JONATHAN HOMA THE SHANNON LIMIT
BEYOND SHANNON: TAKING FIBRE TO THE LIMIT
The Shannon Limit defines the maximum rate of error-free data that can theoretically be transferred over a channel for a particular noise level. However, the fibre optic communications industry is developing work-arounds to get the largest possible data capacity over the longest distance. Optical Connections editor Peter Dykes spoke with Ribbon Communications’ Senior Director, IP Optical Solutions Marketing Jonathan Homa about the latest developments.
Is it possible to get around the Shannon limit?
capacity by using denser modulations, we have reached the limit of that using probabilistic constellation shaping. For example, Metro applications typically aim to use 16QAM that encodes four bits per symbol for distances up to about 1000 kilometres. After that the physics of the fibre says you can’t send it any further. There’s nothing you can do.
for 1000 kilometres, or you can send a single 800G signal for 1000 kilometres over a 150 gigahertz channel. You haven’t increased the spectral efficiency, because you’re still just sending 800G, but what you have done is you have halved the number of wavelengths. By increasing the baud rate, you’re able to ramp up the speed from 400G to 800G. Instead of two times 75 gigahertz channels, you do this over a single 150 gigahertz channel. What you’ve done is decreased the number of wavelengths, and since most of the cost of your network is in the lasers and the transceivers that are used to transmit these wavelengths so one 800G wavelength will be quite a bit less expensive than two 400G wavelengths. So that primarily is how we work around the Shannon limit, by using advances in silicon technology, to increase the baud rate, and have faster line rates with fewer wavelengths.
PD
What the Shannon limit is telling us is that we’re operating at the edge of the spectral efficiency in
JH
terms of what fibre can handle. If you can’t increase the capacity on the fibre because of the Shannon limit, the question is, where do you go from there? There are really two directions you can go. One direction, and this is almost where all the industry efforts have been taking place over the last few years, is by trading something off. The line rate is directly proportional to the modulation density, that is, how many bits we can encode on a single symbol. Depending on using different types of encoding, most of which is quadrature amplitude modulation today, it’s a matter of how fast you can transmit those symbols, the baud rate, which effectively determines the line rate. This is done with two polarisations, so you’re doubling the line rate, and there is a small overhead that has to be taken off for forward error correction, but based on this kind of approach, we have effectively reached the physical limit of what can be done with modulation. It is not possible to get more
If the limit is impassable, what are the alternatives?
PD
The only way that we can really work around the Shannon limit is by sending these symbols faster,
JH
namely by increasing the baud rate. The way we’re doing that is we’re employing DSP technology to enable us to transmit and recover these symbols more efficiently by exploiting silicon. What we’re able to do is use this to have faster line rates at a particular modulation. However, and this is where the trade-off occurs, this doesn’t increase the spectral efficiency because when you increase the baud rate, you also need to increase the spectral bandwidth. Think about it this way, using the 16QAM modulation we mentioned above, you can send two 400G signals over a 75 gigahertz channel
So how is this achieved?
PD
JH There are two kinds of optimisation taking place and it’s all about the underlying optic technologies. One is trying to work as
12
| ISSUE 34 | Q3 2023
www.opticalconnectionsnews.com
Made with FlippingBook - Share PDF online