29-09-2016 | | By Paul Whytock
A nanoscale electronic device has been developed that may be the missing link in the development of body implants that use electrical signals from the brain to treat medical conditions.
Research at the University of Southampton have shown that memristors could provide real-time processing of neuronal signals, also known as spiking events, and that this could lead to efficient data compression and the potential to develop precise neuro-prosthetics and bio-electronic medicines.
Memristor technology is not a new idea but successfully applying it has proved technically challenging. It’s been around since the early 1970s and is a non-linear passive two-terminal device that relates electric charge and magnetic flux linkage. The memristor's electrical resistance is not constant but relates to the pattern of current that has travelled through the device. In other words it remembers its history.
Southampton University says its use to monitor neuronal cell activity is significant to neuroscience and the development of neuro-prosthetics – biomedically engineered devices that are driven by neural activity. However, a persistent problem is getting the device to process neural data in real-time.
The research team developed a nanoscale Memristive Integrating Sensor (MIS) into which was fed a series of voltage-time samples which replicated neuronal electrical activity.
Acting like synapses in the brain the MIS was able to encode and compress up to 200 times neuronal spiking activity recorded by multi-electrode arrays. Besides resolving the bandwidth constraints this approach proved very power efficient – the power needed per recording channel was up to 100 times less when compared to current technology.
But will memristors prove feasible for use in a far wider variety of applications?
HP had ambitious plans for the device several years ago and said it would build a system called The Machine that would be made up entirely of memoristors and that this would drastically reduce the time it takes for different memory systems to work together.
However, the company backtracked on that plan and decided to use conventional DRAM instead.
It is certainly the case that memristors are seen in theory as replacements for DRAM, flash and disk memory but there is a fundamental and substantial problem with their application.
Integrating them straight into an existing computational architecture is just not possible. Many industry technologists believe the system has to be completely designed to accommodate them.
Those problems aside there is no doubt that memristors will be used to create extremely fast memory devices that can retain copious amounts of data whilst being very power efficient.