IEEE Future Directions
Related Articles
Storage technologies, along with processing and communications technologies, have been the engine of evolution in these last sixty years. The more storage we have and the more storage we need.
However, it is not just about “quantity”. As a matter of fact we have now so much storage in use that the amount of power expended in the storage of data has become huge, with some figures pointing to one fifth of the overall electrical power consumption worldwide tied to data centres and communication by 2025. Storage has a significant share in this power consumption.
RAM technologies are the ones mostly used, they are fast and have seen a tremendous increase in performance making them ubiquitous. Alternative technologies are being investigated with particular interest on those using the magnetic component of the electromagnetic field. Spintronics, mentioned a few times in this blog, is one of them. In lay terms it leverages on the magnetic spin of electrons to represent a bit. Electrons do not have to move around to indicate the values 0 or 1, they can stand still (for whatever standing still can mean for an electron…) rotating on the spot (for whatever rotating may mean to an electron…). Now moving around leads to power dissipation spinning on the spot does not (the mysteries of quantum mechanics!).
Spinning is what give rise to the magnetic field and having materials that can align the spins of their electrons in a certain direction creates a magnetic field (in a magnet the spin of electrons is mostly oriented in a certain direction and this prevalence generates a magnetic field in that direction). By changing the spin we can change the orientation of the magnetic field, hence change the bit value from one to zero.
A team of researchers at the University of New Hampshire have reported in an article of their discovery os a thin film, iron based, material that shows nanomagnetic properties that can be used in spintronic applications. The material shows a perpendicular magnetic anisotropy that is over 24 times the ones achieved so far (basically it creates a magnetic field that is 24 times stronger..).
For the technical geek this is of high interest, for the rest of us it brings the promises of creating very dense storage medium that requires extremely low power, able to preserve data for over 10 years (which is a lot in RAM technologies).
Interestingly, I spoke yesterday to a lead researcher in ST Microelectronics in Italy and he pointed out the shift from a processing in the cloud of sensors data to a processing at the edges, near the sensors (and in some cases even in the sensors themselves). The stumbling block today is the availability of storage that can consume very very little power. The gain in this shift would be the possibility to create intelligence at the edges. The application of artificial intelligence algorithms and structures in the cloud is supported by the huge processing power and energy available in the cloud; the availability of very low power storage would allow AI at the edges, considering that there is processing power at the edges, at least enough for managing the multitude of data generated by sensors.
This evolution is in line with the creation of aware ambient and in general with the rise of autonomous systems.
