Cable Tech

Who's Growing a GPON?

Faster gear for passive optical networking (PON) may be in the works for at least one major respondent to the recently released RBOC "Fiber to the X" RFP (see RBOCs Hungry for Fiber) and A Closer Look at PON Econ).

According to reliable industry sources, who've asked not to be named, Alcatel SA (NYSE: ALA; Paris: CGEP:PA) is at work on gigabit passive optical networking (GPON) technology, a super high-speed network that uses a combination of passive optical components and Sonet framing.

GPON technology would be used to extend cheap fiber links within metro networks (see ILECs' Missing Links ). No one is yet suggesting GPONs for residential fiber.

Specifically, Alcatel is said to be interested in creating a GPON for backhauling traffic to the central office in metro networks -- mainly as a way to address carriers' growing demands to replace old private lines with newer, faster services, such as IP SAN extensions.

Alcatel won't confirm any work on GPON, even though word has it the vendor's touted GPON in analyst forums recently. A spokeswoman says research is always afoot on new technologies, but she insists the direction Alcatel's taking for the foreseeable future is in BPON, or broadband PON.

For the uninitiated, here's a thumbnail tutorial in terms:

  • BPON is the same as an APON (ATM PON), with extra overlay capabilities for broadband services like video. APON is the PON transmission technique based on Asynchronous Transfer Mode (ATM) signaling that was developed by a group of carriers as part of the Full Service Access Network (FSAN) in the nineties. BPON is approved as International Telecommunication Union (ITU) spec G.983x. It supports data rates to 622 Mbit/s out to an endpoint (upstream) and back from the customer to the service provider's remote aggregation point (downstream).

  • GPON uses a different, faster approach (up to 2.5 Gbit/s in current products), encapsulating traffic in a version of the Sonet-compatible Generic Framing Protocol (GFP). GPON is on track to becoming the ITU's G.984x. Final pieces are expected to be put in place at meetings in August and October 2003.

  • EPON stands for Ethernet PON, which, as its name indicates, is a technique that uses Ethernet as the main transmission method for the PON. EPON runs at gigabit rates and has its own standarization process underway at the Institute of Electrical and Electronics Engineers Inc. (IEEE).

GPON fans claim its star is rising. Its blazing speed would make it ideal for "triple play" suites of voice, video, and data services from ILECs and other carriers. In contrast, other forms of PON don't have the bandwidth or multiprotocol support to do that. Its use of Sonet framing enables providers to link native TDM and voice connections into the PON without adding Internet Protocol (IP), making it more efficient than EPON.

Of course, for their part, EPON vendors such as Alloptic Inc. have much to say on the relative merits of their approach (see Chip Startups Bank on EPON and Alloptic Delivers FTTP in Wash.), particularly in keeping down the cost of fiber-to-the-home PONs.

So far, the only vendor that's shipped a GPON product is FlexLight Networks, which says its Optimate PON runs at 2.5 Gbit/s downstream and 1.25 Gbit/s upstream (see Giga-PON Ships Quietly). Flexlight hasn't announced any customers, though the box is said to be in trials at France Telecom SA (NYSE: FTE).

Flexlight's CEO, Gary Lee, says he expects competition to emerge fairly soon, mainly because there's so much interest in replacing private lines in metro networks with faster Ethernet services -- such as ones that extend storage area networks (SANs) from site to site via IP. GPON can also start replacing Sonet rings, Lee says.

"With GPON, you have standardized voice carriage like a Sonet or TDM link, with clocking guaranteed," he says. Carriers can save up to 50 percent on up-front capital costs using GPON links instead of Sonet OC48 (2.5 Gbit/s) rings, Lee claims. And the passive nature of a GPON ensures ongoing operational savings.

That kind of cost model is likely to be music to the ears of fiber-hungry carriers, if it can be proven out. Even EPON proponents, like Alloptic, acknowledge that for businesses where GFP is usually required to handle a variety of data traffic such as Sonet, Frame Relay ATM, voice, and Ethernet, GPON might be the most efficient choice, provided it was running at 2.5 Gbit/s or higher.

But cost is everything in PON, which explains why some vendors haven't run out to add a GPON to their BPONs just yet. A spokesman for Terawave Communications says his company will wait till the standard solidifies and component costs are driven down, probably 2004, to release product.

Ditto others. "At the present time the world's largest carriers, represented by the FSAN membership, have chosen to deploy BPON, by and large," writes Jeff Gwynne, senior VP of marketing and business development at Quantum Bridge Communications Inc. If demand goes up and costs drop, that could make the cost of GPON implementation more attractive, he says.

— Mary Jander, Senior Editor, Light Reading

Page 1 / 2   >   >>
canada1 12/4/2012 | 11:49:31 PM
re: Who's Growing a GPON? AFCI is also working on a PON card to be added to their UMC 1000 DLC, although they are somewhat limited with a max oc12 uplink.
gea 12/4/2012 | 11:49:31 PM
re: Who's Growing a GPON? So...does anyone know how the services are mapped into the GPON payload? Since it seems to be using GFP, I'm assuming that even voice is voice-over-IP-over-GFP. Is trhis correct? Or can standard TDM be supported? (If TDM is available, at what levels? DS1? DS3? STS-3c?)

Sounds like it has an advantage over APON in that services aren't mapped over ATM.
gea 12/4/2012 | 11:49:28 PM
re: Who's Growing a GPON? So what you are describing sounds a lot like any TDM would actually be circuit-emulated, as opposed to true ATM. In other words, for GPON they more or less replaced the ATM layer with GFP, and then of course the line rates are different.

Actually, that would make a lot of sense, in that it would leverage a lot of the existing APON work.

The question I guess remains is how big a TDM pipe they can accomadate, such as an STS-3c. In a case such as this, as good ole TDM a possibility, or will even this be mapped over GFP?
atmguy 12/4/2012 | 11:49:28 PM
re: Who's Growing a GPON?
The way I understand is:

GFP <==> TDM (T1/E1)
GFP <==> IP <==> TCP/UDP <==> Data/VOIP

I don't know how synchronization is done.
doco 12/4/2012 | 11:49:19 PM
re: Who's Growing a GPON? One other thing that is possible is

GFP <=> ATM <=> TDM (AAL1, AAL2, etc)

GFP does not have timing stuff built into it. (Then again - neither does ATM strictly speaking, but AAL1 does have something built in).

If the OLT is a combined system, the timing on the ONT can be derived from the line rate of the PON and there isn't a need for more generic packet based systems like AAL1 provides.
FinBurger 12/4/2012 | 11:49:14 PM
re: Who's Growing a GPON? Gentlemen,
In the interest of clearing up things, let me say two important facts about GPON.

1. GPON does not use GFP. Earlier versions proposed using a variant of GFP, but since the window for influencing the GFP standard was already past, that fell by the wayside. In its place there is a custom built GPON-specific frame transport mechanism.

2. GPON also supports ATM in native mode. The base protocol can carry cells as cells, and frames as frames. The implementer is given both choices.

I hope that helps.

optobozo 12/4/2012 | 11:49:12 PM
re: Who's Growing a GPON? Thank goodness somebody who's actually read (and perhaps help create?) the specifications has chimed in. I'm tired of all the marketing gobbledeegook from EPON folks and outright misunderstanding/info from the cable-plant turbochargers (not that the APON folks are lilly white).

When do we get to see some real peformance numbers between vendors? I'd love to see who's 622Mb (upstream) PON is better than the other guy. Or if the Gigagbit PONs are anything close to it. Or what kind of dynamic range these burst-mode receivers really have.

Perhaps one of Lightreadings senior editors could dig in past the hype and marketing and get their fingernails dirty and find out some day?...
amit1665 12/4/2012 | 11:49:09 PM
re: Who's Growing a GPON? Regarding BW Efficiency/Performance:
One of the main reasons FSAN/ITU selected the current TC scheme for the GPON standard (G.984.3) was due to its superb BW utilization
Before the final decision was taken, extensive bandwidth studies have been done in the committee.
Results showed that GPON (GFP*/PON) yields ~92% of the possible 2.5Gbit/s while EPON has only ~50% of possible 1.25 Gb/s
glasvezel 12/4/2012 | 11:49:09 PM
re: Who's Growing a GPON? Amit1665 - a few points re: your 'extensive' efficiency studies.

- Efficiency has to be considered in both downstream and upstream directions. Note that services over a PON in FTTP applications are overwhelmingly assymetric, i.e., it is really the downstream efficiency that is of more significance here. In a Gigabit PON, even a 50% upstream efficiency would be more than sufficient! The majority of upstream traffic consists of short messages requiring very low bandwidth (Acks etc).

- We have seen more realistic studies that show EPON efficiency of downstream vs. upstream BW of about 70%/65% of 1.25Gbps, or 90%/85% of 1Gbps.

- Of course it remains to be seen whether the new GPON systems can be more cost-effective than EPON in achieving this 20% gain. I sincerely doubt it.
Mr. Mutt 12/4/2012 | 11:49:07 PM
re: Who's Growing a GPON? I'm pretty far from an expert on this, but a thought crossed my mind...

Within the service provider market, there are only a few that are actually deploying new service. Would it also stand to reason these are the same ones that are upgrading the uplink side as well as their downstream lines? If that's the case, isn't the overwhelming majority of new provissioning being done with high speed ethernet? It would seem to me that whatever advantages GPON had over EPON in terms of efficiency (I've seen the number 20% in previous posts), wouldn't the upstream ethernet connectivity be more important to these providers who are now more widely deploying GigE in their networks?

I guess the real question is; what will the service provider market choose? THAT will be the winning technology, not necessarily the best or most efficient.
Page 1 / 2   >   >>
Sign In