Adaptive estimator of signal amplitude in unknown noise environment

The authors propose a new estimator which unifies linear matched filters and matched median filters. The optimality of linear matched filters and matched median under Gaussian noise and Laplacian noise, respectively, makes the new estimator suited to various noise environments. An adaptive algorithm is derived for parameter estimation of the optimal estimator under the mean square error criterion. As an application of the new adaptive estimator (AE), detection of antipodal signals was simulated. The results demonstrate that the AE had almost the same performance as the linear matched filter and the matched median filter under Gaussian and Laplacian noise, respectively, while under contaminated Gaussian noise, it performed better than the two filters