generate.md

Signal Simulation

I implemented the possibility to generate simulated signals for evaluation of the filters on realistic data.

Exemplary simulated signal

% you can create a signal with a generic configuration
[signal,echt]  = generate.recording('generic')

You can also adapt or define directly your own setup parameters:

clear setup
setup.NoiseLevel        = 0.1;

setup.erpMagnitude      = 2; % sum of eoModulation and tacsModulation(1)

% tacs parameters
setup.tacsFreq          = 10;
setup.tacsMagnitude     = 20;
setup.tacsSaturate      = 1; %saturation level in percent
setup.tacsDistort       = 0.2; %periodic signal distortion
%setup.tacsDistort       = 0; %no distortion
setup.tacsPhase         = 'random';
%setup.tacsPhase         = 0; %fixed phase -> phase-dependent ERP

% random fluctuation of the artifact amplitude.
% implemented as a Ornstein–Uhlenbeck, i.e. AR (1), process
setup.tacsModulation    = [1,.5]; %variability, stiffness
% setup.tacsModulation    = [0,1]; %no variability, perfect stiffness

% event-locked modulation
setup.eoFreq            = 10;
setup.eoModulation      = 1;
setup.eoPhase           = 'random';

% signal recording parameters
setup.Fs                = 1000; %in Hz
setup.L                 = 4; %Duration of trial in seconds

[signal,echt,tacs]  = generate.recording(setup)
% signal is the signal as it would be recorded
% (black trace in figure)
% echt is a matrix with 2 rows
% first row is the erp (red trace in figure)
% second row the event-induced oscillations
% tacs is the (stationary, undistorted, sinusoidal)
% stimulator current signal (gray trace in figure)

Consider that when tacsPhase and eoPhase as well tacsFreq and eoFreq are identical, they are phase and frequency-matched. Their superposition can be interpreted as event-related exogenous impedance modulation but also as entrainment of an endogenous event-related oscillation. Whether this is an artifact or a signal is a matter of perspective.