Hybrid Wireless Energy Transfer

• Hybrid Wireless Energy Transfer is a design paradigm that couples RF energy and information delivery through modes like WET, SWIPT, and WPCN.
• It employs diverse signal models and beamforming techniques to overcome CSI limitations, ensuring efficient energy harvesting and minimal interference.
• Its architectures span intelligent surfaces, smart-grid integration, and mobility-assisted charging, emphasizing fairness and coexistence in multiuser networks.

Searching arXiv for recent and foundational work on hybrid WET architectures. Searching arXiv for intelligent-surface and sensing-assisted WET systems relevant to hybrid WET. Hybrid Wireless Energy Transfer (WET) denotes a class of architectures in which RF/microwave energy delivery is not treated as an isolated downlink power link, but is coupled to information transfer, protocol scheduling, spatial cooperation, passive sensing or backscatter, transmitter-side energy cooperation, or mobility. Taken together, the literature spans three canonical modes: pure downlink WET, simultaneous wireless information and power transfer (SWIPT), and wireless powered communication networks (WPCNs), where harvested downlink energy enables later uplink transmission (Bi et al., 2014). Subsequent work makes this notion more concrete through co-channel WET/WIT with simultaneous energy and information over the same band (Che et al., 2015), hybrid data-and-energy access points in massive MIMO (Yang et al., 2014), spectrum-sharing WET/WIT coexistence (Xu et al., 2015), backscatter-assisted channel learning for passive tags (Yang et al., 2015), smart-grid-coupled distributed antennas (Yuan et al., 2016), sensing-assisted reconfigurable surfaces (Luo et al., 11 Mar 2025), intelligent transmitting surfaces (Rosabal et al., 9 Jul 2025), and mobility-assisted charging platforms (Wang et al., 2022).

1. Taxonomy and defining architectures

A useful starting point is the three-part organization of microwave-enabled wireless energy transfer into WET, SWIPT, and WPCN. In WET mode, a transmitter radiates energy in the downlink only. In SWIPT mode, energy and information are jointly transmitted using the same radio waveform. In WPCN mode, a downlink WET phase is followed by uplink wireless information transmission using harvested energy (Bi et al., 2014). This organization suggests that hybrid WET is best understood as a family of joint energy-information systems rather than a single PHY-layer technique.

The architectural meaning of “hybrid” varies across the literature. In smart-city deployment work, a hybrid access point (H-AP) supports both WIT and WET, with users in the WET range receiving both data and energy, while users in the larger WIT-only range receive data only. The implementation is dual-band rather than same-waveform SWIPT: one radio front is used for WIT in a high-frequency band and another for WET in a low-frequency band, yielding nested coverage regions with $r^E<r^D$ (Zhao et al., 2020). By contrast, the spectrum-sharing formulation treats hybrid WET/WIT as coexistence between separately operated systems: a primary multiuser MIMO WET network and a secondary MIMO WIT link share the same band and geographical area, with one-way interference from WET to WIT (Xu et al., 2015).

The survey perspective therefore supports a broad definition. Some systems are hybrid because the same waveform carries both information and energy; some because downlink charging and uplink communication are protocol-coupled; some because energy and information infrastructures coexist; and some because auxiliary mechanisms such as sensing, backscatter, or mobility are integrated into the energy-transfer loop. This suggests that “hybrid WET” is primarily an architectural designation defined by coupled functions, shared resources, and energy causality.

2. Canonical signal models and protocol couplings

A representative co-channel model appears in multi-antenna wireless powered communication with simultaneous WET and WIT over the same frequency band. The information receiver observes

$y = g^H w s + f^H x_E + n,$

with achievable rate

$$R=\log_2\!\left(1+\frac{|g^H w|^2}{f^H Q f+\sigma^2}\right),$$

while the wireless device harvests

$E=\eta h^H Q h+\eta |\varphi^H w|^2,$

leading to the self-powered causality constraint

$w^H (I-\eta \varphi\varphi^H) w \le \eta h^H Q h.$

This formulation makes the central hybrid coupling explicit: the ET covariance $Q$ is beneficial through $\eta h^H Q h$ and harmful through $f^H Q f$ (Che et al., 2015).

The WPCN formulation in massive MIMO introduces a different coupling structure. Each frame is partitioned into an uplink CE phase of duration $\tau T$, a downlink WET phase of duration $\alpha T$, and an uplink WIT phase of duration $y = g^H w s + f^H x_E + n,$0, with $y = g^H w s + f^H x_E + n,$1. Users reserve a fraction $y = g^H w s + f^H x_E + n,$2 of harvested energy for future pilots and use $y = g^H w s + f^H x_E + n,$3 for current uplink data, so CE quality, harvested energy, and uplink power are recursively linked across frames (Yang et al., 2014). The downlink harvested energy $y = g^H w s + f^H x_E + n,$4, the CE relation $y = g^H w s + f^H x_E + n,$5, and the uplink power

$y = g^H w s + f^H x_E + n,$6

together define a tightly coupled hybrid protocol.

SWIPT with receiver-side power splitting yields yet another canonical model. In a MISO distributed antenna system, the received signal is

$y = g^H w s + f^H x_E + n,$7

the harvested power is

$y = g^H w s + f^H x_E + n,$8

and the achievable rate is

$y = g^H w s + f^H x_E + n,$9

For fixed $$R=\log_2\!\left(1+\frac{|g^H w|^2}{f^H Q f+\sigma^2}\right),$$0, both $$R=\log_2\!\left(1+\frac{|g^H w|^2}{f^H Q f+\sigma^2}\right),$$1 and $$R=\log_2\!\left(1+\frac{|g^H w|^2}{f^H Q f+\sigma^2}\right),$$2 increase with the same coherent received-power term, which is why the design decomposes into transmit-side power management followed by power-splitting selection (Yuan et al., 2016).

Across these models, the common feature is not identical signaling but identical structure: energy delivery and information delivery are coupled through shared RF resources, receiver architecture, or inter-frame causality. That coupling is the defining technical content of hybrid WET.

3. Beamforming, CSI regimes, and channel acquisition

The strongest recurring design theme is that hybrid WET is fundamentally a beamforming problem under nonstandard CSI regimes. In co-channel WET/WIT, the optimal ET covariance is rank one: $$R=\log_2\!\left(1+\frac{|g^H w|^2}{f^H Q f+\sigma^2}\right),$$3 while the WD beamformer is

$$R=\log_2\!\left(1+\frac{|g^H w|^2}{f^H Q f+\sigma^2}\right),$$4

The ET beam is therefore neither pure energy maximization nor pure interference nulling; it is the closed-form optimum of the energy/interference tradeoff (Che et al., 2015).

In massive-MIMO WPCN, estimated-CSI energy beamforming asymptotically takes the form

$$R=\log_2\!\left(1+\frac{|g^H w|^2}{f^H Q f+\sigma^2}\right),$$5

with fairness-optimal weights

$$R=\log_2\!\left(1+\frac{|g^H w|^2}{f^H Q f+\sigma^2}\right),$$6

In the large-$$R=\log_2\!\left(1+\frac{|g^H w|^2}{f^H Q f+\sigma^2}\right),$$7 regime, the asymptotic harvested energy becomes $$R=\log_2\!\left(1+\frac{|g^H w|^2}{f^H Q f+\sigma^2}\right),$$8, and the system achieves massive MIMO degree-of-rate-gain $$R=\log_2\!\left(1+\frac{|g^H w|^2}{f^H Q f+\sigma^2}\right),$$9 with ZF and $E=\eta h^H Q h+\eta |\varphi^H w|^2,$0 with MRC, matching the ideal perfect-CSI benchmark (Yang et al., 2014).

A different CSI regime arises in backscatter-assisted WET. There, forward CSI is difficult because tags are too energy- and hardware-constrained to estimate or feed it back, so the ET estimates the cascaded backscatter channel $E=\eta h^H Q h+\eta |\varphi^H w|^2,$1 and uses normalized BS-CSI: $E=\eta h^H Q h+\eta |\varphi^H w|^2,$2 The resulting harvested-energy loss is reported as less than $E=\eta h^H Q h+\eta |\varphi^H w|^2,$3 relative to perfect F-CSI with optimized training, shifting complexity from ERs to the ET (Yang et al., 2015).

CSI scarcity also motivates energy-feedback and statistical-CSI formulations. Distributed energy beamforming with only power feedback from the ER uses phase adaptation rather than explicit CSI; the sequential protocol is analytically shown to converge to the perfect-CSI optimal EB solution as training time grows, and $E=\eta h^H Q h+\eta |\varphi^H w|^2,$4 is optimal for both phase-search algorithms considered (Lee et al., 2016). In clustered mMTC WET, the BS uses a hybrid deterministic/statistical channel description consisting of the LoS component and first/second-order statistics of the scattered component; the optimal unconstrained beamformer is the dominant singular vector of a stacked LoS-plus-covariance matrix, and the achieved sum-power is quasi-optimal relative to full-CSIT (Monteiro et al., 2021). At the opposite end of the CSI spectrum, rotary antenna beamforming couples a fixed CSI-free beamformer with mechanical ULA rotation and yields average gain $E=\eta h^H Q h+\eta |\varphi^H w|^2,$5 under LOS, without channel training (López et al., 2021).

Taken together, these works show that hybrid WET has no single CSI model. Instead, it spans perfect local CSI, reciprocity, statistical CSI, backscatter-derived CSI, power-only feedback, and fully CSI-free operation. The beamforming law changes accordingly, but spatial control remains central.

4. Multiuser fairness, interference, and coexistence

Because hybrid WET typically serves multiple users or coexists with communication links, fairness and interference are first-class design variables. In spectrum-sharing coexistence, the rank of the energy covariance directly determines the WIT degrees-of-freedom loss: $E=\eta h^H Q h+\eta |\varphi^H w|^2,$6 where $E=\eta h^H Q h+\eta |\varphi^H w|^2,$7. The key result is that fairness-oriented multiuser WET often needs $E=\eta h^H Q h+\eta |\varphi^H w|^2,$8, but a single-beam time-sharing construction preserves the same harvested energy at every ER while reducing the WIT DoF loss to

$E=\eta h^H Q h+\eta |\varphi^H w|^2,$9

The average WIT rate under this time-sharing scheme is never worse than that under simultaneous multi-beam WET (Xu et al., 2015).

Fairness also appears as the “double near-far effect” in WPCN. Users farther from the H-AP harvest less downlink energy and also require more uplink power. The max-min problem in massive MIMO is designed precisely to neutralize this effect through $w^H (I-\eta \varphi\varphi^H) w \le \eta h^H Q h.$0, while asymptotically equalizing user rates (Yang et al., 2014). In backscatter-assisted WET, weighted-sum-energy maximization is extreme-point and uses only one energy beam, whereas proportional-fair-energy maximization is biconvex and induces a water-filling-like beam allocation

$w^H (I-\eta \varphi\varphi^H) w \le \eta h^H Q h.$1

illustrating the sharp distinction between efficiency-driven and fairness-driven objectives (Yang et al., 2015).

Sensing-assisted WPCN extends the fairness problem to massive-user downlink charging. The RISS-based massive WET scheme selects beam coverage by thresholded beam stitching and minimizes waiting cost through ordering beams according to

$w^H (I-\eta \varphi\varphi^H) w \le \eta h^H Q h.$2

Numerically, the paper reports uplink capacity improvement by $w^H (I-\eta \varphi\varphi^H) w \le \eta h^H Q h.$3, worst-energy improvement ranging from $w^H (I-\eta \varphi\varphi^H) w \le \eta h^H Q h.$4 to $w^H (I-\eta \varphi\varphi^H) w \le \eta h^H Q h.$5, and waiting-cost reduction from $w^H (I-\eta \varphi\varphi^H) w \le \eta h^H Q h.$6 to $w^H (I-\eta \varphi\varphi^H) w \le \eta h^H Q h.$7 (Luo et al., 11 Mar 2025). At the infrastructure level, smart-city H-AP placement formalizes the WIT/WET tradeoff through the B-deployment problem, which maximizes WIT efficiency subject to $w^H (I-\eta \varphi\varphi^H) w \le \eta h^H Q h.$8, with $w^H (I-\eta \varphi\varphi^H) w \le \eta h^H Q h.$9 tuning the emphasis on harvested energy (Zhao et al., 2020).

These results indicate that hybrid WET cannot be evaluated solely by total harvested power. Once multiple users, multiple beams, or coexisting information links are present, covariance rank, user ordering, beam scheduling, and fairness constraints become structural.

5. Expanded architectures: distributed energy, intelligent surfaces, non-terrestrial platforms, and mobility

Hybrid WET also expands beyond the canonical H-AP model into transmitter- and platform-centric variants. In distributed antenna SWIPT, each RAU is powered by locally harvested renewable energy and may trade energy through a smart grid. The optimal transmit-side policy is full power when the grid is profitable, and otherwise follows a double-threshold structure

$Q$0

This turns the problem into joint RF-energy-network optimization across RAUs, the smart grid, and the SWIPT receiver (Yuan et al., 2016).

Intelligent-surface transmitters constitute another variant. An ITS-equipped power beacon combines a digital feeder with $Q$1 RF chains and a passive surface with $Q$2 elements, jointly optimizing feeder precoders and unit-modulus surface phases under a consumed-power objective that explicitly includes Doherty HPA nonlinearities, ITS control power, and feeder-to-ITS losses. Relative to fully digital, HBFC, and HBPC architectures, the ITS transmitter yields the lowest consumed power in the main sweeps and becomes especially favorable as $Q$3 grows (Rosabal et al., 9 Jul 2025).

Non-terrestrial and embodied platforms extend hybrid WET into new spatial regimes. Space-to-ground WET from a grid of $Q$4 satellites with $Q$5 antennas per satellite yields an upper-bound harvested-energy expression proportional to $Q$6, with milli-joule-level energy during grid visibility and strong dependence on altitude, charging frequency, inclination, and phase alignment (Shehab et al., 5 Apr 2026). Robotic WET addresses the opposite regime: short-range RF charging is made area-wide by mounting the transmitter on a mobile robot and jointly optimizing anchor selection, routing, charging time, and beam choice. The mission objective is

$Q$7

subject to routing and harvested-energy constraints, and the reported optimized mission time is $Q$8 s in the tested scenario (Wang et al., 2022).

These architectures show that hybrid WET is not limited to energy-information waveform design. It also includes transmitter-side energy cooperation, physically large passive apertures, non-terrestrial cooperative arrays, and mobility-assisted charging. The commonality is joint optimization across RF propagation and an additional control layer.

6. Assumptions, limitations, and research directions

Across the literature, several modeling assumptions recur. Perfect CSI or perfect local CSI remains common in co-channel WET/WIT, DAS-SWIPT, coexistence analysis, and smart-grid cooperation (Che et al., 2015, Yuan et al., 2016). Linear harvested-energy models are still standard in many formulations, even though several papers explicitly note that real rectifiers are nonlinear and saturating (Bi et al., 2014). Full-duplex WD operation, ideal cancellation of direct/self-interference, perfect synchronization across distributed ETs or satellites, block-static fading, and narrowband assumptions also appear repeatedly (Che et al., 2015, Lee et al., 2016, Shehab et al., 5 Apr 2026).

Recent work makes these limitations more explicit rather than less. RAB identifies rotor cost, motor energy, and limited implementation detail as open issues, and points directly to nonlinear EH models, other array topologies, and full/limited-CSI rotating-array schemes as future directions (López et al., 2021). The satellite-grid model leaves CSI acquisition, inter-satellite synchronization, interference management, and regulatory exposure outside the analytical framework (Shehab et al., 5 Apr 2026). The ITS transmitter assumes known channels and unit-modulus phase control, while much of its optimization remains local because of SCA sensitivity to initialization (Rosabal et al., 9 Jul 2025). Robotic WET identifies multi-robot collaboration and improved map sharing as open directions beyond the single-robot HIL framework (Wang et al., 2022).

The broader survey view is that future systems will be “a mixture of energy and information transfer,” with extensions including full-duplex information/energy operation, interference-aware SWIPT, cooperative relaying, distributed power beacons, and cross-layer redesign in WPCN (Bi et al., 2014). The accumulated evidence suggests that hybrid WET is evolving along three axes simultaneously: richer CSI surrogates when instantaneous CSIT is too costly, richer physical embodiments of the transmitter and network, and richer fairness/coexistence constraints when charging and communication must share spectrum, hardware, or infrastructure.