Fighting technological obsolescence


For many, the term ‘obsolescence’ has become synonymous with the industrial conspiracy to make consumer goods unsustainable. But in addition to planned obsolescence, technological obsolescence is of great concern.

Iconic but controversial, the light bulb that has burned virtually without interruption since 1901 in a fire station in Livermore, California is proof, according to many, of manufacturers’ planned obsolescence of products. Certainly, a cartel of major light bulb manufacturers called Phoebus existed in the 1930s. But did it really tell its engineers to design bulbs that burn out after 1,000 hours? Given the economic context of the Great Depression, it’s possible that the cartel did so to revive stagnant production, but after a public inquiry it was found guilty not of planned obsolescence but price fixing. The question of Phoebus’s guilt aside, the Livermore bulb’s efficiency has declined significantly ; originally generating 60 watts, it now musters only four while using the same energy. In other words, to generate the original light intensity, it would have to consume much more electricity. There is therefore a trade-off between product life and energy consumption.

In any case,’ explains David Bol, a professor at the Louvain School of Engineering’s Institute of Information and Communication Technologies, Electronics and Applied Mathematics (ICTEAM), ‘this type of obsolescence, if it’s proven, doesn’t really interest technology researchers, because if it’s been programmed, from a technical point of view all you have to do to solve the problem is deprogram it. Instead, we’re working on technological obsolescence, for example, wear, that is, the degradation of devices, or obsolescence related to technological evolution, which renders devices incapable of supporting new generations of products with which they must interact.’ For Prof. Bol, the latter type of obsolescence, which is very common in information and communication technologies, is mainly a consequence of Moore's famous 1965 law, which states that the number of transistors (the building blocks of an integrated circuit such as a processor) that can be implanted on an integrated circuit doubles every 18 to 24 months, rapidly making products uncompetitive. Computer scientists know that every two years, electronics will provide a new processor with more memory and better performance; they therefore program more complex, energy-guzzling applications.

Internet of things

Prof. Bol and his team focus on the Internet of things, that is, the interconnection via the Internet of everyday objects. This requires a miniaturised electronic system that acquires, processes and transfers sensor data to the cloud, where it can be accessed by the user or used to act on the physical world. The central piece of this system is a processor, specifically a microcontroller (a specialised microprocessor at the heart of an autonomous system that manages devices such as sensors, actuators, etc.). A simple example of such a system is a boiler controlled remotely. ‘Our task,’ Prof. Bol says, ‘is to identify and reduce the negative environmental impacts of such systems.’ One challenge is that current microcontroller consumption is too great for small, low-cast batteries to last more than two or three years. We know that in ten years there will probably be several hundred billion connected objects, thus changing (or recharging) all their batteries poses an obvious problem. The Louvain-la-Neuve team has worked on reducing microcontroller energy consumption and developed a design that consumes about 50 times less energy than designs currently on the market. Called ‘SleepRunner’ – because its active consumption is of the same order of magnitude as that of other microcontrollers when they’re on standby – it was just presented at the International Solid State Circuit Conference (ISSCC) in San Francisco.

Low tension

The goal is to power connected objects not via battery but ambient energy recovery (see a recent ScienceToday article), using either solar microcells or thermoelectric recuperators that recover the energy of ambient heat. To achieve this, SleepRunner was designed to operate particularly with a much lower than standard supply voltage: 0.4V instead of 1V. Lower voltage means a significant reduction in energy but requires meeting several technical challenges. The microcontroller will thus be more sensitive to environmental factors, especially temperature variations. Similarly, at low voltage, the impact of the manufacturing process is more pronounced.

Rebound effect

The success of Prof. Bol's team doesn’t stop him from looking ahead. ‘Avoiding battery replacement and extending the life of connected objects are obviously good things. But the rebound effect is very important in information and communication technologies: the cheaper, easier-to-use and more sustainable an application, the more it will be used and deployed. For example, systems that no longer require a battery are likely to multiply. Is this a good thing? We risk losing all the benefits of lower energy consumption. ICT greenhouse gas emissions are growing by 9% per year. In this area, as in many others, I believe restraint is in order.

Henri Dupuis

A glance at David Bol's bio

David Bol earned a bachelor’s degree in civil engineering in electromechanics from UCLouvain in 2004, to understand how technologies work. While writing his PhD thesis, which he defended in 2008, he became interested in low consumption for electronic systems, especially hearing aids, which consume a huge number of batteries. He did his postdoc in the field of video compression at both UCLouvain and the startup PIX, then conducted research in Berkeley on the carbon footprint of electronic circuit manufacturing, which heightened his awareness of the general unsustainability of electronics: ‘Not only does it take several kilos of material to make a chip of a few grams, it also requires a lot of energy, especially since the size of the transistors that constitute it decreases. In addition, the transfer of a gigabyte of data causes the emission of ½ to 1 kg of CO2.’ Appointed professor in 2012, he helped create the spin-off e-peas semiconductors. Alongside his research and teaching, he continually raises young people’s awareness of the environmental costs of Internet use.

Published on June 13, 2019