A Distributed optimization Algorithm for Model Predictive Control Problem of Linear Systems Under Communication Noise

We study the distributed model predictive control (DMPC) problem for a group of linear discrete-time systems with both local constraints and global constraints in the presence of stochastic communication noise. The dual form of the DMPC optimization problem is transformed into a stochastic distributed consensus optimization problem by modeling the exchanged variables as stochastic ones and a novel stochastic alternating direction multiplier method (ADMM) is proposed to solve it in a fully distributed way. The effectiveness of the proposed stochastic ADMM algorithm is verified through an simulation example.