The Internet of Things will involve a huge number of sensors. Periodically collecting data from all IoT sensors will waste a large amount of communication and storage resources, in addition to a large amount of energy, which impedes the scalability of IoT systems. Moreover, it introduces significant privacy risks. The goal of the project is to experiment with procedures for efficiently collecting IoT data while achieving target requirements in terms of data accuracy, timeliness, energy efficiency, and privacy protection. The procedures dynamically adapt the IoT data request density in time (frequency of requests to a particular IoT sensor) and space (requests for the same type of data from IoT sensors located in the same area), as well as the corresponding results. The adaptation of the requests will be performed by exploiting the temporal and spatial correlation of measurements, which will be estimated based on the history of data measurements and the location and type of the IoT sensors, whereas the adaptation of the results will be performed by considering privacy-accuracy tradeoffs. The data collection procedures will be applied to two application domains: environmental monitoring (air and water) and IoT network monitoring (link quality and power consumption). A key outcome of the project will be the development of application-aware closed-loop IoT workload models, which will be made available as open source software.