Melbourne 4G Shootout: The 3G Reality

Melbourne 4G Shootout: The 3G Reality

Optus officially launched 4G services in Melbourne last Saturday, providing a rival to Telstra’s existing service just ahead of an expected flooding of those networks with 4G-equipped iPhone 5s. Comparing the performance of those services generates some interesting numbers, but just what should you compare?

Picture by Adam Sellwood

Last week, I wrote a brief post comparing the mobile broadband performance at Sydney Airport Terminal T2. As it happened, on the day I did the testing, my Telstra 4G hotspot couldn’t find any 4G services (even though the airport was one of the first locations covered in Australia when Telstra launched), and so “failed over” to the 3G network. Optus’ device could find the Optus 4G network, so I reported the numbers from each.

That led to several readers saying my comparison was misleading and wasn’t a real-world test. Similar objections were raised when I compared the recently updated 4G networks in Sydney train tunnels, where my Telstra device found 4G on a northbound trip but not on a southbound one.

Asking for an “apples to apples” comparison might be accurate in a purely technical sense, but to my mind it doesn’t reflect a much more important point: service availability and how these products are actually used. No-one buys a 4G hotspot to use purely with a 4G network, because we aren’t even close to having universal availability of 4G. Telstra has much broader coverage than Optus right now, but even it only claims to cover 40 per cent of the population. Anyone using a 4G device will end up on 3G a lot of the time, even when they’re in the same locations. Part of the reason telcos have deployed 4G is to reduce 3G network saturation, but that doesn’t guarantee 4G all the time.

Equally, if you have invested in 4G, you want higher speeds whenever possible. So it’s worth knowing exactly what speeds you’ll get, as well as running real-world checks on whether coverage maps reflect reality.

Nonetheless, for the sake of argument, I ran 4G-only tests in the one location I encountered on a trip yesterday across Melbourne that actually offered 4G from both providers: Southern Cross Station. I checked when I landed at Melbourne Airport, but (as expected) Optus isn’t offering 4G there. I checked in the south-eastern suburbs where I’m staying, which is on the very edge of Optus’ 4G coverage map, but got 3G for both Optus and Telstra.

Even Southern Cross, right in the heart of the CBD, was a challenge. At the northern end of the station, I could only scare up 3G on the Telstra hotspot. Further south, both had 4G, so that’s what I tested. The result was a convincing victory for Optus in terms of download speeds. (For download and upload speeds, higher is better; for ping scores, lower is better.)

Provider Ping (ms) Download (Mbps) Upload (Mbps)
Telstra 58 1.09 1.46
Optus 73 11.33 1.16

It’s worth pointing out that those 4G download speeds from Telstra are slower than the DC-HSPA+ speeds I saw from it in Sydney the week before. A weak connection to 4G won’t work as well as a solid connection to 3G.

To forestall another argument: yes, Optus’ network is much newer and has fewer customers, so logically you might assume there be less demand and hence higher speeds. But that won’t always be the case either. When I ran tests on Optus’ trial network in the Hunter Valley earlier this year, Telstra still came out with faster speeds even though it was a full commercial network, while Optus’ was only being used by a small group of testers. That underscores a key reason to do testing: to see if the assumptions made about network performance are actually valid in specific contexts.

I’m in Melbourne for some time, so I will do some more testing to satisfy my own curiosity. But the big lesson of all the 4G testing I’ve been doing recently, the first tests in Melbourne included, is already clear: 4G networks can offer a significant speed increase, but it’s not guaranteed. As with any mobile network, you’ll see variable performance based on location, time of day, weather, the number of people around and other factors. Remember you pay extra for 4G hardware, but not for the actual connection. That’s not surprising: if payments were based purely on having 4G available, it would be a very tough sell.

Under the circumstances, having a 4G-capable device remains preferable to having one that’s 3G only. You’re already competing with hundreds of thousands of people for 4G, but with millions for 3G. Just don’t buy one on the assumption that you’ll always see 4G. You won’t.


  • Hate to be a negative nelly but one set of data points from a single location is hardly a ‘shootout’ now is it? Would be great if you could add to this article over time to give people an approximate idea of speeds they can expect on different networks in popular locations.

    • Hi Ed, As the article notes, I’ll be doing more testing in Melbourne over the week (will add pointers into the article but might not change the main content).

  • Just pointing out, you’ve listed both columns as “Download (mbps)” when I suspect the right one should be “Upload”. Also, a funny test would be to trial them both at the Telstra building on Exhibition Street.

    Also, Southern Cross Station isn’t really “right in the middle of the CBD”. Perhaps around the corner of Bourke and Swanston Streets would be more “the middle”.

    I’m quite curious to see the results of the rest of your comparison over the week.

      • Angus – re 4G networks, why is there such a focus right now about the “network is much newer and has fewer customers, so logically you might assume there be less demand and hence higher speeds”… Is 4G so technologically different from other networks that more customers means slower speeds? Is there not the same equivalent issue with 3G (or “NextG”)?
        Or is your point that 3G has reached saturation point so whatever speed you get right now is a good representation of what you’ll always get, so 4G speed right now is not indicative of future performance?

  • If you want an interesting area for comparisons, try Richmond station or South Yarra at peak hour. Most days I’m lucky to get any data, even though it shows full reception. Telstra on a 4S.

    • Most of Melbourne CBD is dead during business hours if on Telstra 3G. Since the iPhone 4S launched, I’ve had nothing but trouble with 3G data in the CBD. In fact to the point where I don’t even bother trying to refresh twitter anymore.

    • So it’s not just me, sometimes I wonder whether it’s to do with coming out of the City loop. But invariably I am unable to get any data throughput from Telstra at all near Richmond station.

      I come in on the Sandringham line each day and as soon as the train hits around Balaclava don’t expect any real data throughput, makes working on the laptop tethered to the iPhone 4S useless.

      Hoping with the iPhone 5 and LTE I will be able to get a little more throughput but not expecting miracles.

    • As you’d see from some of the previous tests I’ve done, I don’t deny it _can_ be much faster. However, it can also be slower. “I’ve seen it faster therefore it must always be faster” is not a reasonable conclusion.

  • I’m based in the Docklands and have very turbulent Telstra reception (although it has improved somewhat in the last couple of months).

    Southern Cross Station however is always awful, to the point where I don’t bother trying to get on the Internet via my phone there at all, generally by the time a page loads my train has rocked up.

  • Telstra isn’t worth testing in the Melbourne CBD. No matter what your phone connects to, it will be so brain numbingly slow that you’ll have flashbacks to dial-up. I can almost hear the blips and bings now…

  • Telstra 3G sucks in the Melbourne CBD, however they are saying that 5 new base stations should be live before the end of the year. By then, the SGS3 and iPhone 5 will both have a lot of customers and will mean a lot of customers off the 3G network, so the two should combine pretty well to give us 3G users decent data performances again.

    We can hope anyway.

  • Angus,
    Thanks for these updates to the original story.
    As a fellow SE Suburbs resident, my daily commute goes via Camberwell station to one of the City Loop stations, depending on whether I’m at a client or in the office.
    I have a 2 year old android device on 3G with Telstra. Southern Cross to Parliament seems usable at the stations, but not in between. Richmond to Glenferrie is often unusable in the evening. Richmond is also about where the phone reception regularly dies in the morning heading in. This is the cause of some of my rants about the crapness of the RSS feeds (I suggest you try using google reader on a pre 9am commute and see how the experience changes your mind about the usefulness of RSS).
    Bourke and Elizabeth goes dark at about 5:30pm for 30-45 minutes, and from about 12:30 – 1pm.
    I wonder whether it would be possible to get some speed tests done on several of the train lines on say the SE Suburbs lines, and around the city loop in peak times, where more that every second person is on the phone reading some page or other.
    Speed correlated with GPS and time, as well as multiple readings / journeys would be cool. I wonder whether there’s an app for that?

    I’d be really interested in seeing you take some care with methodology – you’ve suggested a bunch of anecdotal reasons (“as with any mobile network, you’ll see variable performance based on location, time of day, weather, the number of people around and other factors.”) but these don’t seem to follow from any of the research that you’ve presented in the article. You haven’t mentioned the factors that played on your single data point measure. What devices were in use, what time, what day, what train line, how busy was it in comparison to normal, was it raining or clear, …?

    I think my primary concern about a single data point reporting is that you’re reporting about something that will be in use by people on a regular basis. I do care about transient differences in availability (e.g the Camberwell 3G tower broke last weekend), but it’s systematic problems that make me the most frustrated (like the ones listed above).

  • Just did a test in the Melbourne CBD (indoors) with telstra 4g htc one xl

    1 bar 4g reception – 6,105k d/l / 7,580k u/l

    full 3g reception – 14,386k d/l / 502k u/l

    in my experience 4g coverage is very patchy – when it works it is good, however my phone is more often on 1 bar 4g coverage or full 3g reception – and I work and live in the CBD.

  • I can concur with the above- Telstra 3G dead in Melbourne CBD during peak hours. Cnr Collins and Elizabeth all the way through to Crown Casino. Only had 4G coverage in Bourke St Mall (20mb symmetric) and bottom Telstra building on Collins St (23mb symmetric) so far. Still a way to go…

Show more comments

Log in to comment on this story!