Identification of saturated networked systems

We address the challenge of parameter identification in networks of neural mass models. We consider a system of n interconnected populations, structured in successive layers, where measurements are available from only one. Traditional offline methods struggle with parameter ambiguity and time-dependent variability. We propose an online approach that exploits the system’s nonlinear characteristics, particularly saturation functions, to enable effective parameter recovery. By designing specific control inputs applied to targeted nodes, it becomes possible to isolate and retrieve parameters effectively. Crucially, we show that this process must be conducted online to achieve full identification, as control inputs must adapt to previously unknown states and parameters.