While PON leads the charge in the fiber field, several other major trends are
making noise. The most prominent? Look no further than the 40- and 100-Gbit/s
movements. Primary drivers for creating a faster Internet, of course, are video
and graphics.
YouTube already tops 25 Gbits/s on a regular basis, as users upload and download
videos 24/7. The expected forthcoming video on demand (VOD) and high-definition
Internet Protocol TV (HD IPTV) glut is another driver, as are the growing number
of Internet access applications and broadband-connected homes.
To handle this load, equipment based on the Sonet/SDH OC-768 standard, which
runs at 39.812 Gbits/s (what most users call 40 Gbits/s), is already in use.
There was also some discussion about a 40-Gbit/s Ethernet version, but the IEEE
recently committed to produce a 100-Gbit/s Ethernet standard by the 2009/2010
time frame. This maintains the x10 speed increments traditional with Ethernet.
Verizon recently announced a data-rate increase in some sections of its ultra-long-haul
(ULH) optical network to 40 Gbits/s, from the current 10 Gbits/s beginning with
the link between New York and Washington, D.C. The link uses 80 wavelengths
in a dense wavelength-division multiplexing (DWDM) system. The next step is
100 Gbits/s, with trials to begin in about 18 months.
Avago demonstrated a parallel vertical-cavity surface-emitting laser (VCSEL)
optics product, which may be the best solution for 100 Gbits/s based on cost
and reliability. IBM unveiled a prototype parallel optical transceiver chip
set that’s capable of 160-Gbit/s speeds. And, Siemens reported new electronic
hardware that reached 107 Gbits/s over a single fiber channel.
Achieving 40 and 100 Gbits/s isn’t easy, so R&D efforts have stepped
up on many fronts. New modulation methods are just one example. Most modulation
to date is just amplitude shift keying (ASK) or on-off keying (OOK), where the
data just switches the laser beam off and on.
To beat the various dispersion problems, modulation methods like differential
phase shift keying (DPSK) and differential quadrature phase shift keying (DQPSK)
are being tested and adopted. We’re even seeing the resurrection of the
decades-old duobinary technique. These methods extend the reach of 40-Gbit/s
transmissions. We can expect to see further experimentation with various modulation
methods, as the push toward 40 and 100 Gbits/s progresses.
The other problems to overcome, if we expect to use the installed fiber base,
are polarization mode dispersion (PMD) and chromatic dispersion. New types of
fiber solve these problems, but we still have tons of dark fiber that can be
had for pennies.
That’s why electronic dispersion compensation (EDC) techniques were
developed. Equalization methods at the receiver lead the pack. Something called
Maximum Likelihood Sequence Estimation (MLSE) equalization appears to offer
the most promising solution to extend range at high data rates.
The IEEE formed the High Speed Study Group (HSSG) in July 2006 to come up
with a 100-Gbit/s Ethernet (100GE) standard. Work is proceeding, but there’s
an internal squabble about whether to produce a 40-Gbit/s Ethernet interim standard.
Hopefully, that will be resolved to prevent the HSSG from being disbanded or
otherwise sidetracked on the way to that desirable 100GE standard.
Most experts believe the standard will be a parallel transmission system.
For example, 10 10-Gigabit Ethernet (10GE) streams could be sent simultaneously
over 10 different wavelengths on a single fiber. Reach targets are up to about
10 km on single-mode fiber or up to 100 meters on multimode fiber.
Interested companies have formed an industry organization called the Road
to 100G Alliance to speed the development of the 100G standard. A key concern
is that even if we achieve 100G, other parts of the network system must also
scale in speed, including memory, processors, backplanes, switch fabrics, and
all of the equipment (e.g., routers and switches) that use them.
10-Gigabit Networking Products
Though it’s been around for years, 10-Gbit/s Sonet is seeing some new
growth as lower-cost transceivers become available. With wide-area-network (WAN)
and metropolitan-area-network (MAN) expansions demanding rates higher than the
current 2.488-Gbit/s levels, 10-Gbit/s Sonet is growing fast. The lower-cost
transceivers also make practical the planned 10-Gbit/s Fibre Channel systems
that dominate the storage-area network (SAN) scene.
Then there’s 10GE. With the major increase in 1GE ports—they’ve
been built into every PC and laptop for years—10GE local-area network
(LAN) aggregation is a necessity. Now, 10GE switches and routers are more affordable
than ever.
LightCounting recently predicted that the market for 10GE optical components
used in enterprise network equipment is expected to grow by a 35% compound annual
growth rate from 2006 to 2010. 10GE is the new “sweet spot” in optical
fiber systems, and its growth will no doubt accelerate until 100-Gbit/s systems
are available, probably beyond 2010.
POF Is Hot
Plastic optical fiber (POF) has been around for years, but it ran out of steam
as a transport medium when data rates passed 100 Mbits/s. Its high attenuation
and dispersion more than offset its ultra-low cost. But things are changing.
POF is rebounding because new equalization techniques offset the attenuation
and dispersion problems, making it useful up to about 300 meters and data rates
up to 3 Gbits/s. Its low cost is still its main benefit, as alignment with the
laser and optical receiver isn’t as critical due to its larger size. Low-cost
650-nm (red) visible lasers can be used.
You can actually cut POF with scissors. In some cases, you don’t even
need a connector. This makes installation fast, easy, and safe. POF is widely
used in auto systems, but will soon find its way into more mainstream applications
such as Ethernet and home networks.