Distributed Constrained Optimization for Nonlinear Stochastic Multi-Agent Systems: Application to Resource Allocation
Quanxin Zhu, Haokun Hu · Complex Systems Stability & Control · 2025
This paper investigates the resource allocation problem (RAP) for nonlinear stochastic systems subject to random disturbances. The communication network is modeled as a weight-balanced digraph, where each agent can only access its own differentiable and strongly convex local cost function. A fully distributed adaptive state-feedback algorithm is proposed, and rigorous analysis shows that the decision variables converge almost surely to the optimal solution. Unlike existing studies on deterministic RAPs, the system considered here is affected by two types of stochastic factors-Brownian motion and unknown nonlinear dynamics—which significantly increase the difficulty of algorithm design. Finally, numerical simulations on a resource allocation example are provided to demonstrate the effectiveness of the proposed approach.