Skip to Main Content
In this paper, a novel optimization methodology is investigated to optimize the resource allocation in a satellite system where variations of fading conditions are added to those of traffic load. A neural approximation technique is studied to exploit on tine optimal reallocation laws, as functions of the state of the network. No closed-form expressions for the system dynamic equations and the functional cost are needed. Simulation results show how the proposed technique outperms two other optimization strategies, based on a certainty equivalent assumption and on the application of perturbation analysis, respectively.