Abstract
In the past decade, many detection and estimation algorithms have been reported for estimating a desired Bernoulli-Gaussian signal which was distorted by a linear time-invariant system. The well known Kormylo and Mendel's single most likely replacement (SMLR) algorithm, which works well and has been successfully used to process real seismic data, is an offline signal processing algorithm. The paper proposes a recursive SMLR algorithm which has online data processing capabilities and requires much less computational effort than Chi and Mendel's recursive algorithm and Goussard and Demoment's recursive algorithm. Simulation results show good performance.