Achieving resilience in distributed optimization against adversarial attacks

This paper provides an in-depth investigation into developing distributed optimization algorithms that are resilient against adversarial cyber-attacks that compromise nodes and disrupt the optimization process. A resilient distributed optimization algorithm with trusted agents (RDO-T) is proposed that utilizes a subset of trusted nodes to constrain and overcome the impacts of attacks on the network. Theoretical analysis on the convergence properties and performance optimality of RDO-T is provided.