Leech heartbeat neural network on FPGA

Abstract

Most of rhythmic movements are programmed by central pattern-generating networks that comprise neural oscillators. In this article, we implement a real-time biorealistic central pattern generator (CPG) into digital hardware (FPGA) for future hybrid experiments with biological neurons. This CPG mimics the Leech heartbeat neural network system. This system is composed of a neuron core from Izhikevich model, a biorealistic synaptic core and a topology to configure the table of connectivity of the different neurons. Our implementation needs few resources and few memories. Thanks to that, we could implement network of these CPG for instance to mimic the behavior of a salamander. Our system is validated by comparing our results to biological data. © 2013 Springer-Verlag Berlin Heidelberg.