Bias-compensated identification of quadratic Volterra system with noisy input and output

An iterative approach to identification of a quadratic Volterra system with noisy input-output is proposed, whereby the bias-compensated least-squares method of identifying a noisy FIR model is utilised with some modification to estimate input/output noise variances and bias-removed Volterra system parameters. In particular, the proposed identification approach yields better performance even in cases of fewer input/output data than conventional methods, and it can be also extended to identification of noisy higher-order Volterra systems.