The world is getting more and more connected. Other than traditional telecommunication applications such as cellular communication and wireless broadband, many new technologies such as the Internet of things (IoT) and automotive vehicles are using wireless spectrum to establish the connection to the network. With these increasing number of devices, the data rate requirements are increasing enormously, which in turn, necessitate a more efficient utilization of available resources. In an effort to enable higher productive utilization of the available spectrum, the Federal Communications Commission (FCC) has recently opened up many spectrum bands such TV white space, 3.5~GHz Citizen Broadband Radio Service (CBRS) band and 5~GHz The Unlicensed National Information Infrastructure (U-NII) band among others for commercial, unlicensed and shared access. This has led to various innovations in cognitive radio technology, coexistence mechanism among heterogeneous technologies, spectrum sensing as well as novel proposals for dynamic spectrum access. Over a large part of the last decade, there has been a tremendous amount of research on spectrum policy as well as the theory and practice of cognitive radio networks including dynamic spectrum access (DSA) algorithms, networking protocols, and software radio platform development. Generally, the available spectrum to the users in a cognitive environment is not always contiguous and becomes fragmented due to either the presence of primary users or other secondary users. In this thesis, we specifically study the opportunity and challenges associated with the access of such fragmented spectrum.

This thesis first presents the problem of spectrum allocation for cases when the available spectrum is fragmented. As an example, the problem of allocating TV white-space to provide wireless backhaul communication is presented. In the second part of this thesis, the NC-OFDM based communication is introduced as a technique for agile and flexible access to a fragmented spectrum. The advantages of using NC-OFDM for spectrum allocation in a power constrained system are highlighted. In the latter part of the thesis, the challenges associated with the implementation of an NC-OFDM based communication system are discussed. Synchronization among different transmissions, and a separate control channel to enable frequency and time offset calculations are identified as two of the main challenges. The performance results are presented for an NC-OFDM-enabled asynchronous network, and solutions are provided to address this issue. A low-powered underlay channel is designed to help the NC-OFDM-enabled system in the transmission of control information, and frequency and time offset estimation. At last, a physical layer security analysis of NC-OFDM is presented for a low probability of exploitation (LPE) design.