Joint estimation-detection of cyber attacks in smart grids: Bayesian and non-Bayesian formulations

Smart grid operations face a significant threat from the presence of cyber attacks or bad data that may contaminate the system observations. Therefore, in this paper, we are interested in introducing a new strategy for detecting the presence of bad data in smart grids and we also try to simultaneously estimate it in order to be able to separate the bad data from the system observations. We aim to obtain the attack free observations which reflect the true state of the smart grid. This can be done by defining a joint detection-estimation strategy based on Bayesian and non-Bayesian settings where the costs in general will be functions of the observation. We start with Bayes approach and derive the detector (which, in general, may not be a LRT) and then we set the problem by defining some maximum constraint under the null hypothesis based on the derived detector and minimize certain cost under the alternative hypothesis. Our results reveal that the proposed model is applicable on some cases that other models reported in previous works failed to deal with.