Modeling And Analysis Of Fpga Hippocampus Pyramidal Neurons

The research on the nervous system which is high dimension, nonlinear and strong coupling is an important part of life science. Neuron model can replace the animal experiments to study the dynamics of nervous system. Numerical simulation has been the important tool to research the dynamics of the nervous system as a complement for physical experiments since the digital computers came out. However, the current general-purpose computing performance of the computer still can not meet the large-scale, complex biological neural network research requirements. Design of digital logic circuits parallel computing method can solve the contradiction between the complex model and the computing power. It can calculate the complex numerical solution of nonlinear ordinary differential equations accurately and quickly. Field programmable gate array (FPGA) is a digital integrated circuit chip and reconfigurable. Application of FPGA to study the dynamics of large-scale neural networks is significant.In this paper, we use FPGA to calculate the hippocampus CA3 pyramidal neuron model which is the reduced ion-channel model with two compartments including soma and dendrite. We study deeply in many issues such as the structural design of the pipeline, fixed data bit-width and FPGA resource optimization in the modeling process. We propose the method of modeling based on the blocks in the software of DSP Builder and establish a wide range of applied neural network analysis of network dynamics of functional blocks. We also use register arrays to implement the neuron networks coupled by chemical synapses in FPGA.Our designs are implemented on the hardware circuit. FPGA computing results can be uploaded via USB data bus to the general-purpose computer and also can be converted to real-time analog voltage output with real time scale using the convention technology between digital signal and analog signal. The results show that, FPGA results and computer simulation results are consistent.The FPGA modeling system not only provides a new solution for the large-scale neuron networks study, but also make the true implementation of artificial neurons and real neurons interact friendly. The most important, it provides new technological means for the neural repair of medical and non-animal experiments.