вход по аккаунту



код для вставкиСкачать
OFC/NFOEC Technical Digest © 2013 OSA
Differential link-loss compensation through dynamic
bandwidth assignment in statistical OFDMA-PON
Iván N.Cano, Ángel Peralta, Victor Polo, Xavier Escayola, María C. Santos, Josep Prat
Department of Signal Theory and Communications, Universitat Politecnica de Catalunya, Jordi Girona 1-3, E-08034, Barcelona, Spain
Abstract: Flexible bandwidth allocation strategies in a statistical OFDM-PON are experimentally
shown to compensate for differential link loss allowing for proper detection of ONUs separated up
to 55km (11.4dB).
OCIS codes: (060.0060) Fiber optics and optical communication; (060.2330) Fiber optics communications
1. Introduction
Orthogonal frequency division multiple access (OFDMA) has become recently an intensive research topic for
passive optical networks (PONs) due to features like dispersion tolerance and fine bandwidth granularity by using
subcarrier allocation with time division multiplexing [1, 2]. This makes it an interesting proposal for the second next
generation PONs [1]. Still, the introduction of OFDMA technologies into the PONs market is expected to be largely
determined by the availability of competitive cost devices. As an alternative to reduce the expenses in the optical
network units (ONUs), a statistical network with non-preselected light sources and wavelength control was proposed
in [3, 4].
Another system requirement demanded by operators for a future PON is the ability to reuse the existing fiber
infrastructure [5]. This includes the need for broad enough receiver dynamic range to handle differential losses
caused by spread users with respect to the central office. In order to manage this situation, this paper proposes and
experimentally investigates for the first time to the best of our knowledge the use of dynamic bandwidth allocation
in OFDM subcarriers to increase the differential loss margin for acceptable BER in the receiver. The study is based
on a statistical network with identical ONUs with the purpose to keep implementation simple.
2. Network topology
An OFDM-PON tree architecture based on power splitters with spectrum breakdown indicated by the services
supported such as the Accordance network proposal [6] is considered as shown in Fig. 1. In this work we focus on
the upstream transmission since generally it is more challenging than downlink. Furthermore, we study the statistical
OFDMA-PON concept consisting on cheap direct DFB laser modulation at the ONU and direct detection in the
OLT. The emission wavelength of the ONU lasers is random in a frequency band and it can be controlled through
temperature and current changes. The ONUs wavelengths are kept at an spectral distance that allows to maintain the
Optical Beat Interference (OBI) at the OLT below a quality threshold. The temperature control allows for a shift of
up to 0.8nm with steps of 0.1nm in the wavelength of each ONU. The OLT contains an intelligent algorithm such as
that proposed in [3] to control the λ of the ONUs lasers.
Fig. 1 (a) Network architecture, (b) spectra of the two ONUs with original ∆λ=0.5nm (upper), and ∆λ=0.1nm (lower) after control, enough
for correct detection
978-1-55752-962-6/13/$31.00 ©2013 Optical Society of America
OFC/NFOEC Technical Digest © 2013 OSA
3. Experimental setup
The experimental setup consisted of two ONUs which transmitted data simultaneously towards the OLT as depicted
in Fig. 2. For each ONU, 218 bits were generated randomly and mapped into quadrature phase shift keying
modulation (QPSK). In order to get a real signal, the QPSK symbols were accommodated to have Hermitian
symmetry in the iFFT. To test the bandwidth allocation, the total subcarriers were divided between ONU1 and
ONU2, which used the lower and higher frequencies respectively. The OFDM signals were loaded into an arbitrary
wavelength generator (AWG) which produced samples at 5GSa/s with 8 bits resolution in two separate channels.
The output signal then directly modulated two non-preselected DFB lasers launching a power of 0dBm for each
ONU. The transmitted (tx) optical power of each ONU was controlled by a variable optical attenuator (VOA). The
two ONUs emitted light initially at 1554.5nm and 1555.03nm. ONU1 was connected to the PON and afterwards
ONU2 was turned on; then ONU2 was tuned to displace its emission λ rapidly by 0.1nm to 1554.6nm where
reception can still be carried out without high interference. The combined spectra of both ONUs is shown in Fig. 1b.
Both optical signals were then joined by a 50/50 coupler and the complete OFDM uplink signal passed through an
9dB optical attenuator. The signal then travelled through 25km of single mode fiber (SMF) and was detected by a
single 10GHz PIN photodiode (PD) preceded by an erbium doped fiber optical amplifier (EDFA) which kept the
input optical power to the PD at -9dBm. The signal was then sampled with a 50GHz real-time sampling oscilloscope
which also processed the sampled data immediately in a pseudo-real-time way. Finally the bit error ratio (BER) for
each ONU was computed.
Fig. 2 (a) System setup schematics, (b) subcarriers (sc) allocation for each ONU
4. Results
Firstly, each ONU had half of the total subcarriers. The tx optical power was measured just after the VOAs and was
adjusted to be equal for both ONUs. Then, the tx optical power of ONU2 was lowered in 0.5dB steps while keeping
constant the ONU1 power. The same procedure was carried out with ONU1 keeping ONU2 optical power fixed. Fig.
3a plots the average BER of ONU1 and ONU2 obtained against the difference in power measured considering ONU1
as reference (∆P = PONU1 – PONU2). The BER floor appearing in the ONU2 curve is due to the strong noise at high
frequencies. For a target FEC limit BER of 10-3, it can be noticed that the dynamic range margin of the receiver is
extremely limited (around 1.5dB). The main reason for this restraint is the optical noise of the very close Tx spectra
added in the merged signal which cannot be filtered out easily. Thus, both ONUs have to transmit nearly at the same
power level to keep balanced the signal for proper detection. The network design would be then highly constrained.
To avoid such limitation, the subcarriers were asymmetrically allocated to the ONUs. The spectrum of ONU2 was
reduced and the freed bandwidth was added to ONU1. Then the power of ONU2 was varied until both ONUs were
detected at the defined BER limit. An extreme case when ONU2 has only 8 subcarriers and ONU1 the remaining
248 is plotted in Fig. 3b. ONU2 curve falls sharply because of the lower number of symbols and higher power in the
subcarriers compared to the noise. On the other hand, ONU1 presents a BER floor which is explained due to
quantization and higher electrical noise. Notably, the dynamic range of the receiver increased, and the ONUs can
then transmit up to a limit of 6.4 dB power difference and still both being detected with a BER below the FEC limit.
This can translate into 32km of additional fiber for ONU2 or a four-fold splitting ratio.
OFC/NFOEC Technical Digest © 2013 OSA
Fig. 3 BER against differential received power between ONUs for (a) 50%-50%, and (b) 93.75%-6.25% BW allocation. The insets are the
electrical spectra of ONU1 (orange), ONU2 (black)
The receiver dynamic range (∆P) for detecting both ONUs with different subcarriers share is plotted in Fig. 4.
The tolerable differential link-loss between the ONUs is seen to raise from 1.5dB to 11.4dB when the relative BW of
the farthest ONU (ONU2 in this case) varies from 50% to 2%. This allows to detect properly ONUs that can have up
to 55km fiber length difference or with an 8-fold splitting ratio difference. Naturally, the penalty for the low power
ONUs is a lower bitrate as noticed in the right side of fig. 4. This shows the possibility to use a flexible bandwidth
allocation to manage different link-loss present in a PON.
Fig. 4 Differential power allowed between ONUs against the BW allocation for each user
5. Conclusions
An experimental evaluation of flexible bandwidth allocation strategies in an OFDM-based statistical PON has
shown that differential link loses among users may be compensated through reduction of the subcarriers allocated to
the weakest ONU. With the flexible BW allocation, the differential link-loss increased from 1.5dB for equal sharing
of subcarriers to 11.4dB for a totally asymmetrical distribution of subcarriers. This allows a PON to properly detect
ONUs separated up to 55km.
This work was supported by European project ACCORDANCE and CONACYT grant 185291
6. References
[1] N. Cvijetic, “OFDM for next-generation optical access networks,” J. Lightw. Technol. 30, 384-398 (2012)
[2] D.Qian, et al., “A novel OFDMA-PON architecture with source-free ONUs for next-generation optical access netwroks,” in IEEE Photon.
Technol. Let., 21,1265-1267 (2009).
[3] I.Cano, M.C.Santos, V.Polo, and J.Prat, “Dimensioning of OFDMA PON with non-preselected independent ONUs sources and wavelengthcontrol,” in. Proc. of ECOC, Geneva, Switzerland, 2011, paper Tu.5.C.2
[4] W.Poehlmann, T. Pfeiffer, “Demonstration of wavelength-set division multiplexing for a cost effective PON with up to 80Gbit/s upstream
bandwidth,” in Proc. of ECOC, Geneva, Switzerland, 2011, paper We.9.C.1
[5] P. Chanclou, et al., “Network operator requirements for the next generation of optical access networks,” in IEEE Network 26, 8-14 (2012)
[6]K.Kanonakis, et al., “An OFDMA-based optical access network architecture exhibiting ultra-high capacity and wireline-wireless
convergence,” in IEEE Commun. Mag. 50, 71-78 (2012)
Без категории
Размер файла
649 Кб
ofc, oth3a, 2013
Пожаловаться на содержимое документа