- The paper systematically benchmarks BLE against sub-GHz RF for ingestible applications, finding that BLE with a front-end module (FEM) compensates for 2.4 GHz attenuation and achieves lower power consumption and latency.
- BLE configurations consistently maintained a 50 kbps throughput across different tissue depths, with a significant 3x power advantage over 915 MHz and improved latency compared to sub-GHz systems.
- BLE integration includes compact chip antennas and smartphone/hospital IT compatibility, simplified mechanical tenability, and direct smartphone/hospital IT interoperability for potentially superior in-body electronic sensing systems.
Motivation and scope
Ingestible electronics have historically favored sub-GHz RF links (400–915 MHz) over Bluetooth Low Energy (BLE), on the assumption that severe tissue attenuation at 2.4 GHz makes BLE unsuitable for in-body communication (2601.19241). This paper challenges that frequency-centric view with a systematic experimental benchmark of BLE against representative sub-GHz schemes across five axes: power consumption, throughput, tissue-induced attenuation, end-to-end latency, and system-level integration (antenna form factor, ecosystem compatibility). The central claim is that BLE, when paired with a front-end module (FEM) to compensate for 2.4 GHz attenuation, is the more energy-efficient and lower-latency option for most ingestible sensing applications—specifically those requiring less than 100 kbps.
Experimental setup
The BLE platform is Nordic's nRF54L15 DK, operating at 8 dBm natively or up to 20 dBm effective radiated power with an external nRF21540 FEM; the receiver is an nRF21540 DK. The sub-GHz baseline is the Texas Instruments CC1310, chosen for its wide frequency range (287–1054 MHz), low power, and high PHY rate (up to 4 Mbps), tested at both 433 MHz and 915 MHz at 8 dBm. Tissue attenuation was emulated by submerging transmitters in a water tank (40 × 30 × 25 cm) at depths of 0–10 cm, with receivers in air at 1 m; water was justified as a tissue-mimicking medium given its dielectric similarity to biological tissue. Transmitter power was profiled at 100 ksamples/s using the Nordic Power Profiler Kit II. Throughput–power characterization spanned 50 bps to 2300 kbps via duty-cycle adjustment, and latency was measured oscilloscopically via GPIO edge timing between transmitter assertion and host-side reception (Raspberry Pi for both paths).
A methodological caveat worth noting: the water-tank model is a simplified homogeneous phantom, and the 433 MHz RSSI measurements are confounded by antenna mismatch—the CC1310 board used a 915 MHz antenna at 433 MHz, introducing substantial mismatch loss. The authors acknowledge this explicitly, so absolute 433 MHz link numbers should be interpreted cautiously even though relative trends remain informative.
Attenuation and link quality through tissue
All four configurations sustained approximately the 50 kbps target throughput across all depths, but their link-quality behavior diverged. BLE at 8 dBm exhibited an RSSI attenuation slope of roughly −2.26 dB/cm; adding the FEM preserved the slope while providing 12 dB additional margin. The 915 MHz configuration attenuated at about −0.79 dB/cm, and 433 MHz showed nearly constant RSSI across depths, confirming superior penetration at lower frequencies. Despite this, the key result is that BLE with the FEM achieved RSSI comparable to the 915 MHz system (~−40 dBm) at 10 cm depth—demonstrating that amplifier compensation closes the link-margin gap at clinically relevant depths. Sub-GHz throughput was essentially flat over the measurement window, attributed to lower noise floors and more stable propagation than the 2.4 GHz band, whereas BLE showed fluctuations beyond 6 cm depth, smaller at 20 dBm.
Power consumption was nearly depth-independent for all configurations. Critically, the 915 MHz setup consumed roughly three times the power of BLE at 20 dBm, and 433 MHz consumed nearly ten times more. This means that for low-to-moderate throughput ingestible sensing, FEM-augmented BLE delivers reliable communication through tissue at substantially lower energy cost than sub-GHz alternatives—an implication that directly undermines the standard justification for sub-GHz selection.
Throughput–energy trade-off
The power–throughput curves reveal a crossover near 100 kbps: below it, BLE at 20 dBm consumes less power than the 915 MHz system (about 12 mW at the crossover point); above it, 915 MHz becomes more efficient and reaches an effective throughput of 2300 kbps—nearly twice BLE's 2M PHY capability. In the low-throughput regime (50 bps–10 kbps), BLE achieves roughly one order of magnitude lower power consumption than 915 MHz. Against 433 MHz, BLE holds a decisive advantage across 50 bps to 50 kbps, and the 433 MHz system cannot sustain effective throughput beyond 50 kbps, making BLE the only viable evaluated option for higher-rate applications in that comparison.
An important nuance concerns the FEM's duty-cycled behavior: although the 20 dBm configuration draws a peak current of 118.73 ± 1.17 mA versus 28.25 ± 0.15 mA at 8 dBm, the FEM sleeps during idle periods, so time-averaged power at low throughputs is nearly identical between the two BLE configurations. This explains why amplifier-equipped BLE retains its energy advantage precisely where most ingestible sensors operate (pH, temperature, pressure, gas sensing at 10–200 bps; vital signs around 2 kbps; electrogastrography at tens of kbps). Capsule endoscopy, requiring Mbps-class throughput, falls outside BLE's favorable regime—a boundary the paper states plainly rather than obscuring.
End-to-end latency
BLE achieved approximately 7.45 ms average end-to-end latency (STD 0.82 ms) with a 7.5 ms connection interval, essentially identical at 8 and 20 dBm. The 915 MHz path averaged 18.49 ms despite its higher raw throughput, because data must traverse a USB-attached sub-GHz transceiver before reaching the host; 433 MHz fared worst at 59.20 ms. The implication is that BLE's native interoperability with general-purpose computing infrastructure yields better real-time performance than higher-throughput sub-GHz radios burdened by gateway overhead—relevant for ingestible applications where timely alerting matters, though well within the ~250 ms tolerance typical of vital-sign monitoring rather than surgical-grade (<10 ms) requirements.
Mechanical integration and ecosystem
Antenna volume strongly constrains capsule layout, battery capacity, and packaging. Surveyed 2.4 GHz ingestible systems use millimeter-scale chip antennas (0.185–2.56 mm³), whereas 915 MHz designs require tens of cubic millimeters and 433 MHz designs rely on bulky helical structures (e.g., ~Ø9 × 7 mm to Ø10 × 15 mm). Compact chip antennas reduce geometric coupling with the capsule layout, simplifying integration and manufacturing repeatability. At the ecosystem level, BLE offers direct smartphone and hospital IT interoperability without proprietary gateways, plus standardized encryption, authentication, and over-the-air firmware updates—capabilities that are vendor-specific or custom-implemented in sub-GHz platforms but essential for medical-device cybersecurity and lifecycle management.
Limitations and open questions
The paper's conclusions rest on several assumptions. The water-tank phantom is homogeneous and does not capture the layered, heterogeneous dielectric environment of the GI tract, nor multipath effects from organ boundaries or motion; in vivo validation remains absent. The 433 MHz comparison is weakened by the antenna mismatch noted above. Latency measurements used a fixed 7.5 ms connection interval under benign conditions and did not characterize behavior under interference in the crowded 2.4 GHz ISM band, coexistence with Wi-Fi, or worst-case body positions. The 100 kbps crossover is specific to the particular SoC/FEM pair and duty-cycling scheme evaluated; other radio implementations could shift it. Finally, regulatory considerations for 20 dBm transmission from an ingestible device (SAR compliance, interference with other medical devices) are not addressed experimentally.
Conclusion
This work provides quantitative evidence that the conventional dismissal of BLE for ingestible electronics is not supported when system-level factors are considered. With FEM compensation, BLE maintains reliable links through centimeters of tissue-mimicking medium while consuming substantially less power than sub-GHz alternatives for the sub-100 kbps applications that dominate ingestible sensing, achieving ~3× lower latency than USB-gatewayed sub-GHz paths, and offering decisive advantages in antenna miniaturization and ecosystem maturity. The results indicate that sub-GHz communication is not universally necessary for ingestible devices, and that radio selection should be application-driven rather than frequency-driven.