News Researchers Set New Fiber Optic Speed Record of 178 Tbps

Endymio

Notable
Aug 3, 2020
403
117
870
2
Quote: "According to UCL, that's close to the maximum theoretical data transmission speeds defined by American mathematician Claude Shannon. "

To correct this misleading statement in the article, Shannon's Theorem imbues no hard limit on the maximum data transmission rate in a fiberoptic cable. What was meant is that the particular encoding scheme used was near the Shannon Limit for the particular signal bandwidth and SNR of the apparatus in question.

We are nowhere near the theoretical limits for fiber itself. At some point in the future, this 178 Tbps milestone will be exceeded by a factor of 10, 100 or even more.
 

Oldcompsci

Commendable
Apr 30, 2019
20
5
1,515
0
Completely agree. The genious of Shannon and others like him wasn't that they established "hard limits". What they did was to discover a new framework of looking at things and with the tools they developed, exploit what is "naturally available" in more efficient and useful ways.



Quote: "According to UCL, that's close to the maximum theoretical data transmission speeds defined by American mathematician Claude Shannon. "

To correct this misleading statement in the article, Shannon's Theorem imbues no hard limit on the maximum data transmission rate in a fiberoptic cable. What was meant is that the particular encoding scheme used was near the Shannon Limit for the particular signal bandwidth and SNR of the apparatus in question.

We are nowhere near the theoretical limits for fiber itself. At some point in the future, this 178 Tbps milestone will be exceeded by a factor of 10, 100 or even more.
 

comedichistorian

Distinguished
Jun 26, 2010
36
9
18,535
0
Quote: "According to UCL, that's close to the maximum theoretical data transmission speeds defined by American mathematician Claude Shannon. "

To correct this misleading statement in the article, Shannon's Theorem imbues no hard limit on the maximum data transmission rate in a fiberoptic cable. What was meant is that the particular encoding scheme used was near the Shannon Limit for the particular signal bandwidth and SNR of the apparatus in question.

We are nowhere near the theoretical limits for fiber itself. At some point in the future, this 178 Tbps milestone will be exceeded by a factor of 10, 100 or even more.
I was going to reply to the article with some bitchy, uninformed (mostly lol) quip about it being ____ decades before we see anything close to that speed in our homes/on our devices but you appear to actually have some knowledge on the subject so I must ask...how long do you think it'll be?
 

InvalidError

Titan
Moderator
To correct this misleading statement in the article, Shannon's Theorem imbues no hard limit on the maximum data transmission rate in a fiberoptic cable.
Shannon does not set a hard limit on electrical transmission either, electrical transmission simply has more losses, more noise and less bandwidth to work with so practical limits get reached much faster.

Sooner or later though, you will run into practically insurmountable SNR issues mainly from various dispersion modes (contaminants, mechanical vibrations, fiber geometry imperfections, thermal expansion/contraction, crosstalk, stray photons, etc.) even on fiber.
 

Endymio

Notable
Aug 3, 2020
403
117
870
2
I was going to reply to the article with some bitchy, uninformed (mostly lol) quip about it being ____ decades before we see anything close to that speed in our homes/on our devices but you appear to actually have some knowledge on the subject so I must ask...how long do you think it'll be?
I appreciate the vote of confidence, but its a rather complex issue since it depends on economic as well as technical factors. I could pithily dodge the question and reply "we'll see it when the demand is there", but in all honesty, attempting to predict anything more than ten years out (as this will certainly be) is about as accurate as fortune-telling from tea leaves. :)

Sooner or later though, you will run into practically insurmountable SNR issues mainly from various dispersion modes (contaminants, mechanical vibrations, fiber geometry imperfections, thermal expansion/contraction, crosstalk, stray photons, etc.) even on fiber.
Oh sure .. but Shannon of course has nothing to say on those issues. From the pure perspective of information theory, unless you set some upper frequency limit on the lasers involved, there is no theoretical maximum.
 

InvalidError

Titan
Moderator
Oh sure .. but Shannon of course has nothing to say on those issues. From the pure perspective of information theory, unless you set some upper frequency limit on the lasers involved, there is no theoretical maximum.
Lasers aren't the only bandwidth limit on fibers. For starters, single-mode fiber is only single-mode for wavelengths long enough to keep it that way. Making the fiber core smaller would enable the use of shorter wavelengths at the expense of making the numerical aperture that much smaller (more difficult to get the light in and out) and increasing losses. Any given dopant and refraction index profile also has limits on wavelengths they can achieve total internal reflection at, compromises on group delay, attenuation profile, etc. A lot of the fancy stuff being added to optical equipment is to compensate for the fact that optical fibers aren't ideal similar to how electrical transceivers for things like PCIe4 get fancier equalization circuitry to compensate for non-ideal traces.

For very long range applications where you may want to use erbium-doped optical amplifiers instead of electrical ones, you are further limited in wavelengths to 1525-1610nm.

Optical fiber may have theoretically infinite potential but practical limits are being reached.
 

Endymio

Notable
Aug 3, 2020
403
117
870
2
Lasers aren't the only bandwidth limit on fibers. For starters, single-mode fiber is only single-mode for wavelengths long enough to keep it that way.
You're restricting your thinking to current telecom-grade fiber. The bulk of active research today relies on spatial multiplexing. The team which set the record to which this article refers, for instance, used tri-mode fiber (S, C, & L bands). This work would naturally and easily extend to include O and L bands, and possibly 850nm as well.

Any given dopant and refraction index profile also has limits on wavelengths they can achieve total internal reflection at...
But for the long future, we don't have to assume NIR lasers on fibers based on total internal reflection. Take a look at the work being done on microstructured waveguides, particularly HC/PCF fiber. We're not going to be putting this type of glass into the ground for at least another decade, but fiber such as this operating in, say the 300-400nm UV bands with DWDM would allow petabit/sec bandwidth levels.

But all this is getting far afield of my original point, which was simply that these researchers were "approaching the Shannon limit" through their use of constellation shaping, and not by reducing dispersion, attenuation, coupling losses, or anything else which would simply raise that limit in lock-stop with any bit-rate increases.
 
Last edited:

ASK THE COMMUNITY

TRENDING THREADS