Optimum signal input distribution design in the presence of random pointing
jitter for intersatellite optical communications
Xin Li
n
, Jing Ma, Siyuan Yu, Liying Tan, Tao Shen
National Key Laboratory of Tunable Laser Technology, Institute of Opto-electronics, Harbin Institute of Technology, Harbin 150001, China
a r t i c l e i n f o
Article history:
Received 2 February 2012
Received in revised form
7 May 2012
Accepted 8 May 2012
Available online 29 May 2012
Keywords:
Input distribution
Channel capacity
Satellite optical communications
a b s t r a c t
Channel capacity is widely investigated for free space optical links to approach high-speed data-rate
communication. Instead of traditional equiprobable binary symbol input distribution, an optimum
input distribution is proposed with respect to channel capacity by maximizing mutual information for
intersatellite optical communications in the presence of random pointing jitter. It is shown that the
optimum input distribution varies with the variance of pointing jitter s and laser beam divergence
angle w
0
and the normalized intensity threshold I
T
. For traditional normalized intensity threshold
I
T
0.5, the optimum input distribution ranges from about p(x0)0.52 for weak pointing jitter to
about p(x0)0.24 for strong pointing jitter given the same laser beam divergence angle. The results
obtained in this paper will be useful for intersatellite optical communication system design.
& 2012 Elsevier Ltd. All rights reserved.
1. Introduction
With increasing demands for real-time communication
throughout the whole world, intersatellite optical communica-
tions have attracted considerable attention for potential high-
speed data-rate and global coverage access [15]. Channel
capacity, which describes the maximum achievable code rate
allowing reliable communication between transmitters and recei-
vers, is the key parameter for optical links and has been
researched by various studies [610]. The report by Boroson
shows the channel capacity limits for free space optical links [7].
And the impacts of pointing and tracking error on channel
capacity are also investigated for intersatellite optical commu-
nications [8,9]. The effects of misalignment and atmospheric
turbulence on terrestrial free space optical channel capacity
have been studied by Farid and Hranilovic [10]. However in all
case it is assumed that the equiprobable binary symbols input
distribution is applicable to the atmospheric or non-atmo-
spheric free space optical channels. However, equiprobable
input distribution only serves symmetric channels [6]. In fact
the channel for intersatellite optical communications is subject
to random pointing jitter, and the channel is asymmetric [11].
As a result the optimum input distribution is no longer
p(x0) p(x1) 1/2. The input probability is considered max-
imizing mutual information in order to nd the capacity
boundary for a photon-counting Poisson channel by Wyner
[12]. And the optimum input distribution is also given for ber
optical channel numerically [13]. However, an analytical opti-
mum input distribution for intersatellite optical channel has
not been presented as far as the authors know.
The optimum signal input distribution with respect to the
channel capacity for intersatellite optical communications in the
presence of random pointing jitter is the issue of concern in this
paper. First the channel matrix is modeled based on the statistic
characteristic of normalized optical intensity in the presence of
random pointing jitter. Then the optimum input distribution is
analytically derived by maximizing the mutual information.
Moreover, the channel capacity introducing optimum input dis-
tribution is also presented.
2. Channel matrix modeling
A typical intersatellite optical link is sketched in Fig. 1.
The transmitter modulates data onto the instantaneous intensity
of an optical beam, and it is de-modulated at the receiver. The
channel for intersatellite link is non-atmospheric free space. For
intersatellite optical communications the channel state is subject
to random pointing jitter, which induces optical intensity uctua-
tion at the receiver and has a signicant impact on the link
communication performance.
In the presence of random pointing jitter, the probability
density function (PDF) of the received optical intensity follows a
Beta distribution expressed as [14]
f I
w
2
0
4s
2
I
w
2
0
=4s
2
1
1
Contents lists available at SciVerse ScienceDirect
journal homepage: [Link]/locate/optlastec
Optics & Laser Technology
0030-3992/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
[Link]
n
Corresponding author.
E-mail address: cammilee@[Link] (X. Li).
Optics & Laser Technology 45 (2013) 705707
where IA[0,1] is the normalized intensity, w
0
is the divergence
angle of Gaussian laser beam, and s
2
is the variance of random
pointing jitter.
According to the Shannon capacity theory, the channel matrix
model must be developed to investigate the channel capacity. In this
paper, we consider the intensity-modulated/direct-detection (IM/DD)
with onoff keying (OOK) system, which is widely employed in
practical optical terminals [15,16]. With OOK, the transmitted signal
is xA{0,1} and received signal is yA{0,1}. In the presence of random
pointing jitter the received signal demonstrates different statistics
depending on whether 0 or 1 is transmitted.
With the development of photodetector, the detector noise can
be well constricted and in practical system design, threshold is
higher than the detector dark noise level. Hence, when the trans-
mitted signal is 0 the impact of detector noise can be neglected by
an appropriate threshold. In this case, we can assume that the
received signal can be identied as 0 when the transmitted signal is
0, which can be expressed by the conditional probability
py 09x 0 1 2
Furthermore, when the transmitted signal is 1, generally the
detector shot noise raises the received signal intensity level. Thus,
the impact of detector noise can be also neglected. However the
random pointing jitter results in optical intensity uctuation
which would decrease the intensity level at the receiver, which
is the main reason we investigate the inuence of random
pointing jitter on channel communication performances in this
paper. In this case, the probability of falsely identifying the
received signal as 0 when the transmitted signal is 1 is
determined by threshold level, which can be derived by the
conditional probability
py 09x 1
Z
IT
0
f I dI I
w
2
0
=4s
2
T
3
where I
T
is the normalized threshold of optical intensity at the
receiver.
Generally the channel matrix is described as
P
py 09x 0 py 19x 0
py 09x 1 py 19x 1
" #
4
Therefore, introducing Eqs. (2) and (3) into Eq. (4), the channel
matrix for intersatellite optical links in the presence of random
pointing jitter can be obtained as
P
1 0
I
w
2
0
=4s
2
T
1I
w
2
0
=4s
2
T
" #
5
3. Optimum signal input distribution
Channel capacity is dened as the maximum of the mutual
information MI(x;y) between the channel input and output [6,16].
The maximization is done over the input distribution p(x)
C max
p x
MIx; y 6
MIx; y
X
2
j 1
py
j
log
2
py
j
X
2
i 1
X
2
j 1
px
i
py
j
9x
i
log
2
py
j
9x
i
7
X
2
i 1
px
i
1 8
where C denotes the Shannon capacity, the range of which is [0,1]
for OOK system, Eq. (7) is the expression of mutual information,
and Eq. (8) is the constraint of Eq. (6).
The maximum mutual information MI(x;y) can be found
through Lagrange multiplier method, which is widely used in
mathematical optimization and provides a strategy for nding the
maximum or minimal of a function subject to constraints.
Introducing a new parameter l called the Lagrange multiplier,
we dene the Lagrange function as
LMIX; Yl
X
2
i 1
px
i
1
" #
9
The critical values of L occur where its gradient is zero. And the
partial derivations are
@L
@p x
i
X
2
j 1
py
j
9x
i
log
2
py
j
py
j
9x
i
log
2
e
X
2
j 1
py
j
9x
i
log
2
py
j
9x
i
l 0 10
@L
@l
X
2
i 1
px
i
1 0 11
Solving these equations yields the values of p(x), which
maximize the mutual information. Dening e I
w
2
0
=4s
2
T
for simpli-
cation, the optimum p(x) obtained from Eqs. (10) and (11) can be
described as
px 0
1e
1=1e
11ee
e=1e
12
px 1
e
e=1e
11ee
e=1e
13
It should be noted that eA[0,1] since I
T
A[0,1], therefore the
probabilities we got here are non-negative. And it is obvious that
the optimum signal input distribution satises Eq. (8). Conse-
quently, the results obtained here are validated.
Eqs. (12) and (13) show the relation between optimum input
distribution and channel state analytically. It is also indicated that
the optimum input distribution is dependent on normalized
threshold I
T
, divergence angle w
0
, and especially the standard
variance of random pointing jitter s. Introducing the system
design parameter bw
0
/s, which is dened as the ratio of
divergence angle to the standard variance of pointing jitter, we
assess the optimum signal input distribution for various channel
and system conditions based on Eqs. (12) and (13).
Fig. 2 shows the optimum signal input probability of x0 as a
function of normalized intensity threshold I
T
for various system
design parameters b. It is observed that optimum probability of x0
is different with traditional 1/2 for various channel and system
conditions. In the presence of random pointing jitter, the channel is
asymmetric, and the input distribution that maximizes the mutual
information is no longer p(x0)p(x1)1/2. It is also observed
that the optimum input distribution varies with the jitter strength
Transmitter
Receiver
Random point jitter angle
Free space channel
Fig. 1. Sketch of intersatellite optical link.
X. Li et al. / Optics & Laser Technology 45 (2013) 705707 706
and system parameter. Given the same divergence angle w
0
, the
stronger the random pointing jitter, the smaller the b. For strong
random pointing jitter b1, the optimum input probability of x0
decreases with increasing I
T
. And for weak random pointing jitter
b2,5, the optimum input probability of x0 rst increases slowly
and then decreases fast with increasing I
T
. It is implied that the
strength of random pointing jitter has a dramatic impact on
optimum input distribution. For traditional normalized intensity
threshold I
T
0.5, the optimum input distribution ranges from about
p(x0)0.52 for weak pointing jitter to about p(x0)0.24 for
strong pointing jitter with the same laser beam divergence angle.
Substituting the optimum input distribution Eqs. (11) and (12)
into mutual information Eq. (6), we can obtain the channel capacity,
which is the maximum mutual information, expressed as
C log
2
11ee
e=1e
h i
14
Eq. (13) shows the dependence of channel capacity on channel
state. According to Eq. (13), Fig. 3 shows the channel capacity as a
function of normalized intensity threshold I
T
for various system
design parameters b. It is observed that the lower the normalized
intensity threshold is, the higher the channel capacity. This is
because we assume that detector noise is neglectable. In fact the
detector noise exists, hence the normalized intensity threshold
cannot be so low to approach channel capacity of 1. It is also
observed from Fig. 3 that the higher the system design parameter b,
the better the channel capacity. It results from the fact that for a
given laser beam divergence angle, the higher the b, the lower the
variance of pointing jitter, in which case the channel is more stable.
4. Conclusion
The optimum signal input distribution is investigated in this
paper with respect to channel capacity for intersatellite optical
communication in the presence of random pointing jitter. First the
channel matrix is obtained in the presence of random pointing jitter.
And then the optimum input distribution is derived as a function of
normalized intensity threshold system I
T
, laser beam divergence
angle and variance of pointing jitter by maximizing the mutual
information. Numerical simulation shows that the optimum input
distribution varies with the jitter strength and system parameter.
For traditional normalized intensity threshold I
T
0.5, the optimum
input distribution ranges from about p(x0)0.52 for weak point-
ing jitter to about p(x0)0.24 for strong pointing jitter with the
same laser beam divergence angle. Finally the channel capacity is
presented introducing the optimum input distribution. The results
obtained in this paper will be useful for intersatellite optical
communication system design.
Acknowledgments
This work is supported by the National Natural Science Foundation
of China (NSFC) nancially with Project no. 10904026.
References
[1] Kaufmann JE, Chan VWS. Coherent optical intersatellite crosslink systems.
Proceedings of the IEEE 1988;2:053340.
[2] Mahdieh MH, Pournoury M. Atmospheric turbulence and numerical evalua-
tion of bit error rate (BER) in free space communication. Optical Laser
Technology 2010;42:5560.
[3] Chan VWS. Optical satellite networks. IEEE: Journal of Lightwave Technology
2003;21(11):281127.
[4] Toyoshima M. Trends in satellite communications and the role of optical free-
space communications. Journal of Optical Networking 2005;6587:30011.
[5] Cesarone RJ, Abraham DS, Deutsch L. Prospects for a next-generation deep
space network. Proceedings of the IEEE 2007;95(10):190215.
[6] Shannon CE. A mathematical theory of communication. Bell System Technical
Journal 1948;27:379423.
[7] Boroson DM. Channel capacity limits for free space optical links. Proceedings
of the SPIE 2008;6951 69510A-1-11.
[8] Anguita JA, Djordjevic IB, Neifeld MA, Vasic BV. Shannon capacities and error-
correction codes for optical atmospheric turbulent channels. Journal of
Optical Networking 2005;4(9):586601.
[9] Barron RJ, Boroson DM. Analysis of capacity and probability of outage for
free-space optical channels with fading due to pointing and tracking error.
Proceedings of the SPIE 2006;6105 61050B-1-12.
[10] Farid AA, Hranilovic S. Outage capacity optimization for free space optical
links with pointing errors. Journal of Lightwave Technology 2007;25:
170210.
[11] Anguita JA, Djordjevic IB, Neifeld MA, Vasic V. Shannon capacities and error-
correction codes for optical atmospheric turbulent channels. Journal of
Optical Networking 2005;4(9):586601.
[12] Wyner AD. Capacity and error exponent for the direct detection photon
channel. IEEE Transactions: Information Theory 1988;34:144971.
[13] Ho KP. Exact evaluation of the capacity for intensity modulated direct
detection channels with optical amplier noises. IEEE Photonics Technology
Letters 2005;17:85860.
[14] Toyoshima M, Jono T, Nakagawa K, Yamamoto A. Optimum divergence angle
of a Gaussian beam wave in the presence of random jitter in free-space laser
communication systems. Journal of the Optical Society of America A
2002;19(3):56771.
[15] Li J, Uysal M. Optical wireless communications: system model, capacity and
coding. Proceedings of the IEEE 2003;1:16872.
[16] Khalighi MA, Schwarts N, Aitamer N, Bourennane S. Fading reduction by
aperture averaging and spatial diversity in optical wireless systems. Journal
of Optical Communications and Networking 2009;1(6):58093.
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
I
T
P
r
o
b
a
b
i
l
i
t
y
o
f
x
=
0
= 1
= 2
= 5
Fig. 2. Probability of x0 as a function of normalized intensity threshold I
T
for
various system design parameters b.
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
I
T
C
h
a
n
n
e
l
C
a
p
a
c
i
t
y
(
b
i
t
s
/
s
y
m
b
o
l
)
= 1
= 2
= 5
Fig. 3. Channel capacity as a function of normalized intensity threshold I
T
for
various system design parameters b.
X. Li et al. / Optics & Laser Technology 45 (2013) 705707 707