A robust SDP approach to system identification with roughly quantized data

This paper proposes an identification method for linear systems with roughly quantized outputs. Measurement data sampled from low resolution sensors have large quantization errors, which deteriorate the identification accuracy. While the identification problem is formulated into quadratic programming with uncertainty, a proposed method provides an approximate optimal solution via semidefinite programming. Numerical examples demonstrate that we can estimate both plant parameters and true outputs in practical time and show the effectiveness of the proposed method.