How to dimension an LTE network?

[mgrayson]

In the recent Heavy Reading video webinar "4G Mobile Broadband: The Network Impact" (http://www.lightreading.com/webinar.asp?webinar_id=29250) I discussed a scaling use case for dimensioning an LTE network.

I took a look at the Minesota Internet Traffic Study (http://www.dtc.umn.edu/mints/home.php) which estimates the 2008 per capita monthly Internet consumption ranging from 3.2 GB/month in Western Europe through to 24 GB/month in South Korea. I then took a look at the anticipated growth in traffic due to video adoption (see the Cisco Visual Networking Index (www.cisco.com/web/go/vni ) - but this time I also looked at the OFCOM/Analysys Mason report on video adoption(http://www.ofcom.org.uk/research/technology/research/emer_tech/hqvs/analysyshqvs.pdf). The latter analyses the UK market, estimating that video adoption will cause the Internet consumption in the UK to rise from 5.2BG/month at the end of 2008, to between 63 GB/month and 143 GB/month by 2018. I ended up with a figure used in the LTE use cases of 10 GB/month - what would you have used?

In a world where today's iphone user may be consuming up to 1 GB/month and tomorrow's fixed line is supporting >50 GB/month, how should we be dimensioning LTE networks?

[ribansal]

The following statements seem to stand true in LTE:

• Scenario-1: Bandwidth at the radio link will continue to be the bottleneck, especially for Tier-1 operators. These operators will continue to build/leverage on the backhaul and core. For this scenario, radio link will guide the dimension of the network.
  
• Scenario-2: Radio link and Backhaul together will guide the dimension of the network. This will hold true for Tier-2 operators. Backhaul is directly proportional to Opex spending and it does not make sense to over-build if not required. Further these operators will tightly control the users' bandwidth using PCEF.
• Market radio "heat-map" will remain the same as 3G for most operators.

So the dimension exercise should answer:

• What is the operator's existing market view and and radio heat-map?
• Analyze which of the above scenario is the operator.

[rsynnest]

Interesting to note a difference in Tier 1 vs Tier 2 approach. I'd have expected all operators to look at the full network, trying to assure balanced dimensioning. Strikes me as peculiar that a Tier one would focus exclusively on the radio ... sounds like recipe for disaster in the backhaul ;-)

[ribansal]

Absolutely - network dimensioning is an end-to-end phenemenon. Operators who do not look at end-to-end capacity planning, those operators always keep fighting fires. Operators who plan end-to-end capacity have peace of mind.

Though the question is - What amount of traffic is the operator planning for? Is the operator being reactive - wait for the traffic to show, project for next 12 months and grow accordingly? Or more proactive - evaluate the "absolute bottleneck point" of the network and plan up and down the network to meet that bottleneck. The "hockey-stick" growth phenemenon that we are seeing in data traffic can not be caught with reactive approach. No operator relies entirely on one or the other approach, but its what approach the operator relies on more.

So for proactive approach, what is the "bottleneck point" for operators. Tier-1 operators are building large enough backhaul that its the radio again that is the bottleneck point. For Tier-2 operators, since they can not build their own backhaul and lease it, its an opex and they have to carefully plan radio and backhaul as the "bottleneck points".

Again once the operator determines what is the amount he is planning for. Then the entire network needs to be planned for that as an end-to-end system.