Pushing data traffic onto small cells operating both in unlicensed (e.g. WiFi) and licensed bands (e.g. pico or femtocells) is important for alleviating congestion in the overloaded cellular network. However, the ultimate potential of such offloading and its effect on overall system performance is not well understood. The goal of the presented work is to develop realistic yet tractable frameworks to model and analyze offloading dynamics while incorporating the uncertainty and heterogeneity of these networks.