Three-Fold Increase in Number of Active Users in an
O-CDMA System Using Novel Code-Cycle Modulation
P. Vijay Kumar, Reza Omrani, Joe
Touch, Alan E. Willner, and Poorya Saghari
EE–Systems,
(213)740-4683, Fax (213)740-8729, Email: vijayk@usc.edu
Abstract: A novel, spectrally-efficient modulation
scheme, code-cycle modulation (CCM), that permits a threefold increase in the
number of active users is presented along with a supporting receiver
architecture.
Ó2005 Optical Society of
OCIS codes: (060.2330) Fiber optics communication, (060.2360) Fiber optics links and subsystems
1. Introduction
There
has been renewed interest in optical code-division-multiple-access (O-CDMA)
systems due to their potential for enhanced data privacy as well as “plug-and-play”
operation [1]. However, a key drawback
for O-CDMA has been that for the kinds of data rates demanded by current
practical applications, and the number of users desired to be supported, conventional
O-CDMA systems require an excessively high chip rate.
One approach towards partially alleviating
the high chip rate requirement has been the introduction of two dimensional (2-D)
O-CDMA architectures, in which the quasi-orthogonal spreading codes of the
different users are spread over both time and wavelength. [2]. However, even
under the time/wavelength approach using a reasonable number of wavelengths and
practical chipping rates, O-CDMA systems are still unable to accommodate an abundance
of active users [3].
We present here, a novel O-CDMA modulation
scheme, termed CCM for Code-Cycle Modulation.
CCM can be used in conjunction with either 1-D or 2-D O-CDMA
systems. While O-CDMA permits the
different users to operate asynchronously, in order for a particular
transmitter and receiver to exchange information, a synchronous link needs to be
created, usually accomplished through code acquisition at the receiver. CCM
takes advantage of the synchronous link that has so been created to send log2(T) bits of information across the link in the time that
it takes an OOK scheme to communicate a single bit, where T denotes the
length of the spreading code along the time axis. More specifically, under CCM,
the transmitter selects and transmits a particular cyclic (or wrap-around)
shift of the spreading code assigned to that transmitter. The relative spectral
efficiency of CCM permits a vastly increased number of users to be supported.
For instance, with 16 wavelengths, T=64 chips per message symbol,
CCM will permit the number of users supported to be increased more than
threefold in comparison with traditional OOK
2. Code-Cycle Modulation
While
CCM can be employed in conjunction with either a 1-D or 2-D O-CDMA systems, the
discussion here will focus on the 2-D case. The spreading code of each user in
a 2-D OCDMA system with L wavelengths and T time slots, can be
visualized as a 2-D (L×T) {0, 1} array [4]. Let Tc
denote the chip time, corresponding to the duration of a time slot and a chip rate
of 1/Tc chips per second. The duration Tb of a
message symbol is then given by Tb=TTc.
Each user in an O-CDAM system operates
asynchronously. Under On-Off Keying
(OOK) modulation, the presence or absence of the signal marks a 1 or a 0
respectively (Fig. 1).
Prior to demodulating data, each receiver
in an O-CDMA systems uses the relative orthogonality between a spreading code
and time-shifted versions of the code to align the phase of the locally
generated replica of the kth-user spreading code to that of the incoming
signal. Once the code has been acquired, the receiver then proceeds to
demodulate data. Thus in the post-acquisition phase, a synchronous link exists
between transmitter and receiver. The
CCM transmitter takes advantage of this synchronous link and transmits one of T
distinct cyclic or wrap around phase shifts of the user's spreading code. Thus
the size of the message alphabet under CCM is T and to send the ith
symbol in the modulation alphabet, the transmitter sends the ith cyclic
shift along the time-axis of the 2-D spreading
code
allocated to it. The right portion of
Fig. 1 shows the 8 cyclic shifts of an example 2-D code corresponding to
L=4 wavelengths and T=8 time slots.
|
|
|
Fig. 1.Comparing OOK with the new modulation scheme,
Code-Cycle Modulation (CCM) introduced here.
The higher spectra efficiency of CCM arises from its larger modulation
alphabet. The figure shows how CCM employs the 8 possible cyclic
(wrap-around) shifts of a single 2-D spreading code with 4 wavelength and T=8
time slots to create an 8-ary modulation alphabet, thereby permitting 3
binary digits to be transmitted in place of a single bit under OOK. |
3. CCM Receiver Architecture
In
an OOK system, at the receiver end, the pulses along the different wavelengths
are delayed by the appropriate integer multiples of chip delay Tc,
needed to bring all the pulses into time alignment. If in place of the spreading code, the
incoming spreading code were a cyclic shift of the spreading code, then the presence
of the cyclic shift would cause some pulses to be misaligned relative to others
by T time slots. As a result of this decoding one bit can take up to two
bit-times (2Tb). So we need a circulating optical shift register to
memorize the arriving bits within one bit time for duration of up to 2 bit
times. One means of bringing the resultant two sets of pulses into alignment is
to feed the pulses into a circulating optical shift register (COSR) (Fig. 2).
In practice, one would need two banks of
COSR's so that while one bank of COSR's is processing the current bit, the
other COSR completes decoding of the previous bit. It means that the receiver
consists of two copies of COSR bank shown in fig.2. In each bit time while one
of the COSR banks is connected to the input and saving the current data the
other COSR is connected to the output to decode the previous bit.
3.
Probabilistic Performance Analysis of CCM
We
will define the spectral efficiency (SE) here by:
There are three aspects of CCM to be
considered in determining its spectral efficiency: (a) under CCM, one is able
to transmit log2(T) bits of
information every Tb seconds as opposed to 1 bit every Tb seconds under OOK, (b) under OOK the transmitter of
each user is silent roughly 50% of the time, whereas, under CCM however, the
transmitter of each user is never silent and (c) under CCM, the receiver must
test for the presence of all possible phases of the spreading code as opposed
to just checking for the presence or absence of the code. A probabilistic model of signal and
interference was developed taking into account these observations and used to
analyze performance. Fig. 3(a) shows the distribution of interference under OOK
and CCM respectively. As it can be seen
|
|
|
Fig.
2. Receiver architecture for CCM including the time-aligning circulating
optical shift registers |
by
using CCM the probability of interference grows. Fig. 3(b) shows the increase in
number of users that can be supported for two different wavelengths 16, and 64.
It should be noted that in this graph the interference model of the CCM is incorporated.
So it shows that although the interference of the CCM is higher than regular
OOK, the number of active users in the CCM still increases up to 4 times while
using CCM. Fig. 3(c) shows the increase
in the spectral efficiency using CCM. As
can be seen, the SE of CCM is between twice and three times that of comparable
systems employing OOK. This allows CCM
to support a threefold or greater number of active users. Note that the number
of wavelengths does not affect much in Fig 3. b, and c.
|
|
|
|
|
|
(a) |
(b) |
(c) |
|
|
Fig. 3. Performance Gains Under CCM: (a) comparison
of the pdf of a CCM and OOK(b) The first curves plot the ratio of number of
users that can be supported under CCM to that supported under OOK as a function of the length along the time
axis of the spreading code used in the
OOK scheme. (b) the ratio of spectral
efficiencies of CCM and OOK as a function once again of the time length of a
spreading code |
|||
References
1.
J. A. Salehi, et al., IEEE
Trans. on Communications, 37,
824-842 (1989)
2.
G. C. Yang, et al., IEEE Trans.
on Communications, 45, 1426-1434 (1997)
3.
P. Saghari, et al., Optical
Fiber Communications (OFC 2004), paper FG7
4.
R. Omrani, et al., 41st Allerton
Conf. Comm., Control, Comp. (2003)