Signal approximation using GA guided wavelet decomposition
Gabriel Oltean, Laura-Nicoleta Ivanciu, Botond Sándor Kirei · 2015
Signal approximation is a matter of great interest, as working with complete time-sampled signals requires great memory and computational resources. In order to diminish these requirements, signal compression and signal approximation methods are widely used. The paper proposes a signal approximation method, using a genetic algorithm that guides the wavelet decomposition process, by providing specific information, such as: the mother wavelet, the number of selected coefficients, and the decomposition level. The tradeoff between the quality of the signal approximation and its complexity is addressed in the objective function of the genetic algorithm. The method is validated using three test signals, specific to analog circuits. Simulation results prove that the method provides substantial dimensionality reduction, with increased accuracy, which makes it a viable candidate for applications that employ signal storage, transmission, and processing.