarious [14]. Moreover, general turbulence distribution
transmission obtains many advantages over radio frequency model Double Generalized Gamma (DGG) turbulence model,
(RF) transmission. Some of the advantages are: narrow beam was recently proposed [13], [15]. This model is very general
of emission of power, resistance to EMI interference, for since it encompasses many existing turbulence channel models
Frenzel zone is not neccessary, full duplex transmission is as its special cases, because it provides an excellent fit to the
enabled, practical realization is simple, and transmission is measured magnitudes of plane and spherical FSO waves. By
license-free. Main advantages of FSO transmission are caused setting the predefined values to the DGG model parameters,
this model is transformed to other above mentioned turbulence distributed random process with variance expressed as [17],
models. [18]:
In this we will analyze properties of FSO transmission of
halftoned image through DGG channel. The paper is composed sh2 = 2qg 2 FA RPmI
t f (5)
in the following way: section II describes halftoning method
and DGG FSO channel model. Based on simulation results and
performances analysis is carried out in section III. Section IV with constant q representing an electron charge. Parameters g
presents conclusion of the paper. Main contribution of the and R stand for the gain and responsivity, respectively, while
work is that it provides approach proposed for estimation of Pt denotes transmitted optical power. Parameter m denotes
boundary values of FSO link parameters, that are neccesary in modulation index, variable I represents accounted normalized
FSO link designing proccess for obtaining required values of irradiance, while FA denotes the excess noise factor expressed
quality of service (QoS), when simmilar last-mile FSO
transmission scenarious of halftoned image are observed. as [18]:

II. SYSTEM MODEL FA = k A g + (1 k A )(2 1 g ) (6)

Received FSO signal from the aperture plane can be


expressed as: with k A denoting the ionization factor.
Now, total noise variance could be expressed as the sum of
Er (t , r ) = us (t ) exp( j 2 f c t + (t )) exp[ (r ) + j (r )] (1) noise variances:

T
with r denoting the position vector on the receive aperture 2 = 2 + 2 = 4k B Fn f + 2qg 2 FA RPmI
t f (7)
plane and fc defining the optical carrier frequency. Part of the n Th sh
RL
expression us(t)exp(j(t)) stands for the complex envelope of
the modulation signal. Finally, in the exponent of the Information carrying part of signal can be expressed as:
expression, the turbulence-induced amplitude fluctuations and
the phase variations of the channel are introduced through (r) e
and (r), respectively. xT (t ) = AD 2 us (t ) Re { I exp( j 2 f IF t + IF (t ))} (8)
2hf c
Total output photocurrent at the receptor can be expressed
as:
with fIF = fc - fLO denoting the carying signal frequency, and
yT (t ) = xT (t ) + nT (t ) (2) fLO denoting local oscillator frequency. Equivalent signal
phase and effective fluctuation of the channel are represented
as IF and respectively. Simultaneous influence of
with nT(t) representing total noise at the receiver obtained as turbulence-induced amplitude fluctuations and phase
the sum of thermal noise, nTh , and active component shot aberrations, could be modeled through random irradiance
fluctuations distributed as Double Generalized Gamma
noise, nsh , [17], [18]:
process, expressed as [15]:

n(t ) = nTh + nsh (3) 2 pp m 1/ 2 q m 1/ 2 2 1 ( p + q )/ 2 I 1


2 1

fI (I ) =
(m1 )(m2 )
Thermal noise is mainly considered to be zero-mean Gaussian (9)
p p q q 1q (q :1 m1 ), ( p :1 m2 )
p

distributed random process with variance expressed as [17], G p0,+pq+,0q 22 ;


[18]:
q
I m1 m2
p

T where G pm,,qn (x) stands for the Meijers G-function defined in


Th2 = 4k B Fn f (4)
RL [19], (x) denotes special Gamma function [19], p and q are
positive integer numbers that satisfy relation p/q = 1/2, and
where k B stands for the Boltzmann constant while Fn denotes ( j; x ) x / j ,, ( x + j 1) / j . Turbulence parameters 1, 2,
amplifier noise figure. Parameters T and RL define 1, 2, m1, m2, are defining nature of turbulence process,
temperature and load resistance respectively, while f denote namely m1, m2 are shaping parameters defining the turbulence-
induced propagation, while parameters 1, 2 are defining the
the effective noise bandwidth, a function of bit rate, Rb , severity levels of independent irradiances forming observed
f = Rb 2 . turbulence. Analytically their values could be tracked by using
Shot noise is mainly considered to be zero-mean Gaussian the moments of small and large-scale irradiance fluctuations.
Values of DGG model parameters along FSO link are directly been provided and images were reconstructed.
tied to the atmospheric parameters (the ratio of Fresnel zone, Step 5: Qualities of reconstructed images are analyzed.
Rytov variance, 2Rytov, scintillation index, wavenumber, As the quality measure for the image transmission Peak
k = 2/, the refractive index structure constant, and the Signal-to-Noise Ratio (PSNR) and Bit Error Rate (BER) are
propagation distance) with relations presented in [14]. Power used, defined as:
defining parameters 1 and 2 are related to atmospheric
conditions as explained in [14]. Assuming a plane wave with 2n 1
PSNR = 10 log10 , where
inner scale effects considered, the Rytov variance (2Rytov) can MSE
(x yij )
be presented as: 2 (12)
ij

MSE = i = 1...M , j = 1...N


ij
2Rytov
= 1.23Cn2 k 7 / 6 L11/ 6 (10)
M N

where L - propagation distance and Cn2 refraction index, while


k=2/ and denotes the FSO wavelenght.
( x ij l) ( yij )l
2 2
BER =
ijl
, (13)
Now, instantaneous signal-to-noise ratio (SNR) at the M N
reception can be expressed as: i = 1...M , j = 1..N , l = 1...n.

Px Px where: xij, yij -pixel of original image and transmitted image, n-


= = = number of bits, M N -the size of the image, and denotes
2 n 2
2( Th + sh2 )
2

EXOR operator over each of n pair of bits from xij and yij.
e The values of the Rytov variance 2Rytov are obtained for
AD 2 I s (11)
2hf c FSO wavelenght of l = 1550 nm and different values of
= I 2
T refraction index Cn2 and propagation distance L for each
2(4k B Fn f + 2qg 2 FA RPmI
t f )
RL source image from fig.1. Images presented on fig. 2 are
halftoned images obtained by using error diffusion method [4].
where Is=|us(t)|2 denotes the average intensity of the optical
field, while Px stands for the output signal power.
Algorithm for simulation of FSO halftoned image
transmission is accomplished as follows:
Form halftoned digital image A* is created vector with with
binary elements D dimension 24MN. On vector D is applied
Binary Phase Shift Keying (BPSK) modulation X = 2 D 1 .
Now signal is transmitted trough turbulence channel:
Y=H*X+N, with H being vector that introduces the influence
of turbulence and is DGG distributed, while vector N
introduces level of total noise in the channel through defined
SNR level. At reception, first Grey decoding with hard a) b)
decisioning has been provided on Y, and reconstructed
halftoned A* image is obtained from vector D .

III. SIMULATION AND PERFORMANCE ANALYSIS


FSO transmission of the halftoned images over the DGG
distributed turbulence channel has been simulated through
following further procedure:
Step 1: Source images from fig. 1 are imported from the base
and halftoned with error diffusion method (fig. 2).
Step 2: BPSK modulation was performed over obtained binary
representation of images. c) d)
Step 3: The BPSK-signal, X, is transmitted trough turbulence Fig 1. Images used in the paper: a) Lena, b) Girl, c) Baboon
channel: Y=H*X+N, with H being vector that introduces the and d) Peppers
influence of turbulence and is DGG distributed, while vector
N introduces level of total noise in the channel through Reconstructed image after transmission through DGG FSO
defined SNR level. Procedure is repeted for different values of channel are presented at fig. 3.
the Rytov variance.
Step 4: At reception, Grey decoding with hard decisioning has
a)
Fig 3. Quality measurement BER for transmitted halftoned image Lena,
for constant values of turbulence shaping and severity parameters (1, 2, m1,
m2), and varied values of propagation distance and refraction index.

b)
Fig 2. Transmitted halftoned image Lena with apameter values: a)
2
1 = 2.0, 2 = 2.0, m1 = 1, m2 = 1, Rytov =2, and b) 1 = 2.0, 2 = 2.0, m1 = 1,
2
m2 = 1, Rytov =5.

Quality measure values for observed images are given on fig. 4


and fig. 5 and in Table I. After analyzing PSNR and BER
Fig 4. Quality measurement BER for transmitted halftoned image Lena,
values obtained for the halftoned image, transmitted through
for constant values of propagation distance and refraction index and varied
DGG turbulence channel (presented on fig. 4 and fig. 5, and in values of turbulence shaping and severity parameters (1, 2, m1, m2).
Table I), we can see that the transmitted image is reached very
good quality with PSNR values for PSNR = 66.19 and values
of BER smaller then 10-6 for DGG channel parameter value TABLE I. QUALITY MEASURE PSNR
1 = 2.0, 2 = 2.0, m1 = 2.0 and m2 = 3.5. Further performance
Parameters PSNR
improvement could be obtained by designing FSO link 1 = 2.0, 2 = 2.0, m1 = 2.0, m2 = 2.0 60.87
parameters (optical link distance, transmitted signal power, 1 = 2.0, 2 = 2.0, m1 = 3.5, m2 = 2.0 57.16
optical bandwidth, carrier signal wavelength, aperture 1 = 2.0, 2 = 2.0, m1 = 2.0, m2 = 3.5 66.19
diameters at the transmitter / receiver) in a way that quality 1 = 0.85, 2 = 2.0, m1 = 2.0, m2 = 2.0 34.47
measurement PSNR to be higher and BER to be smaller. 1 = 2.0, 2 = 0.85, m1 = 2.0, m2 = 2.0 33.92
IV. CONCLUSION through turbulent media" Optical Engineering, vol. 40,no. 8, pp. 1554
1562, 2001.
FSO transmission of the halftoned image over the DGG [10] R. Barrios, F. Dios, Exponentiated Weibull model for the irradiance
turbulence channel has been considered through the prism of probability density function of a laser beam propagating through
received PSNR and BER values. FSO channel parameter atmospheric turbulence, Optics & Laser Technol., Issue 45, 2013, pp.
values 1 , 2 , m1 and m2 are determined for which is obtained 1320.
that quality measures PSNR and BER reach very good quality [11] M.A. Al-Habash, L.C. Andrews and R.L. Phillips. "Mathematical model
of PSNR and satisfactory values of BER. Capitalizing on for the irradiance probability density function of a laser beam
propagating through turbulent media", Opt. Engineering 40, 15541562,
proposed method FSO link parameters could be efficiently 2001, ISSN:0091-3286.
determined in the FSO link designing process, in order to [12] A. Biswas, V. Vilnrotter, W. Farr, D. Fort, and E. Sigman, "Pulse
reach demanded PSNR and BER quality of halftoned image at position modulated ground receiver design for optical communications
the reception. from deep space," Proc. SPIE, San Jose, vol. 4635, 2002, pp. 224-235.
[13] A. Belmonte and J. M. Kahn, "Performance of synchronous optical
receivers using atmospheric compensation techniques" Optics Express,
vol. 16, no. 18, Sept. 2008, pp. 1415114162.
ACKNOWLEDGEMENT [14] A. Juardo-Navas, J.M. Garrido-Balsells, J.F. Paris, et al. "A unifying
statistical model for atmospheric optical scintillation", invited chapter in
This research was partly supported by the Ministry of "Numerical Simulations of Physical and Engineering Processes," 2011,
Education, Science and Technical Development of Republic pp. 181-206.
of Serbia, grant no. III44006 [15] M. Kashani, M. Uysal, M. Kavehrad, A novel statistical model for
turbulence-induced fading in free-space optical systems, in Proceedings
of the 15th International Conference on Transparent Optical Networks,
REFERENCES Cartagena, Colombia, June 2013, pp. 15.
[16] S. R. Panic, B. P. Prlincevic, P. C. Spalevic, A. Amniesi, M. S.
[1] Y. Gao, O.C. Au, J. Zhou, K. Tang and X. Fan, Halftone image Radojkovic, "FSO transmission of double watermarked image over
watermarking via optimization, Signal Processing: Image Double Generalized Gamma turbulence channels", IEEE Conference
Communication, vol. 41, pp. 85-100, 2016. Proceeding, ERK 2016, pp. 97-100.
[2] P. Freitas, M. Q. Farias, A.F.Arajo,Hiding color watermarks in [17] N. D. Chatzidiamantis, H. G. Sandaliadis, G. K. Karagiannidis, S. A.
halftone images using maximum-similarity binarypatterns, Signal Kotsopoulos and M. Matthaiou, New Results on Turbulence Modeling
Processing:ImageCommunication48(2016)111, 2016. for Free-Space Optical Systems, 17th International Conference on
[3] X. Li, Edge-directed error diffusion halftoning, IEEE Signal Process. Telecommunications, pp. 487 - 492, 2010
Lett. 13 (11) (2006) 688690. [18] D. A. Luong, T. C. Thang, A. T. Pham, "Effect of avalanche photodiode
[4] Y. Saika, Inverse-Halftoning for Error Diffusion Based on Statistical and thermal noises on the performance of binary phase-shift keying
Mechanics of the Spin System, International Conference on Neurel subcarrier-intensity modulation/free-space optical systems over
Information Procesing, (2007), pp. 663-672. turbulence channels," IET Communications, Vol. 7, No. 8, 2013, pp.
738744.
[5] J. Guo, Y. Liu, Improved dot diffusion by diffused matrix and class
matrix co-optimization, IEEE Trans. Image Process. 18(8) (2009) [19] I. .S. Gradshteyn, I.M. Ryzhik, "Table of Integrals, Series and Products",
18041816. Academic Press, NY, USA, 2000.
[6] Y. Liu, J. Guo, New class tiling design for dot-diffused halftoning, [20] B. Asharafzadeh, E. Soleimani-Nasab and M. Kamandar, "Performance
IEEE Trans. Image Process. 22(3) (2013) 11991208. analysis of mixed DGG and generalized Nakagami-m dual-hop FSO/RF
transmission systems," 2016 24th Telecommunications Forum
[7] P. Goyal, M. Gupta, C. Staelin, M. Fischer, O. Shacham, J. Allebach, (TELFOR), Belgrade, 2016, pp. 1-4.
Clustered-dot halftoning with direct binary search, IEEE Trans. Image
Process. 22(2) (2013) 473487. [21] B. Asharafzadeh, E. Soleimani-Nasab and M. Kamandar, "Outage
analysis of mixed RF/FSO cooperative systems with multiple antennas
[8] J. Guo, Y. Liu, J. Chang, J. Lee, Efficient halftoning based on multiple and apertures," 2016 24th Telecommunications Forum (TELFOR),
look-up tables, IEEE Trans. Image Process. 22(11) (2013) 45224531. Belgrade, 2016, pp. 1-4.
[9] M. Al-Habash, L. C. Andrews, R. L. Phillips, "Mathematical model for
the irradiance probability density function of a laser beam propagating

Вам также может понравиться