The optimum management of local resources and storages, inside a Smart Micro Grid (SMG), is a relevant issue related to energy efficiency and sustainability. In this paper, a centralized optimization tool for real time power dispatching in SMG, based on linear programming, is presented. The optimization objective is to make SMG a self-sustainable system, that means a nearly zero energy-impact system on the main grid, thanks to the maximum exploitation of the local resources including storage. Moreover, a communication protocol to enable power management is proposed, so as how it can be implementable via Carrier Sensing Multiple Access (CSMA)-like protocol. A numerical application is reported and the results validate the real effectiveness of the proposed approach.