Italian operators maintain consistent quality while implementing energy-saving practices

In this analysis, Opensignal examines some of the energy saving practices implemented by national mobile operators in Italy. The analysis looks at the use of spectrum throughout the day on 4G and 5G networks, with its resulting impact on end user experience.

At night, when there is low network congestion, we can see that operators are reducing spectrum bandwidth. On all five national operators — Fastweb, Iliad, TIM, Vodafone and WindTre our users see fewer 4G spectrum bands available at night. On 5G networks this practice is predominantly adopted by Iliad. Despite these changes to spectrum bands our users enjoy consistent connection quality during the time these practices are in effect. 

Spectrum usage practices in place during night time hours

To understand Italian operators’ energy saving practices, we looked at the proportion of readings on different types of frequency bands across the country. Typically higher frequency bands — 2600Mhz and 2100Mhz on 4G — are used to add capacity and may not be needed at night when there is less network usage.

During the night, at around 1am, our users see a sudden fall in readings on the 2600MHz and 2100Mhz bands which is consistent with a network change. The reason for this drop is most likely an operator network configuration change made to reduce energy usage, although it could also be caused by short term network maintenance. Changing spectrum availability during periods of lower demand is one of the most direct ways in which operators can reduce energy usage on their networks. While lower frequency bands provide wider coverage through improved signal propagation and have lower overall capacity it can be sufficient to maintain users’ network experience at night.       
 

      
At night, our users make increased use of lower frequency spectrum bands when connected to 4G. This occurs on all five national operators — Fastweb, Iliad, TIM, Vodafone and WindTre. They all exhibit the same transition away from the 2100MHz and 2600MHz bands, towards the lower capacity and lower frequency 800MHz and 1800MHz bands.

Iliad has the greatest shift in 4G spectrum use at night. The share of 4G readings coming from our Iliad users in the 2600MHz band decreases by 32 percentage points against the daytime average (8 a.m. - 12 a.m.). The 1800MHz band has an increase of 38 percentage points in the proportion of 4G readings seen at night causing it to account for as much as 69% of total 4G readings, compared to its daytime average of 31%.

TIM shows the smallest daily shift in the proportion of 4G readings across different types of spectrum band. The operator’s use of 2600MHz declines by approximately five percentage points at night, while 800MHz and 1800MHz increase by up to two and six percentage points, respectively.       
 

      
The story is similar but slightly different on 5G. Iliad again has the most decisive shift in spectrum use out of all Italian operators. The share of Iliad users’ total 5G readings that are on the 3500MHz band declines by up to 30 percentage points during the night, coupled with a rise in the use of the 700MHz. 

Fastweb and WindTre users also experience changes in the spectrum bands at night compared to the day, but to a lesser extent. Users from both operators see their use of the 3500MHz band decline, in favor of the 1800MHz band. The shift recorded by Fastweb and WindTre measures up to eight and nine percentage points during the night respectively. 

Fastweb and WindTre have agreements in place where the two operators share the radio access network (RAN). Fastweb has its own 3.5GHz and 26GHz spectrum assets, but its infrastructure operates in conjunction with WindTre, as set out in an agreement from June 2019 that stipulates the provision of fiber backhaul by Fastweb, while WindTre provides FastWeb mobile network coverage via national roaming. It is therefore not surprising that the two operators exhibit similar patterns in this analysis.

Despite night time energy saving, users’ experience remains consistently excellent

To assess the impact that energy saving practices employed by operators have on end users, we have analyzed the consistency of our users’ experience across all hours of the day. Excellent Consistent Quality measures the share of tests on operators’ networks that meet the thresholds for demanding applications – such as watching HD video, completing group video conference calls and playing games. Our users see a robust level of experience on their 4G and 5G networks during the night hours, when the energy saving practices are in place.       
 

      
5G is much better at maintaining a consistent level of experience. During both periods of reduced demand at night and increased demand during the day, our users measure almost no statistically significant difference in the share of tests meeting Excellent Consistent Quality thresholds. 

For example, Iliad users see the greatest change in spectrum use at night, likely because of energy saving practices. However, the proportion of our Iliad users’ tests that meet the Excellent Consistent Quality thresholds during all hours of the night is unaffected by changes in the available spectrum. 

Energy efficiency is a key strategic priority for operators

Improving energy efficiency is one of the mobile industry’s top priorities due to operators' net zero targets, coupled with the need to keep operating costs down amid continued growth in data traffic and uncertainties around energy security. Energy ranks among the largest operating costs related to network operations — accounting for up to 90% when site rental costs are excluded, according to the GSMA. Furthermore, Ericsson is projecting that mobile data traffic will continue its uninterrupted exponential trajectory, with the growth being driven by 5G networks after 2025. 5G networks were designed with energy efficiency in mind, further reinforced by the vision of 5G networks to break the rising energy curve that was observed with every deployment of the previous generations of networks.

Appendix: