The increasing exploitation of renewable energy sources for power generation introduces a significant instability into the power grid, which has to be addressed with appropriate management strategies. Energy storage is a costly and inefficient solution. Demand-side control mechanisms can help mitigating the unbalance between available supply and demand. This includes both direct and indirect control, depending on the degree of controllability of demand-side loads. In the latter, congestion on the shared resource is managed using a price signal, exchanged throughout the power grid and reflecting the resource availability. This requires the timely exchange of information between energy consumers and producers, namely power and phase measurements to be used for the resource pricing. Furthermore, more fine-grained usage data is progressively becoming available to utilities thanks to the deployment of smart meters. Such an information is also relevant to facility managers and users, to become aware of the energy footprint of daily activities and seek a more efficient usage process. This thesis deals with different applications of high-resolution power usage data for energy management in smart microgrids. To this end, the first stage included a measurement campaign in selected households in Italy and Austria. The resulting dataset, named GREEND, contains more than 1 year consumption data at 1 Hz. GREEND was released to the research community for open use, as well as used throughout the thesis. We elaborate on the design of a data infrastructure capable of collecting data from heterogeneous data sources in highly dynamic environments. Specifically, architectural requirements are identified to achieve interoperability at the level of electrical devices as well as exchanged data. The proposed solution offers a single interface to query for status changes, which eases the application development process. In addition, we propose an ontology modeling both static and dynamic information of household appliances. This allows for the full integration of smart and non-smart devices, whose behavior can be tracked and recorded in a sort of datasheet to be exchanged across the network. The availability of energy usage data allows for the provisioning of value-added services to both end-users and utilities. To this end, we investigate on the possibility of an interactive system to timely inform users on their energy usage, in order to promote an efficient use of local resources. In particular, advices are returned to consumers based on their usage behavior and building occupance. Using the GREEND, this solution alone was quantified as potentially yielding up to 34% of savings. However, the effectiveness of demand response programs is greatly affected by the possibility to automate specific devices. Towards this vision, we introduced the HEMS market simulator, which allows for training appliance controllers. Because of the strictly competitive setting, pure market mechanisms do not offer a complete solution for automatic load management. Accordingly, competition is limited to a specific trading day, and has the potential effect of yielding service interruption. To solve this issue, we propose a microgrid power broker that acts as a retailer of available supply. The broker seeks profit by forecasting the price of different power provisioning durations. The three different approaches are independent and give an individual contribution to the research community. The results provide the basis for future research in the field of energy management systems for microgrids and smart buildings.