System Identification Based on Quantized I/O Data Corrupted with Noise
Hiromi SUZUKI, Toshiharu SUGIE
This paper proposes an identification method for discrete-time linear systems based on the quantized input-output (I/O) data in the presence of measurement noises. First, the parameter estimation error due to quantization of I/O data is analyzed from the statistical viewpoint when we adopt an instrumental variable (IV) method and the least squares (LS) one. Then, based on the result, an IV method suitably adjusted for quantized data is proposed. The validity of the proposed method is evaluated through a numerical example.