Nowadays, an increasing number of popular consumer electronics is shipped with a variety of sensors. The usage of these as a wireless sensing platform, where users are the key architectural component, and ubiquitous access to communication infrastructure has established a new application area called public sensing. We present an opportunistic public sensing system that allows for a flexible and efficient acquisition of sensor readings. This work considers the usage of smartphones as a sensor network in a model-driven sensor data acquisition. We focus on efficiency of query dissemination to mobile nodes, while retaining high effectiveness regarding defined sensing quality of collected data. We adopted and extended an existing geographic routing protocol to design an efficient com- munication system that executes model-driven data acquisition and is robust to changing sensors availability. We use in-network processing paradigm to efficiently distribute queries to mobile nodes and to collect results afterwards. The developed approach was simulated using OMNeT++ network simulator. To verify implemented algorithms and test the overall system performance, we run simulations in different scenarios and evaluate them using adequate cov- erage metrics. Moreover, we verify our intuitive extension to adopted routing protocol and show that it can have a strong impact on the efficiency of protocol in question.