Stealth Data Injection Attacks With Sparsity Constraints

Sparse stealth attack constructions that minimize the mutual information between state variables and observations are proposed. The construction is formulated as design of a multivariate Gaussian distribution aims to while limiting Kullback-Leibler divergence under without attack. sparsity constraint incorporated support distribution. Two heuristic greedy algorithms for first algorithm assumes vector consists independent entries, therefore, requires no communication different attacked locations. second considers correlation entries which results in larger disruption smaller probability detection at expense coordination We numerically evaluate performance proposed on IEEE test systems show it feasible construct attacks generate significant with low number compromised sensors.