Abstracts Category : Other

Operation and reconstruction of signals based on integrate-and-fire conversion using FPGA

by Luis Leito Guilherme

Neurons are diverse in terms of functionality and behaviour; however a family of them can be modeled as an electrical circuit aggregating input currents from neighboring neurons. Integration-and-Fire Neuron (IFN) is a simplified view of a neuron operation that reduces the model to a first order fire-and-reset system. IFN can be used as a modulation technique to transmit data in a very efficient way. It is also possible to perform some signal operations in the IFN space, just looking into the time space in-between IFN spikes, and there are several approaches to reconstruct back the analog signal, at the receiver side, either in time-domain or frequency-domain. After the analysis of the different methods for signal reconstruction based on IFN modulation technique, and also after surveying different methods of doing mathematical operations in the IFN space, the main goal of the thesis is then to synthesize, on an FPGA, the fundamental mathematical operations over the spike based signals, together with a reconstruction method that will convert the IFN signal back to a magnitude/time signal, with a reasonable bit resolution.Advisors/Committee Members: Faculdade de Engenharia.