3D city models provide a three-dimensional representation of cities and are used in visualization, navigation, urban planning and many other areas. They can, for example, be used in positioning algorithms such as shadow matching to improve GNSS position accuracy in urban canyons. The correctness of these models is essential, and validation methods are needed. GNSS observations carry information about the receiver surroundings and could be used for that purpose. Furthermore, GNSS data can be crowd sourced, which could provide huge amount of data and a global coverage. Nowadays, connected devices such as smartphones, watches, cars, etc. are all equipped with GNSS receivers. This provides a good opportunity for crowd sourcing of GNSS observation data. However, the receivers in the above-mentioned devices are often cheap and observation quality is much worse than in professional devices. Our aim is to investigate the feasibility to use crowd sourced GNSS observations from such devices for validation of 3D city maps. GNSS signals are affected by different types of obstacles before reaching receivers. In urban areas the most common obstacles are high buildings. These obstacles can cause significant signal deterioration in the form of signal attenuation, blockage or multipath. All those negative effects are reflected in the GNSS observations. We will collect data using professional receivers and Android smartphones that provides raw GNSS observations. Data will be collected from locations that includes new buildings that are not yet present in the 3D models. A reality index is obtained by comparing which signals are shadowed according to GNSS observations and according to a 3D model-based estimation. If GNSS observations show blocked signals and corresponding 3D model-based predictions do not, the reality index is low, and this can indicate lack of a building in a model. We expect to detect shadowed signals by using signal-to-noise ratios, which should significantly drop when signal is being blocked, and pseudo ranges that increases when signal is first reflected and then reaches the receiver. GNSS based validation of 3D city maps can help with maintaining up-to-date models. It can be used to detect anomalies and errors and indicate potential areas that need re-mapping. On top of that, utilization of crowd sourced GNSS data offers real-time, continuous data with global coverage.