4G: Pretty fly for a wi-fi

Given that I was using a wireless connection at the time, I assumed — as you do — that it was going to take many, many minutes to complete, if it was going to work at all. So you can imagine my surprise when, cappuccino in hand and feet up on the opposite chair, I glanced at the screen a few seconds later and saw the progress bar sailing across the screen with all the fervour of Julia Gillard chasing after good poll numbers.

It was truly epic: 24MB uploaded, via wireless no less, in 12 seconds. That's 120MB per minute, 2MB per second, or an upload channel of around 18Mbps — many times faster than the upload channel on my good old fixed-HFC service, which strains to exceed 0.5Mbps at the best of times and requires many minutes to upload the same file to the same FTP server.

Was this validation that 4G wireless can be every bit the speed demon that Telstra and others say it is? You betcha: After months of lacklustre wireless performance, particularly my unfruitful efforts in the coverage abyss along Bourke and Collins Streets, it was a stunning result (and yes, I did repeat the upload to see if it was just a fluke; it was not).

For a few moments there, my faith in the future of wireless was summarily restored — until I went online at another location and reverted to 3G service, with its flaky performance and snail's-pace uploads. No matter the bursts I was able to get at Fed Square, I remembered that this is the reality of wireless across the majority of the country: 3G at best-effort speeds, which are generally better than yelling really loudly, but aren't, in the scheme of things broadband, really that great at all.

While it was nice, this experience didn't quite convince me — as the ever-amusing Alan Jones believes — that wireless is the future of the NBN. But it did show that there is a valuable role for 4G if — and this is a big "if" — it can get the regular and predictable capacity to allow good coverage to become the norm rather than the exception.

4G's role in the long term will be as a way of ensuring that we can still access NBN-like speeds, even when we leave our homes. The rollout will be complementary, so we will no longer have to plan ahead to compensate for 3G's lousy performance.

Where I envision 4G's role in the long term will be as a way of ensuring that we can still access NBN-like speeds, even when we leave our homes. The rollout will be complementary, so we will no longer have to plan ahead to compensate for 3G's lousy performance, particularly when we have the gall to upload information.

There are two points I'd like to make about 4G.

Firstly, although equipment makers used to talk about building 3G capabilities into laptops, tablets, and the like, the subsequent move to 4G shows the problems with inbuilt technology. This is particularly the case with Telstra and its competitors pursuing non-standard 4G in the 1,800MHz band — and a perfect reason why portable wi-fi hotspots, which share a 4G connection between laptops, iPads, smartphones, and any other device, have rightfully become the de rigueur method for getting online.

Not only do they shift the burden of connectivity away from the individual devices, but they can be easily replaced or upgraded — allowing Telstra and its rivals to deploy 4G services in the 900MHz or any other band they like. As their portfolio of 4G-capable spectrum expands, it will be easy to roll out new wi-fi devices and let customers keep the same laptops, smartphones, and tablets they've always had. They'll work just the same — only a lot faster.

In other words, the shift away from built-in 3G or 4G wireless has proved to be a smart move for frequency-hopping carriers. It could also spell disaster for this month's 4G spectrum auctions, which were originally predicated on there being no real alternative spectrum to support viable 4G services.

Secondly, while I copped a bit of flak for suggesting that it wasn't all bad that Telstra might boost overall service performance by throttling back unchecked BitTorrent downloading, it seems to me that the emergence of wireless services makes deep packet inspection (DPI) even more appealing.

I say this not because carriers need to prevent users from choking wireless networks with BitTorrent; wireless data charges are still way too high for most downloaders to hog 4G bandwidth. I say it because tomorrow's networks are going to need a way to adapt to the different kinds of traffic we are regularly uploading and downloading.

Say you go to a concert, and like the thousands of people there brandishing your smartphone, you record a video that you want to share with friends. You start uploading it, and the phone crawls along as your poorly-balanced, asymmetric data connection contends with the thousands of other people doing the same thing.

In the long term, carriers are going to have to figure out new ways to better match users' activities to the available bandwidth.

Wouldn't it be great if the network could recognise that you're uploading a video, and temporarily reverse your connection so you get a fast and furious upload channel with slow downloads until your upload is done?

Doing so would require the network to be peering a fair bit more closely at the type of traffic you're carrying, which — if the reaction to Telstra's P2P filtering is anything to go by — will have the civil libertarians crying bloody murder. Yet in the long term, carriers are going to have to figure out new ways to better match users' activities to the available bandwidth; given high-speed download streams accessible anywhere, it's only a matter of time until some joker starts saturating the service. Heck, it happens to HFC all the time.

While I was able to get a good upload channel from my perch in Fed Square, this was partly due to 4G's ability to better interweave multiple data streams than 3G can — and because few others in the area would have been using 4G.

If 4G were the default wireless connectivity mode rather than the exception, I rather suspect I would have had the same congestion issues that I and everybody else experience in the CBDs of Melbourne, Sydney, and elsewhere. And this, sadly, would irrevocably compromise the potential of 4G — not as an alternative NBN, but as the mobile extension of whatever NBN is implemented in the future.

What do you think? Would you support better content intelligence to ensure strong 4G performance? Or are we always going to have to put up with wireless broadband as a best-effort network topography?