If one thing was clear at Broadband World Forum (BBWF) in Berlin last week, it was that nobody can afford to be religious about their choice of broadband technology any more. Fiber, copper, wireless – all of these will not just be used in parallel by the same operators, but will increasingly be converged under access-agnostic platforms to support network slicing, network-as-a-service and software-defined access.
In this scenario, some of the arguments about the survival of copper, and the pros and cons of G.fast, already sound old-fashioned. G.fast is not fiber, but it is hard to imagine it will not play a valuable role for many operators. While the UK’s BT and Germany’s DT have been early trialists, G.fast is about to be put through its paces in even more challenging conditions by Australia’s National Broadband Network (NBN).
The government-backed universal broadband initiative has been plagued with setbacks and sometimes hostile press since it was first announced in 2007. It pulled back from its original universal fiber objectives to connect many users with copper or wireless, a decision which has come in for heavy criticism.
NBN’s troubles have been well documented, but this week saw revelations of the true extent of its problems. The total budget sits at A$49bn (US$38bn), and of that, around 35% will be spent connecting the final 10% of users, including the cost of wiring to remote areas, erecting many of the required 2,600 wireless towers and, most expensively, putting two satellites in the sky. Last week, prime minister Malcolm Turnbull called the NBN a mistake and a “calamitous train wreck”.
Against this backdrop, the importance of G.fast delivering its promises of near-fiber speeds for far more affordable costs cannot be overstated.
At BWWF, NBN’s executive manager of corporate media, Tony Brown, rolled off a refreshingly honest reel of costs associated with the mammoth fiber-to-the-home (FTTH) scheme – spiralling expenses which led to the hunt for cheaper technologies, Brown cited examples such as deploying fiber to a single premise in Tasmania at a cost of A$91,000, or to a bowling green for A$86,000. Even some less demanding deployments have been disastrous financially, such as a housing estate in Brisbane where the cost per premises was expected to be A$1,500, but eventually came in at A$5,500.
In the light of this, NBN has slashed its initial plan of delivering FTTH to 93% of Australian households, instead opting for fiber-to-the-curb (FTTC), plus G.fast – for which US vendor Adtran is supplying its newly certified G.fast products. NBN also plans to spend an additional A$5bn on wireless in its new roadmap, A$3.8bn on fixed wireless and the rest on satellite, which is bringing 25Mbps speeds to the most rural areas of Australia.
While much of the press coverage has portrayed the shift as an embarrassing defeat for NBN, it can also be seen as a move to embrace an emerging technology which could recoup some of the losses, and which was not available back in 2007 (when much of the talk was of filling gaps in fiber with fixed 3G or 4G).
NBN plans to roll out G.fast next year, with an initial focus on MDUs and eventual expansion to FTTC. Using FTTC or fiber-to-the-node (FTTN) has accelerated deployment – NBN can use the existing cable TV HFC network and can connect 12,000 homes a week, instead of 7,000 with FTTH. An estate comprising around 25 houses with particularly poor broadband speeds recently had a micro-node installed to bring 25Mbps speeds to the neighborhood. The cost per premises was A$5,000, whereas a fiber deployment would have set NBN back A$15,000 per premises.
Granted this isn’t delivering the gigabit speeds Australians were promised, but G.fast, as well as DOCSIS 3.1, do provide the opportunity to get gigabit speeds at a fraction of the cost of FTTH. And anyway, while NBN is hoping that demand for ultra-fast broadband speeds will drive up ARPU, Brown mentioned that 84% of the FTTx footprint is opting for speeds of 25Mbps or lower – not quite the demand for gigabit speeds we might expect.
NBN achieved speeds of 600Mbps when it first trialed G.fast technology in 2015 on a 100-meter stretch of copper cabling installed 20 years previously. It also tested multiple passive optical network (PON) protocols from Nokia, getting a single fiber line to carry 102Gbps. The operator’s chief strategy officer, JB Rousselot, said: “Using G.fast means we will be able to deliver ultra-fast internet capability to our FTTC and FTTB customers, giving them access to transformative, ultra-fast services more rapidly and cost-effectively than over a FTTH connection.”
Adtran’s SVP of technology and strategy, Jay Wilson, said: “In applying G.fast, NBN is ensuring its users will more quickly realize the benefit of these services without sole dependence on FTTH. Since demonstrating the industry’s first fully sealed FTTdp solution in early 2010, Adtran G.fast solutions now supports open SDN principles ensuring rapid plug-and-play deployment capability within multivendor networks.”
As well as tapping Adtran and Nokia for G.fast services, homegrown vendor NetComm Wireless has been announced as another supplier for NBN’s 2018 roll-out.
Meanwhile, Broadcom’s approach to G.fast is starting to clarify, as it put out a chip last week that combines 212 MHz G.fast with DSL fallback and GPON. The idea is to make a single chip – and a single device with the chip inside it – to terminate fiber, as a FTTH CPE as well as either a VDSL or G.fast CPE.
Broadcom described its new offering as the highest performing integrated broadband CPE system-on-chip. The BCM63158 combines G.fast, xDSL and GPON into a single integrated device, enabling telcos to deploy copper or fiber broadband services with a single, unified platform, a move it insists will simplify platform management.
The chip is built around a 1.5 GHz multi-core CPU running at 15,000 DMIPS with an additional packet offload engine, and it is built to keep up with an 802.11ax WiFi chip. The emerging 11ax standard (see separate item) has full duplex channels of 5 GHz, so that outward and inward traffic can both run at full speed. It will also allow multiuser in both directions. So this Broadcom part will have to support something over 5Gbps.
Broadcom launched its own BCM43684 802.11ax chip in the summer.
“Combining both G.fast and GPON into a single device enables service providers to future proof CPE product offerings as telecom operators deploy both copper and fiber-based solutions,” said Greg Fischer, SVP and general manager for broadband carrier access at Broadcom.
“Further, the advanced processing capability of the BCM63158 enables secure, modern containerized software architectures along with the benefits of home networks supporting 802.11ax.”