The development of a signal classification method based on Chua’s oscillator within the reservoir computing framework

Practical aspects of signal processing using Chua’s chaotic oscillator as a computational substrate are considered. The research is inspired by a growing interest in the framework of unconventional computations that involve the inherent properties of different complex systems known as the «reservoir computing framework». The study proves that Chua’s oscillator meets such a requirement for being used as computation media as the ability to non-linearly transform input data and possesses a short-term memory. To control Chua’s oscillator a special control parameter is introduced to enable the circuit in the chaotic mode to produce non-linear oscillations, for which the form of the attractor in the state space is definitely determined by the control parameter. Besides, the symmetry of the attractor is used to estimate the external influence on the oscillator. As a result, the control and readout methods are developed to apply Chua’s oscillator as the so-called reservoir according to the reservoir computing framework. To exemplify the implementation of the signal processing method according to the reservoir computing framework a classifier of square, triangle and sinusoidal waves is developed. The simulation as well as prototyping of the electronic device show prospects to use Chua’s oscillator as the basis of an analog computations accelerator to perform narrow tasks in hybrid digital-analog control systems for non-industrial robots and smart devices.