x
Optical/IP

Optical Crossconnects

Before reading this you may find the following tutorials useful:
Optical Networks, Wavelength Division Multiplexing (WDM)

Huge amounts of information traveling around an optical network need to be switched through various points known as “nodes.” Information arriving at a node will be forwarded on towards its final destination via the best possible path, which may be determined by such factors as distance, cost, and the reliability of specific routes. The conventional way to switch the information is to detect the light from the input optical fibers, convert it to an electrical signal, and then convert that back to a laser light signal, which is then sent down the fiber you want the information to go back out on.

It all sounds unnecessarily complicated, don't you think? It's as if to give money to a friend, you would have to convert it into Euros, then back into your own currency before your friend can use it. Now, even though your information is safer in electrical form than your money may be in Euros, it still seems a strange way to work. What if we could just move the light itself around, without all this ridiculous conversion to electric signals? Well, my friend, what you need is an optical crossconnect (OXC).

The advantages of being able to avoid the conversion stage are significant. “Optical switching” should be cheaper, as there is no need for lots of expensive high-speed electronics. Removing this complexity should also make for physically smaller switches. Additionally, optical switches are relatively future-proof. An electrically based switch will have electronics designed to detect the incoming light signal. If you increase the speed at which the light signals operate (increasing the “bit-rate”) then the electronics will need to be upgraded to handle the faster speeds. If you are just rerouting light, however, it doesn't matter how fast the data is coming — so you can accommodate any future upgrades of bit-rate without needing to upgrade the switch (this is called “bit-rate transparency”). Optical crossconnects are just now coming onto the market with these benefits and more.

Optical crossconnects are very much designed with simplicity in mind. You've got some light in one fiber that you want to move to a different fiber, so just redirect the light somehow and that's all you need — it's child's play. Unfortunately, the technologies used seem to come out of science fiction rather than a child's bedtime story. There is a wide range of wild and wacky ways to switch light between optical fibers. Semiconductor amplifiers, liquid crystals, holographic crystals, and tiny moveable mirrors are just a few. Truly buttock-clenching switching developments are anticipated in the future. One of the most common techniques being developed is that of the tiny moveable mirrors known as micro-electro-mechanical systems (MEMS).

MEMS Mirror MEMS Optical Cross Connect MEMS consist of mirrors no larger in diameter than a human hair. They can be arranged on special pivots so that they can be moved in three dimensions, and several hundred such mirrors can be placed together on mirror arrays no larger than a few centimeters square. Light from an input fiber is aimed at a mirror, which is directed to move the light to another mirror on a facing array. This mirror then reflects the light down towards the desired output optical fiber. It perhaps sounds a little bizarre, but it does seem to work. MEMS mirror arrays are even used successfully in some of the modern digital projectors used for computer-based presentations.

Key Points

  • Crossconnects forward signals to their destination by specific routes
  • Traditional crossconnects convert light to electricity then back to light
  • Optical crossconnect advantages include cost, size, and bit-rate transparency
  • Redirecting light from one optical fiber to another, without electrical conversion
  • Most advanced optical switching technology is MEMS, tiny moveable mirrors


Further Reading

Fiber Bragg Gratings (FBGs), Arrayed Waveguide Gratings (AWGs), Semiconductor Optical Amplifiers (SOAs)

<<   <   Page 2 / 5   >   >>
Belzebutt 12/4/2012 | 8:27:48 PM
re: Optical Crossconnects Having optical lambda conversion would make it possible to have an all-optical network with lambda granularity end-to-end.

If you're using today's all optical switches, you have some DWDM equipment at both endpoints. These DWDM boxes aggregate lambdas into one single fiber, and shoot it off towards the core. When that fiber gets to the first OOO OXC, _all_ the lambdas on that fiber must get routed together to the same output. The OOO cannot separate the lambdas by itself, and send lambda 4 to one output and lambda 5 to another. If you want to switch lambdas you need to demultiplex all the lambdas before the fiber gets to the OOO switch, then put each lambda on a separate fiber, send them all to a different port on the OOO switch, and then the OOO switch will route them to the proper output ports. There, they will have to go through another DWDM box that will aggregate several separate lambdas and put them on the same fiber.

So basically, you could do all OOO if you didn't use DWDM to its full extent by managing your labmbas as logical "pipes". That is what most carriers do however, from what I understand.

So to truely use OOO OXCs with lambda management in your network, you'd need to have OXCs that can optically demultiplex all the lambdas coming from a fiber, switch them optically, convert the original wavelenghts to different wavelengths as required so that you don't get "lambda blocking" at each output, multiplex all these lamdas that should go on the same output fiber, and send that traffic off to the next OOO OXC or to the end point.

It's all done optically in the above example, while today you need to put one DWDM box in front of every fiber coming to an OOO OXC (Unless you don't plan to extract any lambads from any fiber at that particular OXC).
optinuts 12/4/2012 | 8:27:48 PM
re: Optical Crossconnects thanks bezlebutt
manoflalambda 12/4/2012 | 8:27:47 PM
re: Optical Crossconnects FiberSystems International, May 01, pg 14....
1296 port OXC, its coming!

Salute,
Manoflalambda
tony1athome 12/4/2012 | 8:27:46 PM
re: Optical Crossconnects Most of the practical applications for an OXC are going to reside within a single POP, where they interconnect OEO devices.

The value add is the ease and speed of provisioning, which is a strategic advantage that should not be underestimated.

The control plane for an OXC is likely to be GMPLS. The routed overlay on top of the optical network understands and can optimize the traffic assignment thru lamba assignment computations.

While all-optical switching thru a POP would be a wonderful thing, simplicity says that the first practical step will be DWDM units that use standard SONET wavelengths for their common interconnect. The ability to connect ANY port in the POP to ANY other port is a major improvement in flexibility.

Others will argue that this is more expensive. Granted. Carriers already pay more so that they can have operational flexibility. A fine example is the set of carriers that prefer to use single mode fiber throughout their POP, paying much more than multimode just so that they never have to deal with having the wrong type of fiber on hand.

Tony
Belzebutt 12/4/2012 | 8:27:45 PM
re: Optical Crossconnects "The control plane for an OXC is likely to be GMPLS. The routed overlay on top of the optical network understands and can optimize the traffic assignment thru lamba assignment computations."

In that case, the lambda assignments will all have to be done on the OEO boxes, controlling the OOO OXCs, would they not? So the only real advantage of the OOO OXC would be the bitrate independence and the ease of provisioning. In that POP scenario, the OOO switch would not really be able to function as an all-optical unit, but more like an automated patch pannel connecting a bunch of OEO DWDM terminals.

The way I pictured a "true" OOO OXC is a box would replace a whole OEO cross-connect and any adjacent DWDM line systems, and you could just send a lambda from one end of your network and terminate that lambda anywhere else on your network, switching optically all the way. For that, you would need all-optical lambda converters inside the OOO OXC.
networking_legend 12/4/2012 | 8:27:44 PM
re: Optical Crossconnects Joining any optical networking startup isn't as easy as before, given the downturn in the industry as a whole.

However, if you have the right skills (for example, optical/photonic engineers are always in demand) there shouldn't be too much difficulty.
Otherwise, you might be screwed.

Take a look at:
1) Calient networks
2) Movaz networks
3) OMM
Belzebutt 12/4/2012 | 8:27:44 PM
re: Optical Crossconnects "That, or the OXC's could also participate in GMPLS."

Since you brought up GMPLS, what's your take on that? Will it be useful and well implemented by transport and router vendors? I've heard that Juniper for example is not too excited about it, even though they're working on it. Are there competing standards coming up?

"Yup. This is not a bad thing. It is far more likely to succeed than something that is much more aggressive and much more complicated. The technology is hard enough as it is."

I guess it just seems like it's far from the full glory of the all-optical network some vendors are pitching, where it's actually useful to have line systems that can reach 5000 km. They would be able to use that long range to do some OOO switching at the core...

"But now, that's much more than an OXC. That's an entire God-box transmission system. Yes, that would require all optical lambda converters. The history of technology suggests that this will not be the first product brought to market, especially since we still really haven't deployed a simplistic OOO OXC yet."

That's true, the all-optical scenario I described is pretty much science-fiction. Oh well, it's interesting to consider the possibilities.
tony1athome 12/4/2012 | 8:27:44 PM
re: Optical Crossconnects "In that case, the lambda assignments will all have to be done on the OEO boxes, controlling the OOO OXCs, would they not?"

That, or the OXC's could also participate in GMPLS. That's not out of the question, but it muddies the water in that now some of the control traffic will arrive optically and get an OE conversion. Some poor marketing person will get confused.


"So the only real advantage of the OOO OXC would be the bitrate independence and the ease of provisioning."


Yup. This is not a bad thing. It is far more likely to succeed than something that is much more aggressive and much more complicated. The technology is hard enough as it is.

"The way I pictured a "true" OOO OXC is a box would replace a whole OEO cross-connect and any adjacent DWDM line systems, and you could just send a lambda from one end of your network and terminate that lambda anywhere else on your network, switching optically all the way. For that, you would need all-optical lambda converters inside the OOO OXC."


But now, that's much more than an OXC. That's an entire God-box transmission system. Yes, that would require all optical lambda converters. The history of technology suggests that this will not be the first product brought to market, especially since we still really haven't deployed a simplistic OOO OXC yet.

Tony

move 12/4/2012 | 8:27:44 PM
re: Optical Crossconnects Hi Everyone,

I would like to know what is the prospects of
joining a small company working on OOO OXC
based upon MEM technology ?

Based upon the potential market demands for
OOO OXC and the feasibility of the MEM technology
and others, could someone kindly give me the opinions on this ?
move 12/4/2012 | 8:27:43 PM
re: Optical Crossconnects Hi Network_legend and others,

I was wondering whether it is too late
to join an early stage startup working
on OOO based upon MEM technology at this
point ?

I appreciate any of your opinions.

<<   <   Page 2 / 5   >   >>
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE