Wireless Optical CDMA

Communication Systems

Jawad A. Salehi, Babak M. Ghaffari, and Mehdi D. Matinfar

Optical Networks Research Lab.

EE Dept., Sharif University of Technology
Tehran, Iran

Introduction
Amidst revolutionary communication technologies of recent decades such as op-
tical fiber and wireless systems other forms of communication technologies are
popping up in order to complete or complement the ever increasing and insa-
tiable need for communications links in today’s society. Among many such niche
technologies optical wireless communication technology, in particular, is begin-
ing to enjoy a wide range of applications and attentions from many industries
being for short range interchip applications or interplanetary space applications.
Furthermore, optical wireless systems have attracted even further considerable
growth in research and development since they can be appropriate alternatives
for wireless or fiber-optic communication systems in some specific applications.
Indoor wireless LANs, atmospheric optical links and submarine optical wireless
systems have acquired in importance where lightwave communications is pre-
ferred to radio communications. This preference in various applications can be
originated from security requirements, radio interference avoidance, no need for
reserving frequency bands and cost of development [1]– [9].
In this chapter we study and analyze a particular and advanced form of
optical wireless communication systems namely optical code-division multiple-
access (OCDMA) in the context of wireless optical systems. As wireless optical
communication systems gets more mature and become viable for multi-user
communication systems, advanced multiple-access techniques become more im-
portant and attractive in such systems. Among all multiple-access techniques in
optical domain, OCDMA is of utmost interest because of its flexibility, ease of
implementation, no need for synchronization among many users and soft traffic

1
2

handling capability.
Intensity modulation/direct detection (IM/DD) schemes are widely used in
wireless optical applications. The interest on using IM/DD schemes stems from
the fact that they can be easily implemented in transmitter and receiver sides.
For instance, in optical wireless indoor systems design portion of a wireless
OCDMA transmitter and receiver based on IM/DD can be established easily
using existing and off the shelf technology. On the other hand it is shown that
in the optical scatter channels it is more efficient to employ incoherent optical
modulations [1]. In these cases system structure can be designed based on on-off
keying (OOK) or pulse position modulations (PPM) familiy [2]. Optical orthog-
onal codes (OOC), that have been introduced for use in OCDMA systems are
binary unipolar codes that have been shown to be optimum in OCDMA systems
by mean of limiting the multiple-access interference (MAI) noise [10]– [13]. By
employing OOC in the OCDMA systems, each user transmits its corresponding
data bits on a sparse sequence of optical pulses which constitute its dedicated
code. Such pulsing scheme allows users to communicate simultaneously on the
shared channel without the requirement to be synchronized among each other.
On the other hand in this case the maximum allowable peak power of optical
pulses can be greater than ordinary systems due to multiple pulse OOC struc-
ture. This feature is effective in optical wireless systems where the peak power
is limited due to the eye safety conditions.
In this chapter we study deployment of OCDMA communication systems
in both indoor and outdoor (free-space optical link) environment. In applica-
tions where electromagnetic interference is to be avoided, optical networks are
preferred to radio networks. The networks inside the passenger planes and hos-
pitals are categorized into such group applications. Intra-satellite networks for
small scientific spacecrafts is another example that follow this rule too. Employ-
ing infrared communications as the intra-satellite network not only it prevents
electromagnetic interference but also it reduces the volume and the weight of
the spacecraft. This point is critical since there are several hundreds of wire
bundles and connectors that about one half are dedicated to data transmis-
sion [14]. Usually there are several transmit/receive nodes in these networks
that need moderate bit rate. Employing a multiple-access technique in this case
in order to support communications in the shared physical channel is necessary.
OCDMA is an excellent choice since the implementation of the receiver is simple
and all optical devices operate on the same wavelength and there is no need to
synchronize various nodes.
Free-space optical (FSO) links are extremely important in many applications
especially when installation of fiber between various nodes is expensive or is not
feasible for various reasons. In fact a FSO link can be part of a bigger optical
network that is using OCDMA as the multiple-access technique. We study
and investigate the performance of an OCDMA FSO link. In our analysis we
consider a realistic model that has been implemented recently. We use 1550 nm
wavelength in order to integrate FSO link with high speed fiber optic network.
In this wavelength, EDFA is a good choice to being used as the post- and
(or) pre-amplifier. However in the case that the amplifier gain is high enough,
 3

Figure 1: OCDMA general block diagram

employing high power post-amplifier (booster) can be more efficient than pre-
amplifier since post-amplifier only amplifies the signal but not the noise as the
pre-amplifier does.
In what follows we study OCDMA principles along with the structure of
OOC and follow up with the introduction two receiver structures for wireless
OCDMA systems.

OCDMA System Description

In OCDMA systems each user transmits it’s corresponding data bits by its
corresponding code. The codes in optical systems may be constructed in differ-
ent domains such as time, wavelength, polarization, space or a combination of
them [15]. Due to special conditions of wireless optical channels such as scin-
tilation and phase fluctuations, employing temporal codes in wireless OCDMA
system in the context of intensity modulation and direct detection is shown to be
the most attractive [17]– [19]. Hence from now on we consider wireless OCDMA
systems with their user codes designed in time domain. In such systems a data
bit is mapped on a sequence of ‘0’ and ‘1’s in time domain which constitute the
corresponding code. The codes are designed such that are semi-orthogonal. In
this case the correlation at the receiver side can distinguish data bits belonging
to different users. We have shown the general block diagram of an OCDMA
system with N users in Fig. 1. For each user at the transmitter there is an
encoder which generates the coded data stream. We assume optical orthogonal
codes (OOC) as the signature sequence in the system. Data stream of users are
combined, i.e., added, in the shared media. At the receiver side for each user
there is a decoder that attempts to regenerate the data bit of the corresponding
user among all users data bits.

OOC
Optical orthogonal codes (OOC) are family of (0,1)-sequences with good auto-
and cross-correlation properties providing asynchronous multiple-access com-
munications with easy synchronization and efficient performance in OCDMA
communication networks.
An optical orthogonal code (L, w, λa , λc ) is a family C of (0, 1)-sequences of
4

length L with constant Hamming-weight w satisfying the following two proper-

ties:

1. Auto-correlation property: for any codeword x = (x0 , x1 , ..., xL−1 ) ∈ C,

L−1
the inequality i=0 xi xi⊕τ ≤ λa holds for any integer τ ≡ 0 (mod L)

L−1 property: for any two distinct codewords x, y ∈ C, the

2. Cross-correlation
inequality i=0 xi yi⊕τ ≤ λc holds for any integer τ ,

where the notation ⊕ denotes the modulo-L addition [20]. The small value for
λa and λc guarantees easy synchronization (of each transmitter-receiver pair)
and limited multiple-access noise, respectively. When λa = λc = λ, we denote
the OOC by (L, w, λ) for simplicity. The number of codewords in a family is
called the size of optical orthogonal code. From a practical point of view, a
code with large size is desired [11]. To find the best possible codes, we need
to determine an upper bound on the size of an OOC family with the given
parameters. Let Φ(L, w, λa , λc ) be the largest possible size of an (L, w, λa , λc )-
OOC. It is easy to show that if w(w − 1) > λa (L − 1) then Φ(L, w, λa , λc ) = 0
and if w2 > λc L then Φ(L, w, λa , λc ) ≤ 1 [21]. However, based on the Johnson
bound for constant-weight error correcting codes [22], we have the following
bound [20],
       
1 L−1 L−2 L−λ+1 L−λ
Φ(L, w, λ) ≤ ... ... (1)
w w−1 w−2 w−λ+1 w−λ

where the notation  denotes the integer floor function. Also, it is clear from
definition that Φ(L, w, λa , λc ) ≤ Φ(L, w, λ) where λ = max { λa , λc } [20].
From the mutual interference point of view, the most desirable on–off sig-
nature sequences are OOCs with λc = 1. However, when we look at the system
with a few number of users interacting on each other’s OOCs with λc = 1 is
not necessarily the optimum case. The main reason being that there is a trade
off between the code weight, w and λc . Larger λc allows more interference be-
tween users but on the other hand allows having larger code weight which by
itself increases the coding gain hence decreasing the total interference. Briefly,
the number of users in the network, receiver structure and target bit-error-rate
define the optimum values of OOC parameters [12].
In what follows we study two main structures for OCDMA receiver and
investigate their characteristics in an optical channel.

Correlator Structure
The simplest structure for an OCDMA receiver, namely, correlator includes a
matched filter, corresponding to its code pattern, and an integrate and dump
circuit. We have shown the structure of OCDMA transmitter including encoder,
and the correlator receiver in Fig. 2. In this block diagram decoder is matched to
the corresponding code generated by the encoder. In fiber-optic CDMA systems
the matched filter can be implemented by fiber tapped-delay lines at the receiver
 5

Figure 2: OCDMA transmitter, and correlator receiver

side. Delay lines in each branch serve to compensate for the corresponding
delays induced at the encoder which constitute the corresponding OOC. The
properly delayed replicas add to construct the output correlator which is then
fed to the photodetector, integrate, dump, and thresholder electronic circuit.
In Fig. 2 the white pulses at the input of the desired decoder, are the received
signal from undesired users. The major advantage of the correlator structure
is that the limiting high-speed chip time decoding operation is done passively
and optically. On the other hand in digital electronic implementation, delays
can be generated by employing flip flops. If we assume the sampling rate per
chip equal to ns , ns L flip flops are sufficient to construct the matched filter of
the corresponding OOC code [18]. If we denote the code length, the weight
and the chip duration by L, w and Tc respectively, and if the positions of the
pulsed mark chips corresponding to the code sequence is c1 ,c2 ,. . . , andcw , then
the optical matched filter (passive optical tapped-delay line) consists of w fiber
delay lines each of which make a delay equal to (L − c1 )Tc , (L − c2 )Tc ,. . . , and
(L − cw )Tc respectively. Considering only one bit duration, the output of the
matched filter which is the combination of all w lines output signal has a peak
at the time LTc if the transmitted bit is on. However, when the number of
interfering users is equal to or greater than w, we may observe such a peak in
the absence of the desired user’s signal, i.e., the transmitted bit is off, and this
causes an error.
Although the correlator receiver seems trivial, in the sense that it is optimum
receiver in the case of single user communication with no interference, it can
be much improved taken into account other interfering users using some other
OOC codes on the same optical channel.

Chip-level AND Logic Gate Structure

The correlator structure is not resistant enough against multi-access interference
since it accumulates all the received energy from the marked chips independent
6

Figure 3: Chip-level AND logic gate receiver

of the number of interfering pulses on each specified marked chip. The first
improvement to the correlator structure was suggested in [11]. The idea was
further studied in-depth in [23] and [24]. In this method in every definite chip
time just that amount of intensity transmitted for a ‘1’ bit may carry informa-
tion and any excess intensity is due to interference. Thus an optical hard-limiter,
which limits such excessive amount of intensity, if placed prior to the conven-
tional correlator receiver will block some interfering patterns in causing errors
and improve the system performance by at least two orders of magnitude. This
implies that we first decide on the marked chips by placing the hard-limiter. In
the second stage, the receiver bases its decision on the transmitted bit based
on the detected marked chips. One possible scheme for the decision is that we
decide on bit ‘1’ if all the marked chips are detected as pulse present, and bit
‘0’ is detected otherwise. Interestingly, this is the best method for bit detection
in many cases. The reason is that the dominant noise in OCDMA systems is
MAI which converts chips ‘0’ to ‘1’ but not the reverse, i.e., ‘1’ to ‘0’. Since all
marked chips are needed to be detected as ‘1’ or pulse present in order to result
to the detection of bit ‘1’, we call such receiver structure as chip-level AND logic
gate and that can be implemented by AND gate logically. We have shown the
receiver structure in figure 3.

Indoor Wireless Optical CDMA LAN

In this section a typical wireless OCDMA Local Area Network (LAN) is investi-
gated. A typical wireless OCDMA communication system with a few number of
stations communicating to a single base station through the optical channel is
considered. Usually it is difficult to implement an irregular and adhoc network
architecture for a wireless optical LAN due to the limitations of optical wave’s
propagation properties. Especially, one or more base stations are needed to act
as a bridge when a local network needs to be connected to a global network
such as internet. To perform the communication of multi users in a single share
channel there is a need for a multiple-access technique such as OCDMA.
Fig. 4 shows a typical system architecture which includes one base station,
with a single pair of optical transmitter and receiver, and N user stations, each
having a pair of optical transmitter and receiver. With respect to the maximum
number of permitted active users in the network and the desired code length,
a set of OOCs is defined and each code is assigned to each active user. In
this system each active user is assigned a signature sequence from a set of
 7

Figure 4: Typical indoor wireless optical CDMA LAN containing a base station
and n users

OOC which is specified by (L, w, λa , λc ). All active users send their data
bits asynchronously on the common channel. Therefore, the received signal is
comprised of the analog summation of all users’ optical intensity transmitted
signals. Hence, the received signal can be written as follows,


N
Y (t) = y (k) (t − τk ) (2)
k=1

where, y (k) (t) is the received signal due to the kth user and τk indicates it’s
delay time with respect to the base station reference clock and it can be written
as follows,
∞ 
 L
(k) (k)
y (k) (t) = ui cj PTc (t − iTb − jTc ) (3)
i=−∞ j=1
   L
(k) (k)
where ui is the bit stream of user k, cj denotes the kth OOC code
i j=1
pattern, and PTc (t) is the pulse function with a duration equal to Tc .
In the design of any wireless CDMA communication network near-far prob-
lem needs to be addressed [26]. The near-far problem in wireless OCDMA LAN
requires specific algorithms to mitigate its effect such as the use of proper au-
tomatic gain control (AGC) circuit. Hence, in this section it is assumed the
existence of power control algorithms. Therefore we can safely ignore the near-
far problem such that all users have equal power in the uplink channel. In what
follows we discuss the transmitter and two viable and relevant receiver struc-
tures for a wireless OCDMA system. Further we discuss the pros and cons of
such transmitter and receiver structure in the context of their digital design and
their implementation. Finally their performance is obtained and discussed.

Wireless OCDMA Transmitter

In an OCDMA network, using OOC, network’s active users transmit their cor-
responding data bits asynchronously. But at the receiver side prior to detecting
data bits it is first essential to obtain bit level synchronization between the cor-
responding transmitter-receiver pair. OOC coding takes place in the physical
8

layer. So by adding some redundant bits, namely training sequence, to infor-

mation data bits prior to coding, receiver will be able to synchronize its code
with its corresponding transmitted code using the knowledge of the training bits
pattern.

Wireless OCDMA Receiver

In the design of the receiver and its implementation more complexity is en-
countered when compared to the transmitter especially in the synchronization
module. Similar to radio CDMA systems synchronization in OCDMA receivers
precedes the data detection block and it is applied in two stages namely ac-
quisition and tracking. In a typical synchronization process prior to tracking
state, the received data is first passed through the acquisition stage. Following
acquisition, the receiver is synchronized to the spreading code with an accuracy
which is within one-half of chip duration.
Tracking circuit performs more precise timing using, as its input, the syn-
chronized code from acquisition block. Note that acquisition circuit is active
only at the startup phase or when the receiver exits the synchronization state,
while the tracking loop remains online at all time. The correct phase code
produced by synchronization block is forwarded to detection block. Prior to
studying the synchronization algorithms it is more suitable to investigate the
two viable and relevant receiver structures for the detection block of a wireless
OCDMA system. In our analysis of the performance of the receiver detection
scheme it is assumed that the synchronization has been established successfully.

Correlator

Correlator receiver is the most propounded structure for OCDMA systems. In

this structure, decision is made based on the sum or the accumulation of the
received energy in all marked chips of the corresponding OOC code.
From implementation point of view this simple receiver involves a matched
filter, corresponding to its code pattern, and an integrator followed by a sampler.
In optical fiber CDMA systems the matched filter can be implemented by fiber
tapped-delay lines at their receivers. However, in a typical wireless OCDMA
LAN system where the speed of operation is not very high and the distances
are relatively short correlation may take place after the photodetection at the
receiver. In such a receiver the signal at the output of the receiver’s photode-
tector is sampled by an analog-to-digital converter (ADC) at a rate equal to ns
times the OOC chip rate, i.e., 1/Tc , as shown in Fig. 5(a). The correlation of
the received signal with the desired users’ code can be evaluated by the summa-
tion of wns samples of the received signal in all the marked pulsed chips of one
bit duration using a simple adder. A buffer after the ADC operation saves the
results of addition at each sampling time instant and resets at the end of the
bit time. At the end of the bit duration, the accumulated value that is saved in
the buffer, i.e., correlation value, is compared with an optimum threshold.
 9

Figure 5: Digital Structures for an Optical CDMA receiver (a) Correlator (b)
chip-level AND logic gate

Chip-level AND logic gate structure

In this structure, there are two levels or stages of decision. In the first stage,
receiver decides on the marked chips, one by one. In the second stage, decision
on the transmitted bit is made based on the first stage decisions. It is shown
that multi-user interference effect can be reduced using this method which can
be implemented simply by an optical hard-limiter and optical AND logic gate
in a fiber-optic CDMA receiver using unipolar codes such as OOCs, [12, 19].
Using a simple hard-limiter for chip-level decision not only mitigates the unde-
sirable effect of MAI but also reduces the sensitivity of system to power control.
In a typical wireless OCDMA LAN receiver, the hard-limiter on the marked
chips can be placed after the photodetector instead of being placed before the
photodetector. A simple hard limiting takes place using an analog circuit as
shown in Fig. 5(b). Thus, each sample from wns samples has a binary value at
the sampler output. The remaining operations are similar to the correlator re-
ceiver. In this structure, due to hard-limiting, the effect of MAI is substantially
reduced. From a digital design and implementation point of view, it simplifies
the implementation since sampling operation can be done along with other pro-
cesses for example in a Field Programmable Gate Array (FPGA) without using
a separate ADC circuitry.

BER analysis
The BER analysis is studied separately for the two above mentioned receiver
structures and BER expressions for an OCDMA LAN using an OOC family with
λc = 1 as the signature sequence is studied. The wireless channel is considered
to be LOS and is modeled with the DC loss imposed on the transmitted power
signal (we assume that the communication rate is chosen properly such that not
to cause pulse dispersion due to multipath effect). The maximum number of
codes in an OOC set with λa = λc = 1 which limits the network capacity is
10

obtained from (1).

In the digital receiver structure which is considered in this section, a low
pass filter followed by an ADC is placed after the receiver photodetector, as
described in the previous section. Assuming the low pass filter as a square shape
response equal to PTf (t), Tf = Tc /ns , digital receiver acts as an integrate and
dump circuit. Although, for chip-synchronous multi-user interference pattern,
ns = 1 is sufficient for a proper detection but for implementation reasons and
for tracking circuit more samples per chip are required as it will be discussed in
the tracking section. Hereafter, for BER analysis the value for ns is assumed to
be equal to one, i.e., ns = 1.
The vector α  = (α1 , α2 , ..., αw ) is considered as the interference pattern
occurred on the marked chips. In other word αj is the number of interferences
on the jth marked chip.
a. Correlator Structure
In a correlator structure, signal generated by integrate and dump circuit has
a Poisson distribution function with a mean equal as md ,

md (l) = [(wd + l)γs + wγb ]Tc (4)

w
where d ∈ {0, 1} indicates transmitted data bit, l = j=1 αj denotes the total
number of interferences occured on the pulsed mark chips of the desired OOC
Δ Δ
code, γs =( 2L η η idc
w )Pav hν and γb = Pb hν + q . Pav is the average received power on
the photodetector area, Pb is ambient noise power, idc and η are the photodetec-
tor dark current and quantum efficiency, respectively. Also q, ν, λ, h indicate
electron charge, optical frequency, optical wavelength and Plank constant, re-
spectively.
The probability of error for the correlator structure conditioned on l is ob-
tained as follows [27],

∞
1  e−m1 (l) (m1 (l)) 1  e−m0 (l) (m0 (l))
th n n
P (E|l) = + (5)
2 n=0 n! 2 n!
n=th

since the received energy is considered to have Poisson distribution. In the

above equation th is an optimum threshold that minimizes the total probability
of error. The total probability of error, PE is computed by averaging over l as
follow,

N −1
PE = pl (l)P (E|l) (6)
l=0

where l has a Bionomial distribution [27].

  2 l  N −1−l
N −1 w w2
pl (l) = 1− (7)
l 2F 2F

b. Chip-level AND logic gate structure

 11

For chip-level AND logic gate receiver, probability of error depends on inter-
ference pattern in addition to their accumulation. We can write the conditional
error probability as follows,

w
P (E|l, −
→ nj (1−nj )
α ) = 1/2 |
n|=w j=1 p0 (αj ) [1 − p0 (αj )]

w  (8)
nj (1−nj )
+1 − |n|=w j=1 p1 (αj ) [1 − p1 (αj )]

where n is the binary vector (detected after hard-limiter) caused on w marked

chips, |n| is the Hamming weight of n and we have,

md (αj ) = [(d + αj )γs + γb ] Tc , d ∈ {0, 1}

md (αj ) denotes the average number of incident photons on the jth pulsed mark
chip assuming data bit d ∈ {0, 1}, and we have defined:
∞
 
T H −m1 (αj )
e−m0 (αj ) (m0 (αj ))
n n
e (m1 (αj ))
p0 (αj ) = , p1 (αj ) =
n! n=0
n!
n=T H

where T H is the threshold of hard-limiter and α  indicates the interference pat-

tern caused on the corresponding OOC’s marked chips.
The probability of error for chip-level AND logic gate structure is computed
as follow,

N −1 
PE = P (l, α
 )P (E|l, α
) (9)
l=0
w
αj =l
j=1

where P (l, α
 ), the probability to have l active interfering users with the inter-
ference pattern α  , is obtained as follows [27],
 l  N −1−l
(N − 1)! w2 w2
P (l, α
) =
w 1− (10)
wl (N − 1 − l)! j=1 (αj )! 2L 2L

Using the typical values L = 128, w = 5, N = 6 as the OOC code pa-

rameters and λ = 870nm, Ib = 100μW/cm2 , id = 10nA respectively as the
light wavelength, background light irradiance and photodetector dark current
and η = 0.8, Ad = 1cm2 as the photodetector quantum efficiency and area re-
spectively, performance of the two above receiver structures is sketched in Fig.
6.

Acquisition Algorithms
In order to extract data bits properly from a spread stream of pulses, which
constitutes the desired OOC receiver, one needs to know the correct initial
reference time or the phase of the specified code pattern that carries the desired
information bit.
Assuming chip-synchronous scenario, receiver encounters an ambiguity equal
to L cells, each corresponding to one possible cyclic shift of the indicated OOC
12

0
10
Correlator
−1
Chip level AND logic gate
10

−2
10 L = 128
w=5
N=6
−3
10

BER
−4
10

−5
10

−6
10

−7
10
−60 −55 −50 −45
P (dBm)
av

Figure 6: Performance curves due to various types of receivers vs. received

optical power

code. At the end of the acquisition level the correct cell, or equivalently the
correct shift is found. This can be obtained for example by using training
bits which are inserted at the beginning of each frame. Some algorithms are
suggested to be used in acquisition state. For each algorithm, different number
of training bits is required to meet acquisition with a probability near one. It
is evident that the less training bits required the more efficient the algorithm
in use will be. In the following, three recently suggested algorithms for OOC
based OCDMA are presented.
a. Simple Serial-Search Method
In this approach, one cell is randomly selected and it is assumed to be the
correct cell. A correlation between the received data and the selected code over a
bit time duration (dwell time) is obtained and the outcome is compared with an
optimum threshold. If the output of the correlation is greater than the optimum
threshold, then the first randomly selected cell is the correct cell [28]. Otherwise,
the next cell is examined. By next cell it is implied a chip-sized rotation replica
of the initial cell or phase of the specified code. So by continuing this algorithm
and in maximum L stages the correct cell is obtained in the ideal case. This
method is relatively slow but its implementation in ultra-fast fiber optic systems
is viable.
b. Multiple-shift Method
To reduce the average time of acquisition, multiple-shift method was pro-
posed recently [29]. In this method L cells are divided into Q groups each
containing M cells. In the first stage of the algorithm, all M cells of group 1
are examined simultaneously. If the output of the correlation is greater than
a predetermined optimum threshold then it is proceeded to the second stage,
otherwise it will proceed to examine cells which belong to group 2. In continuing
this process and from [29] in maximum Q steps the true group is obtained. In
the second stage the M possible cells will be examined one by one similar to
the simple serial-search method. So in the ideal case the corrected cell can be
 13

Figure 7: An example of digital matched filter for an OOC stream with w = 3

found in maximum Q + M stages. So acquisition time is reduced considerably √

when compared to simple serial search method from maximum L stages to 2 L
stages.
c. Matched Filter Method
Assume a matched filter containing w delay lines. Each, delay line value
is proportional to the position of pulsed chips corresponding to its respective
OOC. When training bits enter the matched filter then the output signal is
a periodic training pulse with a width equal to Tc and a period equal to Tb .
This implies that in one bit duration the peak correlation of the incident signal
and code pattern can be observed without any knowledge of the correct initial
time reference. As an example the digital implementation of a matched filter
is depicted for an OOC stream with w = 3 in Fig. 7. Considering this idea,
matched filtering automatically searches all the L cells within one bit duration
instead of L bits duration as in the simple serial-search method. This method
is extremely fast and can be easily implemented in wireless OCDMA systems
that have been implemented in digital electronics.

Tracking Circuit
Tracking circuit performs two key operations in a typical OCDMA receiver. In
general, the received code from acquisition block has an ambiguity in phase or
initial time reference which is less than Tc /2. The first task of the tracking circuit
is to minimize this ambiguity. The second task for the tracking circuit is to alarm
the out of synchronization state to the acquisition block. Early-late method is
usually used for implementation of the tracking circuit [30]. In this method
two separate early and late replica of the reference code, which is produced by
the acquisition block, is generated by the tracking block. If c (t) denotes the
reference code, then c (t + Tc /2) and c (t − Tc /2) are the corresponding early
and late codes, respectively. Now, correlation action takes place between the
training bit stream and the early and late codes. If the output of both early
and late correlators are less than a threshold determined by the detection block
that is computed such that (6) is minimized for L sequential bits, then out of
synchronization state is activated. This signal is passed to the acquisition block
to switch on the acquisition state. To track the original code, the values of early
and late correlator outputs are subtracted, and three actions may be initiated
based on the subtraction result. Selecting a proper threshold, say thc , then a
14

0
10

−5
10 L = 128
w=5
N=4

Error Probability
ns = 3
−10
10
ns = 4

ns = 5

−15
10 ns = 8

ns = 16

Analog case ( ns = infinity)

−20
10
−60 −55 −50 −45
Pav (d Bm)

Figure 8: Performance curves for different number of samples taken at the

receiver for detection process

shifted replica to the early correlator is selected as the new reference code if the
subtraction result is greater than thc . A shifted replica to the late correlator
is selected if subtraction result is less than −thc , otherwise the main code is
selected. To select the proper threshold value, thc , two main issues need to
be considered: 1) thc value must not be so low to cause the tracking circuit
to oscillate and 2) not so high in order for tracking circuit not to be able to
follow the phase shifts fluctuations. Fig. 8 shows the impact of this parameter
in the performance of the correlator receiver structure [18]. By increasing the
number of samples per chip, the performance improves at the cost of increasing
the complexity.

Power Budget
In what follows we discuss the methodologies used to obtain the minimum power
level of an optical transmitter that guarantees proper operation of a typical short
range wireless OCDMA receiver in the presence of multi-user interference signal.
Power budget is obtained under two conditions. First, we consider IEC’s class A
standard to meet eye safety conditions as proposition [31]. Second, in the worst
case misalignment for Tx-Rx from the ideal line of sight in the presence of all
interfering users while the system must operate at a BER less than a predefined
value for instance 10−9 . Received optical power can be considered as follows [5],

Pt R (θ, φ)
Pr = Aef f (11)
d2

where Pt and d are transmitted optical power and Tx-Rx distance, respectively.
Aef f indicates lens effective area and R(θ, ϕ) is the propagation pattern func-
tion of the optical transmitter. θ and ϕ are the elevation and azimuth angles,
respectively. The propagation function of a Lambertian source with order m is
 15

defined as [5],
m+1
R (θ, φ) = cosm θ (12)
2π
If propagation half-angle of optical transmitter is equal to θ1/2 then m is ob-
tained as follows,
ln (1/2)
m=   (13)
ln cos θ1/2
As an example for a practical indoor environment and for a chip-level AND
logic gate receiver structure to obtain a BER less than 10−7 , the received optical
power Pr needs to be at least -52dBm referring to figure 6 results. If the distance
of Transmitter and Receiver is assumed to be equal to 3meters and divergence
angle of optical transmitter θ = 30◦ is considered for the worst case scenario
Transmitter-Receiver distance and misalignment are obtained to be d = 3m and
θ = 30◦ respectively. For such parameters it can be deduced that θ1/2 = 15◦
and using (13) Lambertian order m is equal to 20. Using (12) R(θ, ϕ) = 0.188
for the worst case. From (11) the transmitted average power per bit is obtained
as Pt = 3.01mW . Using results of [31], it can be observed that the maximum
allowable transmission power for a typical optical transmitter operating at 870
nm and a 15◦ propagation half angle is equal to 28mW . The optical power for
each OOC chip pulse in an infrared CDMA transmitter is obtained as follows,

2L
Pt/OOC chip pulse = Pt (14)
w
For typical values of an OOC set with L = 128, w = 5, N = 6, the peak
power of optical pulses can be as high as 1.433W . This shows that in wireless
OCDMA systems the peak power can become higher than ordinary OOK sys-
tems for around one to two order(s) of magnitude. One can improve the system
performance especially for chip-level AND logic gate structure by increasing the
SNR.

Free-Space Optical CDMA systems

In optical CDMA systems, an asynchronous communication between users of the
network, usually in an all-optical structure is established. The coded optical
signal may be carried out by various waveguides such as optical fibers or a
free-space optical (FSO) link. The network structure may consist of fiber and
free-space links simultaneously. The structure of the free-space link(s) can be
a point-to-point or a mesh or ring network [7, 32]. However in the performance
analysis of physical layer it is fairly sufficient to consider a single point-to-point
link.
We have shown the structure of a typical FSO-CDMA link in Fig. 9. The
distance between two nodes for a typical FSO link can be in order of a few
hundred meters up to a few kilometers. To support such ranges it is required to
use relatively high-power lasers or using optical amplifiers. Employing erbium
16

Figure 9: Typical FSO-CDMA system with N users: (a) Transmitter part, (b)
Receiver part

doped fiber amplifier (EDFA) as the post-amplifier at the transmitter side can be
a very attractive option in this application. One reason for this attractiveness is
that there are various kinds of EDFA with different desired gains in the market.
In this case by employing just one high power EDFA at the transmitter side, it
can support the link budget for ranges that can extend to a few kilometers. On
the other hand using 1550 nm wavelength can easily integrate the FSO link with
the other part of network that probably use optical fibers and other available
devices working in the same wavelength band [33, 34].
At the receiver side a telescope with high enough aperture size is employed to
mitigate the channel losses due to the transmitted beam divergence. To analyze
the performance of a FSO-CDMA link, the channel characteristics should be
considered as well as the inherent characteristics of a typical OCDMA system.
In this part we repeat our study on two different structures namely correlator
and chip-level AND logic gate structure in the context of FSO-CDMA systems.
It is usually necessary to employ adaptive power control (APC) for FSO-
CDMA systems. Employing APC in such systems guarantees that the power
budget of the link is in the desired range and on the other hand in clear weather
EDFA transmits at a lower power level. Furthermore in this case the perfor-
mance of the receiver is less sensitive to the threshold set for detecting optical
chips (in chip-level structure) or bits (correlator structure). To implement APC
one assigns a dedicated OOC code for transmitting low rate sampled and dig-
itized received power in each node to the other node in FSO link. In node 1,
the information of optical power is received by node 2 using the dedicated OOC
assigned to APC. By using this information it is easy to tune the gain of EDFA
 17

belonging to node 1 in order to transmit at a power level such that the received
power in node 2 is in the desired range. This scheme is performed for both di-
rections. The power control algorithm for evaluating the optimum EDFA gain
in each node is an iterative algorithm that reduces or increases the EDFA gain
according to the received power information corresponding to the other node
in a step by step procedure. In fact a desired range for the received power is
defined. If the received power is higher than the range, the EDFA gain reduces
by a predefined constant value and if received power is lower than the desired
range, the EDFA gain increases by the same predefined value. As a practical ex-
ample, assume that a variable gain EDFA with maximum 30dBm output power
is employed. For some typical commercial EDFA the third stage variable gain
lets the output power to range between 15dBm and 30dBm according to the
external command that sends to amplifier via its communication port. By using
the aforementioned EDFA, it is possible to consider a desired range of 3dB and
to change the EDFA output power by 1dB in each step.
The transmitter and receiver structures are similar to one used in indoor
wireless OCDMA system which was discussed in the earlier section. In the fol-
lowing subsection, performances of these receiver structures in the context of
FSO-CDMA system are studied. Analytical solutions to the error probability
for various channel models using PIN photodetector are obtained. In the follow-
ing analysis, the effects of atmospheric turbulence, ambient light, thermal noise
and multi user interference are considered in the context of semi-classical pho-
ton counting approach. For the sake of mathematical simplicity, the Gaussian
approximation is used for performance evaluation.

BER Analysis
The following analysis is based on semi classical photon counting where the
photon count statistics is obtained separately in each stage. As it is shown in
Fig. 9, we place an optical amplifier in FSO system which is not necessarily
present in indoor wireless optical CDMA systems. High power EDFA amplifier
along with a low power laser source, e.g., with Pt,1 = 0dBm output power, act
as a high power laser source at the transmitter side. Similar to indoor wireless
OCDMA system OOK signaling is employed.

Low power laser source

The photon count in one chip duration due to low power laser is indicated by
n1 which is assumed to have a Poisson distribution. The mean photon counts of
low power laser pulses representing zero and one are respectively, mt,0,c = 0 and
mt,1,c = Pt,1 Tc /hυ. Where Pt,1 , Tc , h and υ are the peaked output power of the
low power laser source, chip duration, Plank’s constant and optical frequency,
respectively. Therefore, the moment generating function (MGF) corresponding
to the photon count of low power laser output with Poisson distribution is as
follows,
φt,d,c (z) = exp [1 − mt,d,c (1 − z)] (15)
18

Optical Amplifier
The dominant noise source in optical amplifiers is ASE (Amplified Spontaneous
Emission). By assuming G, K = nsp (G − 1) and M as the gain, the noise
factor and the number of spontaneous modes, respectively, the optical amplifier
output photon count, denoted by n2 , is obtained as a Laguerre distribution with
parameters (mt,d,c G, K, M ) [35]. Hence, the MGF corresponding to the photon
count of optical amplifier output is obtained as follows,

1 G (1 − z)
φEDF A,t,d,c (z) = × φ t,d,c 1 − (16)
[1 + K (1 − z)]
M 1 + K (1 − z)

FSO channel model

Atmospheric channel can be simply modeled as an attenuator with random
attenuation coefficient. Channel attenuation can be expressed as the multipli-
cation of two parameters one a deterministic parameter and another a random
variable. FSO channel attenuation is caused by both molecular absorption and
aerosol scattering which randomly affects the light intensity traversing through
the channel. The deterministic part can be written as follows [1],

A −βL0
a=  θL0 2 e (17)
π 2

In the above equation L0 and β are the link distance and atmospheric extinction
coefficient, A and θ are the telescope area and transmitted beam divergence in
radian, respectively. The random attenuation of the channel is induced by the
random variations of the channel refractive index which can be modeled as a log
normal distribution e2x in which x is a normal random
 variable. The random
term is assumed to be normalized such that E e2x = 1. So, the mean and
the variance of the normal random variable x are considered as −σx2 and σx2 ,
respectively. Hence, if we assume that the channel attenuation is deterministic,
i.e., condition on known x, the MGF of the photon count due to the channel
output, denoted by n3 , can be written as follows,

φs,d,c (z|x) = φEDF A,t,d,c 1 − ae2x (1 − z) (18)

Furthermore the ambient light can also be modeled as an additive independent

photon count with Poisson distribution. So, the MGF due to the total photon
count coupled into the multi mode fiber is obtained as follows,

φr,d,c|x (z) = φs,d,c (z|x) φa,c (z) : φa,c (z) = exp [1 − ma,c (1 − z)] (19)

where, ma,c in the above equation indicates the mean photon count of the
ambient light in one chip duration and it is obtained as follows,

W (λ) AΩF OV ΔλTc

ma,c = (20)
hυ
 19

where, ΩF OV is the receiver field of view (FOV) in strediant, Δλ is the optical

filter bandwidth, and W (λ) is the spectral radiance function, defined as the
power radiated at wavelength per cycle of bandwidth in a unit solid angle per
unit of source area.

Photodetector model
Photodetoctor is modeled as an attenuator with attenuation coefficient equal to
its quantum efficiency indicated by η. So, the MGF corresponding to the photo
electron count at the photodetector output is obtained as follows,

φP D,d,c (z|x) = φr,d,c [1 − η (1 − z) |x] (21)

As it was mentioned earlier, the Gaussian approximation is employed for the

BER analysis. The BER analysis is studied hereafter for two different receiver
structures, namely, simple correlator and chip-level AND logic gate.

correlator receiver structure

Usually, the coherence time of fading in FSO links is on the order of several
milliseconds [1], while usual bit durations are less than several microseconds;
therefore, we can assume that the fading is constant during a bit period. Con-
sidering the structure of correlator receiver, independence of counts for different
chip intervals, and the fact that the fading is approximately constant during a
bit period, mean and the variance of the decision variable for transmitting bit
”1” and ”0” can be written as follows,

w
(i)

w
md = mr,d,c , σd2 = 2
σi,r,d,c ; d = 0, 1 (22)
i=1 i=1
 w  w
(i) 2
where, mr,d,c and σi,r,d,c are the means and variances corresponding
i=1 i=1
to w marked chips. Assuming l active interfering users, in the desired user’s
2
marked chips, the decision variable mean md,l and variance σd,l are obtained as
follows,
md,l = (d × w + l) × mP D,d,c + w × mdc,c
2 (23)
σd,l = (d × w + l) × σP2 D,d,c + w × mdc,c + σth
2
d = 0, 1

where, mP D,d,c , σP2 D,d,c are the mean and variance of the phoelectrons in one
2
chip duration. mdc,c , σth are the variance of the photodetector dark current per
chip and thermal noise photoelectron count, respectively [36],
id Tc 2 2Kb Tr Tc
mdc,c = , σth = (24)
q RL q 2
where, Kb , Tr and RL are Boltzman constant, receiver equivalent temperature
and the resistance of the receiver load. Also, id , Tc , and q are the photodetec-
tor dark current, chip time, and the electron charge, respectively. Using the
20

aforementioned mathematical model in each stage, the mean and variance of

the photoelectron count in one chip duration is obtained as follows,
2x
mP D,d,c = ηma,c + (M × K + mt,d,c × G) ×2xηae 
2 2x
σP D,d,c = ηma,c + M × K × ηae
 1 + ηae × K (25)
+mt,d,c × G × ηae2x 1 + 2ηae2x × K

Finally, using Gaussian approximation, the BER of the simple correlator receiver
structure is obtained as follows,

P E (x) =
N −1   2 l  N −1−l    
1  N −1 w w2 m1,l − th th − m0,l
1− Q +Q
2 l 2L 2L σ1,l σ0,l
l=0
(26)

Where, th in the above equation is the optimum threshold for the correlator
that minimizes the total error probability. Q (u) is the normalized Gaussian
 √ ∞ 2
function defined as Q (u) = 1/ 2π u e(−v /2) dv. Note that the above ex-
pression evaluates the conditional error probability given the fading coefficient
x. Hence the total error probability is obtained by averaging P E (x) over the
fading coefficient x as follows,
 ∞
PE = P E (x) fX (x) dx (27)
−∞

where, fX (x) is the pdf of the fading coefficient defined as follows,

2 2
(x+σx )
1 −
fX (x) = √ e 2σx2
2πσx

Chip level AND logic gate

The conditional error probability given the turbulence coefficient x can be eval-
uated similar to the approach used in indoor wireless OCDMA section. The
conditional error probability given the interference number, the interference
pattern and the fading coefficient is obtained as follows,

w
P (E|l, −
→ (1−nj )
α , x) = 12
nj
|
n|=w j=1 p0 (αj , x) (1 − p0 (αj , x))

w 
nj (1−nj )
+1 − |n|=w j=1 p1 (αj , x) (1 − p1 (αj , x))
(28)
where, n is the binary vector (detected after hard-limiter) caused on w marked
chips, |n| is the hamming weight of n, p0 (αj , x) and p1 (αj , x) are the chip level
error probabilities on the jth chip for transmitting bit ‘0’ and bit ‘1’, while α 
and x are the known interference pattern and the fading coefficient on that chip
 21

−1
10

−2
10

−3
10

−4
10

−5
10

BER
−6
10

−7
10

−8 Correlator (σ2=0.01)
10
Correlator (σ2=0.03)
−9 Chip level AND logic gate (σ2=0.01)
10
Chip level AND logic gate (σ2=0.03)
−10
10
−48 −46 −44 −42 −40 −38 −36
Pt (dBm)

Figure 10: Performance comparison of simple correlator with chip level AND
logic gate receiver structures for FSO-CDMA link

duration, respectively. Assuming the photoelectron counts in the jth chip to be

Gaussian, the error probabilities p0 (αj , x) and p1 (αj , x) are obtained as follows,
 
T H − m0,j m1,j − T H
p0 (αj , x) = Q , p1 (αj , x) = Q (29)
σ0,j σ1,j

with
md,j = (d + αj ) × mP D,d,c + ma,c
2
σd,j = (d + αj ) × σP2 D,d,c + ma,c + σth
2
, d = 0, 1
Where, T H is the optimum threshold for chip level detection and it is obtained
to minimize the total error probability. The total error probability is obtained
by averaging the error probability P (E|l, αj , x) over the number of interfering
w
users l, the interference pattern {αj }j=1 and the turbulence coefficient x as
follows,
⎡ ⎤
 ∞ N −1 
⎢ ⎥
PE = ⎣ P (l, α  , x)⎦ fX (x) dx
 )P (E|l, α (30)
−∞ l=0
 w
αj =l
j=1

where, P (l, α
 ), the probability to have l active interfering users with the inter-
ference pattern α  , is obtained as follows,
 l  N −1−l
(N − 1)! w2 w2
P (l, α
) =
w 1− (31)
wl (N − 1 − l)! j=1 (αj )! 2L 2L

As an example, consider FSO link with parameters listed in Table I. Perfor-

mances of the aforementioned receiver structures are evaluated using the above
represented expressions for the error probabilities. The result for σx2 = 0.01 and
σx2 = 0.03 as the fading variance is shown in Fig. 10. In a communication system
22

Table 1: Typical values for the FSO OCDMA link

L OOC length 128
w OOC weight 5
N Active users 4
id Photodetector dark current 10nA
η Photodetector quantum efficiency 0.8
λ Optical wavelength 1550 nm
Tc Chip duration 4 ns
RL Load Resistance 10KΩ
Δλ Optical filter bandwidth 1nm
θF OV Receiver field of view angle 5mrad
θ Transmitter beam divergence 1mrad
G EDFA gain 30dB (1000)
nsp Spontaneous emission parameter 1.1
Tr Receiver temperature 300◦ K
L0 Channel length 3Km
β Attenuation coefficient for clear air 0.1Km−1

the variance of an interference or noise signal is a measure of its strength. For

this reason increasing the fading variance, worse result in BER are presenting
(Fig. 10). The figure compares the performance of simple correlator receiver
and chip level AND logic gate receiver. As it is shown in this figure, simple
correlator outperforms the chip level AND logic gate structures in low power
regions because, the system is shot noise limited. As the transmitted power
increases the system becomes interference limited. So, chip level AND logic gate
structure outperforms the simple correlator.

Modulation
With the advent of wireless and fiber-optic OCDMA technology, a large body of
research activity has been carried out on finding powerful code structures and ef-
fective modulations and signalings that can enhance such systems’ performance
and capacity. In particular, we introduced one such powerful code structure,
namely, OOC. On the other hand for effective modulation on-off keying (OOK)
and pulse position modulation (PPM) are among the most desired and em-
ployed modulation techniques in typical intensity modulation/direct detection
(IM/DD) optical communication systems. The immense interest on OOK and
PPM modulations stems from two basic facts, namely, the ease with which they
can be transmitted and received (implementation), and secondly it is shown that
by invoking divergence theorem in a typical optical channel (Poisson channel)
the aforementioned signal modulations are optimum [2].
In OOK-OCDMA systems bit ‘1’ is mapped on to L (code length) sequential
chips in which w (code weight) chip positions that are determined by the code
 23

Figure 11: M-ary PPM OCDMA symbols, M = 4

structure are on and the remaining L − w chips are off. Furthermore bit ‘0’ is
mapped on to L zero sequential chips. The implementation of this modulation
is simple and the performance is acceptable in most applications. Computing
the optimum threshold in the correlator structure is a necessity for OOK modu-
lation. This implies that after the correlator a hard-limiter should decide on the
transmitted bit by comparing the detected energy to an optimum threshold.
It is shown that in some applications the performance of the receiver is very
sensitive to the threshold especially in such cases that the received power is not
high enough. As an example in free-space optical links there is a considerable
performance difference of systems with and without adaptive thresholding [37].
In M-ary PPM, the other propounded modulation in IM/DD optical com-
munication systems, optical pulses are transmitted in smaller time durations
and in different symbols that are distinguished by separating the time domain.
Inherently M-ary PPM is more efficient in energy but less efficient in bandwidth
when compared to OOK. To implement M-ary PPM in a typical OCDMA sys-
tem, each user’s symbol interval (slot) is divided into M spreading intervals
where each interval could contain the corresponding OOC. To send the mth
symbol the OOC of the corresponding user is placed in the mth interval and
the remaining intervals remain off as shown in Fig. 11. At the receiver end
the correlated energy in each symbol duration is obtained and the maximum is
selected as the transmitted symbol. So if at the receiver side we use correlator
structure there is no need to compute an optimum threshold.
In general the use of M-ary PPM, which is among the class of energy efficient
but bandwidth inefficient modulation methods, reduces the system throughput.
However it’s use in conjunction with OOC based OCDMA could effectively re-
duce the multiaccess interference thereby enhancing the performance of a typical
OCDMA system.
To mitigate the degradation in throughput efficiency of M-ary PPM, Over-
lapping PPM (OPPM) was suggested for optical systems [40]– [42]. Fig. 12
shows a typical OPPM OCDMA system modulation. In this modulation, ad-
24

Figure 12: OPPM OCDMA symbols

jacent symbols are not completely disjoint in time domain but they are over-
lapped. The amount of the overlap is defined by the modulation parameter. The
greater this parameter the better the throughput efficiency, but, on the other
hand, more interference between symbols. Therefore one can make a trade-
off between throughput efficiency and interference level in this modulation. In
OPPM similar to M-ary PPM there is no need for threshold setting in the corre-
lator receiver. M-ary PPM can be considered as an especial case of OPPM if we
set the modulation parameter to unity. Here if we consider an OOC code with
length L and given laser pulse duration Tc , PPM multiplicity M , and index
of overlap γ, we divide a spreading interval of duration τ = LTc into γ smaller
subintervals with width τ /γ and define OPPM modulator unit delay as LTc /γ.
In a single OPPM frame each spreading interval is located at a unit delay with
respect to the previous one. Thus adjacent spreading intervals have (1-1/γ)τ
seconds overlap and since there are in total M symbols, an OPPM frame length
equals in seconds:

τ L
T = (M − 1) + τ = (M − 1 + γ) Tc (32)
γ γ

Note that in order to fit unit delay properly in the symbol interval, the following
constraint should be met.
τ L
= integer × Tc or = integer (33)
γ γ

One should note that M-arry PPM is a special case of M-ary OPPM signaling
in which the index of overlap, γ, equals 1 that means that there is no overlap
between the adjacent spreading intervals. As shown in Fig. 12, to send the
mth symbol we place the OOC of the desired user in the mth spreading interval
 25

and make all remaining chips off in the corresponding frame of information.

Considering ni as the number of collected photons in ith spreading interval,
receiver detects mth symbol if for all i, 0 ≤ i ≤ M − 1 and i = m we have
nm > ni .
Studying the performance of OPPM in OCDMA systems will lead us to
different results than typical optical systems. It can be shown that using OPPM
in OOC-based OCDMA systems improve the performance in both throughput
and energy efficiency. To explain the reason intuitively we can say that OOC
codes are typically sparse. Therefore they allow us to overlap the codes without
increasing the inter-symbol interference by much. Furthermore it is shown that
by choosing proper parameters for the modulation, OPPM outperforms OOK
and M-ary PPM [43].
To compare the performance of the three mentioned modulations in OCDMA
systems it is sufficient to consider the case of interference limited (high enough
SNR) and the chip-level AND logic gate receiver structure.

Performance Analysis
An upper bound on the error probability of the above mentioned modulations
in OCDMA systems using OOC codes and chip-level AND logic gate as the
receiver structure is obtained. For the performance evaluation we only consider
MAI noise in order to obtain the limits or the floor on the performance for
different modulation OOC-based OCDMA systems.
For a system with N users where each user is assigned an OOC with code
weight w, code length L, and cross-correlation coefficient λ = 1 the probability
of error for OOK modulation can be found as follows [12],
   
 w N −1
w
1 k w
Pe/OOK = 1+ (−1) 1−k (34)
2 k 2L
k=1

For M-ary PPM modulations the upper bound of error probability can be writ-
ten as follows:
   
 w N −1
w
M k w
Pe/M −P P M ≤ 1+ (−1) 1−k (35)
2 k ML
k=1

and for OPPM with some approximation the following expression can be ob-
tained [43]:
   N −1 
M 
w
k w γw
Pe/OP P M ≤ 1+ (−1) 1−k (36)
2 k (M − 1 + γ)L
k=1

To compare the performance of OOK and PPM systems assume a fixed value for
chip duration, i.e., Tc . Also for a fair comparison assume that for all systems, the
amount of information per chip that each user carries is constant. If this value
is shown by R0 then R0 = 1/2LOOK bits/chip for OOK, R0 = log2 M/M LP P M
26

Figure 13: BER versus number of users for OOK, M-ary PPM and OPPM

bits/chip for M-ary PPM and R0 = γlog2 M/(M − 1 + γ)LOP P M bits/chip for
OPPM. R0 is directly related to the physical throughput of the system. We
obtain the total throughput of the modulation by R0 × N .
For comparing the probability of error for R0 = 0.002 is kept constant for
all modulation. The results are plotted in figure 13 where the cross-correlation
value λ = 1 for the OOC and M = 16 and γ = 16 for PPM modulations. Results
show that OPPM outperforms the other modulations by means of decreasing
the error probability of the system, i.e., accommodating more users.

Generalized OOC
The first attempt to define and establish optical orthogonal codes was based on
minimizing the amount of interference between each two codes in such a way
that there is no need for network synchronization. In this case it is evident
that the minimum amount of interference or cross-correlation between each two
unipolar binary stream is 1 when there is no timing reference. This is the best
case when one is looking at the mutual interference. However in this case the
value of code weight is restricted by the number of codes according to Johnson
bound. From throughput efficiency point of view it is desired that the code
length be as short as possible. So in the case of strict OOC (λ = 1), the
product of users and throughput per user of the system is limited. To increase
the value of users×throughput one can relax the cross-correlation coefficient. It
can be seen from Johnson bound that by increasing λ the cardinality of codes
will increase if we keep the code weight as before. In the first view one can
deduce that by increasing λ, the multiple-access interference increases and the
system performance degrades. But this is not necessarily true since in this
case we can increase the code weight which that in itself could improve the
performance by increasing the coding gain. Briefly,there is a trade-off between
λ and w and there is an optimum point for the code parameters according to
 27

the system conditions.

There are many studies on using generalized OOC in OCDMA systems and
finding their performance in different conditions. In [12], [16], [44], attempts
were made to explore the performance of OCDMA systems that employ OOCs
with λ ≥ 1. In particular, in [16], by obtaining lower and upper bounds on the
system’s performance, it was hinted that OOCs with λ = 2 could, under certain
conditions, outperform OCDMA employing OOCs with λ = 1, with a cardinality
that could be a hundred to a thousand times longer. Similarly, in [fth], using a
structure based on frame time-hopping ultra-wideband CDMA system, it was
shown that the system’s performance degrades gradually with respect to OOCs
with 1 ≤ λ < w, while offering many possible signature sequences. In [12]
performance analysis of generalized OOC in an OCDMA system with chip-level
AND logic gate receiver structure was obtained. It was shown that the optimum
value for λ is between 1 and 3 depending on the design criteria.
To compare the performance of OOCs in OCDMA systems as a function of
code’s cross-correlation coefficient we follow the approach presented in [12]. One
can assume that the probability of interference of an undesired code on the ith
specified marked chip positions of the desired OOC is pi . It is obvious that the
value of pi directly depends on the code set and the position of the marked chips
belonging to the desired and interfering codes. However as an approximation
we assume that pi is the same for all interfering codes. In this case we have,

λ 
w w2
i pi = (37)
i=1
i 2L

The probability of error caused by MAI is obtained as follows,

1 1
Pe = Pr (error|1) + Pr (error|0) (38)
2 2
Neglecting shot noise the first term on the right hand side of the above equation
is equal to zero. However an error occurs by MAI noise only when the trans-
mitted bit is 0. If we show the number of interference on the ith marked chip
of the desired OOC by αi then the probability of error is as follows [12],
1 1
Pe = Pr (error|0) = Pr (α1 > 0, α2 > 0, ...., αw > 0)
2 2
= 1 − Pr (α1 = 0 or α2 = 0 or .... or αw = 0 )
w 
w
=1+ (−1)k Pr (α1 = α2 = .... = αk = 0) (39)
k
k=1

The λ events that are related to p1 , p2 , ..., and pλ are disjoint. Furthermore all
N − 1 interfering users are independent and hence we can write:
  N −1

λ
k
l
Pr (α1 = α2 = .... = αk = 0) = 1 + (−1) ρl (40)
l
l=1
28

Figure 14: Maximum achievable throughput vs. number of users for OCDMA
chip-level AND logic gate receiver

where ρl is defined as:

ρl = Pr (α1 = α2 = .... = αl = 1 | one interf ering user)

  λ 
w−l w−l w−l
= pl + pl+1 + ... + pλ = pi (41)
1 λ−l i−l
i=l

Substituting (A.4) in (A.3) and using (A.2) we can rewrite (A.1) as follows:
⎧ ⎫
⎨ 
w   
λ 
λ   N −1 ⎬
1 w k w−l
Pe = 1+ (−1)k 1+ (−1)l pi (42)
2⎩ k l i−l ⎭
k=1 l=1 i=l

However it is shown that if OOC codes are chosen such that pk = 0 for 1 ≤ k ≤
λ − 1 and thus only pλ = 0, we obtain an upper bound on the performance of
such an OCDMA system with generalized OOCs.
To analyze the effect of cross-correlation of codes on the system performance
we obtain the maximum achievable throughput of the system versus number of
users in the case which the bit-error-rate does not exceed a predefined value, e.g.,
10−9 . The goal is to find the minimum value for the code length, assuming a
fixed value for number of users by finding the optimum value for cross-correlation
and code weight using Johnson bound. The results have been sketched in Fig.
14 where it can be seen that the optimum value for λ varies between 2 and
3. It is noteworthy that for λ >1 we consider the worst case, i.e., pk = 0 for
1 ≤ k ≤ λ − 1 and pλ = w2 /2Lλ w λ . Such codes may not necessarily exist but
from a mathematical point of view it gives an upper bound on the performance
of all possible and existing codes.
 29

Figure 15: Transmitter/receiver block diagram of wireless indoor OCDMA sys-

tem

Experimental Prototypes

Indoor Wireless Optical CDMA LAN

We have shown the implemented block diagram of wireless OCDMA indoor

prototype in figure 15 [19]. In the transmitter side we have used dual port
memory in which the training bits and matched pattern are put in the header
section. Data bit stream is written in the remaining section of the memory while
to generate the whole frame, including data bits and header, data is read from
the first to last cell of memory. The relation between the speed of the read and
write clocks is equal to the relation of the total length of frame to the payload
part of frame. The corresponding OOC is loaded in a shift register. Chip rate
clock feed the shift register. The output of shift register is combined with the
output of the memory through an AND gate. This structure simply generates
the coded bits.
The outcome of the digital part of transmitter feeds the analog part of trans-
mitter that includes fast transistors as the driver prior to the LEDs operating
in infrared wavelength (here 870 nm).
In the receiver part we have used PIN photodiode following by a three-stage
amplifier circuit. The gains of the three stages are computed such that support
the nominal parameters of the system. In the implemented indoor system we
have designed the analog circuit in the receiver such that supports the distance
range of transmitter and receiver 0.5 to 3.0 meters and half-angle of LED equal
to 10◦ . The second stage of amplifier acts as the hardlimiter, i.e., clips the
received pulses from the first stage. The third stage amplifies the pulses coming
from the second stage to the proper level for the digital part, i.e., FPGA.
Binary detected chips go to the digital section where we have acquisition,
tracking and detection blocks. We described the operation of digital blocks in
the previous sections. A simple de-frame block extract data bits from the frame.
We have shown the implemented prototype in figure 16.
30

Figure 16: The electronic board designed for the base station, supporting up to
four serial data terminals

Table 2: Specifications of FSO link

Parameter Specification
Operating wavelength 1550 nm
Telescope aperture size 11.25 inch
Diversity in transmitter 2 or 4
EDFA output power 15-30 dBm
Coupling in receiver MMF
Beam divergence ±0.4 mrad

FSO-CDMA Link
The block diagram of implemented prototype of FSO system is shown in Fig.
9. In the transmitter side we have used ordinary high speed SFP laser modules
operating in 1550 nm wavelength. The laser is modulated by the data stream
coming from another digital board. A high power EDFA with variable adjustable
gain amplifies the power of laser to the desired power level. The maximum
output of EDFA in the prototype is 30dBm. However we use the range 15 dBm
to 30 dBm using the proposed power control schema. The output of EDFA
is split to four branches, by employing passive SMF couplers, in order to use
diversity in the transmitter side.
To support distance of 5Km we employ a 11.25 inch telescope in the receiver
side. The received light is coupled in a multimode fiber and then goes for further
processes in the following blocks. System supports fast and gigabit Ethernet as
well as STM1. It also supports E1-4E1 with OCDMA and different QoS. A
photograph of the prototype is shown in figure 17 and the system specifications
are given in table II.
 31

Figure 17: Prototype of FS-OCDMA system

Acknowledgment
The authors would like to thank the support of Iran National Science Foun-
dation (INSF) for this research.
32
Bibliography

[1] S. Karp, R. M. Gagliardi, S. E. Moran, and L. B. Stotts, Optical Channels

Fibers, Clouds, Water and the Atmosphere: Plenum, 1980.
[2] R. M. Gagliardi and S. Karp, Optical Communications: John Wiley, 1976.
[3] F. R. Gfeller and U. H. Bapst, “Wireless in-house data communication via
diffuse infrared radiation,” Proc. IEEE, vol. 67, no. 11, pp. 1474–1486, Nov.
1979.
[4] J. R. Barry, Wireless Infrared Communications: Kluwer, 1994.
[5] J. M. Kahn and J. R. Barry, “Wireless infrared communications,” Proc.
IEEE, vol. 85, no. 2, pp. 265–298, Feb. 1997.
[6] J. Strohbehn, Ed., Laser Beam Propagation in the Atmosphere: Springer,
1978.
[7] B. T. Binder, P. T. Yu, J. H. Shapiro, and J. K. Bounds, “An atmospheric
optical ring network,” IEEE Trans. Commun., vol. 38, no. 1, pp. 74–81, Jan.
1990.
[8] M. R. Pakravan, M. Kavehrad, and H. Hashemi, “Indoor wireless infrared
channel characterization by measurements,” IEEE Trans. Vehic. Tech., vol.
50, no.4, pp. 1053-1073, July 2001.
[9] S. Hranilovic, Wireless Optical Communication Systems: Springer, 2005.
[10] J. A. Salehi, “Code division multiple-access techniques in optical fiber
networks–part I: fundamental principles,” IEEE Trans. Commun., vol. 37,
pp. 824-833, Aug. 1989.
[11] J. A. Salehi and C. A. Brackett, “Code division multiple-access techniques
in optical fiber networks–part II: system performance analysis,” IEEE Trans.
Commun., vol. 37, pp. 834-842, Aug. 1989
[12] S. Mashhadi and J. A. Salehi, “Code division multiple-access techniques
in optical fiber networks–part III: optical AND gate receiver structure with
generalized optical orthogonal codes,” IEEE Trans. Commun., vol. 45, pp.
1457-1468, Aug. 2006.

33
34 BIBLIOGRAPHY

[13] S. Mashhadi and J. A. Salehi, “Optimum code structures for positive optical
CDMA using normalized divergence maximization criterion,” IEEE Trans.
Commun., vol. 55, pp. 1414-1421, Sept. 2008.

[14] I. Arruego, et al., “OWLS: A Ten-year History in Optical Wireless Links

for intra-Satellite communications,” IEEE J. of Selected Areas in Commun.
(JSAC), vol. 27, no. 9, pp. 1599–1611, Dec. 2009.

[15] P. J. Winzer, “Modulation and multiplexing in optical channels,” Lasers

and Electro-Optics, 2009 and 2009 Conference on Quantum electronics and
Laser Science Conference. CLEO/QELS 2009. Publication Year: 2009 ,
Page(s): 1–2

[16] M. Azizoglo, J. A. Salehi, and Y. Li, “Optical CDMA via temporal codes,”
IEEE Trans. Commun. 40, 1162–1170 (1992).

[17] J. M. H. Elmirghani, and R. A. Cryan, “Hybrid PPM-CDMA systems

utilising OOC for indoor wireless infrared communication,” Microwave and
opt. Tech. Lett., vol 8, No. 1, pp. 44–47, Jan. 1995.

[18] B. M. Ghaffari, M. Matinfar, J. A. Salehi, “Wireless Optical CDMA LAN:

Digital Design Concepts,” IEEE Transaction on Communication vol. 56,
no. 12, pp. 2145–2155, Dec. 2008.

[19] B. M. Ghaffari, M. D. Matinfar, J. A. Salehi, “Wireless Optical CDMA

LAN: Digital Implementation Analysis,” IEEE J. of Selected Areas in Com-
mun. (JSAC), vol. 27, no. 9, pp.1676–1686, Dec. 2009.

[20] F. R. K. Chung, J. A. Salehi, and V. K. Wei, “Optical orthogonal codes:

design, analysis, and applications,” IEEE Trans. Inf. Theory vol. 35, no. 3,
pp. 595–604, May 1989.

[21] H. Chung and P. Vijay Kumar, “Optical orthogonal codes–new bounds

and an optimal construction,” IEEE Trans. Inf. Theory, vol. 36, no. 4, pp.
866–873 Jul. 1990.

[22] S. M. Johnson, “A new upper bound for error-correcting codes,” IRE Trans.
Inf. Theory, vol. IT-8, pp. 203–207, 1962.

[23] H. M. Shalaby, “Chip level detection in optical code division multiple ac-
cess,” J. Lightwave Technol., vol. 16, pp. 10771087, Jun. 1998.

[24] T. Ohtsuki, “Performance analysis of direct-detection optical asynchronous

CDMA systems with double optical hard-limiters,” J. Lightwave Technol. vol.
15, no. 3, pp. 452457, Mar. 1997.

[25] U. N. Griner and S. Arnon, “Multiuser Diffuse Indoor Wireless Infrared

Communication Using Equalized Synchronous CDMA,” IEEE Trans. Com-
mun., vol. 54, pp. 1654-1662, Sep. 2006.
BIBLIOGRAPHY 35

[26] A. Aminzadeh-Gohari and M. R. Pakravan, “Analysis of power control for

indoor wireless infrared CDMA communication,” in Proceedings of the 25th
IEEE Performance, Computing, and Communications Conference (IPCCC)
(IEEE, 2006), pp. 297–302.
[27] S. Zahedi and J. A. Salehi, “Analytical comparison of various fiber-optic
CDMA receiver structures,” J. Lightwave Technol., vol. 18, no. 12, pp. 1718-
1727, Dec. 2000.
[28] A. Keshavarizan and J. A. Salehi, “Optical orthogonal code acquisition in
fiber-optic CDMA systems via the simple serial-search method,” IEEE Trans.
Commun., vol. 50, no. 3, pp. 473483, Mar. 2002.
[29] A. Keshavarizan and J. A. Salehi, “Multiple-shift code acquisition of optical
orthogonal codes in optical CDMA systems,” IEEE Trans. Commun., vol. 53,
no. 4, pp. 687697, Apr. 2005.
[30] R. L. Peterson, D. E. Borth, and R. E. Ziemer, Introduction to Spread
Spectrum Communications. Prentice-Hall, 1995.
[31] Int. Electrotech. Commission, CEI/IEC 825-1: Safety of Laser Products,
1993. J. D. Rancourt, Optical Thin Film. StateplaceNew York: Macmillan,
1987.
[32] R. M. Gagliardi, “Pulse-coded multiple access in space optical communi-
cations,” IEEE JSAC, vol.13, no. 3, pp. 603–608, 1995.
[33] T. H. Carbonneau and D. R. Wisely, “Opportunities and challenges for
optical wireless; the competitive advantage of free space telecommunications
links in today’s crowded marketplace,” in Proc. SPIE Wireless Technologies
and Systems: Millimeter-Wave and Optical, vol. 3232, pp. 119–128 ,Jan. 1998.
[34] P. F. Szajowski, G. Nykolak, J. J. Auborn, H. M. Presby, and G. E. Tourgee,
“High power optical amplifiers enable 1550 nm terrestrial freespace optical
data-link operating at 10 Gbs,” in Proc. MILCOM ’99, vol. 1, pp. 687–689,
1999.
[35] T. Li and M. C. Teich, “Photon point process for traveling- amplifiers,”
IEEE J. Quantum Electron., vol. 29, no. 9, pp. 2568– Sep. 1993.
[36] M. Razavi, J. H. Shapiro, “Wireless optical communications via diversity
reception and optical preamplification ,” IEEE Trans. Wireless Commun.,
vol. 4, no. 3, pp. 975–983, 2005.
[37] M. Jazayerifar and J. A. Salehi, “Atmospheric optical CDMA communi-
cation systems via optical orthogonal codes,” IEEE Trans. Commun. vol.54,
no.9, pp. 1614–1623 sept. 2006.
[38] B. Hamzeh and M. Kavehrad, “OCDMA-coded free-space optical for wire-
less optical-mesh networks,” IEEE Trans. Commun., no. 12, pp. 2165–2174,
Dec. 2004.
36 BIBLIOGRAPHY

[39] T. Ohtsuki, “Performance analysis of atmospheric optical PPM systems,”

J. Lightw. Technol., vol. 21, no. 2, pp. 406–411, Feb.

[40] G. M. Lee and G. W. Schroeder, “Optical PPM with multiple positions per
pulse-width,” IEEE Trans. Commun., vol. COM-25, pp. 360–364, Mar. 1977.

[41] C. N. Georghiades, “Modulation and coding for throughput-efficient optical

systems,” IEEE Inform. Theory, vol. 40, pp 1313–1326, Sept. 1994.

[42] H. M. H. Shalaby, “Maximum Achievable Throughputs for Uncoded OPPM

and MPPM in Optical Direct-Detection Channels,” J. Lightwave Technol. vol.
13, no. 11, pp. 2121–2128, Nov. 1995.
[43] H. M. H. Shalaby, “Direct detection optical overlapping PPM-CDMA com-
munication systems with double optical hardlimiters,” J. Lightwave Technol.
vol. 17, pp. 1158–1165, July 1999.

[44] J. J. Chen and G. Yang, “CDMA fiber-optic systems with optical hard
limiters,” J. Lightw. Technol., vol. 19, no. 7, pp. 950–958, Jul. 2001.

[45] A. R. Forouzan, J. A. Salehi, and M. Nasiri-Kenari, “Frame time-hopping

fiber-optic code-division multiple access using generalized optical orthogonal
codes,” IEEE Trans. Commun., vol. 50, no. 12, pp. 1971–1982, Dec. 2002.