Resource Allocation in Wireless Networks with Energy Constraints
Panagiotis D. Diamantoulakis
DOI: https://doi.org/10.48550/arXiv.1803.04864
2018-03-10
Abstract:This dissertation focuses on the development of novel scheduling and resource allocation schemes, which take into account and regulate the energy constraints imposed by the levels of harvested energy. To this direction, first, the optimal energy, time, and bandwidth allocation problem for the downlink of energy harvesting base stations (EHBSs) is investigated, with the main focus being on autonomous EHBSs. The presented analysis considers the impact of the energy constraint on users' preferences and the BS's revenue. In order to model the competitive nature of the problem, game theory is used. The next two chapters focus on wireless powered networks (WPNs) and simultaneous wireless information and power transfer (SWIPT) using radio frequency (RF) technology. One of the main contributions of these chapters is the introduction of both uplink and downlink non-orthogonal multiple access (NOMA) for WPNs. Moreover, the individual data rates and fairness are improved, while the formulated problems are optimally and efficiently solved. It is shown that, compared to orthogonal multiple access, NOMA offers a considerable improvement in throughput, fairness, and energy efficiency. Rather than this, proportional fairness is maximized and uplink/downlink of WPNs are jointly optimized, in which cases, except for NOMA, time division multiple access (TDMA) is also investigated. Furthermore, the role of interference is considered, which has been recognized as one of the main reasons of the asymmetric overall degradation of the users' performance, due to different path-loss values, called from now on as cascaded near-far problem. Moreover, SWIPT is investigated and efficiently optimized in the context of multicarrier cooperative communication networks. Finally, simultaneous lightwave information and power transfer (SLIPT) is introduced, while novel and fundamental techniques are proposed.
Signal Processing