From hyperconnectivity to AI: How can we tackle the environmental impact of the new Internet era?

(Nanowerk News) Artificial intelligence training can generate more than 200 tonnes of carbon dioxide, equivalent to the lifetime of five cars.
Wireless communication systems and digital technologies have transformed our society to such an extent that it now seems almost impossible to live without Wi-Fi or a smartphone. The accelerated digitalization of our world has led to many advances, but it has also come with an increase in energy consumption and environmental impact. What if using networks for more than just communications could contribute to real sustainability in the digital world?
The "Evolution and alternative uses of connectivity as a response to the sustainable digitalization of society" (RF-VOLUTION) project, funded by the Spanish Ministry of Science and Innovation's National R&D Plan and led by Prof. Xavier Vilajosana, a researcher at the Internet Interdisciplinary Institute (IN3) at the Universitat Oberta de Catalunya (UOC), aims to address the environmental consequences of digitalization and technological obsolescence by devising solutions that maximize the potential of wireless networks.

The impact of the digital revolution

Data have been with humans ever since the first food counting systems were developed in ancient Mesopotamia (and most likely even earlier). For centuries, they were just another aspect of societies, a value limited in scope to certain sectors. But then digitalization came along and transformed them, allowing them to fulfil their very raison d'ĂȘtre. Our highly connected societies now generate immense amounts of data: Cisco estimates that the internet will move 805 zettabytes (a unit equivalent to one trillion gigabytes) in 2023.
Interest in harnessing this massive amount of data, which will continue to grow in the coming years, has triggered a need for increased computing and communication capacity. This, in turn, requires more and more energy and resources, to the point of upsetting the planet's climate and environmental balances. It is estimated that by 2030, data networks and infrastructures will use 14% of all electrical energy consumed worldwide, and this will have a greater or lesser climate impact depending on how this energy is generated.
In addition to this, the development of artificial intelligence (AI) has taken off in recent years, in particular, machine learning, a technology that needs to process huge amounts of data in order to operate and function. The training process of some AI models alone can generate 284 tonnes of carbon dioxide (the main greenhouse gas). This is the same amount that five cars emit over their lifetime.

Taking advantage of wireless communication

Against this backdrop, the RF-VOLUTION project proposes a radical change in the development of cyber-physical and telecommunications systems. "RF-VOLUTION aims to explore the possibility of giving wireless communication systems more than one function. In other words, beyond transporting information from the devices that make up the network, we want to explore the possibility of using electromagnetic waves as a sensor, turning networks not only into information transport systems but also into elements that collect information from the environment," explained Xavier Vilajosana, lead researcher of the Wireless Networks (WiNe) group.

From environmental sensors to geolocalization

The RF-VOLUTION project builds on the hypothesis that it is possible to develop technological components and communication networks that are sustainable by design by tapping into the alternative uses of wireless networks, which have become ubiquitous in recent years. "It would be ideal to make the most of these infrastructures. Giving them more functions can be seen as improving their efficiency," said Vilajosana, who is also a member of the UOC's Faculty of Computer Science, Multimedia and Telecommunications.
"For example, in terms of energy, if the same energy used to transmit data also allows us to measure a physical parameter of the environment, we are doubling that energy's value," he said. "Other aspects of the project are related to the obsolescence of technology. That is, if we can upgrade an infrastructure to give us more service without having to replace it, we are also giving it new value and even extending its useful life."
The way to achieve this is still open. Networks could be used to provide positioning services, determining the location of an object in the network in relation to the power or direction of the transmitted signal, or as sensors for different physical parameters, recording how the humidity or temperature of the medium influences the electromagnetic signal and opening the door to the design of wireless sensors with simpler, more integrated and power efficient electronics.
With these end uses in mind, the RF-VOLUTION project has four main goals:
  • To develop new passive sensing approaches based on the characteristics of radio frequency systems and different types of antennas.
  • To develop and evaluate approaches to enable the convergence of radio frequency systems for communication and localization.
  • To exploit optimization methods and advances in AI to improve the energy efficiency of networks.
  • To generate scientific impact, both for society and industry.
  • "We have already developed various sensors based on radio frequency identification (RFID) technology, with potential that's relevant for industry, as it is a cheap and widely used technology. We are also working on positioning systems with technologies such as Bluetooth and Wi-Fi. We believe it's necessary to take advantage of existing technologies, because each change or new generation in technology also has its drawbacks, such as electronic waste or the mining of materials needed to manufacture them," concluded Vilajosana.
    Source: Universitat Oberta de Catalunya (Note: Content may be edited for style and length)
    We curated a list with the (what we think) best greentech, sustainability and environmentalism podcasts - check them out!