The new computing paradigm of IoT (Internet of Things) is carried out by globally and massively interconnected devices. This is founded on the unprecedented proliferation of heterogeneous wireless technologies in the last decade, each offering convenience in different aspects of our daily lives – WiFi enables hassle-free Internet access and Bluetooth allows prevalent healthcare with its wireless heart monitors. However, wireless technologies are victims of their own success; dense deployment of wireless devices intensify the interference between them, which recently has become a major cause of performance degradation. My dissertation work is an effort to address the issue with practical and cost-effective solutions. The dissertation consists of two main parts. The first part proposes approaches to achieve high-performance networking under arbitrary interference from unknown sources. Namely, cETX and CorrModel are proposed, where they enable efficient unicast and broadcast under severely interfered channel. This is achieved by observing the different aspects of interference including volume as well as temporal/spatial pattern and dynamics. To maximize the adaptability, both cETX and CorrModel are designed as generic techniques to support networks running different protocols, under a wide range of settings. Indoor and outdoor testbed evaluations performed on twelve unicast and nine broadcast protocols demonstrate that they bring 20-30% performance gain. The second part of the dissertation takes a step further, to explore the opportunity behind the coexistence of heterogeneous wireless technologies. The dissertation introduces FreeBee, which extends direct connectivity beyond wireless technologies to enable collaboration and mutual supplementation. By doing so FreeBee not only enables cross-technology interference mitigation but also brings advanced services such as context-aware smart operation. Evaluations reveals that FreeBee achieves reliable communication in under a second and supports mobility of up to 30mph.