Improved range selection method for evolutionary algorithm based adaptive filtering of EEG/ERP signals

A frame work for Adaptive Filter/Adaptive Noise Canceller (AF/ANC) design through Evolutionary Algorithm (EA) is presented as an application in Electroencephalography /Event Related Potentials (EEG/ERP) filtering. Process of parameter setting for EA is also explored. A concept of bounded or controlled search space is proposed to identify the best range for search space. Statistical analysis over the simulation results has been performed to quantitatively identify the range and its control parameter. Differential Evolution (DE), Genetic Algorithm (GA) and Bacterial Foraging Optimization (BFO) are implemented for the design of AF. Testing of AF has been done through consideration of two types of noise (white noise and ongoing EEG noise) over three ERP signals (Simulated Visual Evoked Potential, Real Evoked Potential and Real Sensorimotor Evoked Potential).