Research Projects

The current memory and computing technologies struggle to meet the demands of AI and IoT due to their separate storage units, resulting in high power consumption and limited computational speeds. The brain relies on parallel low-precision processes, consuming less energy and reducing time and energy penalties. Brain-inspired neuromorphic memory aims to emulate the brain through artificial synapses built from electronic components. However, current synapse designs use transistors, which are reliable, deterministic, and fast but lack intrinsic memory functionality. Skyrmion-based artificial synapses offer a perfect platform for realizing and developing the full potential of neuromorphic memory due to their small size, plasticity, non-volatility, stochasticity, and simpler architecture. Skyrmions are ultra-small, self-closing, chiral magnetic domains that are topologically protected from undesired errors and have fast movement and manipulation at low energy costs. A skyrmion-based synapse device consists of a pre- and post-synaptic region, which can be read through multiple electrical measurements including AHE, The and magnetoresistance (MR). The density, size, and number of skyrmions influence the AHE, THE or MR, enabling the reading of multiple states on these synaptic devices. The proposed prototype consists of skyrmion synapse devices, which could be used as a full-fledged neuromorphic in-memory computing electronics hardware. This hardware would provide the country with the necessary indigenous capability to support rapid advancements in AI, IoT, and other sectors, such as finance, banking, military, and space applications.