Least-squares linear estimators using measurements transmitted by different sensors with packet dropouts

The least-squares linear estimation problem (including prediction, filtering and fixed-point smoothing) from measurements transmitted by different sensors subject to random packet dropouts is addressed. For each sensor, a different Bernoulli sequence is used to model the packet dropout process. Under the assumption that the signal evolution model is unknown, recursive estimation algorithms are derived by an innovation approach, requiring only information about the covariances of the processes involved in the observation equation, as well as the knowledge of the dropout probabilities at each sensor.