Information Freshness and Resource-Efficient Scheduling in Ship–Shore Networks: An ADMM-DRL Framework

Abstract

Maintaining real-time synchronization is a critical requirement in intelligent maritime systems, fundamentally constrained by limited satellite bandwidth and dynamic channel conditions. This paper explored a consistency-constrained resource scheduling problem in ship–shore collaborative networks, where communication delay, computation latency, and information freshness are jointly characterized using the Age of Information (AoI) metric. The problem is formulated as a mixed-integer nonlinear program (MINLP) involving tightly coupled continuous and discrete decision variables. A hybrid framework integrating ADMM with DDPG is proposed: ADMM decomposes the global MINLP into distributed subproblems, coordinating consistency via iterative dual-variable updates; DDPG agents learn adaptive scheduling policies within each continuous-action subproblem. A novel reward function grounded in the augmented Lagrangian embeds AoI constraints directly into the reinforcement learning objective. Simulations under varying sea conditions—including severe Beaufort-scale storms—demonstrate faster convergence, AoI maintenance within thresholds, and reduced satellite bandwidth consumption versus baseline methods (pure DDPG, greedy, random policies).