More thoughts on green issues & cellular CO2. But will femtos help?

As I mentioned the other day, I was chairing a conference session on "Green" base stations and network on Friday. It had some really fascinating presentations from NSN, Ericsson, Vodafone & T-Mobile, with a lot of detail on a variety of different angles on reducing the environmental impact of mobile - and the opex implications of energy efficiency in the radio network.

One thing that unfortunately came across very clearly, however, was the difficulty of producing & reconciling statistics. The environmental side of industry is a long way behind finance in terms of coming up with consistent, auditable numbers. At the moment, it seems like pretty much anything can be proven or disproven - for example by including various "behind the scenes" factors in terms of CO2 emissions, or possible "savings" elsewhere from giving people better communications tools. None of this is particularly wrong in my view, but it just means that comparisons are nigh-on impossible and some of the headline figures need close scrutiny.

So for example, should we measure CO2 emissions per subscription, per customer, per square mile, per cell site, per transceiver, per volume of data traffic, or 101 other options?

However, as a useful starting point, one of the Ericsson representative's slides gave some useful headline numbers suggesting that 2G cellular accounts for about 24kg of CO2 / sub / year , and 3G is about 29kg. (These figures are "fully loaded" with handset manufacturing, older legacy cellsites and all sorts of other stuff included). As a reference, the average European probably emits 4-5000kg a year in total, based on a quick look on the web, so we're talking about perhaps 0.8% of total, even with multiple device ownership being common.

Another very important factor that was raised is that to make any big differences, the industry needs to focus much more on improving the performance of current "legacy" infrastructure, rather than just on the performance of new designs & shipments. No operator is going to throw away functioning base stations, even for useful-but-small improvements in power efficiency. T-Mobile had a really good statistic that showed that because of typical electricity prices in Europe, a saving of 100W in average power consumed saves about €130 in opex a year - so any capex on improvements really needs to be <€600 - including the cost of an engineer visiting the site.

One other thing that really struck me was the ready availability of some "quick wins" - especially reducing the need for air conditioning in base station enclosures. Apparently a lot of equipment works perfectly well at 40degC, so keeping it all nicely chilled is a complete waste of energy. The idea of switching off a chunk of radio capacity in quiet periods, rather than leaving transmitters powered-up "on standby" also seems like a no-brainer.

There was also a lot of discussion of sites that are "off grid", especially in emerging markets. Solar and wind power were discussed in a lot of depth, and are slowly being rolled out along with efficiency improvements to reduce reliance on diesel being trucked to sites. All in all, there's obviously a lot being done on the radio side of the industry, which accounts for a huge proportion of the total power consumption of cellular. It's an area I'll keep a closer eye on, and potentially write a report later in the year.

One awkward question did occur to me though - do femtocells help or hinder the trend? It's very easy to say "they operate at much lower power than macro sites, so it must be good", but it's more complex than that as they're also probably going to be left switched on but unused for most of the time. They'll also indirectly drive power usage in broadband infrastructure, and typically it's less energy-efficient to have millions of small gadgets rather than a few larger, optimised & centralised ones. When I asked the panel, the opinion seemed to be "we're not sure yet, but it's possible that femtos will increase overall energy used, although obviously from an opex standpoint the mobile operators won't be paying for it as it's plugged into the users' own electricity sockets".