In this dissertation, we consider the device-to-device (D2D) communications in downlink simultaneous wireless information and power transfer (SWIPT) networks, where both the cellular user (CU) and D2D receiver (DR) not only decode information but also harvest the energy. For the purpose of protecting the CU from the interference due to the D2D transmission, we propose energy harvesting parameter optimization schemes. Unlike the conventional D2D communications, where the CU's performance loss becomes severe when the D2D transmitter (DT) is close to the CU, with the proposed schemes, it is possible for the D2D pair to communicate without degrading the performance of the CU. This is because of that, by adjusting the CU's energy harvesting parameter, the CU's harvested energy gain due to the DT's signals can compensate for the CU's rate loss due to the DT's signals. In order to consider the various conditions of wireless channels, the energy harvesting parameter optimization schemes are proposed for three channel assumptions: global channel state information (CSI), local CSI under fast fading channels, and local CSI under slow fading channels.
Firstly, by solving the optimization problems for the maximization of D2D rate, we propose energy harvesting parameter control schemes with global CSI. For the time switching (TS) receiver architecture, two schemes changing time duration of the D2D transmission are proposed. For the power splitting (PS) receiver architecture, with the energy signal transmission of the DT, a closed-form optimal scheme is proposed.
Secondly, by solving the optimization problems for the maximization of D2D ergodic rate, we propose D2D power control schemes for fast fading channels with local CSI, which are performed at the DT with the instantaneous CSI of only the link between the D2D pair, i.e., decentralized network control. By exploiting the conservative approximation approach and the Lagrangian dual method, for the PS and TS receiver architectures, non-convex power control problems are approximated as convex problems and efficiently solved.
Thirdly, by solving the optimization problems for the minimization of D2D outage probability, we propose D2D power control schemes for slow fading channels with local CSI, which are performed at the DT with the instantaneous CSI of only the link between the D2D pair, i.e., decentralized network control. By exploiting the conservative approximation approach and the Lagrangian dual method, for the PS and TS receiver architectures, non-convex power control problems are approximated as convex problems and efficiently solved.
Simulation results demonstrate that the proposed schemes perform better when the CU's required energy is large. In this case, the baseline schemes have similar performance as the proposed schemes. On the other hand, when the CU's required energy is small, compared to the baseline schemes, the proposed schemes can greatly improve the D2D performance.