NBN wireless latency ‘the same as 3G’


The National Broadband Network Company today provided further details of how its Ericsson-backed national wireless network was expected to perform technically, with chief technology officer Gary McLaren (see video above) stating the network was expected to perform at least on par with existing 3G mobile networks when it came to network latency.

NBN Co has previously declined to say what levels of latency can be expected from the fixed wireless technology it will use to service most of the small percentage of Australia that won’t be covered by fibre-optic cable, in the wake of concerns the technology might be a step backward from what is currently offered by ADSL2+ fixed broadband.

The ADSL broadband currently used by most Australians typically, according to network experts, delivers a latency of between 10 and 50ms, while fibre broadband delivers a faster response time — between 15ms and 25ms. 3G mobile broadband can deliver much slower latency — for example, recent testing in Sydney’s CBD of Telstra’s Next G network showed an average ping time of 85.5ms to local servers, using one of Telstra’s Elite USB modems.

The Long-Term Evolution technology being implemented by mobile operators like Telstra will also be used in NBN Co’s wireless network, and has the potential to cut latency drastically — as low as 10ms.
Speaking at a press conference today, McLaren said latency on the wireless section of the NBN would be “more or less the same” as it was on 3G mobile networks.

The executive declined to provide numbers until NBN Co conducted trials of the rollout. However, “it certainly won’t be any worse,” McLaren said. “We really expect it to be at least the same, if not a little bit better.”

McLaren also provided further information on the likely speeds of the wireless network, noting the 12Mbps speeds outlined under the Government’s NBN policy were peak network speeds, with some network contention expected to cut that down a bit in reality.

However, the CTO added, users could expect to see dramatically better speeds than were available on some mobile networks. “We will be engineering it so that at least 512kbps will be available for everyone, and we expect to see quite a bit more,” he said, appearing to refer to the portion of the backbone link to each wireless base station that each individual premise connecting would be able to access in a guaranteed way.

McLaren emphasised that the fact that NBN Co was optimising the network for fixed connections on the side of buildings rather than roaming mobile use meant that NBN Co’s wireless network would preform much better in general than comparable 3G mobile broadband networks. “So we’ll actually see very good typical speeds,” he said. “That engineering of the capacity is probably ten times more than many mobile networks today.”

A number of trials of LTE services have shown theoretical speeds of up to 100Mbps. But although McLaren acknowledged NBN Co would look at higher speeds in future, for the moment it wanted to focus on delivering 12Mbps speeds in a predictable and reliable way.

“That figure of 100Mbps or whatever it is is obviously highly conditional on how many people are actually in the cell and using it, how close they are to the cell,” he said. “That lucky user … it may be there for a period of time, but over time as the network grows, they’ll see their service degrade. We’re about predictability, quality of the actual service, over trying to get maybe get access to people at higher speeds.”

Video credit: Delimiter


  1. People who don’t know the difference between latency and bandwidth shouldn’t comment on NBN-related articles.

  2. As I said, this concerns me greatly. The fibre solution they have devised is awesome, but if they do such a crap job on Wireless.

    LTE technology is supposed to have latency on par with current ADSL2+ technologies, so to provide a service using a technology like LTE and provide latency as higher as current 3G technology, you’re doing it wrong.

  3. “We will be engineering it so that at least 512kbps will be available for everyone, and we expect to see quite a bit more,”

    wow, the future will be sooo fast…. almost as fast as ADSL in 2002…

  4. I am in the wireless footprint (JUST), so after reading about the UP TO 12 Mbps, the “at least 512kbps” of the NBN wireless and the latency of 3G, to say I am extremely disappointed is a massive understatement.

    Currently I have ADSL1 at UP TO 8Mbps, but in reality is between 0.5 and 1Mbps.

    How this NBN wireless is going to narrow the ‘digital divide’ is beyond me.

    Its now time to start convincing the missus to consider putting the house on the market before its value plummets in the future due to it not having a fibre connection.

      • Yes, I have already been in discussions with NBNCo about this.

        I am a software dev who works from home, so broadband is rather important to both my work and personal life.

        Where I live is not a town, I live in a recent new development of 150 blocks that is on the edge of a large regional town of approx 25,000 people (which is in the fibre footprint). The development I live in just falls into a different locality (with a different telephone exchange) which other than the 150 blocks of my estate is purely a farming area.

        Other than myself, I know virtually no one else in the neighborhood of 150 blocks would dip into their own pockets to pay for fibre. The local council won’t even spend any money on a playground out here, so the chances of getting them to chip in for fibre is nill.

        I can see myself meeting with my bank manager one day in the future, desperately attempting to get a loan approved to get NBN fibre extended to my area!

        My wife thinks I need a shrink, this NBN is becoming an obsession for me…

    • I wouldn’t presume too much just yet. Based on questioning McLaren this afternoon, I have a great deal of confidence personally that the wireless option will be very good; and certainly much better than what you are currently getting from ADSL.

      The thing is … McLaren and co do not think like mobile execs. They think like hardcore fixed telco ppl, and I assure you they want to give you a much bandwidth, and as little latency, as possible. My impression is they are intentionally being conservative ATM.

      • Thanks Renai, your comment has boosted me from the pessimism i was feeling yesterday.

        Fingers crossed that NBNCo are underpromising and will overdeliver on the wireless.

  5. “We will be engineering it so that at least 512kbps will be available for everyone, and we expect to see quite a bit more,” he said, appearing to refer to the portion of the backbone link to each wireless base station that each individual premise connecting would be able to access in a guaranteed way.

    Why would the 512kbps refer to the fibre backhaul? I was under this impression that the 512kbps was in reference to the sizing based on the number of premises per cell (ie, if they were using cells capable of 100Mbps, then each cell would cover no more than 200 premises). I would have thought backhaul from the tower would be a nonissue as even a single fibre would easily exceed the backhaul requirements.

      • I don’t think 512kbps is the backhaul limit, I think it might be the contention.

        It’s not ideal for a worst case connection, but still, considering it’s wireless, that’s pretty good.

        • I read/heard that as the contention…enough backhaul to guarantee 512kbps to each connected premise, but to provide 12Mbps under normal contended access.

          • Well, either way it’s contention. That’s the problem. Personally I hope it’s not contention when it comes to signal conflicting and low spectrum, because then it’s very hard to rectify.

      • I would expect microwave to be a last ditch effort.
        If a fibre connection is available within a single hop range, I would think they would go that way.
        More than a couple of hops and they would likely do satellite.

  6. So, they’ve now researched it and come up with – nbn over wireless will be crap for gaming. Wonder if the liberals (who i voted for, but dont necessarily agree on all things) will still spout garbage about how good a wireless nbn would be.

    • It is a pretty pathetic effort in terms of latency I agree, but it was never going to be good enough for gaming no matter what technology you used when it comes to wireless.

      • Not necessarily true … Ericsson told me this afternoon the LTE cells could do < 20ms latency. I suspect the killer here will actually be the microwave links. However, given you can get ~ 85ms average latency on a Next G cell with fibre backhaul at the moment, which is already getting close to gaming speeds, I think we will be very surprised by how good it will be. And McLaren & team seem very conscious of this issue ... I get the feeling they are determined to tweak the hell out of this.

      • As a demonstration, I can ping my iPhone on my wireless network (about 1 foot from the antenna) and get average RTT of 60.870ms and then I can ping my laptop wired into the same wireless access point through which the iPhone is connected, and get an average RTT of 0.307ms.

        Same network switching fabric/device, and completely different ping times – the difference is one is wireless and one is not. The biggest issue is that doing it wirelessly introduces a number of other transmission protocols that simply add latency to the stack.

        • I’m aware. It just annoys me that they set the bar so pathetically high on this issue. I hope that this is an absolute worst case, once in a blue moon latency situation, but if it isn’t, well, I consider that a waste of the several billion dollars we are spending to service these people via wireless.

          • In this instance, it’s more about distance from the tower – you can’t have everyone living directly under it, and the further you’re away from it, it worse it’s going to be. From memory NBN Co are designing cells of 7km or less radius – (which is not big) – and should deliver quite acceptable ping times in normal conditions.

            But in really high load situations, they’ll run into problems – but given we are talking about cells containing about 400 premises / 1000 people, “really high load” isn’t going to happen very often.

            Possible, sure. Likely? Not really.

          • @NightKhaos

            “I consider that a waste of the several billion dollars we are spending to service these people via wireless.”

            Well I bet the big two wireless providers Telstra and Optus don’t mind, they will just resell it with a increased footprint to their own coverages courtesy of the taxpayer.

          • No Alain, they won’t, because of the people who are getting this technology in the majority already have NextG or Optus Open Network coverage. It won’t increase their footprint.

            As I was saying to Michael below, they need to be painting a roser picture here. If it will provide better latency in most cases than 3G for the customers in the footprint, why are you telling us it’s on par (LTE has better to the tower latancy of about an order of magnitude).

            So if it doesn’t increase the footprint current 3G deployments cover, and it doesn’t perform better, why are we building it?

            The money could be better spent elsewhere in this case. There is no clear advantage to this unlike with the fibre part of the network.

          • “because of the people who are getting this technology in the majority already have NextG or Optus Open Network coverage. It won’t increase their footprint.”

            Huh? pass that by me again, the taxpayer is paying to duplicate existing wireless foot print why? and who resells NBN wholesale wireless, all ISP’s/Telco’s EXCEPT Telstra and Optus?

          • And I quote:

            …they will just resell it with a increased footprint to their own coverages….

            It isn’t increasing their footprint. They won’t therefore resell it will an increased footprint. They will of course be reselling it in general.

  7. And predictably, the comments are filled with people complaining about how this “isn’t good enough”.

    What did you expect? It’s wireless. If it was “as good as” fibre, it would be used everywhere.

    If you think 80ms (for example) is poor latency. 80ms is not going to make any noticeable difference whatsoever to video conferencing, and 90% of the time gaming either. I suspect most of the people complaining, are comparing to their current experience with overseas servers, which will not change one bit.

    • I really think we haven’t seen the full picture yet about what sorts of levels of performance the wireless network will deliver — and I personally think it will be solid.

    • I don’t think you quite understand the pretense of my objection here. Current generation wireless technology (LTE) has better latency than what currently deployed technology (3G) by almost an order of magnitude. If you’re using this technology in such a way that you don’t take advantage of this improvement, then there is something very wrong with your engineering decisions.

      This isn’t some “I want to be able to game and video conference” or anything like this, this is questioning the competency of the engineers, and also the pretense of the Wireless part of the NBN as they are saying it will not be able to outperform currently deployed technology in terms of latancy.

      With that in mind the wireless networks of Telstra and Optus for people out in the sticks present a very real competitor, so much so it brings into question the viability of deploying the fixed wireless infrastructure.

      You compare the proposed footprint of NBN wireless with NextG and the Optus Open Network, and find me enough people to justify the multiple billion dollar expenditure on this new network if the performance they can give you is only just slightly better?

      Now, granted, they could have just been playing the devils advocate here and telling everybody the worst case to cover their arse, and the technology will perform far better than 3G deployments, however that is not a good idea when you have an opposition with a stated goal of stooping the project breathing down your neck.

      If I can find an objection here, so will the opposition. So, the latency isn’t good enough.

  8. I wonder what will happen to the many people who continue to require voice communications. For example, the latency expected in the NBN satellite serviced areas probably means that a VOIP type call wouldn’t work if you tried to call another NBN satellite service user. Satellite latency averages about 280mS. Latency in fibre cables is about 10mS per 1000Km I believe. A typical VOIP call routing of this type would be:

    1. Calling customer via satellite to NBN earth station about 280mS
    2. NBN earth station to ISP (anywhere in Oz) on fibre @ 10mS per 1000Km
    3. ISP to VOIP providers switch on fibre (somewhere in Oz) @ 10mS per 1000Km
    4. VOIP providers switch to called customer ISP on fibre (anywhere in Oz) @ 10mS per 1000mS
    5. Called customer ISP to NBN earth station on fibre (anywhere in Oz) @ 10mS per 1000mS
    6. NBN earth station to called customer via satellite about 280mS

    The delay will be horrendous. Some people may try to use VOIP for social calls but you couldn’t conduct business over links like this.

    Now it seems that latency in the wireless serviced areas will be about a third of that in satellite serviced areas. A VOIP call will still be routed in a similar way so latency won’t be much better.

    • NBN Co will not be offering “native” voice services over the “last 7%” of coverage – (ie: wireless and satellite). In those areas, the copper will be kept – meaning standard voice services and even ADSL will be available in those areas.

      That doesn’t mean you can’t run internet VoIP – (eg: Engin, Skype, etc) – over your NBN wireless or satellite service, it’s just that NBN Co have – (correctly) – deemed that the latency involved – (particularly for satellite) – is not compatible with providing a quality voice service on those technologies as part of the NBN.

      We must remember, the wireless and particularly the satellite components of the NBN plan are in many many cases about providing people with decent broadband access who can’t get anything at all right now without paying a huge amount for an existing satellite service – (eg: IPStar). Many people can’t get anything but dialup, and some can’t get that reliably.

      Under the NBN, everyone will get 12Mbps of DATA, whether it be fibre, wireless, or satellite. Fibre covered premises will mostly get a 150Kbps/150Kbps channel within one of their services – (remember, and NBN NTU can do six simultaneous services) – dedicated to voice, that is QoS’d right across the network.

      • Michael – I am wondering how adequate voice comms will be maintaned. We know that much of the old copper is in poor condition and in most of remote Australia radio services are used. This equipment won’t last forever. Where the copper is replaced by fibre we won’t have a problem. NBN wireless and mobile coverage provided by Telcos will probably overlap. In very remote places it will never be economic to install mobile bases. This is where continuing a voice service will be difficult.

        Already Telstra loses money on providing a universal service. This was fine when it was owned by the taxpayers but a private company won’t continue to do this. USO Co. will contract Telstra to continue existing services but where will the capital come from for the upgrades that will inevitably be required. To many voice comms is more important than broadband

  9. Personally, I expect he “stuffed up” and meant to say 4G rather than 3G when refereing to latency.

    I cannot understand why Erriccson LTE to fixed points would have significantly worse latency than Erricson LTE to mobiles that are already in service and being in CBD areas, in much higher contention than country cells will be.

Comments are closed.