Resource Allocation in Multicarrier Non-Orthogonal Multiple Access (NOMA) Systems

Non-orthogonal multiple access (NOMA) can improve the spectrum efficiency and enable massive connectivity in future wireless communications systems by multiplexing multiple users in a non-orthogonal manner. Many previous works in power-domain NOMA addressed research problems from the perspective of the channel capacity, assumed perfect successive interference cancellation (SIC) and considered the pairing of users with very distinct channel conditions. This can yield inefficient power allocations in terms of sum-rate. Further, the assumption of perfect SIC is not realistic in practical systems, where SIC error propagation greatly impacts the achievable bit error rate (BER) at the receivers.

By applying NOMA to multicarrier-based schemes, the capabilities of both can be enhanced through resource allocation, i.e. the assignment of radio resources to users under an optimization objective. However, resource allocation in multicarrier NOMA systems may lead to a nondeterministic polynomial time (NP)-hard problem requiring exhaustive search, which has prohibitive computational complexity. Instead, efficient algorithms that provide a good trade-off between system performance and implementation practicality are needed.The contributions presented in this thesis are two-fold. First, new performance bounds on the BER of NOMA systems are provided. And second, a novel resource allocation scheme is presented, which can achieve a performance close to optimal with low computational complexity. The contributions are summarized as follows.First, theoretical BER expressions are presented for multi-layer, multi-level quadrature amplitude modulation (QAM) in NOMA. To the best of the author's knowledge, this work represents the first attempt in developing such expressions. The optimal value of the power allocation factor in terms of BER is analytically derived. Further, the theoretical BER expressions are used for calculating the ratios of users' channel gains that maximize the sum-rate. Unlike previous research in NOMA, it is demonstrated that, in NOMA systems with QAM, the channel gains of two NOMA users must be of approximately the same order of magnitude in order to guarantee that inter-user interference can be overcome at the receivers. Additionally, accurate BER approximations are presented in the form of exponential functions. These are used for finding numerical boundaries for the values of the channel gain ratios of NOMA users that fulfill the BER constraints.Second, the contributions on BER boundaries are applied to develop of a novel resource allocation scheme for multicarrier NOMA. A user pairing algorithm of quasi-linear complexity with respect to the number of users is proposed, based on the findings about NOMA optimal channel gain ratios and channel gain gaps. In contrast, the complexity of exhaustive search is of the order of the squared number of users. The problem of power and data rate allocation is solved by applying a Lagrangian optimization method based on the previously derived BER exponential approximations. The optimization result is applied to propose a novel iterative resource allocation (IRA)-data rate selection (DRS) algorithm. Unlike existing works, continuous power levels and discrete modulation schemes are considered. Numerical simulations demonstrate that IRA-DRS yields a sum-rate performance close to optimal, providing an excellent trade-off between computational complexity and performance. IRA-DRS benefits from multi-user diversity in terms of achievable sum-rate, number of iterations required for convergence, and degrees of freedom in choosing different combinations of modulation levels.