A Data-Flow Architecture for fast computation of Hodgkin-Huxley Neuron Equations

Abstract

Event Abstract Back to Event A Data-Flow Architecture for fast computation of Hodgkin-Huxley Neuron Equations Marcel Beuler1* and Werner Bonath1 1 University of Applied Sciences Giessen-Friedberg, Electrical Engineering and Information Technology, Germany The Hodgkin-Huxley-Equations allow detailed simulations of neurons in analogy to electrical nonlinear networks. Due to the complex numerics the software-based simulation of this equation set is quite time-consuming and therefore exist several approaches to model those equations with analogue or digital electronic circuitry.The presented Data-Flow Architecture implements a hardware based simulation of an enhanced Hodgkin-Huxley-Model (Huber-Braun-Model). Computing about 2500 arithmetical operations including mathematical functions results a step size of 100 µs. The main advantage of our architecture is the fast and accurate computation speed, which allows the computation of 40 neurons on state-of-the-art programmable logic (FPGA). The developed architecture is shown in Figure 1. To overcome numerical problems, we chose a 32-bit floating-point data format. By a step size of 100 µs and a frequency of 10 MHz there are 1000 clock cycles to compute all model equations during a neuron cycle. At the present state a network topology and the depolarizing of the membrane potential by postsynaptic potentials are still excluded to reduce complexity. When using 200 clock cycles for these operations, there will be 800 clock cycles still available for computing the equations of 40 neurons.A neuron core consists of 6 arithmetical units and an addition unit as a source of uniform distributed pseudo random numbers. Each unit has its own Code-memory with a 28-bit instruction format for individual programming, a FIFO to store the results of operations and own digital memory. To realize the required mathematical functions in a uniform architecture, the CORDIC algorithm in a 32-bit floating-point environment and intensive pipelining to achieve a high throughput is used. Analyzing the divisions in our neuron model shows a maximum quotient of 1, so the CORDIC algorithm can also be used for this operation and no divider is necessary. The presented architecture can be seen as the condition of implementing a network topology in the next turn of development a digital neuron processor for the desired model. Conference: Neuroinformatics 2009, Pilsen, Czechia, 6 Sep - 8 Sep, 2009. Presentation Type: Poster Presentation Topic: Computational neuroscience Citation: Beuler M and Bonath W (2019). A Data-Flow Architecture for fast computation of Hodgkin-Huxley Neuron Equations. Front. Neuroinform. Conference Abstract: Neuroinformatics 2009. doi: 10.3389/conf.neuro.11.2009.08.109 Copyright: The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers. They are made available through the Frontiers publishing platform as a service to conference organizers and presenters. The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated. Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed. For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions. Received: 25 May 2009; Published Online: 09 May 2019. * Correspondence: Marcel Beuler, University of Applied Sciences Giessen-Friedberg, Electrical Engineering and Information Technology, Giessen, Germany, marcel.beuler@web.de Login Required This action requires you to be registered with Frontiers and logged in. To register or login click here. Abstract Info Abstract The Authors in Frontiers Marcel Beuler Werner Bonath Google Marcel Beuler Werner Bonath Google Scholar Marcel Beuler Werner Bonath PubMed Marcel Beuler Werner Bonath Related Article in Frontiers Google Scholar PubMed Abstract Close Back to top Javascript is disabled. Please enable Javascript in your browser settings in order to see all the content on this page.