Simultaneous Energy Harvesting & Sensing

Updated 24 November 2025

SEHS is a dual-purpose system that harvests ambient energy and simultaneously senses environmental conditions for sustainable, battery-free operation.
It integrates diverse transducers—RF, kinetic, thermal, and more—to convert physical phenomena into usable electrical signals and energy.
Advanced architectures combine signal processing, energy-positive design, and machine learning to optimize performance for IoT, wearables, and infrastructure monitoring.

Simultaneous Energy Harvesting and Sensing (SEHS) refers to systems wherein a single physical transducer both scavenges ambient energy to supply a local energy store and concurrently provides a voltage/current waveform encoding information about environmental or user context. In such systems, the dual-use of the harvesting element as a sensor yields “energy-positive” operation: the system harvests more energy than is consumed during signal acquisition, enabling battery-free or self-sustaining deployment. SEHS frameworks span radio-frequency, kinetic, piezoelectric, thermal, triboelectric, solar, and engineered metastructural systems, with representative applications in wireless IoT, wearable health, infrastructure monitoring, and machine condition sensing.

1. Physical Principles and Taxonomy of SEHS

SEHS systems exploit intrinsic coupling between an energy-harvesting mechanism and a measurable signal correlated to the sensed context. The harvested energy $H(t)$ and acquired signal are both functions of the underlying environmental excitation (e.g., RF power, mechanical strain, vibration, temperature gradient). Key categories by source and signal type include:

Energy Source:
- RF (Radio-Frequency Energy Harvesting, RFEH): voltage waveform encodes amplitude/diffraction changes due to motion in the RF field (Ni et al., 26 Aug 2024).
- Kinetic (Kinetic Energy Harvesting, KEH): piezoelectric or electromagnetic harvesters convert motion/vibration to AC voltage/current profiles (Ma et al., 2020, Sandhu et al., 2020, Li et al., 7 Jul 2025).
- Thermal (Thermoelectric, TEH): DC voltage derived from thermal gradients, with temperature or context modulating signal amplitude (Sandhu et al., 2020).
- Solar (Photovoltaic, SEH): output modulations encode illumination, often for coarse environmental or presence sensing (Sandhu et al., 2020).
Signal Points:
- Open-circuit voltage $V_{oc}$ , rectified voltage $V_{rec}$ , storage node voltage $V_{cap}$ , and harvesting current $I_{har}$ each expose different context sensitivities and distortion properties (Sandhu et al., 2020).

By exploiting the transducer’s contextual sensitivity (e.g., diffraction-induced voltage fluctuations in RF, or vibration spectral content in piezo systems), SEHS systems enable single-hardware platforms to simultaneously provide self-powering and context-aware sensing functionalities (Sandhu et al., 2020, Ni et al., 26 Aug 2024).

2. Canonical Architectures and Signal Flow

A typical SEHS node consists of: (i) harvesting transducer; (ii) impedance-matching, rectification, and energy storage circuitry; (iii) low-power microcontroller (MCU) with ADC for waveform acquisition; (iv) data logger or wireless transceiver for communication; (v) optional power management units (PMU). Circuit variants are source-dependent:

RF-based SEHS (“REHSense”):
- Wi-Fi access point emits 2.4 GHz OFDM signals; a dipole antenna receives incident RF; matching network minimizes reflection; multi-stage Schottky rectifier produces $V_{DC}(t)$ ; MCU samples $V_{DC}$ at 12-bit, 200 Hz (Ni et al., 26 Aug 2024).
- Sensing pipeline applies low-pass filtering, segmentation by activity (variance thresholding), normalization, and either variance-peak analysis for respiration or 1D CNNs for activity/gesture recognition.
KEH/Piezoelectric-based SEHS:
- A piezoelectric cantilever is stressed via mechanical loading, routed through a bridge rectifier into a storage capacitor; MCU samples either AC terminals, rectified voltage, or harvesting current (via shunt and op-amp) (Ma et al., 2020, Sandhu et al., 2020).
- Sensing can exploit voltage (with compensation for storage-induced distortion), or current signals cleanly decoupled from storage via converter-based topologies.
Modular Vibration SEHS (ViPSN 2.0):
- Hot-swappable ETUs for PZT, EMG, or TENG harvesters; configurable PMU provides energy flags at storage capacitor thresholds to coordinate task scheduling and checkpointing; peripherals are attached via standardized interfaces (Li et al., 7 Jul 2025).

3. Mathematical Modeling and Performance Metrics

Energy-positive Sensing Formalism

The energy-positivity of SEHS is quantified by the Acquisition Power Ratio (APR):

$APR = \frac{P_{har}}{P_{acq}}$

where $P_{har}$ is the average harvested power, and $P_{acq}$ the power consumed by data acquisition (ADC + low-power processing). APR $>1$ defines the “energy-positive” regime, allowing surplus energy for sensing, computation, or communication (Sandhu et al., 2020, Sandhu et al., 2020).

Source-Specific Dynamics

RF SEHS (Ni et al., 26 Aug 2024):
- Received power via Friis:
$P_{in} = P_t G_t G_r \frac{\lambda^2}{(4\pi d)^2}$ - Rectifier efficiency and harvested DC:

$\eta(P_{in}) = \frac{P_{DC}}{P_{in}},\quad V_{DC} = \sqrt{P_{harv} \cdot R_{load}}$
Piezoelectric/KEH SEHS (Ma et al., 2020, Peralta-Braz et al., 2022, Yao et al., 17 Nov 2025):
- Charged capacitor:
$E_{cap}(t) = \frac{1}{2} C [V_{cap}(t)]^2$ - Electromechanical plate dynamics model via modal coordinates, with coupled voltage and modal equations (see full state-space formulations in (Peralta-Braz et al., 2022, Yao et al., 17 Nov 2025)).
Duty-cycle and storage sizing (Li et al., 7 Jul 2025):

$E_{[V_a \to V_b]} = \frac{1}{2} C_{storage} (V_a^2 - V_b^2)$

$C_{storage} \geq \frac{2(E_{static} + E_{task})}{\eta(V_{Pstart}^2 - V_{Pclose}^2)}$

Quantitative Energy and Sensing Results

Platform	Harvested power	Sensing energy draw	APR (energy-positive?)	Classifier accuracy
REHSense RF (Ni et al., 26 Aug 2024)	4.5 mW (d=1m)	11.3–12.6 mW	APR ≈ 0.36	≈95% (HAR), 4.6% RMS resp. err
KEH (converter) (Sandhu et al., 2020)	13.2 μW (avg)	2.7 μW	APR ≈ 4–10	97% (HAR, CB-i)
Piezo insole (Ma et al., 2020)	164 μW (dual-PEH)	18.1 μW	APR ≈ 9.1	98.7–98.9% recall
ViPSN2.0 Beac. (Li et al., 7 Jul 2025)	125 μW	–	–	100% BLE shot
SHM PEH (Yao et al., 17 Nov 2025)	4 μW (sensing)	4 μW (PEH), 33 μW (accel.)	8.25	93% (PEH-VAE) (+13% vs accel.)

In kinetic/piezo scenarios, converter-based current sensing achieves both high accuracy and strong energy-positivity (Sandhu et al., 2020). For structural health monitoring via PEH, state-of-the-art models demonstrate a 98% reduction in sensing power and a 13% improvement in accuracy over accelerometer-based baselines (Yao et al., 17 Nov 2025).

4. Signal Processing and Machine-Learning Methodologies

SEHS implementations generally require customized pipelines to mitigate signal distortion induced by concurrent harvesting and to extract robust features:

RF SEHS: Savitzky-Golay filtering, sliding-window segmentation, dynamic variance threshold for activity detection, with 1D-CNN for gesture/activity/respiration discrimination (Ni et al., 26 Aug 2024).
PEH/KEH SEHS:
- Amplitude normalization filters compensate for capacitor charging effects:
$\tilde V_{A,B}(t) = \begin{cases} V_{A,B}(t)-V_C(t)+V^* & V_{A,B}(t)\geq V_C(t) \ V^* \frac{V_{A,B}(t)}{V_C(t)} & V_{A,B}(t)<V_C(t) \end{cases}$

where $V_C$ is the capacitor voltage, $V^*$ a reference. - Feature extraction: time/frequency-domain descriptors (RMS, skewness, IQR, etc.), CWT for spectral images, or direct voltage time-frequency representations (Ma et al., 2020, Yao et al., 17 Nov 2025). - Classifiers: BiLSTM and CNNs (activity/gait), AlexNet on CWT images (traffic speed, vibration classes), and unsupervised CVAE for damage detection (Ma et al., 2020, Peralta-Braz et al., 2022, Yao et al., 17 Nov 2025).
Energy/Task Scheduling: Schedulers adapt duty cycles, sampling rates, and task execution based on instantaneous energy, predicted harvesting profiles, and APR (Li et al., 7 Jul 2025, Sandhu et al., 2020).

5. Trade-Offs, Bi-Objective Design, and Optimization

Fundamental to SEHS system design is the energy-vs-sensing trade-off: maximizing harvested power and sensing accuracy are not always concordant objectives, as demonstrated in structural health monitoring and environmental contexts (Peralta-Braz et al., 2022, Yao et al., 17 Nov 2025). Bi-objective frameworks employ multi-objective optimization with Pareto analysis, frequently using NSGA-II or surrogate metamodeling (kriging):

Design Variables: harvester length, width, aspect ratio, tip mass, external load resistance, PMU thresholds (Peralta-Braz et al., 2022, Yao et al., 17 Nov 2025).
Pareto Fronts: distinct device geometries or circuit topologies optimize one objective at the expense of the other; e.g., short beams yield higher $E/A$ , longer ones richer spectra for sensing (Peralta-Braz et al., 2022).
Context Sensitivity: Road roughness, noise, damage location, and ambient excitation spectra alter the Pareto surface, necessitating adaptable or context-specific designs (Yao et al., 17 Nov 2025).

Bi-objective optimization thus directs practitioners to select device parameters and signal features that best match deployment requirements along the energy–sensing axis.

6. Scheduling, Adaptation, and Control Algorithms

SEHS systems, particularly for IoT and intermittent-compute nodes, require energy-aware scheduling and adaptation that jointly manage harvesting variability, transient storage, and deadline-constrained tasks (Li et al., 7 Jul 2025, Sandhu et al., 2020):

Scheduling Approaches:
- Dynamic Voltage and Frequency Scaling (DVFS), task decomposition/combination, duty-cycling, lazy scheduling, and online greedy heuristics balance energy arrival against computational demands (Sandhu et al., 2020).
Energy Indication and Checkpointing:
- Use of hardware comparators, ADC-based polling, and explicit voltage flags to trigger transitions among cold-start, accumulation, task-execution, checkpoint, and shutdown (Li et al., 7 Jul 2025).
Adaptation Strategies:
- Dynamic sampling/braking, threshold retraining, and CNN model fine-tuning accommodate environmental drift, SNR drop, and context migration (Ni et al., 26 Aug 2024).
- In battery-less intermittent setups, PMU-controlled checkpoint/restart cycles allow seamless operation under unpredictable harvest (Li et al., 7 Jul 2025).
Open Problems:
- Real-time prediction of harvest in KEH/TEH/RFEH, multi-source fusion scheduling, MPP tracking under dual-use constraints, cross-layer OS for energy-pool sensor integration (Sandhu et al., 2020).

7. Application Domains and Extensions

Representative SEHS deployments include:

RF-based Sensing: Battery-free wireless respiration/activity/gesture recognition with up to 95% accuracy and a 98.7% reduction in front-end power versus conventional Wi-Fi receivers (Ni et al., 26 Aug 2024).
Wearable Gait and Motion Recognition: SEHS-enabled insoles attain 98.9% recall and energy harvest of ~164 μW per step (Ma et al., 2020).
Infrastructure Monitoring: Combined energy and unsupervised damage sensing in bridges achieves PEH-based CVAE accuracy up to 13% > accelerometer baselines, with 98% power savings (Yao et al., 17 Nov 2025).
Distributed IoT: Modular platforms (ViPSN 2.0) demonstrate light/heavy/streaming task regimes, dynamic hardware swaps, and checkpointed battery-free imaging under marine/wind/fingertip energy profiles (Li et al., 7 Jul 2025).
Engineered Metastructures: Graded-resonant metasurfaces achieve broadband “rainbow trapping” for local displacement amplification, enhanced piezoelectric voltages, and both distributed sensing and energy harvesting (Ponti et al., 2019).

SEHS principles and platform abstractions generalize across vibration, RF, solar, and thermal harvesters, enabling broad application in autonomous, energy-perpetual sensor networks.

In summary, the advancement of SEHS involves the convergence of physical modeling, circuit optimization, signal-processing pipelines, energy/task co-scheduling, and application-level adaptation. Progress in SEHS underpins sustainable IoT, enhances battery-free structural health monitoring, facilitates low-cost, maintenance-free wearable context sensing, and lays the groundwork for new metastructural energy/sensing materials (Ni et al., 26 Aug 2024, Ma et al., 2020, Sandhu et al., 2020, Peralta-Braz et al., 2022, Yao et al., 17 Nov 2025, Li et al., 7 Jul 2025, Ponti et al., 2019, Sandhu et al., 2020).