SWIPT Framework: Wireless Info & Power Transfer

• SWIPT is a framework that enables simultaneous wireless data transmission and energy harvesting using designs like time switching and power splitting to address real-world hardware constraints.
• The framework optimizes receiver architectures and resource allocation through methods such as convex optimization and deep learning to manage nonlinear RF energy harvesting behavior.
• Emerging innovations in SWIPT integrate safety standards, reconfigurable intelligent surfaces, and near-field techniques to support sustainable IoT deployments and ultra-dense networks.

Simultaneous Wireless Information and Power Transfer (SWIPT) Framework

Simultaneous Wireless Information and Power Transfer (SWIPT) refers to wireless transmission systems that deliver both RF information and usable energy to receiver circuits over the same electromagnetic waveform. SWIPT enables battery-free or energy-sustainable operation of devices in wireless networks while performing conventional information transfer. Such frameworks involve a layered interplay of physical-layer receiver architectures, resource allocation, and optimization, and must account for the nonlinearities of practical RF energy harvesting circuitry, the strict requirements of communications, and, increasingly, safety constraints.

1. Receiver Architectures and Fundamental SWIPT Protocols

The core of a SWIPT system consists of the transmitter, wireless channel, and a receiver tasked with both information decoding (ID) and energy harvesting (EH). Real-world constraints preclude perfect simultaneous ID and EH from all the received RF power, necessitating explicit receiver designs:

1. Separated ID/EH Receivers: Use physically distinct front-ends to enable independent information detection and energy scavenging, incurring hardware duplication and higher circuit power consumption (Zhou et al., 2012).
2. Co-located Receivers: Share a single antenna front-end and partition RF power using:
   • Time Switching (TS): The receiver alternates its front-end connection between EH and ID circuits over scheduled intervals (Zhou et al., 2012, Krikidis et al., 2014).
   • Power Splitting (PS): The received waveform is split in power, with a fraction $\rho$ routed to the ID chain and $1-\rho$ to the rectifier (Zhou et al., 2012, Krikidis et al., 2014, Liu, 2016).
   • Antenna Switching (AS): In multiple antenna systems, certain antennas are dedicated for EH and the rest for ID.
   • Spatial Switching (SS): In MIMO links, spatial eigenmodes are assigned to either ID or EH (Krikidis et al., 2014).

Integrated receiver architectures combine the rectification and baseband conversion stages, requiring sophisticated energy-detection modulation and yielding unique nonlinear SWIPT channels (Zhou et al., 2012), while diplexer-based receivers leverage IF and RF splitting at the mixer output with a fixed split ratio ($0.5$), enabling simultaneous EH and ID with remarkable hardware simplicity but no adaptability (Qin et al., 2016).

2. SWIPT System and Channel Models

Transmitter design in SWIPT exploits the degrees of freedom in waveform, power allocation, and scheduling to jointly achieve communication and energy transfer objectives. Channel models span:

• Flat or Frequency-Selective Fading: SISO/MIMO, including OFDM multi-subcarrier settings (Huang et al., 2012, Zhou et al., 2013).
• Multiuser Downlink: Single-antenna or multi-antenna transmitters allocate resources across $N$ users for joint ID and/or EH (Chynonova et al., 2015, Luo et al., 12 Sep 2025).

The general received signal at time $t$ is $y(t) = h x(t) + n(t)$, with $x(t)$ transmit signal (subject to power constraints) and $h$ the fading channel (Krikidis et al., 2014). For multiuser systems, additional scheduling or beamforming phases may be introduced.

Harvested Power Models

Practical EH circuitry is modeled via three archetypes (Clerckx et al., 2018, Wei et al., 2021):

• Linear Model: $P_{\mathrm{DC}} = \eta P_{\mathrm{RF}}$, valid at low input power.
• Diode Nonlinear Model: $$P_{\mathrm{DC}} \propto k_2 \mathbb{E}[y_{\mathrm{RF}}^2] + k_4 \mathbb{E}[y_{\mathrm{RF}}^4]$$, sensitive to waveform peaks and statistics.
• Saturation Model: $P_{\mathrm{DC}}$ is a sigmoidal function of input RF power, reflecting turn-on and saturation (Clerckx et al., 2018).

The choice of model fundamentally impacts the system-level SWIPT design, especially for waveform and beamformer optimization (Luo et al., 12 Sep 2025, Clerckx et al., 2018).

3. Rate–Energy (R–E) Trade-off and SWIPT Optimization

The central trade-off in SWIPT is between the achievable communication rate $R$ (bits/s/Hz) and harvested energy $E$ (W), formalized as the Rate–Energy (R–E) region $\mathcal{C}_{R,E}$ (Zhou et al., 2012, Krikidis et al., 2014, Clerckx et al., 2018): $$\mathcal{C}_{R,E} = \left\{ (R, E):~ R \leq f_{\mathrm{ID}}(\{P,\rho,\alpha\}), ~E \leq f_{\mathrm{EH}}(\{P,\rho,\alpha\}) \right\}$$ where $P$ is allocated power, $\rho$ the PS ratio, and $\alpha$ the TS duty cycle.

• TS yields a convex, triangular R–E region; PS produces a concave, expanded region; the “ideal” (unattainable) receiver offers a rectangle (Zhou et al., 2012).
• With circuit power consumption, the region boundary is best achieved by on-off power splitting (OPS), hybridizing TS and PS (Zhou et al., 2012).
• Nonlinear harvester models introduce nonconvex, possibly disconnected R–E boundaries, necessitating waveform optimization and joint transmit/receiver design (Clerckx et al., 2018, Luo et al., 12 Sep 2025).

Optimization problems include:

• Sum-rate maximization under energy constraints
• Max-min harvested energy under SINR constraints
• Weighted R–E utility under QoS constraints

These are solved with convex optimization, semidefinite relaxation (SDR), Dinkelbach-type fractional programming, and, recently, deep learning frameworks using transfer and graph neural networks (SWIPTNet) (Han et al., 6 Feb 2025).

4. Multiuser, MIMO, and Resource Allocation Frameworks

In multiuser or MIMO settings, SWIPT introduces additional dimensions:

• Resource allocation—joint optimization over scheduling, user selection, beamforming, and splitter ratios to exploit channel diversity and spatial selectivity (Chynonova et al., 2015, Zhao, 2017, Luo et al., 12 Sep 2025).
• Interference management—explicit in interference alignment networks and collaborative MIMO frameworks, where user roles are dynamically selected between ID/EH to optimize utility (Zhao, 2017, Lee et al., 2014).
• OFDM and Broadband—SWIPT resource allocation in multicarrier systems employs variants of water-filling under EH constraints or greedy inversion when fixed coding rates are imposed (Zhou et al., 2013, Huang et al., 2012).
• Relay and Cooperative SWIPT—power splitting at relays in multi-hop (DF/AF) and space-time coded networks couples the first- and second-hop SNRs, leading to convex (DF) or fractional programming (AF) formulations for splitting optimization (Liu, 2016).

Increasingly, scheduling and resource optimization is conducted under fairness constraints, e.g., proportional-fair (PF) or equal-throughput (ET), which pull the attainable R–E region cognizant of user priorities (Chynonova et al., 2015).

5. Advanced Frameworks: Nonlinearities, Safety Constraints, and Emerging Technologies

Emerging SWIPT research addresses practical nonlinearities, regulation, and advanced applications:

• Nonlinear Energy Harvesting: Dedicated energy beams may offer gains under deterministic sinusoidal waveforms when the received RF power lies in the rectifier's high-efficiency regime (Luo et al., 12 Sep 2025). Under practical Gaussian signaling, information beams usually suffice for both ID and EH unless non-Gaussian, energy-centric waveforms can be exploited. Null-space-based two-stage optimization achieves >90% computational savings with negligible performance penalty (Luo et al., 12 Sep 2025).
• Safety Constraints: Power transfer must comply with Specific Absorption Rate (SAR) and Maximum Permissible Exposure (MPE) requirements. Deep-learning-based beamforming designs ensure probabilistic QoS and compliance, using SAR matrices embedded in quadratic constraints (Psomas et al., 2021). mmWave SWIPT with narrow beams enhances MPE compliance; transmit power must be judiciously chosen to coexist with safety boundaries (Psomas et al., 2021).
• Reconfigurable Intelligent Surfaces (RIS): Active RIS elements with on-board amplification (as opposed to passive phase shifters) bypass the double-fading effect, resulting in superior SWIPT R–E performance. Alternating optimization using SCA and quadratic transforms jointly optimizes RIS coefficients and beamformers, enforcing both IR SINR and ER EH constraints (Ren et al., 2023).
• Near-field SWIPT: Recent work demonstrates that, in near-field (spherical-wave) conditions, a single beamformer can power multiple users without dedicated energy beams. Hybrid digital/analog beamforming via penalty-based algorithms realizes globally optimal transmission in the near field (Zhang et al., 2023).
• Optical SWIPT: High-power, long-range SWIPT is also established through spatially separated optical resonant-beam cavities, achieving multi-watt power delivery and multi-Gb/s spectral efficiency over tens of meters. System-level trade-offs include reflector size and cavity length, with performance traced via transfer-matrix models and ABCD formalism (Fang et al., 2021, Bai et al., 2021).

6. Algorithmic and Implementation Aspects

SWIPT resource allocation and transceiver design is enabled by:

• Convex-optimization (SDR, SCA): SDR is proved tight in various multiuser and MIMO scenarios yielding rank-one beamforming solutions (Qin et al., 2016, Zhang et al., 2023).
• Lagrange duality and subgradient descent: Dual variables associated with energy, power, and fairness constraints drive optimal adaptation of scheduling and power-splitting policies (Chynonova et al., 2015).
• Low-complexity two-stage decomposition: In large-scale or null-space-based scenarios, sequential design of WIT and WET beams cuts complexity by orders of magnitude (Luo et al., 12 Sep 2025).
• Learning-based frameworks: GNN-based SWIPTNet models generalize across both PS and TS receivers with transfer learning, achieving near-optimality with sub-millisecond inference time (Han et al., 6 Feb 2025).

Circuit-level considerations such as the nonlinear dependence of RF-DC conversion on waveform moments, RF chain losses, and circuit power consumption (as in practical modulation and hardware models) profoundly influence the optimal selection of splitting policy, waveform, and even modulation scheme (Zhou et al., 2012, Clerckx et al., 2018).

7. Research Challenges and Future Prospects

Key open challenges include:

• Accurate joint modeling of EH circuit nonlinearities and wireless channel effects, particularly under dynamic environments and imperfect CSI (Clerckx et al., 2018, Wei et al., 2021).
• Cross-layer and application-aware optimization, including co-design with mobile edge computing, federated learning, UAV-enabled networking, and integration with RIS/IRS (Wei et al., 2021).
• Robust and secure SWIPT, with joint secrecy-rate and nonlinear EH optimization against eavesdropping and ambiguity (Clerckx et al., 2018).
• Large-scale, learning-driven SWIPT resource control, leveraging deep neural networks and GNNs to master the high-dimensional parameter space and real-time adaptation (Han et al., 6 Feb 2025).
• Safety-aware design, ensuring compliance with emerging RF health standards using robust and real-time optimization frameworks (Psomas et al., 2021).

SWIPT frameworks continue to evolve, integrating new theoretical models, RF circuit advances, and system-level optimization, and are positioned to enable future ultra-dense wireless networks and sustainable IoT deployments.