ViewsLetter(SM) on Provisioning

Site Map



 Flanagan Consulting             Network Analysts and Consultants
                                                   "We Have the Experience"
ViewsLetter on Provisioning      5 Feb 2004        #35
A fortnightly look at provisioning automation--chips to business software.



    --by Vladimir Kaminsky, Contributing Editor

This ViewsLetter reports on recent progress in Passive Optical Networks (PONs). It is the first in a series that will cover the structure of such networks, standardization processes, and evolving technologies.  PON deployment will contribute to automated service provisioning by delivering "enough" bandwidth for a wide variety of services, which may be introduced over time by upgrading the software in the customer premises equipment (CPE).

The basic principle of a PON network is to share its Central Office (CO) equipment (Optical Line Terminal, OLT) and the feeder fiber (in the local loop) among as many end units (Optical Network Terminations, ONTs) as possible--within constraints set by physical properties, bandwidth,and optical power loss.  So far, these constraints put a typical practical limit on the number users who can share a fiber at 32, but some newer standards support N=64.

 PON places more users on each fiber than traditional point-to-point or ring architectures. Fewer fibers to cover a given service area means less equipment at the CO (one optical interface serves N users).  As a result, the PON solution makes high-speed optical connections economical for business or residential units in scenarios that could not be served by other broadband access technologies.

When using an optical power splitter, the optical fiber transmission channel is dedicated to the N customers. The bandwidth to the customers is shared, reducing costs. As an alternative to a single-splitter 1-to-N "tree" topology, PONs can combine multiple smaller splits in tandem to achieve the eventual 1-to-N split.  PONs can also be configured as rings and buses.  The splitter can be moved closer to or farther from the service provider, to optimize costs and ease of future upgrading without affecting the active terminations or the protocols. Glass is transparent, so to speak.

Because of the lack of electronics except at the ends of the network, reliability can be high and there is no requirement for standby power except at the ends.

The first commercial PON activity was initiated in the mid-1990s when a group of major network operators established the Full Service Access Networks (FSAN) consortium.  The group's goal was to define a common standard for PON equipment so that vendors and operators could come together in a competitive market for PON equipment.  The result of this first effort was the 155 Mb/s PON system specified in the ITU-T G.983 series of recommendations (standards, really).

That system became known as B-PON. It uses ATM as its link-level bearer protocol (the APON protocol).  The name B-PON was introduced since "APON" led people to assume that only ATM services could be provided to end users.  Changing the name reflected the fact that these systems can offer many broadband services including Ethernet access, video distribution, and high-speed leased line services.

APON was later enhanced to support 622 Mb/s rates with protection, Dynamic Bandwidth Allocation (DBA), and other features.  On a parallel track, in early 2001 the IEEE established the Ethernet in the First Mile (EFM) group, recognizing the prospects for optical access.  The group works under the auspices of the IEEE 802.3 committee, which also developed the Ethernet standards. As such, EFM is restricted in architecture to comply with existing 802.3 standards.  Currently, the EFM's work is standardizing a 1.25 Gb/s symmetrical system for Ethernet transport only.

In 2001 the FSAN group initiated a new effort for standardizing PON networks operating at bit rates above 1 Gb/s (GPON).  Apart from the need to support higher bit rates, the overall protocol has been opened for reconsideration, with a goal of efficient support for multiple services; scalability; and operation, administration, maintenance, and provisioning (OAM&P) functionality.

From this latest FSAN effort, a new solution has emerged in the optical access market place--Gigabit PON (GPON).  GPON offers high bit rate and transport of multiple services--specifically packet data and TDM--in native formats and with high efficiency (the ITU-T G.984 series).

Most telecommunications fiber rings use synchronous optical network/synchronous digital hierarchy (SONET/SDH) technology.  These rings, which require optical-to-electrical-to-optical (OEO) conversion at each node, and are optimized for long haul or metropolitan applications.  They are not the best choice for the local access network.  In contrast, a PON uses passive fiber optic splitters/couplers to route traffic, instead of the more expensive active elements (electrical and optical) required for SONET/SDH rings.

Being low in cost, PONs offer a practical solution for upgrading the critical last mile infrastructure for broadband.  Unlike active networks, which require installation of all nodes up front (because each node is a regenerator), PONs can be deployed incrementally.  PONs require less initial investment because carriers need to deploy only the fiber to start.  ONUs can be added incrementally to meet demand for service. 

Further cost reductions can be achieved in a PON through the addition of a wavelength division multiplexing (WDM) layer.  WDM is easier in a PON because the customer nodes sit "off" the backbone. PON fibers may be upgraded one at a time.  In a SONET/SDH ring, WDM requires optical multiplexing/demultiplexing at each node. 

Unlike SONET/SDH, PONs can be asymmetrical, which reduces the cost of ONUs.  For example, a PON can broadcast downstream at the OC-12 (622 Mb/s) rate and upstream at OC-3 (155 Mb/s).  Asymmetric architecture allows the use of lower-speed ONUs, which require less expensive transceivers.  Because SONET/SDH networks are symmetrical, all the line cards in an OC-12 fiber ring would require OC-12 interfaces. 

The downstream point-to-multipoint architecture contributes to another key advantage: efficiency for broadcast applications.  In a PON, an analog or digital video signal can be added easily to the time division multiplexing (TDM) or WDM layer to deliver broadcast services.  The closest competing technology is Hybrid-Fiber-Coaxial access networks, which so far did not gain commercial popularity, mostly due to their cost of deployment and maintenance.  Compared to PON, DSL and its variations cannot compete due to limited bandwidth.

A PON is designed for both Business Customers and Consumers at home.  The magic words for the carrier contemplating PON are "Triple Play."  That is, PONs deliver in one "package" the three main types of service: voice, high-speed data, and video.  This is the reason that when the regulatory climate and cost parameters became favorable, the service providers (RBOCs, Local Exchange Carriers, CATVs, and others will jump on the lucrative PON bandwagon. Long-term, PONs will become the platform for delivering services to homes, businesses, multi-tenant buildings, and so on. 

It is expected that PON will dominant by the 2005-2006 time frame.  In the 2004 time frame, however, PON's primary benefit will be to leverage the installed base of copper and coaxial cable in the local access network.  Rather than competing with DSL, cable modems, and local multipoint distribution system (LMDS), PONs will complement these technologies by serving as a feeder between the local exchange (central office) and remote terminals or pedestal cabinets.  From there copper, coaxial, or wireless systems provide connections to subscribers.

A drawing illustrates general PON connectivity over a fiber tree, a passive point-to-multipoint network of fiber and one or more splitters (in cascade).  Active components, optical transceivers, are required only at the root (an Optical Line Termination (OLT) in the local exchange), and at each leaf or branch (Optical Network Units (ONUs)on the user side).  Bus and ring topologies are considered less suitable for user connections, as they run a higher risk of individual users causing disruptions for other users.

One ONU in a neighborhood can feed many NTs [Network Termination (units)] over short copper loops. NTs have user interfaces such as POTS, Ethernet, and serial.  Taking fiber all the way to the customer calls for an ONT, an Optical Network Termination, that combines the two(ONU and ONT).  The passive network constitutes a shared transmission space. Connections can be made here with various multiplexing technologies in the spatial, frequency, and time domains:
--Spatial, the number and type of fibers determine the possibilities.
--Frequency, as in Wave Division Multiplexing of light sources with different wavelengths.
--Time, the familiar TDM concept, but realized by alternating ATM cells (APON) or Ethernet packets (EPON) from different users.

 More on APON and EPON in future issues.
============ + ======= + ======= + ======= 

Visit when you need independent review,
validation, or verification of network architecture, product positioning,
or your marketing message.

"Flanagan Consulting" and "ViewsLetter" are Service Marks of W. A. Flanagan, Inc.
 Updated:  5 Feb 2004

Flanagan ConsultingSM
W. A. Flanagan, Inc.
45472 Holiday Drive, Suite 3, Dulles, VA 20166
Ph:  +1.703.242.8381
Fx:  +1.703.242.8391