Human body communication transceivers

Taking a cue from the Internet of Things, the Internet of Bodies (IoB) is defined as a network that connects body-centric wearable, implantable, ingestible and injectable devices. IoB devices have become possible owing to simultaneous advancements in microelectronics, signal processing and wireless communications, enabling numerous applications in our everyday life (Fig. 1).

The Internet of Bodies includes devices, wireless systems and applications.  EEG, electroencephalogram; AV, audio-video; HBC, human body communication; RFID, radiofrequency identification; VR, virtual reality.

Full size image

Meeting the diverse demands of IoB applications involves addressing key objectives such as compact size, affordability, low-power consumption, reliability, minimal delay, security and convenience; requirements that directly translate into size, weight and power–cost constraints, mainly driven by battery size and circuit area. Maximizing IoB node lifetime requires low-power transceivers with simple designs. Techniques such as current reuse, duty cycling and power-efficient architectures help reduce power consumption, although they can conflict with the ultra-reliable, low-latency needs of time-sensitive IoB applications¹. Furthermore, IoB devices must limit electromagnetic exposure based on specific absorption rate (SAR) and ensure that currents remain within health regulation standards^2,3. Packaging should be durable and biocompatible for in-body devices, whereas wearable IoB devices should prioritize comfort and safety using technologies such as stretchable or textile electronics.

From a networking perspective, traditionally, IoB devices relied on using radiofrequency (RF) transceivers using protocols such as Bluetooth or Zigbee. However, the increased focus on miniature, low-power IoB nodes resulted in a surge of interest in human body communication (HBC), where IoB nodes network by exchanging messages directly using the human body as a communication medium. The HBC field has been growing focusing on various aspects of IoB systems and wireless body area networks (WBANs), including potential applications, propagation characterization, channel modelling, antenna design, body coupling mechanisms, development trends, challenges, electrode modelling and impedance modelling^{1,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18}. Compared with RF-based health wearables, HBC systems offer several advantages: (1) RF signals are highly radiative, causing energy loss and vulnerability to attacks, unless sophisticated communications and encryption modes are used; (2) RF devices require larger circuits and batteries, by virtue of requiring an RF section and antenna; and (3) RF devices operating in industrial, scientific and medical (ISM) bands face interference. By contrast, HBC, using the body as a transmission medium, ensures secure connections with low signal leakage, lower path loss and reduced power needs^19,20. Moreover, HBC avoids RF shadow fading issues (caused by body movement), making it ideal for ultra-low-power, low-cost IoB transceivers, with state-of-the-art HBC transceivers achieving energy efficiency as low as 6.3 pJ per bit (ref. ²¹).

In this Review, we focus on design approaches and insights unique to HBC transceiver architectures, addressing key challenges such as dynamic channel variations, skin–electrode interface complexities, interference mitigation and energy efficiency. We examine intrabody communication techniques used in WBANs and explore critical design considerations for developing reliable and efficient HBC transceivers. Additionally, we present essential strategies for capacitive coupling-HBC (CC-HBC), including circuit optimizations and adaptive power control. Finally, we discuss the barriers to widespread commercial adoption — such as standardization gaps, security concerns and integration with existing medical technologies — and highlight promising research directions in energy harvesting, ultra-thin soft electronics and deep-learning-driven transceiver optimization.

Effective intrabody communication is crucial for the seamless operation of IoB networks, with different strategies offering different trade-offs in power consumption, data rate, security and interference resilience (Table 1). Although traditional RF-based methods, such as Bluetooth, Zigbee, Z-Wave, adaptive network topology, near-field communication (NFC) and radiofrequency identification (RFID), provide reliable connectivity, they suffer (to various degrees) from high energy consumption, security vulnerabilities and interference in crowded frequency bands. By contrast, HBC leverages the conductive properties of the body to enable secure, low-power and interference-resistant data transmission. This section explores the advantages and limitations of these approaches, with a particular focus on the IEEE 802.15.6 standard, which defines the physical layers for narrowband, ultrawideband and HBC. A comparative analysis highlights why HBC is emerging as a promising alternative for IoB applications, offering reduced signal leakage, enhanced energy efficiency and improved privacy compared with conventional RF-based methods.

IEEE 802.15.6 standard

The IEEE 802.15.6 standard is an international standard for WBANs that focuses on low-power, short-range and reliable wireless communications near the human body²². The IEEE 802.15.6 targets high reliability and robustness in challenging environments, such as those involving body movement and various obstructions typical in medical settings. It encompasses multiple frequency bands, including narrowband, ultrawideband and HBC. Narrowband operates in licensed and unlicensed bands such as 402–405 MHz and 420–450 MHz, among others. Ultrawideband supports a default and high-quality service mode, with frequencies spanning from 3494.4 MHz to 9984.0 MHz, facilitating high data rate transmission and precise ___location tracking²³. HBC uses frequencies below 100 MHz for transmissions directly through the human body, ideal for devices in contact with the skin.

The standard also covers frequencies for specific services, such as the medical implant communication service, the wireless medical telemetry service and the ISM bands. Notably, these services are not part of IEEE 802.15.6 but were developed by the medical community, regulated in the USA by the Federal Communication Commission to minimize interference risks. The medical implant communication service is allocated the 402–405 MHz band for communication between medical implants and external devices, whereas the wireless medical telemetry service uses designated bands (608–614, 1,395–1,400 and 1,427–1,432 MHz) for transmitting medical telemetry data in a controlled environment²⁴. The ISM band, crowded with Internet of Things devices, benefits from ultrawideband high data rate capabilities and power efficiency, enhancing the longevity of IoB devices²⁵.

Other RF communication techniques

Apart from the IEEE 802.15.6 standard, for short-range device-to-device communication, numerous other RF standards have found wide-range acceptance. Bluetooth, introduced in 1998 and operating between 2.40 GHz and 2.48 GHz in the ISM band, supports enhanced connectivity with advancements such as Bluetooth v6.x, achieving to 3 Mbps data rates. Introduced in Bluetooth 4.0, bluetooth low energy (BLE) substantially improved energy efficiency through techniques such as duty cycling, making it fundamental for wearable health devices and real-time patient monitoring^26,27. Zigbee operates over 10–100 m range and excels in low-rate data communication. It is more energy-efficient and has quicker mode-switch latency than classic Bluetooth, although BLE surpasses it in power efficiency^28,29. Secure through access control list and advanced encryption standard, Zigbee is mainly found in network-rich environments such as hospitals, owing to its mesh networking capabilities, rather than in consumer wearables. Z-Wave operates at up to 100 m range with typical data rates of 100 kbps. It supports robust mesh networking ideal for home automation applications such as lighting and security. Known for its low-power use and cost–effectiveness, Z-Wave is mainly used in large network installations for health and wellness³⁰. ANT, an ultra-low-power multicast technology in the 2.40 GHz ISM band, supports various network topologies and adapts to many applications with its efficient power use and flexible network configuration. Although less common in consumer health technology than BLE, the connectivity and the energy efficiency of ANT are favoured for long-term monitoring devices. NFC operates at distances up to 10 cm, supporting data rates up to 848 kbps. Widely used in payment and data sharing, the role of NFC in health technology is growing. This technology is used for patient identity documents and device management via secure, easy communication³¹. RFID technology spans several frequencies from 125 kHz to 5.8 GHz and is pivotal in health care for patient tracking and equipment management. RFID tags, ranging from passive to active, facilitate non-contact data transfer, improving patient care efficiency and enabling remote device monitoring^31,32.

Although specific implementations might vary, Bluetooth generally demonstrates a good combination of higher data rates with lower power consumption, with efficiency in the order of 10 nJ per bit (nRF5340). NFC and ANT are typically around 100 nJ per bit (NTAG), whereas Z-Wave and Zigbee are close to 1 μJ per bit (EFR32ZG23). Bluetooth, BLE and UWB are particularly well suited for scenarios requiring high data rates, such as medical images or audio transmissions. By contrast, NFC, ANT, Zigbee and Z-Wave are more suitable for applications with lower bit rates, such as vital sign monitoring.

Human body communications

HBC can be broadly categorized into two types: magnetic HBC (mHBC) and electrical (eHBC). Furthermore, eHBC can be further classified into two subcategories: galvanic-coupling HBC (GC-HBC) and CC-HBC.

mHBC technology was initially proposed in 2014 (ref. ³³), followed by the introduction, in 2019, of the first mHBC transceiver design³⁴. mHBC uses a transmit coil (TX) and receive coil (RX) to generate and receive magnetic energy for data transmission (Fig. 2a). An effective loop is formed through the body when the feet are connected to the ground electrode and the hands touch the signal electrode, otherwise the loop remains open³⁴. Compared with eHBC, mHBC takes advantage of the fact that the human body exhibits better permeability than conductivity at megahertz frequencies. The magnetic field can thus propagate freely within the body, yielding a lower path loss than eHBC, and can be sensed in a single-ended manner, leading to reduced hardware complexity. Moreover, the inherent filtering characteristics of the human body and inductive coil help reduce external interference from human postures and environmental variations³⁵.

a, Magnetic coupling: a transmitter coil generates a magnetic field that induces a corresponding field in a receiver coil, with the human body influencing the coupling efficiency by shaping the near-field interactions. b, Galvanic coupling: both electrodes are placed on the human body. c, Capacitive coupling: the signal electrode is on the human body, and the ground electrode floats in the air. C_i is the interwinding capacitance. RX, receive coil; TX, transmit coil.

Full size image

The GC-HBC system uses a pair of electrodes, namely, the ground electrode and signal electrode, to transmit signals through the human body using the electric field (Fig. 2b). Both the signal and ground electrodes are in contact with the human skin, offering a stable signal path and channel conditions as both the forward and backward paths traverse the human body. This stability makes GC-HBC suitable for transmitting vital or physiological data in both in-body and on-body scenarios. However, the galvanic coupling operates best in low-frequency bands because of the relatively low conductivity of tissues in the human body, typically 1 S m⁻¹ in the megahertz range³⁵. This limitation results in a restricted bandwidth, lower data rates (<1 Mbps) and a shorter communication range. Furthermore, in galvanic coupling, ionization can occur on the human skin because of electrochemical reactions. This phenomenon is often catalysed by the presence of electrolytes in sweat or body fluids. While facilitating signal transmission in GC-HBC, ionization can raise safety concerns, such as skin irritation or damage, especially when the node is operated for prolonged durations.

Unlike GC-HBC, in CC-HBC, the signal electrode is in contact with the human skin, whereas the ground electrode remains floating in the air (Fig. 2c). Thus, a coupling pathway between the electrodes through the environment and the human body is formed. Compared with GC-HBC, CC-HBC is more susceptible to external interference, human postures and environmental conditions²⁰, but has the advantage of operating in higher frequency bands¹, thus supporting high data rates and enabling communication over longer distances with lower path loss. CC-HBC exhibits superior performance compared with GC-HBC at frequencies above 60 kHz (ref. ⁸). Preliminary investigations into the potential of implanted CC-HBC have also been conducted, using an isolated ground electrode³⁶. Therefore, CC-HBC has received more attention and is currently being extensively investigated.

Each communication technique has a unique profile for channel loss and usage. RF communication, although versatile and capable of long range, faces variable channel losses as it is heavily influenced by frequency and environmental factors and subject to physical layer security vulnerabilities. mHBC generally incurs a complicated hardware profile, resulting in lower adoption. eHBC provides the best combination of efficient hardware and reasonable channel loss, which is why eHBC is the most used mode, particularly in health monitoring and fitness tracking applications. Considering the rise in interest on CC-HBC, in the following we will provide a deeper exploration of its characteristics and discuss the latest advances in research.

The design of CC-HBC transceivers must consider CC-HBC channel characteristics, adhere to safety standards and incorporate effective circuit design strategies. The well-being of individuals exposed to electromagnetic fields is strictly regulated by standards; however, most IoB devices operate well below the safety thresholds, allowing for safe effective communications.

CC-HBC channel

A crucial factor in the design of any communication system is the channel model. Adapting to the variability of the human body as a communication channel — influenced by different postures, environments, external objects and interference — while maintaining reduced transceiver complexity is an elusive goal for HBC systems. Posture changes, such as sitting or standing, substantially impact signal transmission by altering the electrical properties of the body. Environmental factors, such as humidity and nearby electronic devices, introduce noise and interference, necessitating dynamic adaptations for reliable communication. The ___location of the electrodes adds an extra dimension of complexity, as configurations — such as elbow to arm versus elbow to heel — affect the signal path and impedance and must be considered based on specific IoB applications. Finally, operational parameters, such as frequency and type of electrodes, have a direct impact on the channel. Thus, establishing an accurate channel model and understanding its ___domain of applicability are essential.

Modelling of CC-HBC channel can be approached through various methods, including analytical models describing CC-HBC channel features with equations, numerical analysis using computational technique to solve the equations and provide numerical solutions (for example, finite element method and finite different time ___domain) and circuit models representing CC-HBC channel behaviour using equivalent electrical circuit¹. Among these approaches, circuit models offer an effective means of representing CC-HBC by using equivalent resistor–capacitor components, resulting in simple transfer functions that mathematically describe the signal propagation along the transmission path within or on the human body. Moreover, circuit models are advantageous in terms of computational efficiency and accuracy for a wide range of frequencies of interest. However, as the frequency reaches up to 100 MHz, radiative effects reduce the accuracy and alternative means such as full electromagnetic wave simulations might be needed¹.

Operational frequency

The operational frequency is a crucial factor influencing various aspects of HBC, including channel characteristics, data rate, interference avoidance and system stability. Although the frequency range of interest for HBC is typically 10 kHz to 100 MHz, there is no consensus on the operational frequency used in transceiver architectures, and designs often do not adhere to the 5.25 MHz bandwidth centred at 21 MHz recommended by the IEEE 802.15.6 standard²².

At frequencies lower than ~30 MHz, in the electroquasistatic (EQS) band³⁷, the wavelength is orders of magnitude larger than the human body size, and the signal propagates within the body because of the skin’s higher impedance compared with underlying layers, thus it is considered as the near-field propagation ___domain. The far field and near field are defined based on the product between the wave number k and the distance r where the transmitter and receiver are modelled as infinitesimal dipoles^38,39. As the frequency increases to tens of megahertz or higher, the far-field model (kr » 1) becomes more appropriate for describing signal transmission where 1/r becomes the dominant factor in signal attenuation. Compared with near-field EQS propagation, far-field surface waves propagate along the skin’s surface and the signal power becomes radiative. The human body acts as a monopole antenna with a body resonance peak occurring between 10 MHz and 100 MHz⁴⁰. The position of the peak might depend on the properties and postures of the human body⁴¹. Far-field communication offers higher bandwidth, enabling larger data rates and longer distance transmissions⁴². However, at higher frequencies, impedance matching becomes necessary because of the comparable wavelength and circuit dimensions.

Circuit models

The CC-HBC channel can be divided into forward and return paths, which are essential components determining the overall channel quality. The forward path consists of signal electrodes in contact with the skin, whereas the return path is formed by the coupling between the ground electrodes through the air and the body. The distance between the ground electrodes affects the capacitance, impedance and channel quality. As the distance between the ground electrodes increases or decreases because of the human movements, the capacitance changes, consequently impacting the return path impedance and overall channel quality. Environmental factors, such as nearby metallic objects, temperature, humidity, pressure and noise, can also influence the characteristics of the return path. Therefore, depending on the operation frequency, continuous monitoring and compensation of variable losses are crucial for optimizing and stabilizing the channel characteristics.

The basic equivalent circuit model of the CC-HBC channel⁴³ can be extended to include parasitic elements that influence the channel, especially at high frequencies^44,45 (at frequencies higher than 30 MHz) (Fig. 3a and Table 2). At the other extreme (at frequencies lower than ~30 MHz (ref. ³⁷)), the circuit model can be simplified as in the EQS band by ignoring the parasitic components, including the leakages C_leak (~1 pF)⁴⁴, the coupling between the ground electrodes and external ground (~1 pF)⁴³ and interdevice coupling (~60 fF)⁴⁴ (Fig. 3b). The forward path of the HBC channel can be represented by the internal resistance of the voltage source R_TX, the body channel impedance Z and the receiver termination impedance Z_RX. The return path can be modelled with the return path capacitance C_ret. A simplified relationship between the output voltage V_out and the input voltage V_in can be expressed as²¹:

in which s = j2πf is the complex frequency variable, with f the operation frequency.

a, Circuit model diagram for wearable capacitive coupling-human body communication (HBC)⁴³. The black line indicates the forward path for signal transmission through the human body, the red line is the backward path for ground coupling through the air and the cyan line indicates cross-coupling paths between electrodes. The solid lines indicate coupling through the human body, whereas the dashed lines indicate coupling through the air. b, Simplified capacitive coupling-HBC channel circuit models^21,89 in electroquasistatic HBC by ignoring minor parasitic components and reducing the resistors and capacitors. PL, path loss. c, Circuit model of electrode–skin interface⁴⁹. The full description of the circuit model parameters is presented in Table 3.

Full size image

Impedance components have a critical role in characterizing the CC-HBC channel. Within the EQS band, lumped circuit models can be used to describe the channel, where path loss is measured by the ratio of the received voltage to the transmitted voltage (equation (1)). It is desirable to minimize the source impedance and maximize the load impedance to reduce voltage loss. A higher receiver impedance is preferred as it results in more voltage applied at the receiver end. The received voltage V_out increases with higher receiver input impedance (Z_RX), leading to a smaller path loss. Therefore, avoiding a 50 Ω device impedance at the receiver end is preferable, as it might result in higher voltage loss through the body channel⁴⁶. Conversely, a lower transmitter internal impedance should be used to reduce path loss.

Compared with resistive termination, capacitive termination offers several advantages, including lower path loss, wide bandwidth and flat path loss pattern. The channel loss with capacitive termination can be expressed by C_ret/(C_ret + C_load) at the lower frequencies encountered in the EQS band²¹. The internal resistance of the source R_TX is usually 50 Ω⁴⁶. In comparison, typical values for C_ret (in the range of picofarads) and Z_RX (in the range of megaohms) have much larger impedance values in the EQS band, thus rendering R_TX negligible. Clearly, as the frequency increases, the relative value of the impedance changes, and the source resistance should be considered. As the transfer function remains constant, the path loss remains flat across the frequency band of interest (Fig. 3b). Therefore, the channel loss is primarily determined by the ratio of the load capacitance to the return path capacitance⁴⁶. A larger value of C load leads to a higher path loss. Resistive termination exhibits high-pass characteristics because of the resistance and return capacitance. It can be modelled as a high-pass filter with a cutoff frequency given by f_c = 1/(R_loadC_ret)²¹.

Ag/AgCl wet electrodes are the most used electrodes in bio-electric signal monitoring^47,48. These electrodes offer advantages such as good contact with the skin, low contact impedance, affordability, high stability and ease of fabrication^47,48. Another type of electrode widely used is the metal plate dry electrode, which exhibits good conductivity and is often used for capacitive coupling of signal energy into the body. The skin–electrode interface⁴⁹ can be modelled as in Fig. 3c. C_e is the double layer capacitance, R_CT is the charge transfer resistance, R_e is the electrode impedance and E_rev indicates the reversible or equilibrium potential⁴⁹. The skin–electrode contact impedance is subject to variation caused by diverse factors such as human motion and postures, the presence of an air gap between the electrode and the skin, dehydration of wet electrodes, sweat and humidity and dynamic environmental conditions. The stratum corneum, the outermost layer of the skin, greatly influences skin impedance. In the absence of sweat or humidity, its impedance is high. Sweat can hydrate the stratum corneum, reducing skin impedance and improving contact between the skin and the electrode^50,51. This improves current transfer and increases interface capacitance. However, prolonged exposure to sweat can degrade the contact by affecting the adherence of the electrode to the skin⁵². For dry electrodes, long-term sweat exposure can lead to oxidation on the electrode surface, increasing contact impedance⁵³. The overall effect of the contact impedance is modulated by the operational frequency. In the EQS (low-frequency) band, contact impedance is negligible compared with return path capacitance and receiver termination. At higher frequencies, contact impedance increases, reducing the voltage applied to the body and degrading the received signal. Therefore, sweat has little effect at low frequencies but can weaken signal strength at higher frequencies⁵¹.

Interference

Interference in HBC systems can be categorized into three types: inter-symbol interference, which arises from the time smearing of symbols during transmission; structured co-channel interference, such as amplitude modulation and frequency modulation (FM); and interference from surrounding electrical devices emitting electromagnetic radiation, including air conditioners, fans and smartphones⁵. These signals can be received by the human body acting as a broadband antenna in the frequency range of 40–400 MHz (ref. ⁵). Modulation techniques significantly influence signal integrity and resilience to interference. On–off keying is widely used in HBC for its simplicity and low-power consumption but is sensitive to additive noise. Orthogonal frequency-division multiplexing is more complex but offers better interference resilience owing to its use of multiple orthogonal subcarriers and guard intervals to mitigate inter-symbol interference and crosstalk⁵⁴. Orthogonal Walsh codes can also be applied to enable different signals to share the same frequency⁵⁵. Structured interference from amplitude modulation and FM signals can be addressed using receiver designs such as integrating receivers that sample at periodic zero-crossings of the interference²¹. Adaptive frequency hopping can dynamically select frequencies with lower interference levels⁵⁴, whereas filters, such as the fourth-order Sallen–Key low-pass filter, target specific frequency bands⁵⁶. Additionally, decision feedback equalization with multiple taps helps counteract low-frequency path loss and eliminate inter-symbol interference⁵⁷.

Clearly, the parameters that best describe the HBC channel depend on many factors, including the frequency of operation and the bandwidth. For low-frequency operation, characterization is typically performed by measuring the path loss at different frequencies, based on received power or voltage. In wideband channels, which support higher data rates and wider bandwidths, the impulse response method is commonly used to capture the time-___domain characteristics of the human body channel. This method involves applying custom channel-sounding signals to the body using a battery-powered device and measuring the received signal’s power under various conditions. In addition to path loss, metrics such as mean path loss, root mean square delay spread and mean and maximum excess delays are also measured^58,59. The impulse response method provides insights into both the time-___domain behaviour of the channel, including the impact of inter-symbol interference, and the channel’s frequency response. These metrics become increasingly important as carrier frequencies and data rates increase.

HBC safety standards

To ensure human safety from electromagnetic fields, international standards, such as the ones defined by the International Commission on Non-Ionizing Radiation Protection (ICNIRP)^2,3,22,60 or the IEEE C95.1 (ref. ²), define limits on contact current, field intensity and SAR for exposure up to 300 GHz. For example, the IEEE 802.15.6 (ref. ²²) outlines wireless communication parameters around the human body.

For eHBC systems, controlling electric field and contact current ensures that the current along the body stays below safety limits. Excessive current can damage nerves or tissue. The IEEE C95.1 defines touch and grasp contacts with areas of 1 cm² and 15 cm², respectively, although these do not fully apply to HBC devices. The most conservative limit can serve as a ceiling for HBC devices. The standard covers a freestanding individual insulated from the ground, using root mean square current averaged over 0.2 s. Painful heat may occur at 46 mA between 100 kHz and 10 MHz with over 10 s exposure. ICNIRP sets 10 mA and 20 mA limits for children and adults, respectively³. Most eHBC systems operate well below these standards, with microampere-level currents.

Managing electric field exposure in eHBC is essential to avoid electrostimulation and heating. Electrostimulation occurs up to 5 MHz, whereas heating is significant above 100 kHz. Standards limit in situ electric field strength across body parts for 0–5 MHz frequencies, with a 30-min averaging period². Local exposure limits may be stricter than whole-body limits and should be followed. When multiple HBC devices operate simultaneously, their combined field strength and power density percentages should not exceed 100%.

ICNIRP and IEEE standards do not cover individuals with conductive implants, which can alter electromagnetic field and SAR distribution^2,3. Implant factors such as size, shape and orientation affect SAR. Smaller implants concentrate electromagnetic field flux, increasing absorption, whereas larger implants scatter electromagnetic fields, raising SARs, especially when resonating with field wavelengths. The implant length of about λ_T/2 causes the highest SAR enhancement, in which λ_T is the wavelength in the media⁶¹. Implants may reduce transmission gain by acting as sinks for the electric field⁶². Technologies such as titanium shielding and bandpass filters can be used to reduce electromagnetic interference, but issues persist⁶³.

Several CC-HBC transceiver architectures have been proposed to target HBC-specific design goals (Table 3). The architectures presented cover a diverse range of parameters, with transmission frequencies ranging from 1 MHz to 550 MHz, data rates spanning from 10 kbps to 150 Mbps and receiver sensitivities varying from −18.87 dBm to −98.9 dBm. Each architecture targets specific design strategies within the extensive HBC design space to achieve efficient HBC transceivers. The table highlights the diversity of such approaches and indicates which specific design choice was chosen for a given architecture.

Low-power designs

Lifetime maximization of IoB nodes is crucial for battery-powered HBC transceivers. On the basis of their energy efficiency, HBC architectures can broadly be categorized into three clusters: high energy efficiency (η < 100 pJ per bit), medium energy efficiency (100 pJ per bit < η < 1 nJ per bit) and low energy efficiency (η > 1 nJ per bit) architectures, respectively (Fig. 4a). High energy-efficient transceivers (η < 100 pJ per bit) prioritize ultra-low-power consumption by using simple modulation schemes such as on–off keying, Manchester code and non-return to zero. Energy per bit is reduced by increasing data rate, suggesting that rate-independent energy usage gets amortized over a larger number of bits. This reduction is maintained to a point, after which overall energy starts to increase with higher data rate, suggesting that an ideal rate versus energy consumption exists. High energy efficiency designs use smaller process nodes (65–90 nm), which improve efficiency. Medium energy efficiency transceivers (100 pJ per bit < η < 1 nJ per bit) balance power efficiency and performance, utilizing modulation schemes such as binary phase shift keying (BPSK) coding, quadrature phase shift keying and polarized orthogonal frequency division multiplexing (P-OFDM) with BPSK, which provide better noise resilience but at the cost of increased energy consumption. Power usage is significantly higher than in high energy efficiency transceivers, with TX/RX power ranging between 0.6 mW and 9.4 mW. Low energy efficiency transceivers (η > 1 nJ per bit) tend to use older process nodes (180 nm) and more complex modulation schemes such as frequency shift keying (FSK) and discontinuous phase FSK, which lead to higher power consumption (up to 81 mW). They typically operate at lower data rates (<3 Mbps), with some designs as low as 0.2 Mbps. Owing to the reduced data bandwidth, they can achieve higher sensitivity reaching −98.9 dBm in some cases.

a, Power consumption as a function of the data rate; modes include the following: ETM, entertainment mode; HBC, health-care mode; HI, hub IC; NI, node IC; SM, standard mode. Data points are divided into three classes according to their energy efficiency. Red squares indicate energy efficiency >1 nJ per bit (ref. ⁷²): (NI)⁷², (HI)^67,76 and (SM)¹¹⁵; blue circles indicate that energy efficiency lies between 100 pJ per bit and 1 nJ per bit (refs. ^83,66): (HCM)^{49,66,71,95,108,113} and (ETM)⁷³; yellow triangles indicate energy efficiency 100 < pJ per bit (refs. ^{21,35,65,68,74,90,92,96,97,98,109,110,111,112}). b, Receiver power versus normalized sensitivity to 100 kbps reference data rate. Red squares indicate energy efficiency >1 nJ per bit (refs. ^67,76): (SM)¹¹⁵; blue circles indicate that energy efficiency lies between 100 pJ per bit and 1 nJ per bit (refs. ^66,83): (HCM)^66,71, (ETM)^67,73 and (ETM); yellow triangles indicate energy efficiency 100 < pJ per bit (refs. ^{21,35,65,74,90,92,112}). References are to be read from left to right on the graphs.

Full size image

The data show that there is an interplay among power consumption, bandwidth and sensitivity, where sensitivity sets the minimum input signal power required to produce a specified signal-to-noise ratio at the output port of the receiver. Sensitive receivers can operate at lower transmission power, thus reducing the overall network transmitted power, at the cost of higher circuitry power consumption at the receiver. Normalized sensitivity is defined as⁶⁴:

in which D is the data rate and r is the reference data rate.

To have better insight in this trade-off, we plotted the receiver power versus the normalized sensitivity for a reference 100 kbps data rate (Fig. 4b). The choice of 100 kbps as the reference data rate is arbitrary, selected as a basic data rate considered sufficient for the transmission of vital signals. Higher sensitivity requires more gain stages with better noise figures at the front end of the receiver and therefore a larger power consumption. Most published receivers achieved a normalized sensitivity less than −80 dBm with an energy efficiency under 100 pJ per bit. A clear pattern emerges in which higher sensitivity (−80 dBm or lower) is typically associated with higher receiver power consumption. This trend is expected, as improved sensitivity requires additional gain stages and noise-reduction techniques, which lead to increased power usage. However, some high energy efficiency transceivers successfully achieve strong sensitivity (less than −60 dBm) while maintaining low receiver power, demonstrating that optimized circuit design can balance power efficiency with performance⁶⁵. Power-saving techniques such as duty cycling and resource sharing are commonly used in HBC transceiver architectures to optimize power consumption and maximize component use. Duty cycling involves periodically turning on and off specific components or functions to reduce overall power consumption. Resource sharing enables multiple components or functions to share common resources, thereby optimizing their use and minimizing the power consumption. Typical components that could be shared include oscillators, phase-locked loop (PLL) and so on^66,67,68. In some designs, duty cycling is implemented through wake-up mechanisms. For example, the power-expensive envelope detector and the injection locking oscillator can be shared in the main receiver in both amplitude shift keying and FSK paths⁶⁹. The wake-up RX operates in stand-alone mode when no data are received and is enabled only after receiving data. At the end, the operation is handed over to the main receiver.

By contrast, medium and low energy efficiency transceivers consume more power to achieve similar sensitivity levels, indicating a trade-off between energy efficiency and robustness. This aspect could be attributed to the use of more complex modulation schemes, older process nodes or less optimized architectures that prioritize reliability over power savings. A second trade-off, in which transceivers achieving better sensitivity (−90 dBm or lower) require increasingly higher power consumption, is also evident (Fig. 4b). The balance between power and sensitivity is crucial when selecting transceivers for different use cases, emphasizing the importance of optimizing circuit architectures to minimize energy loss while maintaining reliable communication.

Low-power building blocks

Various techniques can be used to achieve low hardware complexity which directly leads to lower power transceivers. For instance, receivers can use wake-up receivers which are ultra-low-power circuits designed to remain in a low energy listening state and activate the main receiver only upon detecting a specific wake-up signal, improving energy efficiency⁷⁰. Furthermore, single-ended structures and simplified processes can be used to realize low hardware complexity⁷¹. Another approach is to replace bulky and power-consuming crystal oscillators by using crystal-less designs. For example, injection-locked oscillators as frequency dividers are used to replace power-consuming PLLs⁷². This approach, coupled with the reuse of the received signal strength indicator for on–off keying demodulation, results in substantial reduction in power up to 70%.

Other approaches to reduce power by improving circuit design strategies include using, among others: frequency ___domain processing⁴⁹ via efficient fast Fourier transform/inverse fast Fourier transform structures which reduce computational overhead; pulse-shaping with an injection locking ring oscillator and digital-to-analogue converter for sinewave modulation⁷³, providing lower power alternative compared with traditional PLL and Wien bridge implementations; adiabatic communication to minimize dynamic power dissipation by recycling energy during switching events⁷⁴; and combinatorial pulse position modulation to optimize the use of time slots for pulse placement, reducing the power requirements for data transmission⁷⁴.

Architectural choices that contribute to reduced complexity include direct conversion architecture by eliminating intermediate frequency stages⁶⁷; successive interference cancellation for multi-access⁷⁵; and discrete-phase FSK architecture utilizing a single PLL instead of continuous-phase FSK with multiple PLLs and oscillators⁷⁶. These approaches reduce hardware complexity while maintaining functionality.

Energy harvesting

Energy scavenging harnesses ambient energy through various technologies including thermoelectric generators for body heat conversion⁷⁷, piezoelectric devices for mechanical motion⁷⁸, triboelectric nanogenerators⁷⁹, biochemical processes⁸⁰ and RF energy harvesting⁸¹. Advances in this field feature flexible materials and hybrid systems that combine multiple harvesting methods to improve efficiency and reliability⁸². Innovations in body-coupled ambient electromagnetic energy harvesting have shown potential in powering HBC systems without the need for high input voltage or complex setups, offering lower path loss compared with traditional RF transmission⁴⁷. Additionally, HBC systems can facilitate simultaneous power and data transfer using advanced technologies such as resonant EQS-HBC with maximum resonance power tracking and device capacitance cancellation⁸³. Other notable advancements include tri-mode buck converters for photovoltaic energy⁸⁴, single-inductor dual-output boost converters for thermoelectric harvesting⁸⁵, single-inductor piezoelectric harvesters⁸⁶ and dual-input buck converters for triboelectric energy harvesting⁸⁷.

Return path compensation and equalization

To enhance channel quality and to design reliable HBC transceivers, return path compensation and equalization techniques are essential. An auto-loss compensation system that uses 5-bit digitally controlled inductors was proposed to adjust the resonant frequency of an inductor–capacitor tank to match the carrier frequency^88,89. Tuning is regulated by a proportional-integral controller. Using a capacitive termination at the low noise amplifier of the receiver expands the bandwidth and prevents the compensation strength from deterioration caused by the resistive interface⁹⁰. Self-adaptive compensation can be achieved by estimating the capacitance between transceivers based on their ground electrode distance using a backward capacitance model coupled with digitally controlled tunable inductors to adjust for varying capacitances⁹¹.

In terms of equalization, a decision feedback equalization scheme with eight taps can be used to address low-frequency path loss, thereby improving bandwidth and reducing inter-symbol interference⁹². Alternatively, a shunt capacitor can be used to tune a filter to counteract centre frequency drift caused by environmental changes, enhancing channel selection accuracy⁹³.

Environmental variation and human motion compensation

To address the challenges posed by varying human motion, postures and environmental changes, compensation techniques typically focus on detecting, compensating and calibrating based on path variations caused by variable contact impedance in eHBC. To monitor the contact impedance and compensate for the resulting loss, an RC relaxed contact impedance monitor can be used⁴⁹. This monitor uses different time constants obtained from various suspension distances to detect impedance changes that are used to adjust the low noise amplifier (LNA) gain. Contact impedance sensors can be introduced to mitigate the influence of contact impedance on distance measurements between transceivers, in which the sensor continuously switches between the HBC mode and impedance sensing mode⁷³. It operates on chopper modulation principles, with impedance measurements based on the level of the chopper-demodulated and filtered signal. The sensor reliably covers contact impedance up to 1 kΩ. To ensure robustness against environmental changes, a wideband signalling receiver was proposed⁹⁴, with a self-tuning circuit for each threshold voltage tuning branch, enabling the hysteresis threshold voltage of the Schmitt trigger to remain stable under changing environmental conditions.

System reliability

To ensure consistent and dependable communication, clock reliability and voltage offsets are critical issues to address. For clock reliability, the frequency of oscillators in HBC transceivers needs real-time monitoring and calibration. Injection locking oscillators commonly utilize oscillator delay cells to automatically calibrate the operation frequency to the free-running frequency⁷³. Self-calibrated voltage-controlled oscillators are used to maintain the oscillation frequency within the normal range⁹⁵. Voltage offset can occur in single-ended circuits or at the skin–electrode contact interface. The single-ended receiver chain might experience saturation caused by DC offset⁷¹. To mitigate this problem, a self-calibration technique is used to reduce the deviation of the DC point at the receiver side. This involves selecting the output of the low-pass filter first stage as an inner node, detecting it and providing feedback to the front-end amplifier for automatic adjustment of its DC point. Another approach involves down-converting the received signal and up-converting the polarization voltage induced by the electrode–skin interface and DC offset of the front-end circuit⁷⁶. The up-converted polarization voltage and DC offset are then filtered by a subsequent limiting amplifier, which also increases the amplitude of the demodulated baseband signals. Moreover, using effective coding schemes can enhance system reliability by achieving low jitter, improved synchronization, low BER and scalability. For example, Manchester coding incorporates clock information within the coded data and provides better noise immunity⁹⁶. This coding scheme improves time synchronization between the transmitter and the receiver, thereby enhancing data transmission reliability. The reconfigurable Hamming coding scheme can result in up to 7× increase in data rate compared with traditional HBC coding, allowing the transceiver to operate within a wide channel range from 40 MHz to 100 MHz (ref. ⁹⁷). To prevent DC offset build-up, a bipolar return to zero coding scheme is adopted⁹⁸. Furthermore, AES encryption coding is applied to enhance security performance in HBC systems⁶⁵.

Noise and interference suppression

Different frequency bands offer varying levels of interference resistance and noise immunity. For instance, the near-field EQS band is suitable for short-distance and low-frequency communication, exhibiting less sensitivity to human gestures and environmental variations, resulting in higher bandwidth and reduced interference^38,42. Surface wave propagation within specific frequency ranges, such as 350–550 MHz (ref. ⁴²) or 402–614 MHz (ref. ⁹³), can also been used, offering decreased interference compared with other transmission methods. Adaptive frequency hopping techniques can be used over multiple channels to improve interference resistance and compensate for varying backward paths^49,65,99. By carefully selecting frequency bands and using frequency hopping mechanisms, interference from sources such as FM radio can be avoided^66,73.

Various filtering techniques have been used to enhance signal quality in HBC systems. For example, matched filter-based reshaping processes can be used for noisy channel mitigation, accompanied by phase error detection and correction schemes¹⁰⁰. Using steep roll-off filters helps suppress interference at specific frequency bands, particularly in cases in which it may disrupt low-frequency vital signals⁷¹. Moreover, adaptive duty cycle integration techniques act as notch filters used to separate the desired signal from interfering noise in the time ___domain²¹. Decision feedback equalization with multiple taps compensates for low-frequency path loss and eliminates inter-symbol interference⁹². In the signal processing ___domain, mixing techniques are utilized to up-convert polarization voltage and DC offset, which are then filtered out by limiting amplifiers during down-conversion of the received signal to the baseband⁷⁶.

Several standard-compatible transceiver architectures have been proposed to meet the standard mask requirements defined in the IEEE 802.15.6 standard. An active digital-bandpass filter was proposed, consisting of an eighth-order bandpass filter implemented with a Butterworth infinite impulse response filter, an 8-to-256 decoder and a thermometric digital-to-analogue converter¹⁰¹. A second-order intermodulation cancellation technology was proposed for the body surface driver in the low-frequency band¹⁰². This technique effectively meets the stringent requirement for the transmit spectral mask, achieving a mask of −122 dBr at 1 MHz. Another approach uses an analogue active filter instead of a digital-to-analogue converter¹⁰³ to achieve wider bandwidth at lower power. Another work proposes using an infinite impulse response and a bandpass filter with a fifth-order high pass and a third-order low pass to improve mask rejection at frequencies lower than 2 MHz for the BPSK stream⁵⁶. In addition, using a sigma-delta modulator to replace the digital-to-analogue converter reduces the precision to 8 bits instead of 14 bits.

Since the introduction of HBC technology in the mid-1990s¹⁰⁴, HBC has matured substantially, offering key advantages in secure and energy-efficient communication. However, its commercial adoption has lagged because of several challenges. One major issue is the complexity of modelling the human body as a communication channel. Early research, particularly in wearables, overlooked crucial factors such as grounding and parasitic effects, resulting in large channel estimation errors, as highlighted in works as late as 2017 (ref. ⁴³). These challenges delayed the development of reliable systems and hindered progress towards widespread commercialization.

Another issue is the lack of industry-wide standardization. Although IEEE 802.15.6 offers some guidance, more comprehensive protocols are needed to ensure interoperability with medical devices and networks. Moreover, although the standard provided a foundation, it has not fully unified the community, and the market remains fragmented with proprietary technologies creating barriers to widespread adoption. By contrast, technologies such as Bluetooth and BLE have succeeded owing to unified frameworks that enable seamless interoperability. Achieving industry-wide standardization will be key to driving HBC’s commercial success.

Finally, although security is a key advantage of HBC owing to its closed-system nature, there are scenarios in which it remains vulnerable. For example, under specific conditions such as inter-body attacks (for example, physical contact or close proximity), eavesdroppers can intercept HBC signals¹⁹. Moreover, off-body data transmission to wireless nodes or the cloud increases exposure to potential threats, highlighting the need for encryption and secure off-body communication protocols. Additionally, CC-HBC systems are susceptible to interference from surrounding electronics, making robust interference mitigation essential for reliable operation. To ensure seamless integration into existing infrastructures, further standardization is needed to address these vulnerabilities effectively.

Beyond modelling and standardization challenges, the key strength of HBC lies in its exceptional energy efficiency, which can be further enhanced through innovative power solutions such as energy harvesting, enabling self-sustained and minimally powered devices for continuous and reliable operation. Coupled with advancements in deep-learning techniques that optimize signal processing, security and adaptive communication, HBC is poised for a transformative future with expanded applications in health care, human–computer interaction and beyond.

Energy harvesting for IoB devices powering

Three common methods can be used for powering IoB nodes: batteries, in which a rechargeable battery is integrated into the system to be charged periodically through an electric outlet similar to smartwatches and cell phones¹⁰⁵; power harvesting, in which a power harvester module is integrated into the HBC sensors to harvest power from the environment and motion, used independently or in a hybrid arrangement with a small battery¹⁰⁶; and body-coupled power transmission which is achieved using the human body channel where a power signal is transmitted through the body⁴⁷.

Although battery-powered systems are common, they are bulky and costly. Transmitting power through the body raises potential health concerns. Power harvesting is an interesting alternative that seems to be increasing within grasp, where with the advent of IoB-based devices, a whole new range of possibilities can now be investigated owing to the reduced area and power requirements. At picojoule per bit efficiency, energy harvesting becomes possible, resulting in self-powered or minimally powered nodes. Companies such as Powercast and others have already commercialized RF energy-harvesting solutions, and advancements in flexible rectenna design further push the boundaries of EH technology. For example, a rectenna¹⁰⁷ designed at 2.4 GHz band achieved 78.2% efficiency at −5 dBm input power and uses a biocompatible polyimide substrate with a parasitic element to reduce SAR, making it a good candidate for wearables. Other studies and surveys¹⁰⁶ have illustrated flexible rectennas harvesting energy from multiple bands, reaching efficiencies up to 76% at −10 dBm. These advancements, combined with commercial efforts, are paving the way for more autonomous and reliable IoB devices, enhancing continuous health monitoring and other IoB applications.

Deep-learning solutions

Deep learning is transforming HBC through enhanced channel modelling, signal processing, data transmission optimization and security improvements. Convolutional neural networks have a crucial role by filtering noise and improving signal clarity, which is vital for reliable data transmission across human body channels. Deep-learning algorithms also adapt to fluctuating channel conditions to optimize data rates, thus enhancing energy efficiency and throughput. A notable innovation is the integration of end-to-end autoencoders in HBC¹⁰⁸. Autoencoders compress high-dimensional data into compact formats, easing data transmission over HBC’s limited bandwidth, which is especially beneficial for transmitting complex physiological data effectively. Autoencoders also excel in noise reduction and feature extraction from raw HBC signals, facilitating accurate activity recognition and health state diagnostics by focusing on essential data characteristics and eliminating noise. This capability is critical for precise monitoring and diagnostic applications, further boosting the signal quality and reliability of HBC systems.