Some pioneering work at Orange UK was presented by Adrian Pike from their radio planning department at the Avren basestation conference recently. They’ve been looking at how and where specific locations could be identified for small cells and femtocells, how accurately they need to be positioned and the contribution they would make to network capacity and performance.
It’s all about data
Confirming reports from many other networks, Orange has witnessed phenomenal growth in data traffic in recent years. Data traffic has multiplied 15 times in the last 5 years and is now 94% of their total traffic. Of this, around 50% is mobile broadband versus 50% smartphone. The directional split is 90% downlink and 10% uplink, with a projection that this will remain stable at around 88% downlink by 2015.
Small cells are crucial
Orange recognises the importance of small cells, characterising these to have a maximum RF power level of 5 watts. The presentation discussed carrier installed small cells, both in public outdoor areas and in businesses (which others might term enterprise femtocells). A crucial factor is locating them in the correct place. A tolerance of 50 metres or less was thought to be a good target.
How to identify the right place for each small cell installation
Orange have been experimenting with analysis of the RNC trace data. This is a stream of measurement data reported from every active mobile device on voice or data calls, passed up through the basestation to the RNC and consolidated. Recently, large basestation vendors such as Ericsson, Huawei and Nokia-Siemens have made available these data feeds which contain an enormous amount of useful “real world” data (as compared with theoretical projections from RF planning tools).
The analysis provides detailed coverage maps which identify hotspots of data traffic (i.e. areas where there isn’t enough basestation capacity). It also identifies areas where there is too much (i.e. where too many different basestations interfere with each other, causing pilot pollution and resulting in poor coverage). The analysis predicts traffic over each 24 hour period to establish which areas which are consistently poorly served.
So far Orange UK have evaluated software analysis tools from vendors such as Arieso and Actix, although there are other players in this space.
Such data analysis uses the measurement reports from mobile devices to estimate their actual location. This is based on distance from the nearest cellsites. A series of measurements typically achieve accuracy of some 50 metres or less, depending on the proximity of the macrocells. Unlike costly drive testing which is confined to the streets outside, accurate measurements can be captured from users inside buildings where the majority of traffic originates. An example was given of a street where the drive test showed adequate coverage, but the measurement reports showed heavy use inside a building.
Accurate location for small cells is essential
The study found different results depending on how far the small cells were located from the nearest macrocell:
| Distance from macrocell (metres) | Impact |
| Less than 50m |
Adds little capacity – it just adds interference. |
| 100m | Achieves 50% overall capacity gain, but small cell still adds some interference |
| 200m | Creates a small cell radius of 30m, 100% capacity gain which offloads macro traffic (i.e. doubles the capacity of the macrocell). There is also improved QoS to both macro and small cell users |
Where two small cells were located both >200m from the macrocell, a mean user throughput increased 3x and data rates increased from 360kb/s to over 1Mbit/s
Asking for a green requirement too
France Telecom/Orange also has an aggressive target to reduce their total energy consumption. Small cells could contribute to this by incorporating a sleep mode when not required for capacity. There should be a fast auto-powerup to cater for unexpected increased traffic – a target figure of a few seconds was suggested. If it took (say) 2 minutes to power up, then this would be far too much. The idea is that the small cell passively monitors mobile device use in its vicinity when sleeping, and quickly powers up to announce itself and allow devices to handover to it when required.
It has to be noted that there was some consternation among the audience at this suggestion, but I don’t see why it shouldn’t be on the industry roadmap. After all, there are already features in today’s macrocell basestations which shutdown additional capacity (e.g. the 2nd and 3rd carriers) during quiet overnight periods.
Summary
Simply sprinkling large numbers of small cells liberally around dense traffic areas isn’t enough. In order to achieve maximum impact cost effectively, a performance measurement approach described above needs to be used. Accurate targeting of each small cell can dramatically increase the capacity of existing cellsites at a fraction of the cost. Small cells should be located at least 50 metres away from existing macrocell sites, and located as close to the identified problem areas as possible. Drive testing may still have a contribution to make, but the insights from accurate large scale measurement reports are far more useful.
Comments
RSS feed for comments to this post