It's not that we can't wait for 2013 to end, but here's a look at what optical communications technologies should be hot in 2014.
What can we say about 2013 that hasn't been said already? It seemed the optical communications space spent much of the year waiting for the economic turmoil of 2012 to blow over. That finally seemed to start happening in the middle of the year, if recent revenue estimates by market research firms can be believed.
From a technology perspective, silicon photonics became a bogeyman for some and a fairy godmother for others, depending on whether you were working on the technology or not. Software-defined networking (SDN) vied with silicon photonics for "Buzzword of the Year," while network functions virtualization (NFV) tagged along for the ride. And all the noise made last year about colorless/directionless/contentionless ROADMs died down.
The deployment of 100-Gbps technology kicked into high gear – except in the data center, where 40 Gbps is just getting established. Nevertheless, the IEEE determined that it's not too early to begin thinking about what comes next and decided it's 400 Gigabit Ethernet. In carrier networks, a small handful of companies offered 400 Gbps, but carriers started to talk about breaking that in half and deploying 200 Gbps.
As is the tradition here at Lightwave, the fact that we've reach the last two months of 2013 means the time has come to determine which of the current hot topics will continue to burn brightly in the next 12 months and which ones will cool off. Our discussion will encompass the following application areas:
- Networking (except for the access)
- Fiber to the home
- Cable-operator applications
- Equipment design
- Test and measurement
Our prognostications rely on basic research, conversations at trade shows, interviews, and a well-worn dart board. This article contains statements about expected future events and financial results that are forward-looking and subject to risks and uncertainties. Your results may vary. But that's what makes it interesting, isn't it?
Optical networks in the abstract, quickly
If you thought you couldn't get away from SDN and NFV last year, you may find 2014 even worse. Many of the efforts toward carrier-friendly
SDN and NFV that began this year will reach fruition over the next 12 months, which may lead these concepts out of the petri dish and into optical networks.
And those efforts are numerous. The most significant is that of the Optical Networking Foundation's Optical Transport Working Group, which is examining use cases and specifications for fiber-optic networks. It should complete its task toward the middle of 2014; with luck, the working group will provide a guidepost others will follow toward the realization of provisioning and orchestration of multivendor network resources between the routing and transport layers. Meanwhile, we'll continue to see parallel efforts such as Open Daylight as well as ecosystems and demonstrations generated by individual vendors.
Focusing on the actual optical transport layer rather than its abstraction, we'll see more 100-Gbps offerings customized for the metro in the second half of next year, thanks in large part to new coherent optical transceivers (about which we have more to say in our discussion of equipment design trends below). We also should expect more efforts to promote interoperability among vendors' systems, with the recent interoperability announcement between Acacia Communications and NEL a step in that direction.
We may also see the first deployments of 200-Gbps technology, which Verizon already has taken for a spin. The interest in 200G indicates that while some carriers may have already outgrown 100G on certain links, they're not quite ready to make the leap to 400G right away. That said, with the first 400-Gbps deployment announced this year, we'll likely see a few more in 2014, with more vendors signaling their ability to support such data rates when their customers require them.
The maturation of FTTH
Overall, 2013 was a quiet year for fiber to the home (FTTH) technology. The next 12 months likely will be more of the same. That's because the major building blocks for FTTH networks are now firmly in place.
A carrier has a choice among EPON, GPON, and point-to-point Ethernet architectures, with multiple vendors available for each option. There is a wide variety of CPE to meet the services expected on the subscriber end (even if that end is a tower). And when the carrier runs out of capacity on their chosen architecture, there's 10-Gbps versions of PON and point-to-point ready to be deployed.
The same could be said for the fiber, with dozens of offerings that promise some degree of bend tolerance. There's specialized cabling for multiple-dwelling-unit (MDU) applications that make running fiber inside the building more efficient and less intrusive.
So in 2014 expect incremental improvements to these major pieces, aimed at lowering costs and streamlining deployments. The innovation will come at the intersection of copper and fiber.
In the most obvious sense, that means the copper coming after the fiber in fiber to the node (FTTN) and fiber to the cabinet (FTTC) architectures. We saw vectoring roll out in 2013, and this noise-cancellation technology will become more ubiquitous next year. We'll also hear more about G.fast, the vectoring successor that will take copper into data rates of 200 Mbps and above – perhaps even 1 Gbps, if the nodes are close enough to the subscriber. Expect systems houses to aid carriers to deploy VDSL2 with vectoring now and an upgrade path to G.fast.
But even carriers intent on wringing the last megabit out of their copper lines know that fiber will be necessary in the future. So we also can expect to see more emphasis placed on platforms that support the migration from copper to fiber when recalcitrant carriers finally see the light.
Cable MSOs find more use for coax
The use of FTTH technology is more ubiquitous among U.S. cable operators than they're willing to admit. But new technologies promise to enable U.S. cable MSOs – and others around the world in the future – to meet future competitive requirements via their existing hybrid fiber/coax (HFC) networks.
DOCSIS 3.1 holds the most promise along these lines. The technology, now in the specification development process within CableLabs, targets 10 Gbps downstream and 1 Gbps upstream, much like the asymmetrical variant of 10G EPON. CableLabs has targeted this year for the completion of specifications (the company was expected to report its progress at SCTE Cable-Tec), which means chipset samples sometime in 2014. First hardware prototypes could follow within the next 12–18 months as well. DOCSIS 3.1 is expected to be backward compatible with DOCSIS 3.0 to the point that signals using both DOCSIS variants could travel together. DOCSIS 3.1 signals will leverage orthogonal frequency-division multiplexing, while DOCSIS 3.0 uses quadrature amplitude modulation.
If that wasn't bad enough for FTTH proponents, the IEEE has embarked on development of a standard that would enable EPON capabilities over coax. In-building use of existing coax, particularly in MDUs, would be a natural application of the technology. However, there's a chance that cable operators could use EPON over coax in the outside plant as well.
By the time this technology reaches the field in a few years, cable operators may have more widely deployed EPON in their networks, thanks to the fact that DOCSIS Provisioning of EPON (DPoE) has finally matured to the point where it's approaching deployment. Bright House Networks recently announced it will use Alcatel-Lucent's DPoE offering to support business services. The systems vendor says it has at least one more North American DPoE customer. With this success, we can expect more DPoE offerings.
Integration dominates equipment design
Many of the technology design advances we'll see in 2014 will have integration as a foundation. Wafer-level, hybrid, photonic, digital, analog, vertical – it's just a question of which kind.
Silicon photonics will graduate from being exotic and hyperbolic and become a better understood approach in 2014. We should see technology from such startups as Aurrion and Skorpios Technologies in the upcoming year, as well as (finally) from Intel and IBM. It seems clear that the upcoming wave of silicon photonics technology will wash across the data center first; the question then becomes what, if anything, it can do for line-side applications. We'll also learn whether integration via silicon photonics naturally leads to vertical integration among systems houses; with Cisco, Mellanox, and Huawei having purchased silicon photonics companies already, others may think they need such capabilities in-house as well.
Another integration question that developers will debate in 2014 is whether upcoming coherent CFP2 transceivers should integrate the necessary DSP inside the module or whether those chips should remain on the host board. There are points in favor of both options: An integrated design is simpler, whereas leaving the DSP outside the module opens the door to sharing a large DSP device with multiple modules and enables companies that use ASICs developed in-house to continue to leverage that investment while enjoying the benefits of a smaller module and a wider ecosystem (since we can expect companies that have so far sat out the coherent module business to jump in if they don't have to worry about supplying the DSP). Both approaches will have adherents; it may turn out that one finds more favor in long haul applications at certain systems houses.
Meanwhile, as this topic and some of the earlier sections of this article indicate, the importance of electronics and software to optical communications technology will continue to grow.
Putting simplicity to the test
As optical communications technology developers have tackled demands for higher bit rates, the avenues they've pursued have become increasingly complex. This fact presents a pair of challenges for test and measurement-instrument suppliers. First, they have to stay ahead of current requirements, which means coming up with instruments that can handle the growing complexity of optical communications techniques. Second, while the technology within these test instruments arguably is at least as complex as the elements under test (if not more), the test sets must perform their functions as simply as possible.
In field applications, the drive for simplicity has begun to translate into an increased use of automation – what could be thought of as smart test sets. The ability not only to perform measurements but also analyze results has become significantly important as carriers transition from copper to fiber faster than installers can be retrained. We see this trend most strongly in test instruments for FTTx applications, but a similar case can be made for mobile backhaul as fiber to the tower (FTTT) becomes more prevalent as well as within data centers. So in 2014 we will continue to see instruments reach the field that automate test functions, ease collaboration, and leverage platforms with which technicians are becoming familiar outside of the job, such as mobile phones and tablets.
Of course, the most complex optical communications technology now in deployment supports 100-Gbps wavelengths. The topic of what kind of test equipment will be necessary to handle coherent transmission has been a point of debate for the last two years. This debate has several parts, but two elements have attracted the most attention.
One question is whether you still need to create dispersion maps, particularly if you plan to run wavelengths other than 100G down the same fiber and whether you're using Raman amplification. This question will continue to be asked in 2014, but I believe you'll see more carriers abandon dispersion mapping on greenfield links, particularly if they're not using Raman.
The other debate point is whether you need a new class of field test instruments designed to examine signals at the phase and amplitude level. That debate also will continue in 2014, but increasingly it appears that the conclusion will be that you don't. It seems that most carriers don't want to deal with issues at the phase or amplitude level, partly because the transmission schemes their systems vendors use are highly proprietary and therefore something of a mystery. So the consensus approach has become the placement of Ethernet or OTN test sets on the client side of the coherent systems, injection of a signal on one end, and a determination of whether it comes out okay on the other. If there's an issue in between, it's the systems vendor's problem.
And systems vendors are stepping up to the responsibility – and helping with the simplification of fiber-optic-network test – by incorporating test functions into their hardware. Many, if not all, coherent transmission equipment now comes with functions that aid acceptance testing that previously would have required a separate piece of test gear. Similarly, we're seeing the systems houses incorporate OTDR functionality into their equipment, particularly for FTTH applications.
Do these trends spell disaster for test equipment vendors? Probably not. An executive at one test and measurement vendor told me that the number of OTDR sales the integrated FTTH systems were costing his company was hardly worth noting. But it does show that the landscape of field testing is changing rapidly.
Meanwhile, back in the lab, the complexity problem is somewhat different. The technicians here know how to do the tests, but often need several pieces of test equipment to get the job done. That makes test setups cumbersome, particularly if instruments from multiple vendors are involved, either by desire or necessity. These setups also are extremely expensive – a coherent modulation analysis capability, including the necessary oscilloscope, can run in the multiple hundreds of thousands of dollars on its own.
So the emphasis in 2014 will continue to be placed on simplification as a vehicle toward reduced cost. That will mean companies adding to their product lines must be able to meet all aspects of a specific requirement, adding functions to existing systems to lower the number of instruments needed and making it simpler to share equipment and capabilities to spread out the costs.
Meanwhile, with work underway toward a 400 Gigabit Ethernet specification, we'll see more test equipment either introduced or upgraded with this requirement in mind. The 400G interfaces likely will be based initially on 25-Gbps lanes. The good news is that 100G is moving toward 25-Gbps lanes, so the test capabilities are already in place; the bad news is that 400G will require 16 such lanes, meaning a requirement to make testing this many lanes more efficient.
There also has been talk about applying multilevel modulation formats to 400 Gigabit Ethernet signals, which means we'll see a continuation of the recent string of announcements around PAM4 test equipment.
Archived Lightwave Issues