The National Broadband Network (NBN) is one of the current government’s infrastructure investment planned to take Australia firmly into the online economy bridging distances and bringing products and services to everyone with an internet access. Still, it has its critics. Get to know the debate about this.
Its a shame the Philippines had a similar NBN plan but got cancelled after being caught by the reports of graft and corruption.
Fibre vs HFC: Understanding the NBN technology debate
With the NBN debate continuing to make headlines, arm yourself with this breakdown of Fibre, HFC, wireless and ADSL.
This week Communications Minister Stephen Conroy attacked the Coalition at the National Press Club, saying that the Coalition’s criticisms of the NBN were flawed. In particular, he argued that relying on HFC cable networks as part of the National Broadband Network (NBN) was a “dead-end solution”. [Ed: read Shadow Minister for Communications and Broadband Malcolm Turnbull’s rebuttal here]
One issue the comments have raised is that of shared broadband. It’s a big honking asterisk that hangs over any discussion about HFC or 3G/4G as an alternative to fibre.
Okay, so what are our options?
To get a clearer idea of how these different technologies stack up, here’s a breakdown of the different speeds available with each:
- ADSL 2+: 24Mbps maximum, degenerating at range. Dedicated.
- VDSL2 (for fibre to the node): Most likely in the 60-80Mbps range, although technically it has a 250Mbps maximum. Dedicated.
- 3G HSPA+: 42Mbps. Shared.
- 3GPP LTE (sometimes called 4G or 3.9G): Telstra’s recently launched services works at around 100mbps with practical speeds (according to Telstra) of between 2Mbps and 40mbps. Theoretically, LTE can go to 326Mbps in the future. Shared.
- LTE-Advanced (true 4G): the specification isn’t finalised, and rollout is quite a ways away, but it’s expected to support more than 1Gbps per tower. Shared.
- Hybrid fibre/coaxial (“cable Internet”): 100Mbps currently; the DOCSIS 3.0 specification supports speeds up to (and beyond) 300Mbps and even the NBN corporate plan acknowledges that 240Mbps is possible on the current network with HFC node-splitting. Shared.
- Fibre to the premises: 1Gbps now; unknown upper limit (terabit speeds are certainly possible). Dedicated.
What’s that dedicated/shared thing about?
Wireless and coaxial networks are shared mediums. That means that the bandwidth is shared between every user connected to the same cell or node. You only get the listed speed if you’re the only person using the connection. If other people nearby are also using the same network, then you only get a share of the total. If, for example, you have 100 people on the same node using 100Mbps HFC, then a user’s average connection speed is only 1Mbps.
It’s even worse with wireless, as a good chunk of the bandwidth will be lost to noise and signal degradation. A 100Mbps 3G service might only deliver, say, a total of 50Mbps to its collective user base. That’s why Telstra claims a speed of between 2 and 40Mbps for its new LTE services, rather than the theoretical throughput.
This is not entirely unsolvable. One can reduce the number of users per node or cell by installing more nodes and cells. However, with HFC cable, subdividing loops is not a trivial thing, and that’s on a network that was pretty much stopped at 30% population coverage. With wireless, you could potentially have a cell tower on every corner (which isn’t going to be popular), and even then you run into diminishing returns as the mobile signals overlap.
Why does the local loop matter? Isn’t it all just shared deeper in the network anyway?
To a certain extent this is true. Eventually, somewhere on the network, you’re going to be contending with other users for limited bandwidth. The connection from your local exchange or aggregation point to your ISP won’t have enough bandwidth to support everybody running their connection at full speed; your ISP’s connection to the Internet will not be fast enough to support all its subscribers accessing the Internet at once; the Internet itself has a huge number of potential bottlenecks.
But the big difference is that this is fixable. If the link between the ISP and aggregation points or the ISP and the Internet is insufficient, then the ISP can provision extra bandwidth to improve their contention ratios. As a user you can switch to another ISP. You can’t just switch local access technologies.
To put it another way, a fast local connection removes one of the potential bottlenecks to Internet performance. Yes, it’s not the only bottleneck, but it’s the hardest to correct.
Tell me about FTTN
The Coalition strategy has largely crystalised around fibre to the node (FTTN) for the majority of the population. FTTN stretches fibre to street-side cabinets, with the last few hundred meters to be handled by the existing copper networks. It’s designed to solve the range problem of DSL (which gets slower the further away you are from the exchange).
The Coalition’s plan is to “incentivise” private companies to build such a network. According to the Coalition, that will provide practical user speeds in the range of 60 to 80mbps using VDSL technology.
What are the technical merits? Conroy said that 60 to 80Mbps was unfeasible with Australia’s current copper network and would require bonded pair copper wires (basically using two copper wires instead of one to carry the signal).
However several countries, including the UK, have plans to increase VDSL speeds to 80 and even 100Mbps using a technique called vectoring, which mitigates crosstalk on the wire. Tests by Alcatel have shown that 100Mbps at 400m from the node is possible.
The downside of FTTN is that it has very little headroom. In this writer’s opinion, it’s really another stopgap measure to keep us ticking over for another decade or so. Maybe new technologies and techniques will be found to squeeze even more bandwidth out of copper, but in the long run it just can’t compare to fibre to the premises.
The great thing about fibre is that it has so much headroom – we’ve yet to find the upper limits of its capacity. Although it’s being rolled out in Australia with service offerings of 25Mbps, 50Mbps, 100Mbps and 1000Mbps, fibre can actually surpass 1,000,000Mbps (that’s one terabit) on a single cable. As reported here here andhere and here, for example. Obviously those are backbone applications right now – but it shows that fibre is indeed capable of terabit (and a good deal more). For practical purposes, terabit has been demonstrated in the field as early as 2005. Right now, we already have 10Gbps fibre products commercially available.
With fibre, the current limitation is really in the signal processing electronics, not the cable itself. Better processors can add extra wavelengths and increase the carrying capacity of a fibre cable. If we want to upgrade fibre in the future, we just need to replace the electronics; the cable itself can stay.