Distributed Mixed-Integer ADMM for Express Transport Network Design

This paper focuses on decentralized decision-making in Express Transportation Network Design (ETND) for multi-region networks, where collaboration is achieved through package transshipment to minimize total transportation costs. We introduce a two-network representation to model vehicle and package flows separately, enabling efficient handling of transshipment operations. The formulation is presented as a path-based Mixed-Integer Program (MIP) with coupling constraints to capture interregional interactions. To solve the ETND problem in a distributed manner, we develop a Mixed-Integer Alternating Direction Method of Multipliers (MI-ADMM) algorithm, which decomposes the task into region-specific subproblems solved in parallel. Numerical experiments show that MI-ADMM achieves competitive performance compared to a centralized global MIP approach.