Off-line detection and estimation of abrupt changes corrupted by multiplicative colored Gaussian noise

The problem addressed in the paper is the detection of abrupt changes embedded in multiplicative colored Gaussian noise. The multiplicative noise is modeled by an AR process. The Neyman Pearson detector is developed when the abrupt change and noise parameters are known. This detector constitutes a reference to which suboptimal detectors can be compared. In practical applications, the abrupt change and noise parameters have to be estimated. The maximum likelihood estimator for these parameters is then derived. This allows to study the generalized likelihood ratio detector.