Decentralized robust control invariance for a network of integrators

Robust control of networked storage devices is considered. Each network node is a storage element modeled as single-state, discrete time integrator with bounded control input and subject to additive bounded disturbance. Nodes exchange matter through links of limited capacity. We characterize the maximal robust control invariant set and consider a decentralized solution to the robust feasibility problem. We show how to compute the set of link capacities which ensure the feasibility of the proposed decentralized design by using convex optimization. The results allow to guarantee persistent feasibility when decentralized control strategies are used to robustly control the network flow while satisfying nodes and links capacity constraints.