In a joint project with Alcatel-Lucent, BT claims to have a set new record for transmission over a single commercially deployed fiber link by sending data at 1.4 Tbps over a 255 mile circuit between its Adastral Park research campus in Suffolk and its iconic tower in London. Such speeds have been demonstrated in the lab but not so far in a ‘real world’ situation. The companies attribute the speed to use of an ‘alien super channel’, so called because it is superimposed over an existing network, but this is really a misnomer. There is nothing alien about the technology, which is likely to become mainstream over the next few years as Telcos seek to boost capacity of their existing infrastructures to meet growing demand for bandwidth driven largely by proliferating video streaming and downloading.
Indeed the significance of the announcement lies not so much in the headline speed but the fact that it demonstrated a new generation of wave division multiplexing (WDM), which has become the staple of core fiber transmission networks around the world.
There are two dimensions to the transmission capacity of a single optical fiber, the bit rate of individual channels and the number of channels that can be multiplexed onto that fiber via WDM. There has been steady improvement on both counts over the last two decades since the dawn of the public Internet.
Over that time the maximum capacity of individual fiber channels has increased 100 times from 1 Gbps to 100 Gbps. At the same time WDM emerged to allow increasing numbers of channels to be multiplexed onto a single fiber by confining them to specific wavelengths spaced at 50 GHz.
To date these two dimensions have scaled largely independently of each other, but now a crunch is coming as a further increase in the bit rate of a single channel driven by the electro-optical transponders at each end of the fiber would make it hard to contain within that 50 GHz grid space defined by current WDM technology. This emerged through research into the next level of transponder technology, stepping up from 100 Gbps to 400 Gbps, which found that this could not be contained reliably within 50 GHz channels.
One solution would be to increase the grid space to 100 GHz, but this in turn would reduce the number of channels that could be multiplexed, defeating the object. In any case it would be wasteful of spectrum for those services that will continue using the lower bit rate transponders for some time to come. Another option considered by Telcos such as BT, was to keep a fixed grid but with different sized slots to accommodate transponders at a variety of speeds, but this was dismissed as being too complex to manage.
Such considerations led to the development of Flexgrid, which as the name suggests allows flexibility in allocation of spectrum to channels. Total flexibility is impossible to achieve, but what Flexgrid does is break the spectrum up into smaller slots, typically 12.5GHz, with the ability to join them together on demand to construct channels of varying capacity. This is a step change for WDM and so we have dubbed it WDM 2.0.
A key requirement is that channels can be reconfigured in real time as the mix of services changes, which has been achieved. This does raise the question though of why BT has not deployed an existing technology, inverse multiplexing, which appears to have the desired property of allowing flexible channel sizes configurable in real time. Inverse multiplexing would allow 12.5 GHz to be the default channel size, but with the ability to effectively bond them together to create larger channels on demand, seemingly accomplishing the same end.
The answer is that inverse muxing was considered and then dismissed for several technical reasons, largely because it makes less efficient use of the optical spectrum. This efficiency can be expressed in bits per Herz of spectral width, with one of the advantages of the Flexgrid approach being that it comes with very efficient use of spectrum. As a result BT and Alcatel-Lucent have been boasting of a record 5.7 bits per second per Herz with their recent demonstration. For this reason 400 Gbps channels can be accommodated in 75 GHz, which admittedly still busts the original 50 GHz WDM limit but only by 50%, while quadrupling the throughput per channel. By contrast, with inverse multiplexing, a 400 Gbps channel constructed by bonding say four 100 Gbps channels, the largest currently available, would consume 200 GHz of spectrum, according to BT.
All in all the underlying Flexgrid technology looks extremely promising and Telcos around the world will be itching to get their hands on it.