Identification Using Binary Measurements for IIR Systems

Recursive system identification method from binary input/output measurements

“…Most of those algorithms are dedicated to the identification in the case of a one-bit quantization only on the output. They are based on a specific design of the input such as in [1,[3][4][5] or a specific identification criteria as in [6][7][8][9][10][11] or on an assumption on the knowledge of noise distribution [1,[12][13][14][15]. In addition to both being not adapted to real-time identification for some of them, these latest methods require full knowledge of the input; they are consequently not adapted to our context.…”

Impulse response identification from input/output binary measurements

“…These methods address the identification problem in different ways: some of them use a periodic input signal ( [14]), others use the knowledge of the noise distribution function ( [14], [16], [5], etc.) and some approaches are based on a specific identification criteria ( [3], [10], [11], etc.). These previous solutions are dedicated to the identification of Finite Impulse Response (FIR) systems but there also exist some solutions for the identification of Infinite Impulse Response (IIR) systems (see for instance [13]).…”

Learning Dynamical Systems From Quantized Observations: A Bayesian Perspective

“…Several contributions have been also proposed in the more general case of multi-level quantized data. Results on identification of FIR and IIR models are given in [5]- [7], [21] in the set-membership setting, where a bounded description of the quantization error is adopted. Contributions are also available in the stochastic framework.…”