I rather liked the “two cloud”  model of an advanced mobile network, described by Huawei. They’ve suggested that future networks will use macrocells to provide universal coverage everywhere (albeit with a worst case 256 kbit/s speed), supplemented by a large pico/femto layer where high capacity and/or data rates are needed (with a more generous 2 Mbit/s speed).

Lots of capacity means lots of small cells

In order to achieve this, they’ve identified that many pico and femtocells will be required. With the significantly lower cost of femtocells (not just the capital equipment cost, but the simplicity of installation and self-configuration), we’d expect femtocells to form the bulk of this network capacity layer in the future.

This is in addition to the traditional view of domestic and enterprise femtocells installed at home or in the office. These so-called metro-femto cells would be found in areas of very high public use, such as at transport hubs, public squares, high streets etc.

For example, the Spanish Steps in Rome was always considered a nightmare for network operators – few buildings in the area available to site equipment, huge numbers of highly talkative people, lots of nasty granite to reflect or absorb the signal. I’m sure there are plenty of other examples in your country too.

Notspot Products starting to appear

It seems other equipment vendors also share Huawei’s view. Femtocell pioneer Ubiquisys has teamed up with US siting specialist Public  Wireless to create the 3G “colo-node” (sorry, but this name sounds more like a medical complaint to me). This gangs up to 4 Ubiquisys 16 channel femtocells to provide high data rates and capacity at very low cost where needed. Although a range of 2 km is quoted, I’d expect this capacity to be consumed within a few hundred metres of the device. The node can also be equipped with Wi-Fi and/or other radio technologies where required.

The picture below illustrates a unit installed strung between existing street wiring.

Backhaul starts to become a problem with super metro femtos

A growing problem for many mainstream 3G macrocells is the backhaul – the connection back to the operators central switching centres. Ever higher data rates easily exceed the capacity using traditional E1 and T1 circuits, requiring the industry to migrate to Ethernet over fibre or point to point wireless. Larger macrocells can easily require 30 to 50 Mbit/s of capacity as they grow.

With 4 femtocells potentially offering 64 voicecalls or 56 Mbit/s peak data rate (4 x 14Mbit/s), total bandwidth for the backhaul connection would far exceed that available from today’s DSL or most cable broadband modems. The large number of voice calls would require a reasonable amount of uplink bandwidth – something less suited to the asymmetric properties of ADSL. Perhaps a SDSL service (which is symmetric for uplink/downlink) might be used here instead.

While it is statistically extremely unlikely to reach these peak rates of demand, multiple DSL lines and/or other backhaul technologies may be required in these circumstances. Beware that you can’t just “double up” multiple DSL lines to achieve higher speeds – the crosstalk interference between long DSL lines reduces the overall performance and other techniques are required.

This isn’t insurmountable, but needs to be borne in mind for the larger equipment. Smaller capacity metro-femtocells shouldn’t be a problem.

Multiple femtocells for multiple operators

This solution is marketed for use by a single operator. Up to now, femtocells have been positioned as being single network only, prompting me to point out that they aren’t suitable for Starbucks or similar public businesses.

Would it be feasible to run each of these femtocells independently, each operating on behalf of different network operators? This would allow a single location (e.g. a coffee shop, remote hotel) to provide excellent 3G network service for all networks.

Real product means real commitment

This is a real product, not a prototype, and demonstrates that there is a market for metro-femto seen to be worthwhile to invest in. Unless picocell vendors up their game to make their products easier to install and maintain (ip.access are an example of one which has done so, other RAN vendors I’m less sure about), we could see a fairly quick adoption and migration to metro-femto solutions like this for some enterprise and dense urban hotspot areas.