Network Analysts and Consultants
"We Have the
ViewsLetter on Provisioning 21 March 2004
A fortnightly look at provisioning automation--chips to business
PASSIVE OPTICAL NETWORKS EVOLVE
--by Vladimir Kaminsky, Contributing Editor
TOWARD WAVE DIVISION MULTIPLEXING
part of our study of Passive Optical networks concentrates on future
technologies that are still mostly in the R&D stage. As the
title of this paper states, we are looking for PONs beyond the FSAN and
E-PON group of technologies, discussed in previous issues, which either
have been standardized or are close to that status.
Today, the most promising
PON technology still in R&D appears to be Wavelength Division
Multiplexing (WDM) PONs (W-PON). W-PON research focuses on
wavelength splitters (which replace the power splitters found in
standardized techniques and E-PON). Because of the large number
of possible wavelengths, they do not need to be shared. Adding a
new customer may mean adding a laser with a different wavelength in the
central office, but that need have no impact on current subscribers.
LIMITATIONS OF "TRADITIONAL" PON
W-PON can partially overcome issues usually associated with "traditional" PON such as:
--Limitations on the splitting ratio (number of subscribers per OLT port) due to the available power budget.
--Broadcast transmission over PON means both the OLT and ONT must work at the aggregate bit rate.
--Bandwidth sharing limits each subscriber to a fraction of the aggregate bandwidth, and may require contention for bandwidth.
--Security issues arise from delivering all packets to all sites on the fiber tree.
--Network integrity risk: a corrupted signal from one ONT can affect all network users.
These issues, together with desire to work with higher bit rates, stimulated interest in W-PON development.
W-PON can mitigate many of
these issues of "traditional" PONs by allocating a different wavelength
to each subscriber. To add a user, the service provider deploys a
laser with a new wavelength and a matching wavelength selection filter
at the distribution point. With a "private" wavelength, each
subscriber can communicate to the OLT with a different data rate signal
and a different protocol.
A subscriber can be added
simply by deploying common modules at both the central office and the
subscriber's site. This feature simplifies inventory management
A Figure at http://www.viewsletter.com/ illustrates this concept.
In the CO, the "head end"
equipment combines the features of a Wavelength Division Multiplexer
and multiple OLTs. Downstream from there, a
wavelength-multiplexed signal contains wavelengths for all
subscribers. The outside plant contains wavelength splitters
(there may be more than one in a path). Note that a WL splitter
can have less insertion loss than a power splitter used in
"traditional" PONs. After a WL Splitter, individual signals
(selected by frequency) are directed to the particular subscriber
ONU/ONT. It is also possible to combine WL and power splitters on
a path. That is, one wavelength can support a group of users (the
same as a PON) or multiple wavelengths may reach a subscriber (in which
case the receiver is equipped with a wavelength filter that admits only
that subscriber's signal).
In the upstream direction,
each subscriber can transmit on a wavelength pre-defined for his
location during W-PON design as shared or not. Even when shared,
the smaller number of users per WL means it's easier to coordinate
subscribers transmissions than on "traditional" PONs. In the
upstream direction, a WL Splitter acts as a passive wavelength
combiner. CO equipment demultiplexes wavelengths, among other OLT
Note that there are different flavors of W-PONs, including, but not limited to:
--Hybrids of "traditional PON" and W-PON using both power and WL splitters;
--Upgrade of existing fiber infrastructure to W-PON by replacing power splitters with WL splitters.
With W-PON development
concentrated on R&D, it's too early for standardization activity in
this arena. In spite of this, companies are delivering
proprietary equipment (for example, Korean company Novera
Optics). Another approach is coming from the Essex Corp., which
announced manufacturing of passive WDM splitters with very narrow
spacing between channels. For example, its WDM can function with
a data rate of 1.28 Gbit/s over channels spaced only 3.125 GHz apart.
This opens a lot of new opportunities to increase the capacity of W-PON
The issue with W-PON is not
so much the WL multiplexing in the optical links themselves, but the
customer premises equipment. Deploying W-PON would be a large
operational problem if for each subscriber the service provider had to
install a different WDM ONT for each laser frequency supported.
That approach would require a supply of ONTs for all possible
frequencies, with spares, creating an excessive inventory and high OPEX
and CAPEX spending.
Moreover, most W-PON
solutions appear to be based on passive signal splitters, with the
lasers in the end-user's equipment functioning at different
wavelengths. Such a model creates the risk of interference in the
passive signal splitter when subscribers transmit on adjacent
wavelengths unless end points have sophisticated filters.
WHAT TO DO
Three main directions for solutions have been formulated (so far):
1. Fit equipment at one or both ends with tunable lasers.
2. Introduce an optical connecting cable for the ONT that determines the laser frequency.
3. Deploy an ONT containing a modulating mirror rather than a laser.
1. Tunable lasers are more
expensive than fixed lasers and have less power, making them less
attractive when used with the preferred passive power splitters.
These devices reduce signal strength by at least 15 dB under the best
conditions (a ratio of 1:32; -3 dB is half power). In practice,
loss increases to 17 dB or more. Currently, a downward trend for prices
of tunable lasers is becoming obvious. Maturing technology may
create an opportunity for them in W-PON.
2. A specially treated
fiber cable can determine the laser frequency of the equipment in the
home or business. In this case a cable determines the wavelength,
while the laser in the ONT can be the same for all wavelengths,
simplifying provisioning and hardware sparing. If the ONT has
problems, a serviceman can simply replace the ONT with standard
equipment; the existing connecting cable will ensure proper
wavelength of the light.
This can work because a
laser needs two reflecting surfaces to function. One mirror is
only partially reflective and allows part of the light to pass into an
optical fiber (to the output). Mirrors typically are placed at
the ends of a laser, creating a "cavity." This technique replaces
the output mirror with a special cable. The laser has only one
mirror, on the back side.
UV-treatment creates a
pattern of 'light' and 'dark' rings (of different refractive indices)
at regular intervals along the fiber used in the connecting
cable. This pattern works as an optical grating with the
properties of a wavelength-selective mirror that transmits (and
reflects) only one precise wavelength. The cable, as the second
mirror, determines the wavelength of the laser. This technology
is still in development.
3. Lucent tested a
solution that requires no laser in the ONU. A continuous laser
signal sent from the OLT is reflected back by the ONU. The ONU
modulates the returned signal on or off (a "1" and a "0"). The
reflected pulses are detected in the OLT's receiver. Thus the OLT
controls the wavelengths in both directions; the ONUs and ONTs
are uniform. Adjusting the wavelength simplifies to changing a
Altogether, we see a
bright future for W-PON. This group of technologies should have a
place in the general PON infrastructure, particularly if combining
E-PON with CWDM. Analysts are also talking about IP-based
PONs--not a surprise, but they are far from ready.
"Flanagan Consulting" and "ViewsLetter"
are Service Marks of W. A. Flanagan, Inc.
Updated: 21 Mar 2004