Femtocells - Reducing Power Consumption in Mobile Networks
Just 5% to 10% of the energy that goes into a base-station emerges as a useful radiated signal. Femtocells address that problem.
Mobile communications is big business. Around a billion handsets are sold each year, to satisfy a global mobile subscriber base of five billion, according to the ITU.
Along with the benefits come the responsibilities. According to the Greentouch Alliance, an industry group set up at the beginning of 2010 to tackle the issue, information and communications technology (ICT) is now responsible for 2% to 2.5% of global carbon dioxide emissions, a figure that is expected to double over the next decade. Put another way, ICT is responsible for about the same amount of carbon dioxide emission each year as air transport.
The ITU reckons that about 10% of total ICT emissions are attributable to mobile phone networks, with between 55% and 60% of the energy consumed being used in the radio access network. With macrocell base-stations consuming anything from 2.5kW to 4kW each, and national networks using tens of thousands of them, that’s a lot of energy, and a lot of emissions, being used to connect mobile devices.
The power implications of macrocells
Part of the problem is the macrocell-based architecture of mobile networks, which is coming under strain as data traffic rises. Some estimates say that just 5% to 10% of the energy that goes into a base-station emerges as a useful radiated signal. That signal is pumped fairly indiscriminately into a large volume of space in a bid to reach as many users as possible with the least number of base-stations. Macrocells cover a very large area (several km2), but most of that area is empty space: there are lots of gaps between the people you need to connect to. This was fine when all we wanted was to make voice calls, and coverage was the main driver, but is less useful now that we are beginning to rely on constant, high-bandwidth mobile data connections and capacity to users is crucial.
Macrocells have other drawbacks. The complex modulation schemes of modern cellular standards demand that the RF sections of base-stations are run at high power, and less than maximum efficiency, to achieve the necessary linearity - meaning that a lot of energy is consumed ‘just in case’. Most macrocells then drive this high-power RF signal up a long feeder cable to the antenna array, incurring further energy losses. There are ways of tackling these issues, such as better amplifier design and new system architectures, but the equipment in the field can’t benefit from these techniques until it has paid for itself.
Lowering power demands with femtocells
Femtocells, small base-stations that can be installed in homes, offices and even outdoors to serve a handful of users, changes the network to overcome these issues.
People mostly think of femtocells for the home: to give better capacity if you live in a ‘not-spot’ for example. But the technology also means operators can deliver better service to enterprises and high-value customers; or a ‘metro femto’ can deliver lots of capacity to a dense urban area, perhaps for commuters at a busy train-station.
While a macrocell delivers coverage to a wide area, a femtocell delivers a lot of capacity in a locale – just delivering power to where the users are. Because they are closer to the users so it takes less RF power to provide a high-bandwidth connection - a typical femtocell may have a 20mW RF section and consume a total of 2W. That applies in the handset too (your cellphone does not need to shout to connect to the distant macrocell), so battery life is better and you can charge your phone less often.
Using lower RF power also localises signals, so that scarce spectrum can be reused more often than is possible in a macrocell network.
The lower power means more efficient RF technology too, so less energy is wasted there.
Operators benefit from the uptake of femtocells, because customers get better coverage at home without them having to expand their macrocell networks. Network capacity increases without the need for more (scarce) spectrum. Customer churn falls, due to the improved user experience and novel services, such as localized tariffs, made possible by femtocells.
Customers of course benefit from better service, better connections (which means both higher-data rates and clearer voice calls) and better battery life.
Macrocells and femtocells each have their advantages. What happens to total network energy consumption when you combine them in hybrid networks (macro for wide area, femto for where you need it)?
Two different models, considering quite different approaches, have reported on this.
The first was from Ofcom, the UK telecoms regulator, and Plextek, a consultancy. They consider the energy savings made possible by using closed femtocells (reserved for private use as a residential system exclusively), rather than extra macrocells, to achieve good indoor 3G coverage for subscribers.
The analysis compared two approaches. In the first, eight million households (one in four of the UK total) got femtocells, each running all day, everyday and consuming 7W, for a total annual energy consumption of 490GWh.
In the second, a 3G macrocell network was upgraded to achieve the same indoor coverage as the femtocells. The analysis started from the premise that a UK operator would need 30,000 base-stations to offer good geographic coverage, and that the power each used was set to achieve adequate outdoor and marginal indoor coverage at the edge of each cell, so that improving indoor performance would have to be achieved by adding cell sites.
The model estimates that it takes 40 times as much power to deliver a signal to the inside of a UK home as to its outside, and that the macrocell signal strength would fall off by a fourth-power law with the distance between the antenna and the home. Using these assumptions, the distance between macrocell sites would have to be cut by a factor of 2.5 times to deliver the same indoor coverage as the femtocells. This implies a 6.3-fold increase in site density, which translates to an increase in network energy use of the same amount.
With macrocell power consumption forecast to have fallen to 500W by the time the femtocell market is mature, the increase in total macrocell power consumption to achieve the same coverage as femtocells is 700GWh per operator per year, or 3500GWh per year for five operators with similar networks.
The analysis concludes that using femtocells to increase indoor coverage in the UK would have an energy advantage ratio of 7:1 over using macrocells, given five operators.
The value of open-access
Femtocell market 2012 (Courtesy Forward Concepts)
Bell Labs has its own analysis of the likely benefits of a hybrid femtocell/macrocell network, using open-access femtocells (ie they are part of the network like any basestation, and any subscriber can access).
The analysis considered a 10km by 10km urban area of Wellington, New Zealand, with a population of 200,000 people (65,000 homes), 95% of whom were mobile users. The analysis then introduced varying numbers of open femtocells, each able to serve to up to eight users within a 100m × 100m area, at random throughout the sample area. Each femtocell consumed 15W (which is very high for residential, but perhaps right for the more capable metro model) and had to operate continuously to fulfill its role as an open access point. Each macrocell also operated continuously, consuming 2.7kW.
The analysis showed that a relatively small number of open-access femtocells could achieve significant coverage: if an operator with 40% market share managed to get 20% of its customers to use femtocells, they would satisfy about 80% of total demand. This would shift a lot of user sessions on to lower-energy femtocell links, at the energy cost of running multiple femtocells.
The impact on total network energy consumption then depends on the way that the network is used. Macrocells can support many more voice calls than femtocells, so if the femtocells are used mainly for voice calling they don’t have a large impact on total network energy. On the other hand, if customers use fast data then the femtocells are far more effective, and fewer such connections have to be made over relatively energy-inefficient macrocells.
The Bell Labs analysis suggests that for high data-rate users, a hybrid macrocell/femtocell network, in which the macrocells are used to ensure coverage and the femtocells carry most of the load, can cut network energy consumption by up to 60%, compared to a macrocell-only network, given 20% penetration.
As might be expected, operators with a lower market share and therefore less ability to generate femtocell coverage from their user base benefit less from the potential energy savings of a hybrid macrocell/femtocell network. The analysis suggests that such operators should share their femtocell networks.
The Bell Labs paper concludes that the advantages of hybrid networks are likely to increase as the technology matures.
A third analysis from researchers at Aalto University in Espoo, Finland, makes a similar point, suggesting that femtocells, especially those used in open networks, can benefit from energy-saving features such as low-energy sleep modes into which the unit switches when there is no traffic.
A hybrid macrocell/femtocell architecture can significantly reduce the energy consumption of cellular networks in situations where users want to make a lot of high data-rate connections and the capacity of local macrocells is limited.
The effect is expected to increase as the technologies of macrocells and femtocells mature. It is also stronger when femtocells are opened up so that they serve all subscribers, since this gives operators a way to strengthen their coverage without building more macrocell base-stations.
Further power gains
With femtocell rollouts beginning in earnest last year (2010), we’re just at the beginning of a phase of rapid technological improvement driven by increasing volumes.
Companies that have simply repurposed macrocell parts to gain a foothold in the femtocell market will have to rethink their chips to compete with those designed from the outset with energy efficiency in mind.
Power-management techniques are also going to be important, at both the chip level, in terms of how the devices are architected, and at the protocol level, where subtle interpretations of the 3GPP standards will be necessary to make the right trade-offs between service availability and energy saving.
RF sections will improve, with the introduction of more efficient amplifiers and power supplies, as well as the use of multiple antennas to decrease the energy used per bit transmitted, enable higher data rates and help reduce interference. Antenna beam forming techniques may be used to direct the radiated energy to where it is needed.
In the mid-term, femtocells may borrow techniques from macrocells such as macro diversity, in which multiple femtocells collaborate to deliver a signal, and self-optimization, which cuts the cost of running femtocell networks.
As the market grows, expect companies to develop chip variants that are highly optimized for particular uses, so that chips for home femtocells are much simpler and cheaper than those for the so-called ‘metro-femto’ use cases being proposed by some manufacturers.
Greener network, better service
The cellular world has changed from voce and coverage in the GSM / 2G era, to data & capacity in the HSPA+ and 4G world.
Macrocells are struggling to service today’s user traffic, which has evolved rapidly from relatively short voice calls to long sessions of high-bandwidth data connections. Phrases like ‘capacity crunch’ underline this problem.
Femtocells are an obvious answer: put the base-station closer to the user and more of the network’s traffic is carried over shorter, less attenuated RF links, improving coverage, service availability and energy efficiency.
There’s a hierarchy of effectiveness here. If every subscriber installs a femtocell, built around an adapted macrocell chip, and runs it all day every day in order to make a few short voice calls, it’s likely that total network energy consumption will rise. If the femtocells use optimized chip designs and implement sleep modes to keep them in a low-power mode until a call is detected, energy consumption falls. If customers’ traffic patterns change so that they’re using the femtocells to handle long sessions of high-bandwidth data connections, rather than using a macrocell, total network energy consumption falls again. If the femtocells are opened to all subscribers, macrocells end up providing infill coverage, saving further energy. If femtocells are integrated with other home electronics, such as DSL modems, routers and WiFi access points, further systemic energy savings are likely.
It’s worth noting that for operators, many of whom worry more about their energy bills than their networks’ energy consumption, introducing femtocells shifts some of that bill, as well as the cost of providing the backhaul, to the consumer.
The wider issue is the impact of mobile communications on greenhouse gas emissions. The studies have shown that adding femtocells to the network can result in dramatic savings in energy consumption (7x more efficient for example). What is more, this saving is not at the cost of experience: quite the reverse, as consumers benefit from better coverage, higher data rates, better voice quality and longer battery life.
Femtocells can become a key part of the armory of techniques that will control that growth, as well as improving service availability and enabling the next generation of mobile services.