Tu.4.B.1.
pdf ECOC Technical Digest © 2012 OSA
XG-PON1 versus NG-PON2: Which One Will Win?
Frank Effenberger
Futurewei Technologies, 400 Crossing Blvd, Bridgewater, NJ 08807,
[email protected] Abstract By 2010, XG-PON technology had been standardized, but now NG-PON2 is following
closely behind. The question of how these two systems will compete, coexist, or cooperate will be
considered. NG-PON2 can be seen as an augmented version of XG-PON1, with a large amount of
component reuse. So, it’s not win or lose, but an orderly succession.
Introduction A description of XG-PON will be given. Then,
G-PON and EPON are right now the dominant an outline of the recently selected NG-PON2
deployed optical access systems in the world. architecture will be given. Then the likely
This fact comes from the very low cost of the deployment trajectory will be considered.
TDM-PON technology and architecture. The
XG-PON
optical fiber plant is the most cost effective
The FSAN and ITU-Q2/15 communities began
possible, and the opto-electronic equipment has
the consideration of next generation PON (NG-
been cost-reduced to levels that are really quite
PON) in 2007. Rather than approach the
remarkable. This system provides 50 to 100
problem with a preconceived notion of the
Mb/s service to each ONU, which is a quantum
solution, the group took a complete stock of all
step above any other fielded technology.
the possibilities, and worked step-by-step to find
Despite all these good aspects, these
the best solution. The first thing that became
systems cannot last forever. The growth of
obvious was that there were two large classes of
demand will eventually outstrip the gigabit
new systems: those that required the ODN to be
technologies and require an upgrade of the
changed, and those that did not. These two
network. So, interested groups such as FSAN,
classes were of such a difference that the NG-
IEEE, and ITU SG-15 have begun
PON project was split into two phases: NG-
standardization efforts to contemplate what
PON1 would consider the ODN-compatible
comes next. The first crop of replacement
systems, and NG-PON2 would consider
systems were 10GE-PON and XG-PON, with
systems that were either non-ODN-compatible
efforts starting around 2006 and culminating in
or required technologies that were not mature.
standards about 2009. Close on the heels of
While not explicitly restricted, the general feeling
this work, FSAN has carried on with a study
was that the NG-PON1 system would see the
effort named NG-PON2. The standardization of
field sooner than the NG-PON2 systems.
this new system is currently in process, and is
It was determined that 10Gb/s would be the
expected to be completed by 2013. The time
next reasonable step in bandwidth from the
interval between these two standards is
current 2.5 Gb/s of G-PON, given the very large
somewhat shorter than the usual 5 years. This
commercial development of 10G optics, and due
of course raises the question of how these
to the relative cost insensitivity of the
standards will interact in the marketplace.
downstream link. The upstream is far more cost
This paper will consider this important issue.
sensitive; however, and the requirement for
GPON OLT GPON ONU
OLT Tx ONU Rx
GPON 1490nm 1490nm GPON
NNI UNI
OLT ONU
Data
MAC
WDM
1490nm WDM
MAC Data
I/O OLT Rx ONU Tx I/O
Logic Logic
1310nm 1577nm 1310nm
WDM1
XG-PON OLT XG-PON ONU
OLT Tx 1310nm ONU Rx
XGPON 1577nm 1577nm XGPON
NNI UNI
OLT ONU
Data
MAC
WDM 1270nm WDM
MAC Data
I/O OLT Rx ONU Tx I/O
Logic Logic
1270nm 1270nm
Fig. 1: WDM/WDMA coexistence method for XG-PON1
978-1-55752-950-3/12/$31.00 ©2012 Optical Society of America
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on November 22,2020 at 10:54:26 UTC from IEEE Xplore. Restrictions apply.
Tu.4.B.1.pdf ECOC Technical Digest © 2012 OSA
bandwidth could not be set so easily. The conformance work that is ongoing. The OMCI
accumulated experience of the operators told for both G-PON and XG-PON1 are now defined
them that the upstream was in fact not as by the same single document (ITU G.988),
heavily used as the downstream in residential which virtually assures that the same software
applications, by a significant margin. For these implementation that over 20 companies have
reasons, the upstream was set to be 2.5 Gb/s, implemented for G-PON will work perfectly for
to leverage the significant base of optics at that XG-PON1. The BBF documents provide the
well-known SDH rate. This 10G down, 2.5G up guidance of how to apply the TR-101 model of
system was termed XG-PON1, and is described data internetworking. This gives designers the
in the ITU G.987 series. concrete instructions they need to build OLT and
Coexistence was again a major driving force ONU equipment that will connect seamlessly
for the architecture of XG-PON. In this respect with backbone networks and CPE, respectively.
G-PON had an advantage over GE-PON in that The interoperability and conformance work
a recent addition to the ITU G.984 series is what puts all of the standards and
defined three new features that made G-PON specifications into action, because it involves
equipment forward compatible. The first was testing real implementations. Interoperability
the definition of a blocking filter that would involves testing OLTs and ONUs together, in a
protect the G-PON ONU’s receiver from the new double-blind fashion, where the ONU and OLT
XG-PON1 downstream wavelengths. The do not know which vendor is on the other side.
second was the definition of a WDM1 filter that This mimics the real world case of full plug-and-
combines/ separates the G-PON signals from play interoperability. Conformance involves
the XG-PON signals at the CO end. The third testing and ONU against a test equipment that
was the narrowing of the G-PON upstream issues a known good test protocol sequence.
wavelength band. With these in place, the way Proper response and behavior to this sequence
was clear for a full WDM/WDMA coexistence confirms that the ONU has conformed to the
method, as shown in Fig. 1. This allows an standards. These efforts are already complete
operator to leave the entire G-PON system in for G-PON, and as mentioned above, they port
place when upgrading to XG-PON1. directly to XG-PON. In that sense, XG-PON is
A wide range of loss budgets were defined fully interoperable on day one.
for XG-PON1, ranging from 29 dB to 35 dB. All
NG-PON2
of these budgets can be achieved with a single
Immediately after the XG-PON1 project was
ONU type, because all of the changes are made
spun out of FSAN and into the ITU, work began
at the OLT side. Since it is far easier to stock a
on NG-PON2. At the beginning of this work
range of OLT circuit packs rather than a range
project, the basic requirements were for a
of ONUs, this is an optimal scheme for budget
system with at least 40Gb/s of capacity and
flexibility. In addition to the multiple budgets,
40km of reach at 64-way split, but not
there are also two options for the ONU receiver
necessarily backward compatibility with existing
design to achieve the 31 and 35 dB links. One
ODN or even previous PON systems like the
is based on APD components, and the other
video overlay. Based on that very loose scope
uses PIN components. This provides the
of requirements, many different systems were
market with the flexibility to fully explore the cost
proposed, including 40Gb/s TDM-PON, TWDM-
optimization problem, and arrive at the best
PON, five flavors of WDM-PON, and three types
answer. When the XG-PON1 system begins to
of OFDM-PON. Importantly, some of the WDM-
achieve significant volumes, it is expected to
PONs required wavelength selective ODNs,
revisit these budgets and down-select to those
while others did not.
that have proven their economic viability.
After some initial study, it was obvious that
XG-PON1 finds the solutions to all its other
the broad brush approach had gathered
design issues be reusing those from G-PON.
together systems that were in several ways
The TC layer is highly similar to G-PON, with
incomparable. Some systems could maintain
some streamlining of the protocols, and
backward compatibility, while others couldn’t.
expansion of some of the fields to expand the
Some systems were quite close to practical
address range (maximum split ratio) amongst
implementation, while others were further away
others. Unlike 10GE-PON, XG-PON1 could
from development. Some systems had
draw heavily on the accumulated system
moderate performance, while others were
integration progress of G-PON, and fully
superlative.
inherited the ONU management and control
Given this solution space, the network
interface (OMCI), the BFF-156/167 specification,
operator members of FSAN had to reconsider
and the G-PON interoperability and
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on November 22,2020 at 10:54:26 UTC from IEEE Xplore. Restrictions apply.
Tu.4.B.1.pdf ECOC Technical Digest © 2012 OSA
what were their true objectives for the NG-PON2 support 8 wavelengths at 100GHz spacing. This
project. The ability to operate on the existing wavelength plan leverages the development
fiber ODN was the first firm requirement to work that has gone into XG-PON1 optics and
crystallize, followed by the support of up to 8 coexistence filters. It is compatible with G-PON
independent operators, and the compatibility and the 1555nm RF video overlay channel, but
with the video overlay. The compatibility with blocks standardized XG-PON. However, it is
XG-PON was made optional. Also, the time- difficult to meet the 40km passive reach
frame for practical availability was set to be requirement with such a system, as it uses the
2015. higher loss 1270nm band. Other options are
These new or shifted requirements had a being considered such as the O-plus band at
huge effect. Some systems were practically 1350nm, or the guard-band at 1535nm. These
eliminated due to the ODN compatibility would possibly allow coexistence with XG-PON1
requirement. Some could be eliminated due to or 10GE-PON. If the TWDM system could use
the time-frame requirement. In a short time, all only the longer wavelength bands, its long reach
eyes turned to the TWDM-PON system, and capability would be far easier to achieve.
indeed this was selected in Apr 2012 as the Beyond the physical layer changes, TWDM
architecture for NG-PON2. is targeted to completely reuse without change
Figure 2 shows the architecture of TWDM- all the material that was developed for XG-
PON. In this diagram, four XG-PON1s are PON1. The TC layer, OMCI, and interoperability
stacked by using four pairs of wavelengths; documents are all the same. There is the need
however the standard will specify the system to for some small additions to specify the methods
have eight wavelength channels. For simple to manage and control the multiple wavelengths
network deployment and inventory management in the system. These aspects are being
purposes, the ONUs use colorless tunable developed in an ITU standard termed G.multi. It
transmitters and receivers. The tunable is anticipated that the TWDM system’s
transmitter is tunable to any of the upstream standards will be completed by July 2013.
wavelengths. The receiver is tunable to any of
Deployment
the downstream wavelengths. In order to
The current state of affairs for XG-PON1 and
achieve power budget higher than that of XG-
10GE-PON is that products are available from
PON1, optical amplifiers are employed at the
multiple vendors, and these products are being
OLT side to boost the downstream signals as
trialed at low volume. It is anticipated that
well as to pre-amplify the upstream signals.
starting in 2013 there might be the start of some
ODN remains passive since both the optical
moderate deployments. Meanwhile, the TWDM-
amplifier and WDM Mux/DeMux are placed at
PON system will be quickly productized, as it
the OLT side.
can reuse all the electronics that were
Coexistence with previous generations of
developed for XG-PON1. Quite literally, an XG-
PONs in the legacy ODN relies upon the
PON1 product can be turned into a NG-PON2
TWDM-PON wavelength plan. This decision is
product basically by changing the optics. Given
still pending. One option is to reuse the XG-
this, NG-PON2 will indeed be available in 2014.
PON1 wavelength bands. It defines a finer grid
Also, it is hoped that NG-PON2 will carry a
inside of the previously defined bands. There
relatively small cost add compared to XG-PON1.
happens to be just about enough bandwidth to
What is likely to happen, therefore, is that
we will see a minor amount of XG-PON1
deployment (less than 1M ONUs), followed on
by more extended deployments of NG-PON2.
Some might view this as NG-PON2 taking over
the market from XG-PON1. But, as described
above, the majority of work that went into XG-
PON1 will be reused. So, from a vendor’s
perspective, this shift of design is not very hard.
Also, from the operator’s perspective, NG-
PON2’s availability better matches the real need
to upgrade the existing G-PON networks. For
this reason, most operators can go directly to
NG-PON2 and skip the XG-PON1 step. So,
while NG-PON2 will take over, it is as a son
taking over his father’s business.
Fig. 2: The TWDM-system architecture.
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on November 22,2020 at 10:54:26 UTC from IEEE Xplore. Restrictions apply.
