# DSL Oversubscription Vs. 3G Capacity

A fierce competition is raging in Austria between DSL and 3G operators positioning 3G data cards as an alternative for DSL connectivity. Prices are interesting too, so many people are going wireless these days. Which leaves the question of how much capacity mobile networks could have compared to DSL.

Certainly not an easy question to answer so let’s take a couple of assumptions:

Austria has 4 HSDPA networks today. Let’s say in a city like Vienna the average cell inter distance is 1km. Usage is still in it’s early stages so only a single 5 MHz channel is used in a 3 sector cell. Per sector throughput is assumed to be 2.5 MBit/s. Since the cell covers an area of 1 km², the capacity in that area per operator is thus 2.5 * 3 = 7.5 MBit/s. All 4 operators together would thus create a capacity per km² of 30 MBit/s.

On the fixed line side I would say that DSL today offers a speed of on average of 4 MBit/s to housholds in cities like Vienna. Vienna has a a population density of 4000 inhabitants per km². Let’s say the average household has 3 people and DSL penetration is 40%. Thus there are (4000 / 3) * 0.4 = 533 DSL lines per km². With an average speed of 4 MBit/s per DSL line that would be 2.113 GBit/s. Sounds like a lot more than what the 3G calculation results in above. But wait, there’s a catch. The 4 MBit/s are only valid between a subscriber and the DSLAM (DSL Access Multiplexer). The connection to the core network is usually much smaller. I’ve heard the ‘oversubscription’ is anywhere between 1:20 and 1:50. Let’s assume the oversubscription is 1:30. As a result, the DSL capacity per km² would be 71 MBit/s.

30 MBit/s wireless vs. 71 MBit/s via DSL

The example stands or falls with the DSL oversubscription ratio. If you have more details on this please let me know!

## 10 thoughts on “DSL Oversubscription Vs. 3G Capacity”

1. Martin, I don’t have data on DSL operators, but there is some data on cable modem subscribers actual usage here: http://www.broadbandgear.net/archives/bgr/2007/060807/
Once you deal with the access network, an ISP has to buy “IP transit”, i.e. connection to the rest of the Internet. IP Transit is sold in Mbps per month at the 95th percentile. I.e., traffic is measured in Mbps every five minutes throughout the month, then the top 5% of those measurements are discarded and your monthly charge is based on the next largest five minute interval. Measured that way, cable operators in 2006 were seeing and average of 8 Kbps of traffic per subscriber. Put another way, 120 cable modem subscribers required 1 Mbps of IP transit service.

2. Interesting. However I believe the better comparison would be DSL oversubscription vs. 3G oversubscription, or should we believe that 3G providers give each sector cell a clear 2.5 Mbit/sec connection to the rest of the ‘net? At least in this market, advertised rates for consumers have nothing to do with real world performance, especially in peak hours. Perhaps 3G providers are commanded legally to comply with certain QOS, like phone companies? Well here mobile phone companies also show pretty lousy QOS anyway. There is no reason for me to trust a 3G company more than a DSL company; especially when they are owned by the same parent companies, which is usually the case here.

3. jens says:

The oversubscription obviously happens at different places in fixed or mobile networks. Whereas the fixed network needs a DSLAM for each subscriber whether he is using that resource constantly or not, this is different with a mobile network where that air interface is a shared resource.
It would be interesting to understand what would be cheaper to deploy and operate given a certain demand in a certain area – a fixed or a mobile network.
Especially for those new air interfaces that are very cost efficient (I’m thinking of LTE) I would assume it is cheaper to reuse existing antenna sites than installing new DSLAM. (Not having hard facts though.)
However, I would also assume that this is a very fierce discussion within the business units of operators that are providing both – fixed as well as mobile access.

Oh, and I absolutely agree that it is a stupid idea to use the peak traffic capabilities of a technology as competitive numbers in marketing. This will only lead to user frustration. Given the fact that more and more laptops are sold, I would assume that mobility is a value that is much more appreciated by users. I mean, that’s also why everybody is installing a WLAN at home.

4. Zed says:

Way not to go, Martin.

First, your math is totally wrong. An average cell inter distance of 1 km equals over 3 square kilometers of coverage per cell. Subsequently there is less than 10 Mbps of capacity per square kilometer with these figures.

Secondly, a lightly loaded 3G network is likely to feature oversubscription in the backhaul.

Thirdly, the xDSL vs 3G comparison doesn’t take into consideration the technical limitations of 3G as measured by the bandwith latency product and all the assorted downsides of a shared access media.

Furthermore modern xDSL networks are able to operate at wirespeed and xDSL backhaul can always be upgraded to wirespeed, which can’t be said of the aggregate 3G subscriber capacity. xDSL backhaul is denominated in Gbps whereas 3G backhaul is measured in Mbps.

With regards to oversubscription and IP transit, usually only those parts of the xDSL network that require IP transit, ie. transnational or overseas traffic, are oversubscribed to any meaningful degree. This leaves on-net, peering and national traffic as statistically available at wirespeed.

It is to be remembered that oversubscription ratios should only be calculated over large subscriber pools and single POPs should not be bandwidth provisioned using typical oversubscription ratios as intenet traffic is very bursty up to wirespeed.

3G is a complement to wireline broadband and a very poor substitute. 3G is a dialup replacement which happens to be mobile. If that matches your user profile, fine, but one should not expect 3G to be a broadband replacement to the general population.

The difference between wireline broadband and 3G is like the difference between an 18 wheeler an a battle tank. The battle tank will take you more places than an 18 wheeler, but it will be slow going, you might have to make a few trips, it will be slow going, it’ll cost you an arm and a leg and if you go cross border no good will come from it.

– Zed

5. Mark says:

Working for a mobile operator I have to say that there are quite a few limitations to HS which are not usually advertised by the operator. First of all there is often a backhaul limitation (especially from NodeB to RNC) in mobile networks. Just because a site has been upgraded to HS doesn’t mean it has the necessary E1s to carry all that traffic. Second, R99 traffic will always take priority over HS so you might find your throughput restricted by CS usage in your cell. Third, (and quite important in my opinion) the latency of HS is still quite bad and definately noticable if you are used to DSL.

I think it is fair to say that although impressive the wireless technology will never catch up with the wireline. By the time LTE etc are implemented we will probably have fiber to the home so wireless will always be a quite a few steps behind wireline.

As mentioned in other comments the only true advantage is mobility but that is only needed by very few people.

Well done for the blog by the way, it is one of my favourites sources for wireless news and discussion.

6. Hello all,

thanks a lot for the comments so far, please keep them coming! Here are a number of comments from my side:

– Using 2.5 MBit/s as the transmission speed per carrier per sector is well below of the “theoretical” peak of what is advertised today with 3.6, 7.2 or even 14.4. This value is realistic and already contains the overhead incurred by the shared medium. (If you would like to discuss this point further, send me a mail, I am well prepared 🙂

– I agree that when cells are HSDPA activated this doesn’t necessarily mean the backbone link was increased to meet the maximum throughput especially in a three sector configuration. I guess that’s true for most deployments today. For my calculation I assumed that capacity would be added here when required (rising number of subscribers). Long live good network monitoring and appropriate response!

-Bandwidth/Latency: I am a heavy user of both DSL and HSDPA. I agree that for online gamers the extra latency is probably an issue. For web surfing, eMails, VoIP, etc. I can’t detect a difference between my (slow 1 MBit/s) DSL at home and a HSDPA connection when on the road.

-It’s of course correct that DSL can decrease the oversubscription very quickly if required. Here, however, I think the limiting point is the price of interconnection to other networks. I had a look around on the net based on the first comment and saw in a number of places that the price per MBit/s is around €20 a month (VAT included). When you assume a requirement of 70-80 MBit/s (assuming none used for local streamingn content) for 500 subscribers that’s a monthly charge for the operator of around €1600 divided by 500 subscribes = around €3.- Add to that the cost for the own network, DSLAM, customer support, monthly line rental of the last mile, etc. and it already gets narrow as far as the margin is concerned.

– I think an important take away is that the limits of the two technologies are at different points: For DSL the limit is the price for routing the data to other networks. Thus DSL operators can offer TV and multimedia streaming from a server in their own network very cost effectively. On the wireless side the bottle neck is the wireless interface. Thus multimedia streaming even from their own network pretty quickly becomes a bandwidth issue. To my knowledge no wireless operator in Austria competing with DSL is currently doing that. Pretty obvious why not.

-Based on the population density taken into account above it’s clear that you can’t have all people on wireless rather than on DSL. But the exercise was to see to which degree wireless systems can compete with DSL access (excluding TV streaming) today. If they can address 30% of the market than it’s competition and not a joke. And when looking at the numbers, 30% is quite realistic in my opinion.

-It’s correct that DSL can quickly increase capacity. Wireless can do so, too but obviously not as fast. The calculation above was based on the use of a single carrier and 4 operators. That leaves 66% of the 2.1GHz band available for future expansion. And for a bit further in the future there is the 2.5 GHz IMT expansion band and around 200 MHz available in the 3.5 GHz band for WiMAX. But in all fairness, that has to be compared to optical deployments. But these again are limited by the cost for the peering connection (see above).

-One thing that we didn’t have so far: Alternative DSL providers usually have to pay a certain amount per month for the line rental. In Germany that’s around €10 per month. Wireless operators don’t have that basic cost (but a bit of debt due to license costs, but that’s another story…).

-The 3G network to some extent also has to carry voice calls today and this is increasing as more people get 3G phones. I didn’t consider this in the calculation above to make things a bit easier and because I think most voice calls are still handled by the 2G network. And once that changes, capacity is freed in the 900 MHz band which at some point will also be used by 3G networks (refarming…).

O.k. the keyboard is glowing now… Thanks again and keep the comments coming!

7. Siegfried says:

Hm. I wonder if there are not other radio ressource limitations you need to take into account – I am not sure about how the lower layers of the GSM networks work, but I could imagine that if you add thousands of terminals which are all “idle”, they still generate some radio traffic (just to let the network know they are still there). I wonder how many “idle” data equipment an average mobile network could handle.

8. Jaggs says:

Wireless broadband is definitely the future, and anyone who believes otherwise is sadly missing the point. The usage patterns will not be the same – multimedia streaming will initially be for xDSL and cable clearly – but in the long run the mere fact you can use it anywhere at decent speeds means demand is going to go through the roof once more HSDPA devices arrive on the market. For an easy comparison take a look at the almost vertical take up of home wireless once laptops with WiFi built in hit the market. Watch for the same thing with mobile broadband. 1000%. Forget about limitations in the technology, those will be overcome (remember when they told us you couldn’t get more than 56K down a copper wire?)

9. I totally disagree with Jaggs.

Over the last 20 years, average “real world” wired data speeds available to consumers have consistently increased faster than wireless. The gap is still widening and will continue.

Pretty much every DSLAM has one or more fibres available for backhaul. Only a tiny fraction of cellular base stations do – most still rely on a few E1s or microwave. Many existing 3G cell sites don’t even have ethernet ports to enable upgrade to metro ethernet.

That’s the backhaul, but there’s also an access issue. With the progressive rollout of higher-speed DSL or ultimately fibre to homes & offices, there is a clear roadmap to scaling DEDICATED bandwidth to subscribers as fast as the switching infrastructure’s cost can come down. Wireless broadband will always involve SHARED bandwidth unless you use point-to-point links, which is almost impossible for residential subscribers in urban areas as most will not have line-of-sight to a cell tower.

Interestingly, wireless vendors generally agree with this view. I was at an LTE conference last week, and the consensus was that wireless broadband is most suitable for home use either for users beyond reach of copper/fibre, or in markets like Austria where fixed broadband is unusually expensive… or as an adjunct (eg for 2-broadband households).

10. …and another thing…

ISP’s infrastructure and Opex costs (eg backhaul & maintenance) are also supported by their business customers’ use of the same PoPs.

Some will use DSL at much lower contention rates, while others will have dedicated pipes. So any ISP’s covered km2 may also have several 100Mbit/s of business access as well as consumer DSL.

There is no mobile broadband equivalent to ‘dedicated access’ or VPNs – which given that 3G access is aimed in part at corporate users, means it’s not an apples vs. apples comparison