Trust-based secure distributed nonlinear estimation against adversarial attacks

This article studies secure distributed estimation problem under undirected communication networks, where the measurements and updates at some agents may be manipulated arbitrarily by adversarial attacks. Despite the attacks, normal agents make successive local measurements of the unknown parameter of interest and aim to infer the parameter consistently by sharing these measurements with their neighbors. It is assumed that there are some agents, called trusted agents, that cannot be compromised and then a novel secure distributed estimation algorithm is proposed to ensure that the local estimates at all normal and trusted agents converge to the true value of the objective parameter under some mild assumptions. To show the practical effectiveness of our algorithm, a numerical example is presented.