Energy harvesting of ambient and aeroelastic vibrations is important for reducing the dependence of wireless sensing and networks on batteries. We develop a configuration for a piezoelectric energy harvester with the capability to wirelessly communicate vibration measurements while using those vibrations to power the sensing and communication devices. Particularly, we perform experiments that aim at identifying challenges to overcome in the development of such a configuration. Towards that objective, we successfully tested a self-powered real-time point-to-point wireless communication system between a vibration sensor and transmission and receiving modules. The sensing device and transmission module are powered by the vibrating object using a piezoelectric energy harvester. The communication is established by using two XBee modules. In the second part of this dissertation, we address the optimization of the output power of piezoelectric energy harvesters of aeroelastic vibrations. Given the complexity of high-fidelity simulations of the coupling between the fluid flow, structural response and piezoelectric transduction, we develop and experimentally validate a phenomelogical reduced-order model for energy harvesting from wake galloping. We also develop a high-fidelity simulation for the same phenomena. The modeling and high-fidelity simulations can be a part of a multi-disciplinary optimization framework to be used in the design and operation of galloping-based energy harvesters. / Doctor of Philosophy / Energy harvesting of ambient or flow-induced vibrations is important for reducing the dependence on batteries in wireless sensing and networks to monitor deterioration conditions, environmental pollution or wildlife conservation. Balancing the benefits and shortcomings of a specific approach, namely piezoelctric transduction, for energy harvesting from vibrations, we address a specific challenge related to the development of a configuration that allows for communicating measured vibrations using their power. Furthermore, given the low levels of output power from piezoelectric transduction, we address the need to optimize power output levels through the development of predictive models that depend on geometry and speed of the fluid flow.