G.fast aims at providing ultra-high speeds over copper twisted pairs, up to and sometimes even exceeding speeds of 1 Gbps. The planned loop lengths for G.fast are from 50 to 250 meters (150 to 750 feet). G.fast is being standardized as ITU-T Recommendation G.9701. Similar to vectored VDSL, G.fast supports vectoring, which reduces crosstalk that is found in multi-pair cables and at higher frequencies. The first version of G.fast operates over frequencies of up to 106 MHz, and uses linear vector pre-coding to eliminate crosstalk in the downstream direction. A future version of G.fast may operate over frequencies of up to 212 MHz, and may support higher-performance non-linear pre-coding to allow for even higher speeds.
Unlike prior DSL technologies which used Frequency-Division Duplexing (FDD), and similarly to G.now, G.fast uses Time-Division Duplexing (TDD). With TDD, the system transmits only downstream signals for a fraction of time, and transmits only upstream signals for the remaining time. TDD allows the speed asymmetry to be varied at will among all the lines emanating from the same Distribution Point Unit (DPU, which is the name for a G.fast DSLAM). This allows some areas to be served with business-class symmetric service, while other areas can be served with asymmetric service that best addresses consumer needs. Transceivers operating in the same bundle must use the same TDD ratio, and TDD frames must be synchronized to a common clock to eliminate Near-End Crosstalk (NEXT).
G.fast is amenable to low-cost self-install deployment, similar to ADSL. G.fast supports new On-Line Reconfiguration (OLR) techniques, including Fast bit-Rate Adaptation (FRA), to adapt to changes in the transmission environment and overcome the harsh home wiring environment. G.fast supports reverse-powering, which sends power from the customers’ CPE to the DPU, thus eliminating the need for network power. Reverse powering needs to provide approximately 10 Watts, which is sufficient because the G.fast transceivers are very close to the subscribers and require little transmitter power. Reverse powering, and low-power modes planned for G.fast, are expected to lower G.fast deployment costs by eliminating the need for costly network powering and battery back-up of remote DPUs.
G.fast is expected to be deployed in a Fiber-To-The-distribution point (FTTdp) architecture as shown in the Figure above. The “dp” may also be called the “terminal” or “drop-wire terminal,” and is where the Distribution-Point-Unit (DPU) is located. Fiber is fed to the terminal, and from there, very short copper cables and drops, up to about 250 meters (750 feet) long, serve subscribers.
Some service providers will require G.fast DPU equipment to be backwards compatible with vectored VDSL. Such G.fast transceivers can fallback to support VDSL, albeit with lower performance compared to a dedicated VDSL transceiver. This is useful for initial installation of G.fast as an existing VDSL customer can be cut-over to the new G.fast DPU. At a later time, a customer can receive a G.fast-capable modem, which will automatically sync up in G.fast mode.
In addition to higher speeds, FTTdp supports other features that appeal to operators, such as customer self-install, variable asymmetry, reverse-powering, fast On-Line Reconfiguration (OLR), and Remote Copper Re-configuration (RCR). Remote Copper Re-configuration (RCR) means that after initial installation, it is possible to provision service for any customer connected to the DPU with no need for a technician dispatch. RCR is typically enabled by a switching matrix that connects subscriber lines either to G.fast transceiver ports or back to exchange or cabinet-based legacy services.
G.fast can coexist with ADSL and VDSL on adjacent pairs of copper wires by using frequencies above these technologies. However, operation above typical VDSL frequencies implies a minimum G.fast frequency of about 23 MHz, and this incurs a significant performance penalty on longer G.fast loops, beyond about 100m.
Both G.now and G.fast are candidates for Fiber-To-The-Building/Basement (FTTB) deployments, and perhaps for feeding small cells as they proliferate. For other types of deployments, G.now and G.fast entail a very high number of active electronic boxes in the Outside Plant, which raises concerns with regard to operational costs. G.now is available today, while G.fast is currently progressing through the standards. The first version of the ITU-T G.fast standard should be completed in December 2014, with some interoperability testing occurring in 2015 under the auspices of the Broadband Forum.
Read blog post: G.fast, MGfast, and Beyond