Technology For High-Performing And Complex AI Electronics

730
Advertisement

An enhanced transistor design helps provide efficient analogue computation for robotic control, AI and unsupervised machine learning

A device with the ability to overcome long-standing challenges in electronics design, including next-generation wearables and eco-disposable sensors has been developed by researchers from the University of Surrey. It is expected that the device can perform the same operations as more complex circuits.

Called the Multimodal Transistor (MMT), the device has immunity to parasitic effects that reduces a transistor’s capacity to produce uniform, repeatable signals. These have hindered traditional “floating gate” designs ever since their invention in 1967, but this new structure promises efficient analogue computation for robotic control, AI and unsupervised machine learning.

Traditionally, gate electrodes are used to control a transistor’s ability to pass current. With this device, ON/OFF switching is controlled independently from the amount of current passing through the structure. This allows the MMT to operate at a higher speed than comparable devices and to have a linear dependence between input and output, essential for ultra-compact digital-to-analogue conversion. This also gives engineers unprecedented freedom of design, which could lead to greatly simplified circuits.

“Multimodal Transistor is a paradigm shift in transistor design. It could change how we create future electronic circuits. Despite its elegantly simple footprint, it truly punches above its weight and could be the key enabler for future wearables and gadgets beyond the current Internet of Things,” said Dr Radu Sporea, Project Lead and Senior Lecturer in Semiconductor Devices at the University of Surrey.

For more details, visit here.


Advertisement


SHARE YOUR THOUGHTS & COMMENTS

Please enter your comment!
Please enter your name here