As reported yesterday, Adtran and Deutsche Telekom are testing a pair of new technologies designed to improve G.fast. What are they, and how do they work?
One is doubling the available bandwidth to 212 MHz, the other is a new dynamic time allocation (DTA) scheme called collective DTA, or cDTA. The combination of these technologies appears to solve some drawbacks that G.fast advocates have acknowledged even while glossing over some of their more serious practical ramifications.
Sckipio claims to have been first to introduce DTA, and it believes it has at least a one-year lead on the market with cDTA. FierceTelecom spoke with Sckipio VP of Marketing Michael Weissman, who walked us through the progression of technological advances culminating in cDTA.
DSL relies on frequency division schemes, where different frequencies are allocated to the downstream and the upstream, and the allocations are set. G.fast differs in that it is based on time allocations. The entire frequency band could be used by either upstream or downstream at different times, in essence increasing the available bandwidth in either direction.
Furthermore, with DTA, it is possible to dynamically allocate upstream and downstream ratios from once customer to the next. In other words, every individual customer attached to a DPU (a distribution point unit) could get their own individual upstream and downstream rates. The practical result is that a service provider can offer different service tiers to different customers.
Theoretically. DTA has a practical drawback, however. It works fine in situations where there is little to no crosstalk, for example when the run to the customer is coaxial cable, or if the optical connection leads into a single TCP wire. But telcos (Deutsche Telekom, for example) don’t use coax and aren’t going to terminate optical runs with single TCP wires. They run any number of TCP wires from DPUs out to any number of customers, and there’s crosstalk in any bundle of TCP wires.
The practical result, Weissman explained, is that if a service provider offers G.fast-based service that is based on DTA technology, every single customer off a DPU has to get the same allocation ratio—meaning they all have to subscribe to the same service tier.
That’s unlikely.
Sckipio worked to develop a version of DTA that was resistant to crosstalk, and in May announced cDTA. Adtran and DT are testing it to see if it works as advertised, but if it does, it means that service providers will be able to roll out G.fast and they won’t have to touch existing TCP and they will still be able to dynamically allocate broadband resources among different customers, enabling the provider to offer different customers connected to the same DPU different service tiers.
Another practical result is that the service provider doesn’t have to be absolutely precise with its DPU location. DSL and G.fast speeds are highly dependent on loop lengths; everything else being equal, data rates will be increasingly faster on increasingly shorter loops.
Say a service provider needs to have a DPU 100 meters from its customers. As a practical matter, when serving multitenant buildings, each individual customer might be a tens of meters closer or further from the DPU. Previously that might have affected transmission rates, but with cDTA, allocation is dynamic. In other words, the system can compensate for distance variations.
As for bandwidth—to date, G.fast has had 106 MHz of bandwidth. Doubling the bandwidth helps lead to accelerating transmission rates, potentially into multigigabits. Also, by going to 212 MHz, it should be possible to deliver gigabit rates on a single twisted copper pair (TCP) wire.
With previous versions of G.fast, carriers expected to have to get faster rates by ganging multiple TCP wires and aggregating their throughput. While it would have been a boon to continue using dirt cheap TCP, it still would have required rewiring customers’ homes and offices. That’s enough of a task when dealing with single units, but rewiring multitenant dwellings can be a nightmare.