Microgrid data aggregation and wireless transfer scheduling in the presence of time sensitive events

Microgrids enable a network of distributed energy generators to sustain energy needs off-the-grid. Microgrids can experience islanded operational mode, being this mode a time sensitive event that affects costs of power generation and distribution. The detection of time-sensitive events is important because the control unit needs to be aware of changes in the grid to avoid losses in power quality and costs. This requires a quality of service (QoS)-aware data aggregation and queuing mechanism in the core of the network infrastructure to convey microgrid data to a central server (considered as a macro base station). This paper investigates the impact of time sensitivity-based microgrid data aggregation on message delivery under different priority and time-sensitivity levels. Hence, we propose a framework to cluster the electrical data based on the time sensitivity criteria using unsupervised machine learning. We introduce a multi-class queuing system in the pico-cells to ensure that clustering reduces the processing time for high priority data. The results show that the proposed approach significantly reduces the delivery delay of messages carrying time sensitive events from the microgrid.