Toward Biorealistic Silicon Neural Circuits on Reconfigurable Platforms

Reconfigurable analog hardware presents an ideal platform for the realization and analysis of biorealistic neural systems. This work explains the use of subthreshold MOSFETs to model biological channels and proposes energy-efficient, bioin-spired neuron and synapse circuits for reconfigurable analog hardware. The proposed circuits are experimentally evaluated in the context of a two-neuron oscillator on an in-house SoC field-programmable analog array in a 350 nm CMOS process.