Signal estimation with a noisy saturating sensor

Summary form only given. Many biological and man-made sensors saturate outside a relatively narrow range of input signal values. Although such sensors are clearly nonlinear, they often exhibit nearly linear response within their range of high sensitivity. The problem considered in this work is estimating a signal using noisy measurements from such a sensor. The sensor considered has a high-gain (or low noise) central measurement region surrounded by low-gain (or high noise) regions. Three Kalman filter-based estimation schemes that rely on the general signal, sensor, and a particular estimator model are discussed.<
>