This thesis contributes to the planning of ICT infrastructures for smart grids and the optimization of their applications from a research point of view, incorporating communication networks and power grids in the design phase. The particular focus lies on wide area measurement systems (WAMSs), which consist of phasor measurement units (PMUs), phasor data concentrators (PDCs), data centers, and the communication links between them. WAMSs play a crucial role for numerous applications since they collect the high-resolution and accurate synchrophasor measurements and transmit them to energy management systems, where the information is turned into decisions and actions.

In the scope of this thesis, a novel approach is proposed for the joint planning of a WAMS with a hierarchical heterogeneous communication network. The planning approach takes into account the requirements of both the power grid and the communication network. Unlike other works in the literature, the presented mathematical optimization models allow for planning with multiple communication technologies. Furthermore, the design of the hierarchical communication network is handled jointly with the optimization of both the numbers and the locations of PMUs and PDCs in one comprehensive model. It is shown that the deployment costs can significantly be reduced due to the joint optimization. For tackling large instances of the combinatorial optimization problems, a customized genetic algorithm is developed and implemented.

A further contribution of the thesis focuses on the operation of wide area protection applications which use the synchrophasor data collected by WAMSs. In particular, motivated by the operation principles of data concentrators in a WAMS, a novel approach is proposed to improve the performance in machine learning-based fault detection. The originality lies in training the classifiers under consideration of the network delays of individual synchrophasor measurement streams. The presented simulation results show that system faults can be detected with a shorter response time as a result of the proposed training scheme.

Finally, an integrated simulation environment is presented, which has been developed by the coupling of domain-specific simulators and thus enables the investigation of the interdependencies between communication networks and power grids.