In recent years, virtualization in Wireless Sensor Networks (WSNs) has become very popular for many reasons including efficient resource management, proper sharing and using the same WSN physical infrastructure by multiple applications and services. Semantic applications are very much pertinent to provide situational awareness to the end-users. Incorporating semantic applications in the virtualized WSNs can play a crucial role in providing contextual information to understand the situation, increase usability and interoperability. However, provisioning of semantic applications in virtualized WSNs remains as a big challenge. The reason is the data collected by the virtual sensors needs to be annotated in-network, and the pre-requisite of the data annotation process is to have an ontology that needs to be provisioned, i.e., developed, deployed and managed. Unfortunately, annotating sensor data and ontology provisioning in virtualized WSNs is not straightforward because of limited resources of sensors, on-demand creation of virtual sensors, and unpredictable lifetime. As the existing researches do not consider data annotation in virtualized WSN infrastructure level, these solutions are domain specific and lack of providing support for multiple applications. Moreover, the major drawback of the current ontology provisioning mechanisms requires domain experts to develop, deploy, and manage the ontologies in WSNs.   This thesis aims to propose a solution for provisioning of multiple semantic applications in the virtualized WSNs.  
The main contribution of this thesis is twofold. First, we have proposed an architecture to annotate sensor data in the virtualized WSN infrastructure and defined an ontology in sensor domain to perform data annotation. Second, we have proposed an architecture for provisioning ontology in the virtualized WSNs that consists of an ontology provisioning center, an ontology-enabled virtualized WSN, and an ontology deployment protocol. The proposed architectures use overlay network as a foundation. We have built a proof-of-concept prototype for a semantic wildfire monitoring application in the cloud environment using the Google App Engine. In order to evaluate the viability of the proposed architecture, we have made performance measurement of the implemented prototype. We ran a simulation to justify our proposed ontology provisioning protocol.