Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

NGPON2

Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 15

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/272640999

NG-PON2 technology and standards

Article in Journal of Lightwave Technology · March 2015


DOI: 10.1109/JLT.2015.2389115

CITATIONS
READS
230
12,776

1 author:

Derek Nesset
Huawei Technologies
108 PUBLICATIONS 3,037 CITATIONS

SEE PROFILE

All content following this page was uploaded by Derek Nesset on 29 July 2016.

The user has requested enhancement of the downloaded file.


JOURNAL OF LIGHTWAVE TECHNOLOGY 1

NG-PON2 Technology and Standards


Derek Nesset, Senior Member, IEEE

Abstract—This paper provides a tutorial overview of the latest


generation of passive optical network (PON) technology
standards nearing completion in ITU-T. The system is termed NG-
PON2 and offers a fiber capacity of 40 Gbit/s by exploiting
multiple wave- lengths at dense wavelength division multiplexing
channel spacing and tunable transceiver technology in the
subscriber terminals (ONUs). Here, the focus is on the
requirements from network op- erators that are driving the
standards developments and the tech- nology selection prior to
standardization. A prestandard view of the main physical layer
optical specifications is also given, ahead of final ITU-T
approval.
Index Terms—Optical access, optical communication equip-
ment, optical fiber communication, optical fiber networks,
optical network (PON), time division multiple access, wavelength
division multiplexing.

I. INTRODUCTION
HE incessant rise in data consumption by both fixed and
T
Fig. 1. PON Architecture and Terminology.
mobile broadband subscribers drives continual industry
innovation to meet this challenge. Optical fiber access systems
based on passive optical networks (PON) are currently being de- for NG-PON2 were previously reviewed in detail in [7] and a
ployed on a mass market scale by numerous network brief summary of the key points is provided below.
operators worldwide. These systems typically exploit Gigabit- The architecture and terminology relating to PON systems
class PON systems such as G-PON [1] or EPON [2] as is shown in Fig. 1. The key elements are the optical line
standardized by the ITU-T and IEEE respectively. Both these terminal (OLT) or PON headend that is usually located in a
standards bodies have also defined 10 Gigabit-class PON central of- fice, the optical network unit (ONU) that is
technologies (XG-PON1 typically located at the subscriber’s premises and the optical
[3] and 10G-EPON [4]) and early deployments have been re- distribution network (ODN) made up of fiber and optical
ported. As the next step in fiber access evolution, the ITU-T is power splitters located in the outside plant (e.g., ducts and
defining the second next generation PON (NG-PON2) [5]. poles).
NG-PON2 is a 40 Gbit/s capacity PON system that exploits The original view from network operators in FSAN was
both the time and wavelength (λ) domains. In Section II, this that NG-PON2 could be a disruptive technology that might
tu- torial gives an overview of the network operator exploit a new ODN with wavelength (λ) splitters rather than
requirements for NG-PON2. In Sections III, IV and V the power splitters [8]. However, considering the substantial
system technologies considered for NG-PON2 are described. investments to date in power splitter based fiber infrastructure,
In Sections VI, VII, and VIII details of the NG-PON2 system it became clear that a key requirement for NG-PON2 must be
capabilities and stan- dardization progress are reviewed. compatibility with such deployed ODNs. This does not mean
that λ-splitter based ODNs are not within the scope of NG-
II. NETWORK OPERATOR REQUIREMENTS PON2, however, the optical transmission technology must not
Within the full service access network (FSAN) forum [6], require a λ-splitter to work. Thereby ensuring a common
network operators have an opportunity for exchanging ideas transceiver technology can also be used by power splitter
and building consensus on worldwide requirements for fiber based ODNs.
access networks. These requirements can be used as the basis A further requirement driven from the re-use of deployed
for input to the ITU-T, Question 2, Study Group 15 where optical fiber infrastructure concerns the optical path loss. NG-
fiber access recommendations are formalized. Operator PON2 must be able to operate over the same passive infras-
requirements tructure (ODN) previously defined for G-PON and XG-PON1.
This means optical budget classes as defined for XG-PON1
must be achievable i.e., from 29 to 35 dB (maximum) with up
Manuscript received October 6, 2014; revised December 2, 2014; accepted to 15 dB differential loss. The differential loss is defined as
December 18, 2014. the absolute difference between the optical losses of any two
The author is with the British Telecommunications plc., Ipswich IP5 3RE,
U.K. (e-mail: derek.nesset@bt.com).
given paths within the same ODN.
Color versions of one or more of the figures in this paper are available Furthermore, network operators require that the NG-PON2
online at http://ieeexplore.ieee.org. system can co-exist with previously deployed systems i.e., G-
Digital Object Identifier 10.1109/JLT.2015.2389115
PON and XG-PON1. This facilitates progressive migration
0733-8724 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
JOURNAL OF LIGHTWAVE TECHNOLOGY 1
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
2 JOURNAL OF LIGHTWAVE TECHNOLOGY

NG-PON2 will be expected to meet new service demands. A

Fig. 2. Future optical access networks are expected to be truly multi-


service.

of existing subscribers as they move to the new technology,


and addition of new NG-PON2 subscribers, without disrupting
services for customers on the legacy PON systems. These re-
quirements are primary factors in determining the wavelength
plan. Clearly, when installed alongside legacy systems, the
NG- PON2 system should not impact the existing services i.e.,
no (or minimal) impact from linear/non-linear crosstalk or
increase to optical budget on the already operating legacy
PON system.
To differentiate itself, the NG-PON2 fiber capacity must be a
significant advance from XG-PON1 that already offers a
shared 10 Gbit/s capacity in the downstream. However, the
NG-PON2 system must not be so ambitious that it makes the
technology costly or impractical in the near term.
Additionally, network operators are looking to offer 1 Gbit/s
sustainable bandwidth services to any ONU. The target for
NG-PON2 was, conse- quently, set at a baseline downstream
capacity of 40 Gbit/s.
Network operators are constantly concerned with reducing
the operational expenditure (OPEX) involved in running their
networks and this drives an aim to keep inventory to a minimum.
Where possible, the number of network equipment part
variants should therefore be limited as far as is practical. This
results in a requirement for colorless ONUs in NG-PON2—
especially for the highly cost sensitive, mass-market
residential services.
NG-PON2 must be suitable for eventual use in residential
fiber access i.e., it must support anticipated residential access
service requirements at an acceptable cost. Typically, residential
customers do not expect to pay more as broadband service
rates increase and there is no evidence this propensity will
change for NG-PON2 derived services. However, early
deployments may target business/backhaul markets in advance
of residential services. There will be an inevitable technology
learning curve and a strong dependence on worldwide
deployment volumes to achieve lower equipment costs.
Ultimately, NG-PON2 is expected to be a multi-service in-
frastructure enabling network operators to collapse diverse
ser- vice platforms into one (see Fig. 2). NG-PON2 must be
highly scalable, flexible, reliable and efficient in both
bandwidth and power consumption.
As well as conventional business and residential services,
2 JOURNAL OF LIGHTWAVE TECHNOLOGY
reflective semiconductor amplifier based transmitter.
major driver for future evolution of PON will be further The downstream signal is modulated
growth of mobile broadband (e.g., LTE moving to LTE-A)
and the need for corresponding backhaul capacity.
Furthermore, mobile net- work operators are considering
new network architectures such as the Cloud-RAN [9] that
drive a requirement for, so called, mobile fronthaul. Mobile
fronthaul demands high capacity links to carry IQ samples
from the air interface (Remote Radio Head) to the Base
Band Unit with strict latency requirements and NG- PON2
is a prospective solution for such applications.

III. NGPON2 TECHNOLOGY OPTIONS


Before determining the best candidate technology to
meet the above network operator requirements, a
benchmarking ex- ercise was undertaken in FSAN to
compare technology options. These are reviewed briefly
below before describing the adopted approach in more
detail.

A. Time Division Multiplexed PON


In this proposal, the downstream line rate of the PON is
increased to 40Gbit/s using duo-binary modulation [10].
The approach is very similar to current PON systems with
a single wavelength used downstream and a single
wavelength upstream. Thus there are none of the wavelength
tuning/management com- plexities of some WDM based
systems. Exploiting duo-binary modulation in the
downstream allows a 20 GHz bandwidth re- ceiver to be
used at the ONU as well as reducing the dispersion penalty.
The reduced bandwidth also brings into consideration
advanced, higher sensitivity, APD receivers just coming
out of research labs [11]. To provide sufficient link budget,
strong for- ward error correction from low density parity
check codes was suggested. Furthermore, use of a bit-
interleaving protocol was proposed in the downstream to
reduce power consumption and processing burden at the
ONU [12].

B. Wavelength Division Multiplexed PON


There are many concepts concerning how WDM-PON
may be implemented and several were studied in the
context of NG-PON2. These included:
a) Externally seeded WDM-PON [13] whereby a broad-
band light source at the head-end is spectrally sliced
in a λ-splitter based ODN and distributed to
reflective ONUs. The main advantage of this
approach is that it is well understood and relatively
mature, with commercial sys- tems already available.
However, the potential for offer- ing greater than 1
Gbit/s services is less clear and would be beyond
current system capabilities. The commercially
available systems also require a λ-splitter to be used
and operation over a power splitter based ODN is
very chal- lenging for the link budget.
b) Wavelength re-use WDM-PON [14] permits a single
wavelength per ONU to be used for both upstream
and downstream transmission. This is enabled by
the reuse of the downstream wavelength for
upstream transmission by re-modulation in a
NESSET: NG-PON2 TECHNOLOGY AND STANDARDS 3

with an Inverse Return-to-Zero format and the upstream


with a Return-to-Zero format.
c) Tunable WDM-PON [15] using cost reduced versions of
conventional DWDM tunable transmitter modules.
These modules are modified by removing thermoelectric
coolers and wave-lockers from the modules to drive
down costs. Upstream transmitter tuning and calibration
is performed using a shared OLT based wave-locker. In
the downstream, tunable receivers are required at the
ONU to provide col- orless operation. Fig. 3. Illustration of the NG-PON2 architectural approach with TWDM-PON
overlaid with PtP WDM PON to meet the diverse service requirements of
d) Ultra-dense Coherent WDM-PON [16] leverages the FTTH and Wireless Backhaul.
progress in coherent transmission and digital signal pro-
cessing (DSP) in core transport and adapts this to the
fiber access domain. Very tightly spaced channels∼ ( 3
bandwidth, low latency business/backhaul services on the
GHz) are selected through coherent detection. This
same ODN; timeframe for first system availability; relative
provides the benefits of both high selectivity and a very
maturity of components; prospects for low cost residential
sensitive receiver. Each user is offered 1 Gbit/s using a
equipment; port density and fiber management complexity at
dedi- cated differential quadrature phase shift keying
the central office; potential for incremental capacity growth or
modulated wavelength. Due to the complexity of the
channels dedicated to different services; power consumption
envisaged trans- mitter and receiver, developments in
and applicability of power saving modes; and technology risk.
photonic integration (e.g., silicon photonics) are
To reduce the magnitude of the technology step to NG-PON2,
considered necessary to make this cost effective and
it is beneficial to exploit synergies with already developed
practical to implement.
PON equipment. Where possible, this means similar optical
e) Self-seeded WDM-PON [17] discards the broadband light
compo- nents and protocol concepts and schemes. In practice,
source seen in option (a) above and the ONU seed light
the sec- ond aspect points towards considering re-use of the
is self-generated using a reflector placed at the common
XG-PON1 standards as far as possible while accommodating
port of a λ-splitter. This scheme has advantages of
necessary extensions for NG-PON2.
simplicity but it does rely on the λ-splitter to be
A non-technical, but crucial factor, to consider in selecting
present to work and there are limitations on the length
a technology for standardization is actual, or potential for,
of the drop fiber between the splitter and the ONU.
wide industry support. Without this, progress in standards
bodies is likely to be difficult and the supply chain necessary
C.Orthogonal Frequency Division Multiplexed (OFDM) PON for low costs and wide availability will not materialize.
Something that is theoretically possible, but promoted by one
The OFDM proposals considered for NG-PON2 use OFDM
vendor or demon- strated with niche/research components is
both as a spectrally efficient modulation format and a multiple
not a good candidate for subsequent standardization where
access technique. OFDM permits the use of 10 GHz optical com-
viable alternatives exist.
ponents for 40 Gbit/s downstream data rates. The trade-off is
With all the above considerations in mind, the proposal jointly
complexity in the electrical domain with advanced analogue-
taken by FSAN member companies into the ITU-T consisted
to- digital/digital-to-analogue converter and DSP circuits.
of TWDM-PON as the primary solution for NG-PON2 with
Down- stream data destined for particular ONUs can be
the optional addition of coexisting Point-to-Point WDM
allocated to OFDM sub-carriers and timeslots. For 10 Gbit/s
overlay channels (PtP WDM). The reason for two approaches
upstream trans- mission, conventional TDMA by way of ONU
being used in this way relates to the need to offer dedicated,
burst allocations can be exploited or, alternatively, OFDMA
high capacity, low latency links for some services on the same
schemes can also be used [18].
ODN as residential services. This approach is illustrated by
Fig. 3 that shows Fiber to the Home (FTTH) delivered by
The final NG-PON2 technology option considered was
TWDM- PON and a nearby wireless access point being served
Time and Wavelength Division Multiplexed (TWDM) PON
by a PtP WDM channel. Both services are delivered over a
[19], [20]. TWDM-PON is described in more detail below
common power splitter based PON.
as this was the primary technology progressed through to the
ITU-T for standardization.
It is hard to capture in this brief article the reasons behind IV. TIME AND WAVELENGTH DIVISION MULTIPLEXING
the choice of NG-PON2 technology. There is an element of TWDM is the primary solution for NG-PON2 as it was
subjec- tivity about the evaluation and, quite rightly, not all
most
judgments are purely technical in nature. Key factors
compatible with the high volume residential application. The
considered were as follows; loss budgets with passive ODNs;
baseline TWDM-PON uses four, DWDM spaced, bi-directional
compatibility with legacy PON and deployed ODNs; relative
λ-channels, each of 10 Gbit/s downstream and 2.5 Gbit/s up-
complexity; potential for higher line-rates and 10 Gbit/s
stream line rate (see Fig. 4). This results in 40 Gbit/s
services; efficient utilization of system capacity; ability to
aggregate downstream capacity and 10 Gbit/s upstream. With
offer both residential and high
this ap- proach, extensive re-use of XG-PON1 developments is
NESSET: NG-PON2 TECHNOLOGY AND STANDARDS 3
made, as
4 JOURNAL OF LIGHTWAVE TECHNOLOGY

available,

Fig. 4. NG-PON2 system architecture and high level features. Note that the
internal configuration of the WDM is just for illustration and is not defined by
ITU-T recommendations.

well as exploiting the λ-domain to enable incremental increase


of capacity. To facilitate colorless ONUs, λ-tunable
transmitters and receivers are necessary. A key challenge for
TWDM-PON is to realize the underlying tunable components
at low cost [21] and to precisely control and manage the λ-
domain [22].

V. POINT-TO-POINT WDM OVERLAY


Point-to-Point WDM Overlay (PtP WDM) allows NG-
PON2 to meet demanding operator requirements for business and
back- haul services. PtP WDM enables a dedicated λ-channel
to be provided to each ONU. In essence the ideas promoted in
the Tunable WDM-PON proposal have been adopted as a PtP
WDM overlay. In the baseline configuration, eight channels of
PtP WDM are considered to allow full co-existence with
legacy systems (see Fig. 4). Depending on the particular
deployment scenario, a network operator may dedicate unused
spectrum to additional PtP WDM channels in a flexible way.
The ONUs for PtP WDM require similar low cost tunable
transmitter and receiver elements as for TWDM-PON with the
main difference being the continuous mode operation for PtP
WDM (c.f., burst mode for TWDM-PON). Bit-rate classes
have been defined for PtP WDM spanning 1-10 Gbit/s.

VI. ITU-T STANDARDS FOR NG-PON2


NG-PON2 standards have made significant progress in the
ITU-T. The requirements and physical layer specification
have been consented as recommendations G.989.1 and
G.989.2 re- spectively. In development is the protocol (TC-
Layer) recom- mendation G.989.3; targeted for consent in
mid-2015.

VII. NG-PON2 PHYSICAL MEDIA DEPENDENT LAYER (PMD)


At the time of writing, the G.989.2 recommendation for
NG-PON2 was still in the process of resolving comments
prior to final ITU-T approval and publication. The information
given below is provided by the author in good faith based on
long standing and uncontentious material. However, when
4 JOURNAL OF LIGHTWAVE TECHNOLOGY

TABLE I
LINE RATES FOR NG-PON2

TWDM Downstream line rate (Gbit/s) Upstream line rate


(Gbit/s)
Basic Rate 9.95328 2.48832
Rate Option 1 9.95328 9.95328
Rate Option 2 2.48832 2.48832
PtP WDM Downstream/Upstream line rate

(Gbit/s) Class 1 1.2288–1.2500


Class 2 2.4576–2.6660
Class 3 9.8304–11.09
Class 4 6.144 (still under study)

approved and published versions of ITU-T Recommendations


should always be consulted.

A. Line Rates
In addition to the baseline configuration described
above, there are options for symmetric line-rates of both
2.5 and 10 Gbit/s. Symmetric services are typically
demanded in busi- ness/backhaul applications. Bit-rate
classes have been defined for PtP WDM spanning 1–10
Gbit/s as shown in Table I with specific line rates being
derived by considering the client signals to be transported
e.g.: Ethernet, CPRI [23], SDH/SONET and OTN.

B. Wavelength Plan
As might be expected, with each generation of PON, the
availability of unallocated spectrum diminishes. The
wavelength plan for NG-PON2 has been selected as the
best compromise for co-existence with G-PON, XG-PON1
and RF Video; re- use of deployed co-existence WDMs
[24]; ease of filtering; availability of existing C/L-band
components (e.g., transmit- ters, receivers and amplifiers);
and minimizing Raman fiber impairments. There was some
consideration of defining two wavelength plans that would
be selected based on the presence, or not, of RF Video.
However, it was reasoned that a single wavelength plan to
meet the requirements of all FSAN opera- tors would
facilitate a worldwide standard and increase optical
component volumes and ultimately lower costs for all.
The upstream wavelength bands for TWDM (see Table II)
are located in the C-band where components are already
shipping in high volume so this should facilitate lower costs
in the ONUs. The downstream TWDM transceiver
components will be lower volume, and costs are shared
across multiple subscribers, so L-band wavelengths are
allocated.
It may be noted that three upstream wavelength band op-
tions have been specified for TWDM-PON. These options
are motivated by differing capabilities of the ONU
transmitter to control its wavelength. The Wide option may
be useable by a “Wavelength-Set” [19] approach to
channel control, where a distributed feedback laser (DFB)
is allowed to drift in wave- length over a wide range, as the
OLT de-multiplexer has cyclic pass bands. The Narrow
option may be most appropriate for temperature controlled
lasers that can lock onto an assigned DWDM wavelength.
NESSET: NG-PON2 TECHNOLOGY AND STANDARDS 5

TABLE II
WAVELENGTH PLAN FOR NG-
PON2

TWDM Upstream Wavelength Band (nm) Downstream Wavelength Band (nm)

Wide 1524–1544
Reduced 1528–1540 1596–1603
Narrow 1532–1540

PtP WDM Downstream/Upstream Wavelength Band (nm)

Shared Spectrum 1603–1625


Expanded Spectrum 1524–1625
to deliver lower cost optical transceivers at the ONUs.

In the downstream direction, wavelength channels are on a


fixed 100 GHz grid with eight channels having been specified
from 187.1–187.8 THz. In the upstream direction, channel spac-
ing of 50, 100 and 200 GHz are supported but no specific channel
plans have been recommended as it is envisaged that the tunable
ONU transmitters will adapt to whatever grid is employed by
the OLT de-multiplexer.
It is worth expanding on the reasoning for two wavelength
bands being specified for the PtP WDM channels. The Shared
Spectrum option is the useable wavelength band considering
the scenario of full co-existence with legacy PON systems (G-
PON, XG-PON1 and RF Video). The Expanded Spectrum
option exploits the concepts of spectral flexibility in NG-
PON2 by enabling bands not being used in a particular
deployment to be utilized by PtP WDM. The Expanded
Spectrum option may also be beneficial in a Greenfield
scenario with no legacy co-existence limitations.

C. Spectral Flexibility
One of the key features of NG-PON2 that facilitates a
diverse range deployment scenarios, network applications and
evolu- tion paths is Spectral Flexibility. This means that,
whenever a particular subset of optical spectrum is unused by
TWDM and/or legacy PON systems, PtP WDM may make
use of that particular sub-band. Such flexibility can facilitate
the support of different customer types on the same ODN in a
flexible way. Additionally, spectral flexibility facilitates a range
of system co- existence scenarios and allows operators to use
new wavelength bands when legacy systems are
decommissioned.

D. Co-existence
Co-existence with legacy PON systems is facilitated by the
wavelength plan. As for the XG-PON1 system that came be-
fore it, the NG-PON2 system uses wavelength overlay to
permit each system to operate independently on a common fiber
infras- tructure. This permits simultaneous co-existence with
legacy PON (G-PON and XG-PON1) and 1555 nm RF
video. This can be seen in Fig. 5 that shows the
wavelength plan for NG-PON2 along with those of G-
PON, XG-PON1 and RF Video. The TWDM PON
downstream channels fit between XG- PON1 downstream and
the monitoring band (labelled OTDR). The upstream TWDM
channels work in the C-band above the co-existence wavelength
multiplexer [24] edge at 1524 nm and below the 1550 nm RF
video band. Use of C-band is expected
NESSET: NG-PON2 TECHNOLOGY AND STANDARDS 5

The Shared Spectrum wavelength band for PtP WDM is


shown in Fig. 5. This is configured as a mixed upstream
and downstream plan which is assigned according to the
network operator requirements.
Co-existence with RF video signals at 1555 nm is
accommo- dated by the wavelength plan but may need
methods to compen- sate for Raman crosstalk that can
impact lower frequency RF channels. Due to the variety of
RF video implementations, this will require joint
engineering between the NG-PON2 system vendor and
network operator.

E. Compatibility With Legacy ODN


In response to network operator requirements to be
able to re-use their already installed PON fiber
infrastructure, the NG-PON2 PMD has been defined to be
compatible with power splitter based ODNs.
To permit re-use of power splitter based ODNs already
de- ployed for G-PON and/or XG-PON1, the optical path
losses supported by NG-PON2 transceivers must also be
compatible. Furthermore, the need for co-existence with
legacy PON sys- tems drives the NG-PON2 standards to
assign the same nomen- clature and values to the classes for
optical path loss as were used for XG-PON1. These are
shown in the Table III. As can be seen from Table III, the
requirement for compatibility with a 15 dB ODN loss
differential has also been retained. Furthermore, the 40 km
fiber distance difference [25] from the OLT to the fur-
thest and closest ONUs respectively is also to be
supported in NG-PON2.

F. Additional ODN Options


In a Greenfield scenario, where there is no legacy PON
in- frastructure to reuse, NG-PON2 can support a choice
of ODNs i.e., power splitting, wavelength splitting, or a
combination of both. Wavelength splitting offers the
advantage of reduced split- ting loss that may be used to
extend system reach or permit the use of lower
specification transceivers. For some services the physical
layer channel isolation offered by wavelength splitting
may also be attractive to offer virtual private line services
using wavelength channels.

G. Optical Path Penalty


The optical path penalty (OPP) accounts for
transmission impairments due to the presence of an ODN
and the conse- quent apparent reduction of receiver
sensitivity. Thus, OPP
6 JOURNAL OF LIGHTWAVE TECHNOLOGY

Fig. 5. NG-PON2, G-PON and XG-PON1 Wavelength Plans. (a) Downstream direction (OLT-ONU). (b) Upstream direction (ONU-OLT).

TABLE III
TABLE IV
CLASSES FOR OPTICAL PATH LOSS IN NG-PON2
WAVELENGTH CHANNEL TUNING TIME CLASSES FOR NG-PON2

Class N1 N2 E1 E2 Tuning Time Class Tuning time

Minimum Loss (dB) 14 16 18 20 Class 1 < 10 μ s


Maximum Loss (dB) 29 31 33 35 Class 2 10 μ s to 25
Class 3 ms
25 ms to 1 s

includes the impact of inter-symbol interference, fiber chromatic channels


dispersion and fiber non-linearity. An interesting new OPP factor
in NG-PON2 that was not relevant in previous PON
generations is Raman power depletion. In NG-PON2, the
TWDM upstream channels act as a Raman pump source for
the TWDM down- stream channels and so the power of the
upstream channels is depleted during transmission. The
impact can be as large as a
1.5 dB or so power reduction at the OLT receiver input. Some
mitigation of this effect has been necessary in the case of an
E2 class ODN by restricting the power variation of the 10
Gbit/s downstream TWDM channels to 2 dB (c.f., 4 dB
allowed for lower ODN classes).
Similar depletion effects can occur in co-existence
scenarios with legacy PON systems. Namely, for either G-
PON down- stream (at 1490 nm) or RF video (at 1555 nm)
that act as a Raman pumps for the TWDM downstream
channels and are thereby depleted in power.

H. Incremental Upgrade (Pay-As-You-Grow)


In order to enable network operators to defer capital costs
and meet network capacity growth in a cost effective way, the
NG-PON2 systems should support an incremental upgrade ca-
pability; sometimes termed “pay-as-you-grow.” In effect this
means allowing wavelength channels to be added one-by-one
to the system. As the ONUs are colorless, and can therefore
tune to any NG-PON2 channel, they may be redistributed
when new channels are added for load balancing purposes. A
newly added channel may also be used to provide a new
service offering e.g., adding a 10 Gbit/s symmetric line rate
TWDM PON channel for business services to an already
running 2.5 Gbit/s upstream line rate service being used for
residential services.

I. Transceiver Technologies
The most notably new optical components in NG-PON2 are
the tunable receivers and tunable transmitters at the ONUs.
The ONUs are equipped with a tunable transceivers so they
can selectively transmit (receive) upstream (downstream)
6 JOURNAL OF LIGHTWAVE TECHNOLOGY

i.e., the ONUs are colorless and so avoid ONU inventory


issues for network operators. There are several technology
options in scope to implement the required wavelength
tuning functions at the ONU.
For the tunable transmitter, several possible components
based on existing and commercially available technologies
have been identified. The simplest candidate is a thermally
tuned DFB laser with either heating only control [19] or
both heating and cooling. A more complex option is the
multi-section DBR laser either with or without cooling.
Depending on the implementation choice, the tunable
trans- mitter may be wavelength un-calibrated to save on
manufactur- ing costs. In this case, accurate wavelength
alignment will rely on feedback from the OLT.
Wavelength calibrated transmitters are also in scope and
these have several operational advantages providing the
manufacturing costs can remain acceptable.
Whatever technology is adopted for the tunable
transmitter, there will be new challenges for the component
vendors in the PON application. As the upstream
transmission is burst-mode, the operating conditions for the
laser can be widely varying on very short timescales. This is
not typical for DWDM transmitters used in other parts of the
network where the data is continuous mode. Thus, the laser
can experience fast internal transient tem- perature changes
that could result in wavelength shifts outside the acceptable
window without adequate control schemes [22]. For the
tunable receiver, there are components available but the
maturity of actual products is not as great as for the tun-
able transmitter. Nevertheless, there are still several
technology options in production or demonstrated as
research components. To date, the thermally tuned Fabry–
Perot cavity filter technology seems to have received most
attention [26]. Other technologies suggested for the tunable
receiver include angle-tuned Fabry–
Perot cavities, current injection-tuned silicon ring
resonators and liquid crystal filters.
Integrated prototype NG-PON2 transceivers based on
plug- gable form factors have already been demonstrated
[27]. A four channel TWDM OLT transceiver was
integrated in CFP
NESSET: NG-PON2 TECHNOLOGY AND STANDARDS 7

[28] module that included four 10 Gbit/s transmitters, four single wavelength to provide a basic service on a protection OLT
2.5 Gbit/s burst-mode optical receivers a four channel optical in a diverse location. Thus, saving costs for service restoration
multiplexer and de-multiplexer along with optical amplifiers. compared to full duplication of all wavelengths on the protection
For the TWDM ONU, a tunable transmitter and tunable receiver OLT.
were integrated in an SFP+ [29] transceiver module.
IX. CONCLUSION
J. Wavelength Channel Tuning Time Classes
The drivers for a new generation of ITU-T PON system
Three classes for the wavelength channel tuning time of the have been highlighted, the key network operator requirements
ONU transmitter and receiver are specified in NG-PON2 as have been discussed and system technologies in scope for NG-
shown in Table IV. PON2 reviewed. The selected TWDM-PON and PtP WDM
Devices falling into each of these classes open up diverse PON sys- tem adopted for standardization in the ITU-T has
use cases for wavelength tunability e.g., dynamic wavelength also been described. Based on progress to date in the ITU-T,
as- signment [30] and advanced power saving [31] that may and the re- ports already of pre-standards prototype TWDM-
require different tuning speeds. The class boundaries are PON systems [27], [33], we might expect first commercial NG-
broadly de- fined based on known wavelength tunable PON2 systems to be available in the 2015/16 timeframe.
technologies. Class 1 components may include switched laser Deployments may fol- low soon after but there are many
or filter arrays, Class 2 components may be based on uncertainties and challenges to overcome with such a new
electronically tuned lasers (DBR) and Class 3 components technology and these may hinder early adoption. Nevertheless,
could be thermally tuned DFBs. Spe- cific technology this an exciting era for fiber access with 40 Gbit/s capacity,
implementations are not defined in the stan- dards so these are DWDM optics and tunable transceivers appearing at the very
to be considered as just examples. edge of the network.

K. Rogue ONU Behavior and Mitigation ACKNOWLEDGMENT


Rogue ONU behavior in previous PON generations The author would like to thank his research colleagues at
typically related to transmitting in another ONU’s timeslot BT and the years of stimulating discussions and collaboration
[32]. The additional degree of freedom from wavelength within the FSAN/ITU-T.
tunability adds new opportunities for rogue ONU behavior as
an ONU could now emit in the wrong wavelength channel.
Opportunities for rogue behavior will be minimized as far as
possible in defining the standards so that, for example,
interference with already run- ning services when activating and
re-tuning channels is avoided. Even so, the possibility of a
faulty ONU that does not follow correct procedures always
exists. In this case, the accessibility of the wavelength domain
adds a possible mitigation strategy against such a device
through retuning impacted ONUs to an unaffected wavelength
channel.

VIII. WAVELENGTH AGILITY


TWDM PON allows enhanced network functionalities un-
available in previous generations of pure TDM PONs by ex-
ploiting the wavelength domain. New use cases are opened up
by the wavelength agility inherent in the tunability of the TWDM
ONU transceivers.
Internet traffic patterns show strong diurnal patterns which
can be exploited in NG-PON2 to realize network power
savings. For example, OLT side power savings can be realized
during times of low traffic load (e.g., overnight) as all ONUs
on an ODN can retune to a common wavelength and allow
unused OLT ports to be powered down [31]. This is often
termed OLT port sleep.
Wavelength agility can be exploited for network protection,
resilience and service restoration in the event of a fault e.g.,
a fiber break. In the event of a wavelength port failing at an
OLT, ONUs can retune to a predefined or randomly assigned
remaining wavelength port. In the case of protection and full
loss of connectivity to a working OLT, all ONUs can retune to
a
8 JOURNAL OF LIGHTWAVE TECHNOLOGY

You might also like