Spintronics for energy efficient big data meshes


A completely new niche market has emerged in the past 10 years: the small hardware to big-data business. Thanks to the ever expanding telecommunications networks, a big portion of our territories have access to Internet. Using relatively friendly hardware components and affordable data plans, numerous start-ups have been born with the objective of manufacturing meshes of sensors that feed large data centers. One profound business aspect for these newcomers is that most of the value of their final product is the connectivity and, for this fundamental aspect, “Internet is just there” and current data transmission fees are low compared to the costs of the developing the wireless infrastructure. Example: twenty isolated devices that measure temperature are worth much less than a mesh of ten gadgets broadcasting to a data center or your smartphone. While two decades ago one would need to elaborate its own telecommunication solution, nowadays one can just buy an integrated circuit and a SIM card and put them together with the thermometer. And the cost of the required data plan will be orders of magnitude cheaper. Welcome to the “Internet of Things”.


What so far looked like a tempting invitation for either investors and entrepreneurs carries a fundamental problem. Covering large areas with wireless sensors requires to rethink the way each one of those nodes gets energy. Back in the school, we were told that “energy does not create or destroy, it just transforms”. Ironically, it will most likely transform far away from where you are interested in setting up a monitoring station. In other words, there aren’t too many plugs up the hill. Energy is THE primary aspect of the emerging wireless meshes: no energy, no data. I am very concerned about seeing the predictions of how Internet of Things will grow: Forbes forecasts that 1.5 billion devices will be connected in 5 years [1]. Meanwhile, the main reason for which all of us renewed a laptop or a smartphone in the past 2 years is… the short battery life made it unbearable. I truly wish there is a global plan in place to replace billions of batteries. At the moment, the only way to circumvent this problem is to design hardware that is extremely energy efficient.


In the past two years I came across two projects involving magnetic sensing that made possible the monitoring thanks to a hardware design that put the energy management in the center of the project.  I will cover both of them in two respective posts in this stream. The first case brings us to the field of geological monitoring. As you can imagine, monitoring the small displacement in some remote rock masses drives you to beautiful natural landscapes with no energy plugs around. To make it more complicated, the measurements have to be done some tens of meters below the ground to remove the diurnal temperature oscillations that would hinder the displacement measurements. Furthermore, down there, in a dark cave, there is no light and no GPS. Scientists have been in the need of collecting 3 dimensional data so that unidimensional measurements based on elastic wire changes of resistance were not enough. At the moment of writing, the state of the art is to use a pair of Moire patterns printed on crystals and take periodical pictures of them [2]. Deploying hundreds of such automated devices have generated a tremendous data overhead (sending Mbyte-large high-resolution pictures) and a very high energy demand (boot up a computer, take a picture, send it, go to sleep). The analysis of the problem from the big-data and internet-of-things prespective concluded that the measuring princpiple should be changed to reduce both. Recently, we have developed a sensor which is based on spintronics, that is, the change of electrical resistance of a magnetic material caused by the presence of a magnet nearby. This technology perhaps will sacrifice instrumental resolution but will enable the remote monitoring of large meshes. In some case, this trade off is a fundamental make-or-break point in a project. Now, the present technology takes 1 millisecond long measurements and produces a data package of less than 100 bytes containing all the necessary information to reconstruct, in three dimensions, the relative position of the magnet to the magnetic sensor. The tremendous benefits of this devices have been already proven in the pioneering installations done in Canary Islands (Spain) and Eastern Alps (Austria). It is now possible to design broader projects covering more sites in which the scalability of the deployment of instruments is guaranteed [3].

Follow us at igsresearch.com/spintronics


 Xavier Marti,
CTO IGSresearch Ltd.
The Czech Academy of Sciences


[1] Internet Of Things On Pace To Replace Mobile Phones As Most Connected Device In 2018

[2] The monitoring of slow-moving landslides and assessment of stabilization measures using an optical–mechanical crack gauge


[3] Spinterference, using magnetic footprints to measure small geodynamic displacements

https://youtu.be/p98K8XGetdI (1 hour full-lecture)

https://youtu.be/c8NAWagPD1k (3 minutes technical summary)