Simultaneous wireless information and power transfer (SWIPT) offers a way to support communication and intermittent energy delivery to low-power devices. Its design must balance information-decoding quality and harvested energy while accounting for propagation, near-field interactions at the transmit array, and the complexity of coordinating distributed transmitters. This thesis develops analytical models and optimization methods spanning cell-free massive MIMO (CF-mMIMO) networks, reconfigurable surfaces, parasitic antenna arrays, and multi-port circuit theory. First, a beyond-diagonal reconfigurable intelligent surface (BD-RIS)-assisted CF-mMIMO SWIPT system is analyzed using closed-form expressions and resource-allocation methods for joint information-bearing symbol and radio frequency (RF) energy delivery. A stacked intelligent metasurface (SIM)-assisted CF-mMIMO SWIPT system is then developed, together with a multi-agent learning framework that uses reward normalization to stabilize training for joint AP-mode selection, power allocation, and meta-atom phase control. Under the evaluated settings, decentralized execution reduces the decision dimension handled by each AP and provides lower post-training inference time than the centralized alternatives considered. The thesis then moves from network-level models to coupling-aware transmit-array design, accounting for antenna properties including mutual coupling, near-field reactive energy storage, and radiation loss. A multi-port MIMO SWIPT model links active excitation and induced currents, array geometry, parasitic load configuration, and realized gain to SWIPT metrics through impedance relationships. Optimized non-uniform planar arrays (NUPAs) are assessed numerically, and selected NUPA configurations are evaluated through full-wave electromagnetic simulations. Finally, the circuit-theoretic modeling approach is extended to wideband OFDM-SWIPT through a tri-hybrid precoder with an impedance-based antenna $Q$-factor constraint. Numerical results show improvements in the evaluated power transfer efficiency and harvested-energy metrics over the considered benchmarks. Together, these contributions connect MIMO SWIPT optimization with physically-aware transmit array design.
- Simultaneous wireless and information transfer (SWIPT)
- wireless communication
- wireless power transfer
- wireless systems
- deep reinforcement learning
- optimization
- reconfigurable intelligent surface
Next-generation SWIPT networks: reconfigurable architectures and intelligent optimization
Duc Hua, T. (Author). Dec 2026
Student thesis: Doctoral Thesis › Doctor of Philosophy (Thesis with Publications)