Energy efficiency and sustainability in wireless

Deep Dive report: Energy efficiency in wireless: RAN and beyond

Are energy efficiency, net-zero emissions and sustainability just greenwashing and hype? We hope not, because we need them to preserve our environment and our lives on the planet. They can also make wireless networks more efficient and reduce costs.  

In wireless, we have the responsibility and the privilege to be a force of good by optimizing energy use in our networks and enabling energy efficiency in other verticals and in society through connectivity to people and things.

Measuring and optimizing energy efficiency is fiendishly complicated – yet intellectually fascinating – because there are multiple complementary ways to improve it and because performance and quality of service cannot be compromised to reduce energy consumption. As traffic and service expectations increase, we are asking our networks to do more with less.

We have to abandon the traditional approach that energy bills are a fixed cost item over which we have little or no control and that networks need to run at all times at capacity to be ready for peak time. Energy consumption can be managed, and operators are increasingly doing so with the help of new and old technology, automation and AI/ML.

In this report and the conversations that follow it, we look at how we can improve energy efficiency in wireless. We look at energy efficiency in the RAN, where most energy is used, but also in the rest of the network. We look at how to measure energy efficiency and benchmark it. We look at how different choices – e.g., open or traditional RAN, virtualization, chipsets, centralized/distributed architectures, indoor/outdoor infrastructure, cellular/Wi-Fi access – may affect energy efficiency. And, finally, we look at how wireless can make the rest of the economy and society more energy efficient.

The report includes conversations with Wind River, Deutsche Telekom, CommScope, Satellite Applications Catapult, VMware, ConnectiviTree, Fujitsu, Intel and Keysight.