To fully empower sensor networks with cognitive Internet of Things (IoT) technology, efficient medium access control protocols that enable the coexistence of cognitive sensor networks with current wireless infrastructure are as essential as the cognitive power in data fusion and processing due to shared wireless spectrum. Cognitive radio (CR) is introduced to increase spectrum efficiency and support such an endeavor, which thereby becomes a promising building block toward facilitating cognitive IoT. In this paper, primary users (PUs) refer to devices in existing wireless infrastructure, and secondary users (SUs) refer to cognitive sensors. For interference control between PUs and SUs, SUs adopt dynamic spectrum access and power adjustment to ensure sufficient operation of PUs, which inevitably leads to increasing latency and poses new challenges on the reliability of IoT communications. To guarantee operations of primary systems while simultaneously optimizing system performance in cognitive radio ad hoc networks (CRAHNs), this paper proposes interference-aware flooding schemes exploiting global timeout and vaccine recovery schemes to control the heavy buffer occupancy induced by packet replications. The information delivery dynamics of SUs under the proposed interference-aware recovery-assisted flooding schemes is analyzed via epidemic models and stochastic geometry from a macroscopic view of the entire system. The simulation results show that our model can efficiently capture the complicated data delivery dynamics in CRAHNs in terms of end-to-end transmission reliability and buffer occupancy. This paper sheds new light on analysis of recovery-assisted flooding schemes in CRAHNs and provides performance evaluation of cognitive IoT services built upon CRAHNs.