fibresystems.org blog

Big purple truck

2009-06-30T13:51:10Z

Earlier this year I was invited to visit London to find out how easy it is to set up and tear down optical connections, courtesy of the Infinera Express, which was setting off on a European roadshow. Living in a travel "not-spot" (it takes me the best part of three hours and three different forms of transport to get to London ), I had to decline.

But last Monday, while walking back from the sea front in Nice, I found the Infinera truck had come to me. This was great news for a couple of reasons: one, it meant I was in the right place for IIR's WDM & Next-Generation Optical Networking conference; and two, I would get to see this stuff up close after all.

A big purple truck...

]]>
...welcomes me on board.

The ability to add new capacity quickly is often praised by Infinera's customers, and that's one of the things the demo is designed to show. The company claims that setting up a new connection takes just 20 minutes with its kit, compared to 3—4 days using competitor's equipment.

On board the truck, I met with Geoff Bennett, Infinera's director of strategic marketing, who showed me a couple of tricks with the kit. The simplified network on the truck comprised three network nodes (cutely called London, Paris and Frankfurt), and an optical amplifier.

Yoo-hoo, it's behind you...

There was a real-time view of all the equipment and all the capacity on the system. We set up a new connection, and pulled out a few fibres out just to check it was all for real. When a cable was unplugged, sure enough the system restored service by sending the traffic the other way around the network.

Control plane rules, okay!

Then we decided to go for broke and "delete" the network, to show how automated topology discovery could bring everything back up again. And so it did. (At one point the GUI decided to locate all three European cities somewhere just off the world map in what must be a parallel universe, but that didn't matter, because the network was still topologically correct — phew!)

At this point I discovered that I might have overstayed my welcome...

How to deal with the press

Thanks to Jeff Ferry for the images.

Day 3: The headline take

2009-06-26T16:16:40Z

By Mark Lum, independent telecoms consultant

Day 3 brings the end of WDM and a time for delegates to ask each other "What did I learn?", "What was new in 2009?", "What's the big trend?", "Who's up?", and "Who's down?". Some questions can easily be answered by reading newspaper headlines — whilst others are more complex, or working on difficult-to-discern evolutionary timescales. Friends, colleagues and other delegates I've spoken to have a positive view on the conference, in these difficult economic times.

We have all enjoyed hearing the latest 100G developments from many vendors, but your watchful diarist must also report that the operators speaking here at WDM seem to be on different agendas. Deutsche Telekom alone — with some of the highest density routes in Europe — held the 100G banner with a detailed set of requirements and implementation scenarios

National operators such as KPN, Swisscom and Telekom Austria (and also DT) presented their challenges, thoughts and strategies for re-architecting their infrastructure and deploying new OTN WDM elements; handling "legacy" SDH-based services remains a long-term need.

Bharti Airtel, from further afield, reminded us that things are not always as smooth: dust, heat, humidity and an unreliable electric supply provide a much higher stress level for networks. If you're experiencing a high rate of fibre cuts, squeezing all your traffic onto multi-lambda 100G links may not be the wisest strategy! A mesh-based, multi-path protection scheme using lower capacity channels would probably provide a more robust network.

Office consolidation is moving up the agenda, as national operators here express intentions or plans to close local exchange facilities and make use of WDM backhaul — perhaps even WDM-PON — to concentrate active electronics at fewer network locations. Lower opex, improved reliability, reduced power, smaller real-estate: the strategic benefits of such an upheaval can be easily identified, with a serious amount of cost-saving in prospect.

Direct optical routing (IP bypass or router bypass, if you like) has been simmering on the industry agenda for a few years. To a vendor, "IP-Optical integration" means a converged packet optical network element, or DWDM interfaces on routers. However, to operators such as Orange and Telefonica, it means an optimised network architecture based on traffic flows. With operators worried about costs, this issue has now come to the boil: we'll see much more of this.

Alas, there was far more at WDM than I can comfortably even mention. In the FibreSystems Europe Future of Photonic Networks seminar, a focus on power consumption rubbed shoulders with developments in photonic integrated circuits and 100G components.

Today also brought the OIF's global interoperability demonstration to Europe. Linking seven nodes from China Telecom, Deutsche Telekom, KDDI, NTT, Orange Labs, Telecom Italia and Verizon, the OIF showed that Ethernet VPL services can be provisioned, on-demand, across global networks, and supported with end-end service restoration using E-NNI.

Many questions remain as the delegates head homewards, perhaps the most common one: "Will WDM be held in Nice again next year, or return to Cannes?" I tend myself towards Cannes, but like many, have also found Nice to be a refreshing change. Regardless of location, the speakers and contributors at all levels have ensured a remarkable content quality and session engagement: sometimes until 18:30 in the evening. Hope to see you at WDM 2010!

Visit www.optical-transmission.com/wdm for the latest information on WDM 2010.

WDM Nice: Rough Travel Guide to 100G

2009-06-26T11:06:18Z

By Mark Lum, independent telecoms consultant

So here we are at the must-attend 100G showcase, with no fewer than seven vendors jostling for position on the stage in front of a baying crowd*. The only qualification? To have a published field trial with an operator. However, one does not... I leave as an exercise for the alert reader to discover which!

I'll try to encapsulate each trial/vendor in a few highlights, which will doubtless cause trouble for omitting some important aspect — there's no substitute for being here! As the 100G alphabet soup is cooked, I suspect the session will run late.

Alcatel-Lucent: First 100G trial with Verizon 16/11/07. 500 km Tampa-Miami. DQPSK-NRZ/RZ modulation also with 10G channels. Has published 13 R&D papers since 2005. Conclusion: 100G PDM-QPSK (25 Gbaud) with coherent detection and digital post-processing, also compatible with 10G NRZ and 40G.

Ciena: 100G trial with Caltech over 1500 km. Agrees with Alcatel-Lucent on system technology: ADC plus DSP ASICs also photonic integration. 30-50% p.a. traffic growth means capacity exhaust, even 40G will be insufficient in 2012. There will be much secret sauce from different vendors.

Cisco: 1st 100 GbE 802.3ba-based router demo with Comcast (6/08). Key requirement is compatibility with 10G links @ 50 GHz. Conclusion: PM-QPSK modulation (25 Gbaud). Sees 100G in routers for IPoWDM architectures, but also transponders. Vision shows 200G, 400G, 1T, CO-OFDM, software-defined optics.

Ericsson: 112Gbit/s trial with Deutsche Telekom over 633 km including four ROADMs. PM-RZ-DQPSK modulation with fast LiNbO3 polarization tracking plus 10G and 40G neighbouring channels. Conclusion: PM RZ-DQPSK @ 50GHz needing no ADCs or DSP since polarization tracker is used for PMD tolerance. Investigating OFDM, SCM, nQAM.

Huawei: Several options available. OPFDM-DQPSK or OPDM-DQPSK — no ADC/DSP is needed so cost-effective. ePDM-DQPSK provides best performance, >1,000km but is more complex. Market timing is critical, as is ASIC complexity and availability. Sees a five-year delay between 40G and 100G deployments.

Nokia Siemens: 107 Gbit/s trial with AT&T in 2006, 2x 80km direct optical. First CP-QPSK in 2007. 111 Gbit/s trial with Verizon in 2008, 1040 km and mixed 100G, 10G and 40G channels. Believes that coherent CP-DQPSK is the way forward, hero experiments are now finished, and need to make real commercial products.

Nortel: 100G trial with Comcast (3/08), 335km link including ROADM. Many network demos 2008-09, including Verizon 73 km link that could not carry standard 10G. Conclusion: coherent FDM-DP-QPSK (14Gbaud) @ 50GHz. Vision that coherent plus WDM gives next big jump in fibre capacity.

I have the feeling of an industry — at least most of the vendor side — in a hurry, pushed by an invidious mix of public demand from a few high-profile carriers, the prospect of competitive IEEE interfaces, intense vendor competition plus a driving technological rivalry to scale the 100G mountain.

I remain most concerned about gaining sufficient economy of scale, but despite appearances, the industry is largely converged on DP-QPSK with coherent detection. And whilst vendors may be reluctant to highlight it, the OIF's MSA functional standardisation is a critical first step towards economy.

The showcase session has indeed run late and we are now behind schedule: let's see how 100G deployments in production networks fare in the coming years.

[* I mean: an enthusiastic and receptive audience]

Day 2: WDM admires the view to 100G

2009-06-24T17:00:26Z

By Mark Lum, independent telecoms consultant

Another fresh and sunny morning and a short ride on Nice's next-gen ligne d'azur tram brings me to the Acropolis from my sea-front hotel next to the Marche des Fleurs. It's difficult to juggle my diary duty with chairing duty today, but I can first bring you some notes from yesterday afternoon.

From IDC, we hear that 40G network systems have developed though no fewer than four generations of subsystem technology since 1999 (those with good memories may also remember Nortel's 80G demonstration at ITU's Telecom 1999, a 2x40G system). A decade further on, and we witness at least a handful of live 40G traffic links, mostly in the US, whilst 10G deployments are still increasing.

And as for 100G deployments? There is much talk that 2012 will be the year — but realising that's just over 2 years away, perhaps some collective optimism is at work. We learn that 40G transponder prices are, even now, no better than 4x10G (itself a decreasing target), and I feel it will surely be many years before real economy of scale can be provided at 100G. However, one vendor boldly shows its forecast of 100G deployment exceeding 40G as early as 2012. Tune in to WDM 2012 to find out how things have really progressed!

I quote from www.provence-hideaway.com:

"The Grande Corniche [from Nice to Menton via Monaco] was built under Napoleon's reign and pretty much follows the Roman Aurelian Way. One would wish there were more belvederes with parking facilities, because you want to stop at every turn and admire the views, which extend both seaward and inland. Alas, there are few places where you can stop. Be careful, it is the favourite playground of Formula 1 wannabes, bikers with no fear and sightseeing tourists creeping along in their cars — a potent mixture"!

The similarity with 100G network evolution is surely uncanny (excepting that few fibre networks will have been built in Napoleon's time). WDM is playing the role of sightseeing tourist as the majority of delegates attends the packed standing-room only "100G Showcase", whilst I wonder if anyone is left in the eminently worthy parallel stream. I'll have a rough guide in a separate 100G diary special.

Moving forward in time to this morning, and I encourage the conference to search for economy of scale in optical networks. With internet traffic doubling approx. every 18 months according to several incumbents, we as an industry need to figure out the best solution. I suggest that upgrading to 40G at a cost of 4x10G is hardly a convincing business case — especially if associated traffic revenue is either declining, or non-existent in the first place. And with 10G volume cost reduction just hitting its stride, perhaps 10G will be the real sweet spot for economic high-volume networking for many years to come?

Swisscom presented its strategic plan to phase out its very first SDH network, installed in the early 1990's, and replace it with a next-gen SDH/OTH integrated network. After 20+ operational years, I expect many operators will similarly be looking to replace their original SDH infrastructure in the next five years, so there should be plenty of opportunity for new architectures and network build to handle future traffic projections.

A somewhat abbreviated diary today, and I also need to catch up with highlights from the FibreSystems Europe seminar. Tune in tomorrow!

Day 1: WDM Cannes re-emerges in Nice

2009-06-23T16:53:00Z

By Mark Lum, independent telecoms consultant

After a decade in Cannes, IIR's annual WDM conference has moved just along the Cote d'Azur to Nice and its Baie des Anges. We are meeting at the Acropolis — the huge and rather garish alien building beamed down next to Nice Vieux Ville. We are all finding our feet in a new venue, but we have a sun terrace opening from the airy exhibition area and it's a refreshing change from the subterranean depths of the Hilton in Cannes. Welcome to my 2009 conference diary!

We open with a keynote from Orange UK. We have all heard the challenges faced by mobile operators as their rapidly-growing HSPA/3G internet traffic stresses all parts of the network, from backhaul to the converged IP/MPLS core. Orange has adopted an optical offload strategy to deal with the latter, where high-speed internet traffic is logically routed at the edge onto express optical paths, directly to the internet peering point. With this optimisation, internet traffic never traverses the IP/MPLS core and avoids the whole capacity/scale/cost/revenue dilemma.

Optical bypass has been much talked about, often on a per-node basis, but it seems to me that this type of network-wide optimisation — with a rapid payback — holds much future promise. If we are indeed to have on-demand video delivered to consumers on mobile devices or 3G-PCs, mobile (and fixed) operators need to get with the program, if they expect to be profitable.

On the conference agenda, I'm looking forward to the "100G showcase" this afternoon, where 6 vendors who have supported operator field trials will present their systems. I'll have more on this session in a special 100G diary. Coming up tomorrow is the "Future of Photonics Networks" session, led by FibreSystems Europe and including a focus on photonic integration. Also we have the OIF's interoperability demonstration, more on this too. And of course this evening, WDM's signature cocktail reception on the sun terrace to catch up with both new and long-standing industry colleagues.

Aside from the headline 100G linespeeds, all-optical technology continues its advancement: fully tunable sources, optical path tracing, fast adaptable transponders, dynamic PMD compensation, advanced modulation, photonic restoration, digital post-processing — and more. A host of technologies and techniques attempt to wrestle a badly-behaved analogue optical channel into submission, so operators can run high-availability digital services over it. But as Belgacom ICS asks: "When will this be commercially available?" which I interpret as asking "Great science projects, but when will it be affordable?" To me, all-optical still seems a very specialised type of network, when perhaps we should be figuring out how to get some economy of scale into optical network operations.

IP—optical integration was also examined by Telefonica International Wholesale. Its objective is to leverage next-gen optical functions to reduce total capex (optical+IP) of IP networks — IP hardware is taking an increasing proportion of IP product costs and this needs to be rebalanced for scalability. Music to an optical conference's collective ears! For an operator, such an approach requires a high co-ordination between design, operations and maintenance teams. Even simple strategies such as protection at the optical layer, rather than the IP layer, result in substantial capex savings. In common with Orange, a careful analysis of traffic flow is the starting point, allowing a rational removal of intermediate nodes.

So a very interesting first morning. I'm off to look around the vendor exhibition area; WDM has proper exhibition-style booths this year, not just tabletops. It's all change here, and perhaps also a time to consider changes in the next decade of optical networking.

Tune in to WDM Nice tomorrow!Visit the event here www.optical-transmission.com/wdm

Waiting for fibre

2009-06-12T11:05:19Z

By Mark Lum, independent telecoms consultant

Excuse me for being parochial, but being British and chairing at this week's FTTx Summit leaves me feeling somewhat depressed, as I wait for my flight home from Munich. Doubly depressed, if I tell you that I first saw FTTH with BT's early deployment in Milton Keynes back in 1984, I think it was (as a young graduate engineer, I should hasten to add!)

I can report that in the intervening 25 years, the UK has hardly advanced, whilst the rest of Europe — and large parts of the world, too — has marched on, now reaching mainstream fibre deployment in many countries. A brilliant constellation of technologies, investments and businesses — open and closed access, municipal and incumbent players, FTTH and FTTB, P2P and PON architectures, private and state investment, individuals and corporations — is testament to the vision and human ingenuity to pioneer, find the successful and make this thing happen commercially over the past decade.

And as for the UK? Sadly, we are a "no-show" as far as the FTTH league tables are concerned. Many would say that we are the donkey cart of the information superhighway... Given up before we even try to deploy fibre to the home... Several fibres short of a muesli breakfast... Flatlined in the FTTH casualty department: it's just embarrassing!

I fully appreciate BT's market position, but it's latest announced plan is largely a "mend and make do" FTTC strategy. Luckily, there are a few green shoots showing from other initiatives.

For a country where Kao and Hockham first conceived using optical fibre as a communication medium (at ITT's STL Harlow labs in 1966) this is a disappointing situation, quite frankly. One might hope that we would be in a more enlightened position. Perhaps the UK government's imminent Digital Britain report will realise where the rest of the world is headed, help jump-start the country and shake us out of fibre torpor. Let's see.

Now I have that off my chest, I'll gather my wider thoughts and look to report on a few highlights from what was an extremely cheerful, forward-looking and optimistic FTTx Summit. Ironically, next year's event is planned for London! For many countries and people, fibre is coming home — those involved couldn't be happier.

Shock horror: IP networks 100x more faulty than SDH?

2009-04-28T15:01:17Z

By Mark Lum, independent telecoms consultant

I'm here at Packet Transport Networks in Vienna, and a good conference turnout reflects the growing importance of packet transport. I'm sure there'll be much of interest over the next few days.

But in my first blog post for fibresystems.org, I'm startled by a statistic from Huawei comparing a national mobile operator's IP network layer with its SDH layer. We are told that:

• 95% of network element failures are down to the IP equipment, with just 5% from SDH;
• 66% of network failures derive from the IP layer, with 33% from SDH.

It's hardly a happy comparison, but we are further told that the SDH layer had three times as much equipment supporting 30 times more connections. It seems that the fault incidence from IP is out of all proportion to its deployment volume, not to mention the increased opex and staff costs.

Several possible conclusions occur to me (I'm happy to consider other suggestions!)...

• we expect this type of anti-IP propaganda from the dumb-pipe transport people!
• this is just a one-off extreme outlying example;
• the operator must be using the wrong brand of router;
• the IP technicians are not properly certified and need more training;
• the operator should transform its network with a new architecture;
• IP/MPLS is still a new technology and can only improve in the future;
• there's nothing we can do: IP is more complex by definition — get with the program!

No-one is suggesting that SDH and IP are interchangeable, neither as technology nor in function. But as bandwidth consumption continues to grow — both packet and (still) TDM-based — we need to figure out the best way to provide lower cost packetised networks by an order of magnitude.

There's lots of discussion here about MPLS-TP, even as IETF and ITU-T try to decide exactly what that is with heated opinions. Based on this datapoint, I think they have their work cut out to define a packet transport technology that will improve carriers' economics, rather than make them worse.

OFC/NFOEC 2009 report

2009-04-07T19:32:09Z

San Diego Convention Centre

The official statistics are in: OFC/NFOEC in San Diego last week attracted 550 companies and around 9,500 attendees. Numbers are down a fair bit on last year when there were around 12,000 attendees, and some big name vendors chose to book meeting rooms instead of booths.

But while things were certainly quiet, they weren't uneventful. The key decision makers still attended, and business seemed to be getting done. For a show where everyone was talking about how there wasn't a lot going on, I certainly found plenty to interest me.

Here's a round-up of OFC/NFOEC news published on fibresystems.org:

Pre-show coverage:
Agilent gets up to speed on 100G modulation
ADVA tackles metro-optimized 100G
Ericsson trials 100G with Deutsche Telekom

From the event:
JDSU thinks big on tiny tunable
Optical components: can the industry be fixed?
Electronics takes centre stage
From the show floor: CIP
Thermal management: a hot topic
From the show floor: Fabrinet

This article will be updated as new content is added.

My first tweet

2009-04-01T17:38:18Z

I'm a bit of a late bloomer where social media is concerned, but yesterday I finally got around to joining Twitter.

Well, it seems that I'm not a moment too soon. The Guardian is really showing us the potential of this platform by switching to Twitter and in future will tell all its news stories in 140 characters or less. Brilliant! I took that idea seriously for all of about 3 seconds before spraying my keyboard with coffee. All the same, the Guardian may have a point: it could really cut down my workload if FibreSystems were to adopt the same policy.

If you'd like to follow me on Twitter, you'll find me at http://www.twitter.com/paulinerigby. I'm sure that when I have something to say, it'll be worth it.

Your correspondent in Pheonix

2009-03-31T18:37:12Z

Reporting from Phoenix

Plenty of time today to mull on this week's OFC/NFOEC event, as I missed the connection to my UK flight, and am stuck in Phoenix, Arizona, for 24 hours. Trying to land in 62 mph cross winds was one of the more scarey experiences of my life; the pilot aborted the landing at the last minute and we were diverted to Tucson.

We (publisher Susan Curtis and I) arrived in the small hours at a slightly seedy hotel, the sort where you get disposable plates and cutlery for breakfast. Too depressing to stay in the hotel room and work, so we've gone hiking in South Mountain Park.

Cactii

This is the third time in three flights that I've been delayed overnight. The trip before that was by Eurostar, and my train was cancelled in both directions, leading to more delays. My colleagues swear they will make separate travel arrangements from me in future. So if you see me at the airport, you have been warned!

From the show floor: Fabrinet

2009-03-30T16:28:22Z

Nat Mani and the Fabrinet elephant

As the economic situation gets tougher for optics vendors, one company is poised to take advantage of the situation. Fabrinet is an optical foundry that has seen its business expand even as the optical components industry has shrunk from $15 billion a year in revenues down to just $1 billion.

"Our revenue is in lockstep with our customers," said Nat Mani, Fabrinet's executive vice-president of sales and marketing. "But of course [as customers feel the squeeze] outsourcing starts to look better and better to those customers."

"Quoting activity is as high as it has ever been," he added.

Outsourcing has been a key trend among optics suppliers since the telecoms bubble burst in 2001, as they look to save costs. Not only can vendors save on the high cost of setting up the fabrication facility in the first place, but they can offload the risk of periods of low demand onto the manufacturing partner. "The decision process used to take longer; now people come and see us and they get it," said Mani.

Typically these operations are located in countries with low-cost labour rates — in Fabrinet's case it's Thailand, which appears to be home to yellow elephants. That said, the skill of the workforce, not low-cost wages, was the biggest factor in the decision to set up in Thailand, according to Mani. Fabrinet was founded in 1999 by executives from the disk drive industry who saw similarities between disk drive manufacturing and optical manufacturing.

Fabrinet began operations by acquiring a facility from Seagate Technology in Bangkok, and has benefitted from the previous oversupply in the optics industry, picking up additional fabs from the likes of Finisar and JDSU at, presumably, knock down prices.

According to Mani, outsource penetration still has some way to go. Consolidation among Fabrinet's customer base, which has reduced the number of vendors in the marketplace but not their product portfolios, isn't really an issue. "The opportunity to consolidate fabs probably doesn't exist," he commented.

Fabrinet says its success lies in the "factory within a factory" model. Each customer gets a dedicated team to support them, so that they can be comfortable that they will get the attention they need — and that no company secrets are leaked to the factory "next door".

Lean manufacturing techniques — eliminating non-value-add steps in the assembly process, and the overhead associated with them — have been key to the company's strategy over the past six to seven years. Not only can this save time and money; too many steps in the production process leave it prone to lower quality because there are more "touch points". Fabrinet is now in the process of applying those same principles to create a "green" manufacturing initiative. This is a topic that we plan to cover in more detail in FibreSystems Europe in future, so stay tuned.

Thermal management: a hot topic

2009-03-26T19:38:04Z

Components vendors are always trying to shrink the size of optical modules, this week's announcement of the tunable XFP being a good example of this trend. In doing so, the thermal management presents just as much of a challenge as the optics. In fact, the required power dissipation often determines the size of a module.

So I was very interested to meet with Nextreme, a company that's got an extremely efficient method of cooling optical chips. The material that it has developed can pump 10 to 20 times more heat than conventional thermoelectric coolers (TECs), and comes in a much smaller package. "In telecoms most of the money is in the packaging, and the way to get that out is to shrink the packaging. Our focus is to help people do that," said Paul Magill, the VP of marketing and business development for Nextreme.

Nextreme claims that its cooler is the only one that can fit inside a TO-56 can. A TO-56 can is 5.6mm wide, and represents the next step in shrinking optical transmitters; it's a lot smaller than the TOSA package that vendors normally use. Nextreme has also integrated its material into flip-chip solder bumps, and imagines that one day the material might become part of the packaging rather than an additional device that must be bought.

On one level, Nextreme's technology doesn't sound all that new. It's based on thin films of bismuth telluride, a material that's been used for years to make Peltier coolers. On the other hand, the capabilities of this material sound like are a significant enhancement on what was available before, which could make a difference to optics vendors.

Here's a video that Nextreme showed me:

From the show floor: CIP

2009-03-25T19:35:14Z

If there's one company I look forward to talking to, it's the UK's Centre for Integrated Photonics (CIP), because they always seem to have something interesting going on. At ECOC last September, CIP wowed attendees with its demonstration of a 32-channel multi-wavelength laser. The component contained two 16-channel laser arrays, with each channel being directly modulated.

This device is aimed at WDM-PON applications, where a single transmitter could replace 32 separate devices and a modulator, allowing all the optics at the PON headend to be collapsed down onto a single linecard. One of the benefits of GPON and EPON technologies is that they simplify fibre management and economises on equipment space in the central office; the multi-wavelength laser could bring both of those benefits to WDM-PON equipment.

Now CIP has taken that idea further by adding a modulator array to its multi-wavelength laser to obtain better transmission performance. The device that CIP showed at OFC was a 10-channel laser array, containing two arrays of five lasers mounted on the same substrate. An array size of five was picked to optimize yield.

]]> The company expects that multi-channel devices could replace CWDM transceivers in metro networks. "Most people think that for low cost you need to do CWDM," explained chief technology officer David Smith. "The approach we've taken is intrinsically narrow linewidth, and locks onto the [DWDM] ITU grid. This could compete with CWDM, but you end up with potentially much higher capacity."

Other advantages include lower cost, smaller size, and better power efficiency. In other words, this technology boosts the three key performance metrics that are important to vendors.

Reducing power consumption has also become increasingly important over the last year, according to Smith. It's not just environmental awareness; carriers like Verizon are demanding more energy efficient equipment from their suppliers. The problem is that for every Watt of power that's produced by the device, it needs five to six times as much power to remove it, Smith explains.

"One of the critical things we're going to try to do with our technology is get it coolerless," he said. The company is developing uncooled lasers as part of a project funded by the UK Technology Strategy Board, and hopes to partner with another vendor to gain access to athermal (uncooled) arrayed waveguide gratings (AWGs), which combine and separate wavelengths.

CIP is hoping that it can apply its multi-wavelength laser technology to multiple markets — metro and access — to spread the development cost, and access higher volumes to bring manufacturing costs down. The company's challenge now is to make that transition from a grant-funded outfit with really cool R&D, to a commercial vendor with revenues.

Electronics takes centre stage

2009-03-25T19:33:01Z

With 100G and coherent receivers, a lot of the complexity in the components moves from the optics to the electronics.

Ed Cornejo of Opnext tells me that the company is working on a coherent solution for 100 Gbit/s transmission, and will have a demonstration of this capability by the summer. The vendor is doing in-house development for the electronics piece of the puzzle, where much of the complexity lies; in Cornejo's view other optics companies will face a significant barrier if they don't have this expertise. Opnext's electronics know-how came from Stratalight of course, which started out as a subsystems developer, and had been working on digital signal processing prior to the merger with Opnext. "We should have the best cost structure of anyone doing 100G," he claims. "It's going to be lower cost than what's out there today, because we're doing it in CMOS."

One vendor that does have that electronics focus is CoreOptics, which is both a chip maker and a module maker, with experience in digital signal processing (DSP) coming from the development of MLSE algorithms for electronic dispersion compensation. The company issued a press release at the show claiming that it has validated its "coherent" technology for cost-effective 40G transmission (their quotation marks, not mine). What this means, I have absolutely no idea. CoreOptic's senior marketing folks were ensconced in a room with a systems vendor when I turned up to meet with them, so I'm still in the dark.

One thing I've learned this week is that, strictly speaking, "coherent" is a term that can be applied to any modulation scheme that uses phase to encode the information, whether the detection scheme is differential (comparing the phase of one bit with the next using optics) or uses a local oscillator laser and DSP to extract the signal. So CoreOptic's approach could be either.

The ever-shrinking tunable

2009-03-23T12:07:42Z

You might expect one of the usual suspects — JDSU or Bookham — to be first out the gate with a tunable laser in the XFP form factor, but Emcore seems to have pipped them at the post.

Or has it? Emcore's announcement this morning says the company is "planning to release a full-band tunable XFP product line", but doesn't give a time-frame (see Emcore unveils industry's first tunable XFP optical transceiver).

That announcement could be roughly equivalent to the one JDSU made last year, when it unveiled its tunable transmitter chip (see JDSU shrinks tunable transmitter). JDSU told fibresystems.org it was working on a tunable XFP module that would be ready around the end of 2009. Although JDSU has been highly secretive about the press conference it's calling for tomorrow morning, the timing would be about right for that to be a tunable XFP.