Before I start, let me just say I’m not an engineer, so if I get some of the technical bits wrong, forgive me. But I genuinely think I’ve got the ramifications correct!

There’s been a fair amount of debate about how NZ gets better faster broadband. This isn’t a simple proposition and much of it has been made even more complex because of politicalisation, poor press leading to general acceptance that Telecom New Zealand is ‘bad’ (just look at the sensationalist title to this), ignorance and a real clouding of issues.

So lets separate out the issues. Jim Donovan provides a nice synopsis of the issues as Rod Drury sees them. In my opinion there are a lot of issues tied up even in this.

Firstly can we accept the fact that Telecom is a public company and as such will try its best to make a profit for its shareholders, (which incidentally includes just about everyone in NZ with a managed fund, the point being you are doing yourselves out of your retirement dosh!). The implications of this are that they are legally obliged to invest their shareholders money to get the best return possible…which may not always be what people think is ‘good’ for the country as a whole.

Having multiple international trade routes is different from national data speeds. It’s a damn good idea and should just happen ala Google.

Next, lets address national data speeds. There’s a lot of apples with pears comparisons with other nations out there. These nations don’t have our population levels or distribution, so doing this isn’t that valuable. Benchmarks are good, but how about making them meaningful (as opposed to a political weapon).

Lets look at broadband. Everyone is talking about fast ‘broadband’, but what they really mean is fast internet connectivity – give me the webpage I want faster, upload my stuff faster…. Semantic, but important difference.

Broadband in common use means an ADSL connection. ADSL is effectively compressed data travelling down the same copper as you use for a phone line. Broadband in other countries means other things like cable, Fibre and other technologies. The whole broadband thing became quite topical when the government noticed that people might be pointing the finger at them for our slow movement down the OECD averages for income and wealth. They also noted that the countries moving up that list had higher penetration of faster internet technologies than us and viola, it became an election issue.

If we take a stick to Telecom, make them the bad guy people won’t point at us anymore. This conveniently neglected other things like company tax, RnD, incentives and funding for start ups, tax breaks for international tech companies etc. Anyways, the end result was that in a knee jerk reaction to get the monkey off its back Telecom agreed to invest in its fixed line business to the tune of $1.4bn. The point of this investment is to shorten the loop between the point where the core network (fibre) stops and the copper (last 2km – the bit to your home) starts. This shortening reduces the distance impact and will provide greater speeds (10mbps) to the home and …. no one will be happy with the result. It still won’t be fast enough (here for religious screaming from the left, here for business impact).

This type of disappointment will mean more Telecom bashing, closely followed by the realisation that the loop distance needs to be reduced to about 800m. That means more cabinets (I’m told 3 times as many, and another $700mill). And get this, this additional investment will only give data speeds of up to 20mbps… more disappointment !!! Which gets you to the point where you realise you need Fibre to the home. I heard that the last time Telecom looked at that it was going to cost circa $10bn to deliver. So that’s just not going to happen (as Paul Reynolds has already said).

Or, you try a different approach.

Think about the world we live in, most people have telephones they carry around with them, notebook sales outstrip desktops, pda’s, ipods, WiFi in the home… its all about untethering the cable…or simply being mobile. Why then are we trying to solve the issue with redundant technology when the requirement is just for fast internet speeds.

Here’s my hypothesis. The network you think of as mobile, isn’t really mobile. It’s a fixed backhaul network with cell towers attached at the end to deliver the last mile. If you think about it, its identical to your home WiFi network and we love those!

Why then don’t we leverage the fibre to the cabinet programme, turn all those cabinets into a cell site and start thinking about fast mobile data technologies like LTE. I know its not a ratified standard (others are though and they are quite quick!), but it is already demo’d as doing 150mbps. That’s 15 times what cabinetisation is going to deliver to the home…. Isn’t it about fast internet, not copper or fibre. I know there is a pricing issue, but scale will bring that down….


  • The problem is that LTE is still some years away from a commercial reality, and it changes quite a few things thoughout the network including making VoIP to the handset required (under SAE).

    Apart from not being ratified, LTE requires heaps of spectrum to get those speeds and the speeds are the banner headline peak theoretical not what someone in a hilly suburban area will get.

    This is the current state of LTE hardware. Bleeding edge. HSPA battery life is still evolving, which is why the iPhone is not yet 3G (so says SteveJ anyway, I believe him). GSM followed the same pattern and I have no reason to expect LTE will not also be the same.

    For NZ we have a significant investment in UMTS/HSPA by both incumbents, so they’ll be looking for a ROI. They’re also using the radio spectrum required for LTE for other things, and we’ll fall back on the “2100 is crap, 900/850 works better” problem that is being worked around now by UMTS850/900. Getting 20MHz of spectrum per carrier is hard; that’s about the total of what a carrier has now in total for UMTS+GSM and you end up needing multiple carriers per urban site for capacity. Spectrum is a scarce resource.

    Mobile is an important facet and might replace the last mile (see femto-cells) for many users, but wires and fibre are always faster.

    I’m still struggling with what people are going to use >10MBs for. No one can answer me properly; the 24MB/s for video conferencing is, frankly, rubbish given what is done today on far far less. Video conferencing is, after 30+ years, still pretty shocking. I don’t think we’ve solved the UX problem with it; one on one is easy but with more than this it becomes impersonal, error prone, and people on the far end lose out to those who are physically present.

    There has been no killer application identified in countries that have fast internet access. The PirateBay for Sweden is not a good example. It does make things more flexible when you can run a video editing business from home but this isn’t a common need.

    If you want fast internet access for your business, site your business appropriately. You may not be able to run it from home, but I can’t see this as different from other enterprises. If you want fast internet at home then make this a factor in your purchase/rental decision (I did). Just as you may choose to live next to an airport, you may choose to live in an area with crap internet coverage.

  • Issues:

    1) Planning permission for the thousands of additional cell towers.
    2) LTE has an optimal cell size of 5km.
    3) Since these are essentially microcells, the amount of base stations/switches would be impressive.
    4) You still have bandwidth problems because while the demo may show 150mbps, that’s probably with 1 device consuming all of the spectrum in the cell.
    5) Assuming 175Mbit/s max for the cell (wikipedia, from handset speeds), and according to this geekzone discussion:
    You can assume ~200 homes per cabinet.

    That’s 1Mbit/s per house. You’re better off with DSL. Even if you take it to 326Mbit/s which would be using the 4×4 antennas (again from wikipedia), that’s still only 1.6Mbit/s per subscriber.

  • Hi Jason & Bwooce,

    As I said first line, not an engineer but I’ll have a crack at some of this to defend the hypothesis.

    1) The RMA needs radical overhaul
    2) LTE covering 5km is still longer than the current planned cabinet density (2-2.5km), Granted that in mobile there is a contention issue. But how often are all 200 homes going to be online at the same time?
    3) Battery life is a mobile handset issue and getting better. Notebooks are much better…I see your point, you’ve just gone a step further than I planned ie using mobile broadband to be mobile (nice!)
    4) The standards and spectrum thing is outside my knowledge. I know they take time… can’t comment on what they are using it for but love to know more
    5)Femto’s are interesting. I personally think they’re great but when you’re average home owner realises that it’s basically like having a cell site in the home…well the Nimby phenomenon kicks in
    6)Bwooce, I’m with you what are you going to use if for. I’ve asked the same question here on Ben’s blog and got nothing back. Faster is a convenient extravagance… the reality is that there are SHEDLOADS still on dial up (who could be on ADSL) because for them that’s good enough

  • Falafulu Fisi |

    Paul said…
    The RMA needs radical overhaul

    No. Not radical overhaul Paul. It needs to be completely killed. Here is an excellent article of why it should be killed.

    It’s time to drive a stake thru the heart of the RMA

  • 1) Except it’s outside of the problem set here, and therefore part of the environment, not the solution. Fixing the RMA would make pulling cable a lot cheaper too.
    2) Very frequently between the hours of 6PM-midnight, and it will only get worse as take-up grows. You’ll have to find someone who knows what TNZ’s provisioning multiplier is. 🙂

    When we are talking about 300mbps, it doesn’t take many people with ADSL2+ usage (at 25mbps) to fill it up. 12 in fact. That’s 6% of the population covered by the cell. Since it’s pretty reasonable to assume that at least 5% of the customer population will be high usage people, there won’t be enough bandwidth to meet everyone’s expectations. In other words, you won’t be able to deliver better than ADSL2+ to enough people to make it worth the additional cost.

    Woosh is learning the hard way that fixed mobile broadband is a hard sell. It is really for individuals who can’t get wired connections. Even with LTE, it will still be for that niche.

  • A couple of things:

    Broadband over the local loop is not quiet as bad as you are making it out to be. For example:

    1) ADSL2+ does delivers 15 – 24 Mbits/s at 800m and 12-16Mbits/s at 1.6km. (Depending on cross-talk and loop quality etc.)

    2) The natural evolution after ADSL2+ is VDSL2. VDSL2 is almost here and will deliver 30-70 Mbits/s at 800m and 16+ Mbits/s at 1.6 km.

    In other words, there is plenty of juice left in the local loop. Of course, the loop length needs to be shortened to accommodate the above speeds. (Got an article on my site covering ADSL and ADSL2+ vs. loop length, if you are interested.)

    Also, Bwooce is asking what 24Mbits/s will be used for. Well IPTV and video of course! (To start out with.) In other words, triple play. Each simultaneous standard definition channel will use 2-4 Mbits/s and each high definition one 6-12Mbits/s. That bandwidth will be used up! (Reminds me of the old 640kb statement for PCs. Sorry.) Wait until YouTube delivers HD… Also, I am sure there are many bandwidth hungry applications coming, that we have not even thought about today.

    IMHO, wireless will never be able to compete with wired for bandwidth and quality. And of course, wired will never be able to compete with wireless for mobility! The wireless speeds that are promoted by the vendors are mostly the theoretical shared “sector speeds” that will be used by many subscribers. This is not what each individual subscriber should expect to achieve consistently. And then we have the interference and noise to deal with in any wireless solution.

  • A few points (and I hesitate to say that I was a Mobile Engineer in a past life)

    1. If you want to cover 5km radius you’re going to need towers that are about 15 – 20 m high to get above the clutter. I note that Vodafone and Telecom are now deploying towers with lower heights in suburban areas (on light poles about 8m high – good shared used of infrastructure) for the reason that they are controlled activities under the RMA so relatively easy to get permission for.

    2. Laws of physics apply – wireless will never be as effective as wired in terms of bit/customer. It’s shared bandwidth at the access end which is a physics constraint. Constraints at the DSLAM are engineered and can be increased significantly from currently used business cases permitting.

    3. Really important to make a distinction between access speed and capacity. Faster access speed will improve broadband customer experience. Greater capacity (ie the ability to actually fill 24 M/sec consistently) allows different services – in particular IPTV. There are very few people who would pay even current TV rates in NZ to support simultaneous streaming of 2 x SD channels and 1 x TV channel.

    4. With fibre to the cabinet it is possible that DSL cabinets have better backhaul that cellsites in some cases – probably out of date but most sites were on n x 2Mbit streams, with n being up to 4 for a high capacity site….

    5. People dont use all the bandwidth all the time – so to Jason’s point you can probably support 10 times the amount you suggest. Still not 100% and I would concur that the current challenges faced by Woosh would not go away under this model.

    6. There are significant stability benefits with being wired, particularly with IPTV. Even with 802.11 n , streaming video within the home with the reliability we have today is really problematic. meeting the same level of quality we have today (even with digital satellite) is a serious challenge

    The key point to this post is that there is no easy way to make ubiquitous fast broadband happen, not that wireless is not the answer. Either way there will be pain in terms of either huge wedges of cash with long investment retunrs or changes to the NZ landscape with more radio sites. It’s time to bite the bullet, pick one of these options and just go for it. FibreCo is probably as good an answer to this as any.

  • Oh – and I forgot one other point.

    Check out PCCW in Hong Kong if you want to see an example of how fast internet is used. PCCW have managed to successfully disaggregate the pay TV providers and provide SD and HD IPTV.

    Works really well.

    Of course they have the following advantages

    1. They have large number of multiple dwelling units that they run fibre to.

    2. They are geographically quite small and densely populated

    3. Their Pay TV model was quite poor and PCCW were able to negotiate some very good content deals as a result

    4. Their IPTV business unit was not part of their broadband unit (whose strategy was clear – build great broadband everywhere)

  • All,

    Some fantastic debate here. Really good to see so many people engaged in solving the issue. I’m going to refine the hypothesis and come back… hope you will too

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.