International Journal of Electronics Engineering, 2(1), 2010, pp.

137-140

Performance of Turbo Coded CDMA System with PPM

Modulation on Optical Fiber
Puneet Chandra Srivastava1 & R.K.Singh2
1
Associate Professor, EC Dept., Raj Kumar Goel Institute of Technology, Ghaziabad, Uttar Pradesh Technical University,
Uttar Pradesh, India
2
Professor, EC Dept., Kumaon Engg. College, Dwarahat (Almora), Uttarakhand, India

Abstract: Pulse Position Modulation (PPM) is a power efficient scheme that is used with IM/DD.In this paper, performance
of an optical code division multiple access (CDMA) system with Turbo coding is analyzed and simulated. Turbo coding is
applied on an optical channel with PPM and direct detection of the received optical signal then the performance is evaluated.
From the simulation results, it is demonstrated that performance of the optical CDMA system can be improved by increasing
the interleaver length and the number of iterations in the decoding process for a fixed code rate.
Keywords: Turbo Codes, Pulse Position Modulation (PPM), Code Division Multiple Access (CDMA).

1. INTRODUCTION 2. SYSTEM MODEL

CODE division multiple access (CDMA) is an alternative 2.1. Turbo Encoding/Decoding
to the wavelength division multiple access (WDMA) and
The block diagrams of turbo encoder and decoder are shown
time division multiple access (TDMA) techniques in optical
in Fig.1 (a) and (b).The fundamental turbo code encoder is
fiber communication system [1], [2]. By applying CDMA built using two identical recursive systematic convolutional
to an optical communication system, the following benefits (RSC) codes with parallel concatenation [5]. An RSC
can be obtained: (i) Accurate time of arrival measurements, encoder is typically r = 1/2 and is termed a component
(ii) Security against unauthorized users,(iii) Protection encoder. An interleaver separates the two component
against jamming, (iv) Flexibility of adding users, (v) encoders. Only one of the systematic outputs from the two
component encoders is used, because the systematic output
Asynchronous access capability,(vi) Ability to support
from the other component encoder is just a permuted version
variable bit rate and bursty traffic etc. [3], [4]. of the chosen systematic output. Figure 1 shows the
A turbo code can be thought of as a refinement of the fundamental turbo code encoder.
concatenated encoding structure plus an iterative algorithm
for decoding the associated code sequence. The substantial
coding gain by turbo coding has been confirmed for a CDMA
system in an AWGN channel as well as in a wireless channel.
We consider the optical CDMA packet network which can
be modeled as a discrete Markov process. The performance Figure. 1(a): Fundamental Turbo Code Encoder
of this network is evaluated by turbo coded BER Systematic c1 and recursive convolutional c2 sequences
performance in terms of packet throughput delay while the second RSC encoder discards its systematic
characteristics. sequence and only outputs the recursive convolutional c3
sequence.
The rest of the paper is organized as follows: In Section
II, structures of encoder and decoder of turbo code and
system model of an optical network are described. In Section
III, packet throughput and delay are derived. In Section IV,
Simulation results are presented and conclusions are drawn
in section V.

*Corresponding Author: [email protected], [email protected]

Figure. 1(b): Turbo Code Decoder
138 International Journal of Electronics Engineering

The turbo code decoder is based on a modified MAP optical correlator, the correlator output is converted in to
(maximum a posteriori) algorithm that incorporates the electrical signal at the photo-detector. In the PPM
reliability values to improve decoding performance [6], [7]. demodulator, the highest output of two slots is selected as
Interleavers in a structure similar to that of the encoder link the transmitted symbol, and then it passes through the turbo
two component decoders. The MAP decoder computes a decoder to estimate the transmitted information bits.
posteriori probability conditioned on the received sequence.
The iterative turbo decoder makes use of a posteriori 3. PERFORMANCE ANALYSIS
probability (APP) in the form of a log likelihood ratio given
In the performance analysis, the following are assumed:
by [8], [9].
1) Optical components for each user are identical,
Pr{u = 1| x} 2) synchronization between the transmitters and the intended
L(u) = log (1) receiver is perfect in a chip and a bit levels, 3) constituent
Pr{u = 0 | x}
codes of turbo encoder are identical.
= Λ e + Λr + Λs
m
(2)
And, as a simple example for application of the turbo
Where x is received sequence, Λ is extrinsic
m
e
coded performance analysis to an optical CDMA network,
information from mth decoder (m = 0 or 1), Λr is a priori log
the following are assumed: 1) an optical network is modeled
likelihood ratio of the systematic bit u and Λs is log likelihood
ratio of the a posteriori probabilities of the systematic bit. as a slotted CDMA Network, 2) a packet has a packet size
The iteration process continues until a desired performance of L bits, 3) finite K terminals are in the network, 4) each
is achieved at which point a final decision is made by terminal transmits with same intensity, 5) timing information
comparing the final log likelihood ratio to the threshold 0. for synchronization is perfect, and 6) each packet consists
of one codeword of turbo code.
2.2. Optical CDMA Network In the slotted optical CDMA network, at the beginning
of each slot, each terminal is either blocked or unblocked,
depending on whether its previous packet was successfully
transmitted or not. Only the unblocked terminal can generate
a new packet with probability Pn and the blocked terminal
retransmit its backlogged packet with probability Pr in each
slot, the central receiver broadcasts feedback messages to
all the terminals simultaneously.
The dynamics of this network can be modeled by a
multidimensional Markov chain. The number of blocked
Figure. 2(a): Optical CDMA Transmitter terminals can be modeled by a Markov chain where a state
represents the number of blocked terminals. The state space
of this Markov chain is {0, 1, …, K} where K is the number
of terminals in the network. The transition from one state to
Figure. 2(b): Optical CDMA Receiver another is determined by the difference between the number
of new successful transmissions and retransmissions. The
The optical CDMA network consists of K users, each
successful new transmission and the successful
transmitting its own message. Each user is assigned a unique
retransmissions have no influence on the system state.
optical pulse sequence for its address. The optical CDMA
network has the advantages of being able to use simple In a given slot, there are N = Nn + Nr packets where Nn
pulsed lasers without wavelength control and standard photo and Nr are the number of new and backlogged packets ,
detectors without wavelength control and standard wideband respectively. Then the Nn and Nr are independent Bernoulli
photo detectors without narrow optical filters. All of the users random variables with distributions.
operate independently, and no common clock is needed to
 K − i l
synchronize the transmitters of the users. Qn (l | i ) = Pr{ N n = l | X k = i} =  Pn (1 − Pn ) K − i − l
 l 
(3)
In Fig. 2(a) the information bits of each user is fed in to
the turbo encoder. The turbo encoded bit stream is modulated  i
at the PPM modulator where transmitter sends its pulse Qr (l | i ) = Pr{N r = l | X k = i} =   Prl (1 − Pr )i − l (4)
 l
sequence in one of the two time slots to represent each data
bit. where Xk denotes the number of blocked terminals at the
beginning of slot k. Thus, the total number of packets in a
In Fig. 2(b), the received signal is first processed at the
slot has a distribution given by
optical correlator. When the desired signal passes through
Performance of Turbo Coded CDMA System with PPM Modulation on Optical Fiber 139

l Average delay in a steady state is defined as the average

Qr (l | i ) = Pr{N t = l | X k = i} = ∑ Qn (m | i )Qr (l − m | i ) (5) number of time slots required to successfully transmit a data
m=0
packet. From the Little’s formula, the average delay is given
One–step transition probability of this Markov chain at by
the given time slot k, Pi , j = Pr{ X k +1 = j | X k = i} , is given
by Bg
Dav = + 1, (14)
Tp
0, j < i − 1 Where Bg is expected channel backlog and given by

Qn (0 / i )∑ l =1 Qr (l | i ) Ps (l ), j = i − 1
j
K
Pi,j =  (6) Bg = ∑ jf ( j )
Qn (k + 1| i )∑ l =1 Qr (l | i ) Ps (l + k + 1)
j (15)
j =0

+Qn (k | i )∑ l =1 Qr (l | i )[1 − Ps (l + k )], j ≥ i
j
4. SIMULATION RESULTS
For simulation examples, the code length F = 500, the weight
The Markov chain has a limiting distribution denoted
of optical orthogonal code w = 5, packet size L = 1024 bits
by
and two identical four–state recursive systematic convolution
f = [ f (0), f (1),  , f ( K )], (7) codes with code generator polynomials (1 + D2, 1 + D + D2)
Where and memory size V = 2 are assumed. The overall code rate
is chosen as 1/3 where every source data bit is mapped into
f ( j ) = Pr{ X ∞ = j} = lim Pr{ X k + n = j | X k = i} (8) 3 coded symbols, respectively. In a turbo code with rate
n →∞

Since the above Markov chain has finite population and 1/3, the 3 bits at the encoder output can be interpreted as
is irreducible, a steady –state distribution exists. The steady- the sum of one uncoded bit and two parity bits.
state probabilities are found by solving the linear system of In Fig. 3, packet throughput vs. offered load is shown
equations for 8 iterations, MAP algorithm, interleave length =2000,
f = fP, (9) new packet generation probability P n = 0.04, packet
retransmission probability Pr = 0.8 with the varying number
with constraint
of iterations. In the decoder, the MAP algorithm is adopted
K
because it achieves better performance than the other
∑ f ( j ) = 1,
j =0
(10) algorithms. It is shown that the packet throughput is
significantly improved by increasing the number of iterations
Where P = [Pi,j] is a state transition matrix. The above
is the decoding process. The turbo coding improves packet
two equations, (9) and (10), involves solution a set of N + 1
throughput by about more than twice for a wide range of
simultaneous equation.
offered traffic load.
Packet throughput in a steady state is defined as the
average number of successfully transmitted packets per slot,
and given by
K
 K

Tp = ∑ jS ( j )  ∑ f (m)ξ( j | m) 
j =1 m=0
(11)

Where S(j) is the probability of a successful packet

transmission at the state j and given by
e
 L
S(j) = ∑  m [ P ( j )]
k =0
b
m
[1 − Pb ( j )]L − m (12)

e is error correcting capability in a block of length L and

ξ(j|m) is composite (newly generated and retransmitted ) Figure. 3: Packet Throughput for the Different no. of Iterations
packet arrival distribution function (newly and retransmitted) In Fig. 4, packet throughput vs. offered load is shown
and given by for 8 iterations, MAP algorithms, Pn = 0.04 and Pr = 0.8
min( j , K − m ) with the varying interleave size. It is confirmed that the larger
ξ ( j | m) = ∑
i = max( j − m ,0)
B (i, K − m, pn ), B ( j − i, m, pr ) (13) interleave size, the higher packet throughput. So, we can
find a tradeoff between a system complexity due to the
 j interleaver size and the packet throughput. And, it can be
for binomial distribution, B (i, j , ρ) =   ρi (1 − ρ) j − i noted that the long interleaver may not be applicable for the
 i
140 International Journal of Electronics Engineering

optical CDMA network because the high speed (>Mbps) data that turbo coding offers considerable coding gain compared
is transmitted in an optical network. to the uncoded system with reasonable encoding /decoding
complexity. The performance was gradually improved by
increasing the interleaver length and the number of iterations
for a fixed code rate. As a decoding algorithm, the Max-
Log-MAP algorithm is recommendable to be a suitable
choice in terms of both performance and complexity. It can
be concluded that turbo coding is very effective in improving
throughput performance from the viewpoints of network. The
extension of BPPM to higher order PPM (>2) is very
straightforward from the analysis of this paper. The result in
this paper can be applied to an optical CDMA network with
power limitation.

REFERENCES
Figure. 4: Packet throughput for the Different Interleaver Sizes
[1] J. A. Salehi, “Code Division Multiple Access Techniques
in Optical Fiber Networks – Part I: Fundamental Principals”,
IEEE Transactions Commn., 37, pp. 824-833, Aug. 1989.
[2] J. A. Salehi and C. A. Brackett, “Code Division Multiple
Access Techniques in Optical Fiber Networks – Part II:
System Performance Analysis”, IEEE Transactions
Commn., 37, pp. 834-842, Aug. 1989.
[3] L.A.Rusch and H.V.Poor, “Effects of Laser Phase Drift or
Coherent Optical CDMA,” IEEE J. Select. Areas Commun.,
13, No.3, pp. 577-591, Apr.1991.
[4] P.R.Procnal, M.A.Santoro, and T.R.Fan, “Spread Spectrum
Fiber Optic Local Area Network using Optical Processing,”
J.Lightwave Technol, 4, No.5, pp. 547-554 May 1986.
[5] Berrou, C. and Glavieux, A., “Near Optimum Error
Figure. 5: Packet throughput for the Different Decoding Correcting Coding and Decoding: Turbo-Codes,” IEEE
Algorithms Trans. on Communications, 44, No. 10, October 1996,
In Fig. 5, Packet throughput vs. offered load is shown pp. 1261-1271.
for 8 iterations, MAP algorithm, K = 8, Pn = 0.04 and [6] S. Benedetto and G. Montorsi, “Unveiling Turbo Codes:
Pr =0.8 with varying the turbo decoding algorithms. The Some Results on Parallel Concatenated Coding Schemes,”
IEEE Trans. Inform. Theory, 42, pp. 409428, 1996.
MAP algorithm accomplished the best packet throughput
among three algorithms. However, throughput difference [7] L. R. Bahl, J. Cocke, F. Jelinek, and J. Raviv, “Optimal
between the MAP and the Max-Log-MAP algorithms is not Decoding of Linear Codes for Minimizing Symbol Error
Rates,” IEEE Trans. Inform. Theory, 20, pp. 284- 287, Mar.
significant.
1974.
[8] J. Hagenauer and P. Hoeher, “A Viterbi Algorithm with Soft
5. CONCLUSION
Decision Outputs and its Applications,” in Proc. of IEEE
The performance of an optical direct-detection CDMA with GLOBECOM’89, pp. 1680-1686, Dallas, TX, U.S.A, Nov.
turbo coding was analyzed and simulated for the BPPM 1989.
modulation. From the simulation results, it was confirmed [9] A. J. Viterbi and J. K. Omura, “Principles of Digital
Communications and Coding”, McGraw-Hill, 1979.