5G was meant to be power efficient, but it could quadruple RAN energy levels

Many of the expectations that operators had for 5G a few years ago have changed now that commercial reality is close. The first few years of 5G are set to look far more conventional, and 4G-like, than many had anticipated – faster mobile speeds for YouTube viewers rather than autonomous cars. One of the most serious areas where 5G may disappoint is in power consumption.

In 2014, Rethink’s biannual survey of almost 100 MNOs identified greater power efficiency as a top 10 driver to deploy 5G, and it was expected to save one on of operators’ biggest opex costs, while enabling them to meet green targets round the world, as governments become increasingly concerned about this power-guzzling industry. In that same year, the Greentouch consortium, led by Bell Labs, was talking about 1,000-fold improvements in energy efficiency by 2020, compared to 2010.

Short battery life is a far greater brake on mobile data usage than speed limitations, and that alone makes power efficiency an important topic, before we even consider larger environmental and economic concerns.

But it turns out that 5G’s power efficiency is elusive, and this fact is being highlighted by none other than Huawei, which talked about the issue at its recent Mobile Broadband Forum in London (and of course it has claims to have a power saving site solution).

But the challenge is considerable, stemming from the huge number of active components that will exist in a 5G network. In particular, Massive MIMO antennas, replacing two or four transmit and receive elements with 64 or 128 – but many operators are planning to deploy them to increase the capacity and coverage of their macro sites and exploit high frequency spectrum more effectively.

Even densification may be an issue, since a large zone of low power small cells can still have high collective power output if they are all working at once (an issue currently being discussed by the European Commission as a concern for smart cities).

At the Huawei event, Guiqing Liu, an EVP at China Telecom, raised the issue of high power consumption of 5G base stations, predicting they would consume three or four times more energy than 4G networks.

Orange’s SVP of radio networks, Arnaud Vamparys, said reducing these levels would be a “key target for the industry in the next few years”. He said energy-related costs account for as much as 25% of opex costs for Orange’s mobile business and echoed Liu’s forecasts, saying 5G could hog three times more power “if not optimized”.

Some remedies exist in the standards, including an ‘advanced sleep mode’ which turns down power when no device is communicating with a base station. Operators will certainly be pressurizing their suppliers to implement such features actively. And individual MNOs are developing their own solutions for power optimization, including Orange.

Liu says artificial intelligence (AI) – which many telcos are expecting to use for network optimization and planning – could also help to cut energy costs. During trials, China Telecom used AI techniques to reduce power consumption at data centers by around 30%, he said, and so similar tools could be used to optimize the MIMO antennas.

In a 5G network, where RAN and core are increasingly virtualized, there is also the issue of data center energy bills, so this needs to be addressed in conjunction with the antennas. As LightReading points out, the UK’s fourth MNO, Three UK – which, as a challenger, needs to keep costs as low as possible – is building 20 modern data centers as part of its 5G process, and expects these to save it £2m ($2.6m) a year in areas like cooling. Similar IT modernization will be taking place in most advanced operators.

All this is a far cry from the vision of the Greentouch initiative, which in 2015 published an aggressive roadmap for reducing mobile network power consumption by 1,000 times between 2010 and 2020. Its work was heavily focused on virtualization from day one, arguing that flexibility and resource efficiency are the key to 5G.

The consortium was formed in 2010 by Bell Labs (now owned by Nokia, then by Alcatel-Lucent) and others. In May 2013, GreenTouch put some solid goals behind its theories, setting a shorter term target of cutting energy consumption in telco systems by 90% by 2020 – and doing that for one national US MNO would be the equivalent of taking two million cars off the road. The project covered all kinds of telecoms infrastructure, but sees the greatest potential in mobile systems. “The wireless medium is the most inefficient because it wastes a lot of power transmitting signals over a large area,” Thierry Klein, chairman of the group’s technical committee, said in 2013.

In particular, Greentouch believes three key changes will make the difference to cellular networks’ power hunger and CO2 emissions levels. These are the use of smaller, low power metrocells to add capacity in dense areas; wider infrastructure sharing; and techniques to match power consumption flexibly to the level of usage of the network at any time. “We need to make resource usage proportional to the amount of traffic the network is handling,” said Klein.

One of GreenTouch’s key projects was to demonstrate proof of concept base stations that used large arrays of up to 1,000 smart antennas to cut power consumption. The trial found that the energy needed to power each one dropped significantly as more were added, without impacting the range or capacity of the cell. The team was using fairly standard MIMO techniques, but harnessing the arrays not to boost capacity, but to reinforce a single transmission, creating a single strong beam from many low power signals. It is ironic, then, that Massive MIMO is now seen as one of the main contributors to boosting power usage.

GreenTouch also showed how moving in-home devices – such as broadband/WiFi gateways, small cells and set-top boxes – while redesigning the point-to-point optical transceiver, could reduce energy consumption by 46%.