Hot-spot or not?
Social networking is one of the key applications that drives mobile data demand on today’s 3G networks, and keeping up with this demand is one of the most difficult tasks for operators.
So it is interesting to see that data mined from social network updates can also be of assistance to operators as they optimise the network rollout.
Keima look at the geographic information (geotags) embedded in social network updates sent from mobile devices.
By aggregating together this anonymised data, they are able to produce arresting images that show where the busiest parts of the network are.
In cities, the busy streets glow like a night-time satellite photo and landmarks are easily picked out.
The nature of this data mining means that it is the geographical intensity of the uploads from mobile devices that is being studied.
However, downloads are actually the quantitatively dominant data on mobile networks, and these are inherently not geotagged.
So it is important for us to understand whether upload and download are well-correlated when we use tools like this to guide network designs.
Let’s take a look at one of the live networks that CBNL backhauls and see if that is the case.
Each point on the graph shown here represents an HSPA+ node B in this network.
You can see that there is a clear correlation between the mean downstream demand and the mean upstream demand, with about a 3:2 ratio between the two.
There is a reasonable spread, but it’s possible to take that into account when designing the network.
This is really good news – it means that, as operators start rolling out small cells, they can use innovative tools like Keima’s with confidence to target capacity where it’s most needed.
I’ve previously talked about small cell backhaul at the Small Cell World Summit, and it’s a topic we’ll be returning to soon.