Estimation of impulse action parameters using a network of neuronlike oscillators

摘要

Aim of the study is to develop a method for estimating the parameters of an external periodic impulse action using a spiking network of neuronlike oscillators. Methods. The spiking activity of a network consisting of coupled nonidentical neuronlike FitzHugh–Nagumo oscillators was studied, depending on the parameters of the periodic impulse action. To estimate the amplitude of the external impulse signal, we detuned the FitzHugh–Nagumo oscillators, which were in a stable state of equilibrium in the absence of an external action, by the threshold parameter responsible for the excitation of the oscillator. To estimate the frequency of excitatory pulses, we detuned the FitzHugh–Nagumo oscillators by the parameter characterizing the ratio of time scales, the value of which determines the natural frequency of oscillators. We also changed the duration of external pulses. Results. It is shown that the number of spikes generated by a network of nonidentical FitzHugh–Nagumo oscillators has a monotonic dependence on the amplitude of the external pulse signal and a nonmonotonic dependence on the frequency of the pulse signal. The number of spikes generated by the network remains constant over a wide range of external pulse durations. A method for estimating the amplitude and frequency of impulse action is proposed. The method efficiency is demonstrated in numerical simulations and in a radio physical experiment. Conclusion. The proposed method allows one to estimate the amplitude of an external pulse signal, knowing its frequency, and estimate the frequency of this signal, knowing its amplitude. The method can be used in robotics when solving the problems of information processing related to the motion control of mobile robots.