^{1}

^{1}

^{2}

^{†}

^{*}

^{1}

^{3}

^{2}

^{4}

^{5}

^{1}

^{2}

^{2}

^{4}

Current address: Department of Physiology and Biophysics, University of Colorado, Anschutz Medical Campus, Aurora, CO 80045, USA.

This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/).

Intra-Body Communication (IBC), which modulates ionic currents over the human body as the communication medium, offers a low power and reliable signal transmission method for information exchange across the body. This paper first briefly reviews the quasi-static electromagnetic (EM) field modeling for a galvanic-type IBC human limb operating below 1 MHz and obtains the corresponding transfer function with correction factor using minimum mean square error (MMSE) technique. Then, the IBC channel characteristics are studied through the comparison between theoretical calculations via this transfer function and experimental measurements in both frequency domain and time domain. High pass characteristics are obtained in the channel gain analysis

Recently, Body Area Networks (BANs) have been developed to facilitate low power devices operating on, in or around the human body to serve a variety of applications including medical and consumer electronics [

Galvanic-type IBC is a relatively new communication technology using the human body as a channel for data communication. The transmitter converts the transmitted signal into a flow of ionic current within the tissue and the receiver, which is located somewhere on the body, recovers the original signal information by detecting the ionic current flow. The low ionic current conveys the signal successfully and no obvious local body heating [

The issue of utilizing an analytical model to study channel characteristics plays an important role in both understanding signal transmission mechanism and hence later for IBC system design. With the purpose of obtaining channel characteristic, several IBC channel models have been proposed in the recent literature for the two general coupling-types of IBC, namely galvanic-type and capacitive-type. Generally, galvanic-type IBC [

Our work here concentrates on the galvanic-type IBC because of its secure and lesser interference features. Specifically, we based it on the analytical quasi-static EM model from our group in [

The arrangement of this article is as follows: Section 2 briefly reviews the quasi-static field model of a galvanic-type IBC in order to obtain the channel system transfer function up to 1 MHz and also improve the transfer function model semi-empirically by the MMSE technique; Section 3 investigates the channel characteristics in the time domain via pulse response in both baseband and passband transmissions; Section 4 evaluates different possible modulation schemes over IBC. Finally, the conclusions of this paper can be found in Section 5.

Based on our previous work [_{4}, _{3}, _{2}, _{1}), respectively. The electrical properties of each tissue layer are assumed to be homogeneous and characterized by permittivities (_{4}, _{3}, _{2}, _{1}) and conductivities (_{4}, _{3}, _{2}, _{1}). From [_{cs}_{s}_{rs}_{o}_{4}_{n}_{n}_{4kn}(_{4kn}(_{4kn}(_{4kn}(

Using _{4}(_{4}, _{rp}, z_{rp}_{4}(_{4}, _{rn}, z_{rn}_{4}, _{rp}, z_{rp}_{4}, _{rn}, z_{rn}

Similarly, the applied differential signal at the transmitter side of galvanic-type IBC channel can be expressed as:
_{4}(_{4}, _{tp}, z_{tp}_{4}(_{4}, _{tn}, z_{tn}

Using

_{e}_{e}

The calculated results were then compared with

Both calculated and measured results of IBC channel gain are shown in

Even though both the calculated and measured results of the IBC channel exhibit similar high pass characteristics, the calculated values from

In order to take into account of non-avoidable variations, a correction factor _{10}(

Then _{ΠdB}(_{dB}

With the introduction of the constant correction factor

Due to the frequency dependence of the transfer function (7), a transmitted signal would suffer dispersion and amplitude reduction. In digital communication, this phenomenon will cause pulse spreading and overlapping in the transmitted pulses. Hence, this will hinder the data communication rate. For this reason, both baseband and passband transmissions have been investigated by transmitting square waves into the IBC channel to assess the channel dispersion and distortion behavior parameters, which could assist in IBC design. In this Section, we study the IBC time domain channel characteristics with the corrected analytical model discussed in Section 2.2. For simplicity, only Subject B calculation and measurement results are shown throughout this Section.

For baseband experiments, the general measurement set-up block diagram displayed in

The corresponding output results for input square waves in baseband transmission in calculations and experiments are displayed in _{4}) is defined as:

The reason why we take only up to the 4th harmonics here is to have a fair comparison with passband transmission which will be discussed later in this Section. Like the 500 Hz square wave transmitted in the baseband channel, low frequency harmonics suffer higher attenuation, which will cause _{4} higher than that of ideal square wave around 41.4%. The calculation square wave _{4} result is around 96% while the corresponding measurement _{4} results are 99.7% and 99.2% for S = 6 cm and 11 cm, respectively. The _{4} deviation from the ideal case is around (55–58)%, which indicates signal transmitted via baseband technique in IBC is not suitable (pulse-shape → spike-shape).

For passband experiments, two input square waves (one at 500 Hz, the other at 50 kHz) are modulated to the same carrier of 500 kHz. The 500 kHz carrier is chosen because it is located at around the center of low attenuation frequency band as shown in

As seen in _{4} around 41.9% in calculation and around 42.3% in measurement. Comparing with those values (upper ninety %) in baseband IBC transmissions, passband IBC transmission obtains lower distortion and has a much better communication performance. In _{4} in measurement is around 42.4% and 41.7% at S = 6 cm and 11 cm, respectively. All passband _{4} are close to ideal square wave case, indicating rather low distortion for passband transmissions. In addition, all calculation and measurement rise times (defined as the time spent from 10% to 90% during transition) of output pulses (after demodulation) are around 0.6 μs at different distances, as shown in the insets of

From the relationship between rise-time and bandwidth, the corresponding system bandwidth would be inversely proportional to 0.6 μs. This indicates that system bandwidth is in the order of MHz, providing enough bandwidth for possible low distortion in passband transmission. Note that the maximum bandwidth of the transmitted signal is limited by the carrier. In summary, passband transmission with proper choice of carrier frequency would have rather good communication performance in terms of both attenuation and distortion.

The application of efficient digital modulation can improve power efficiency and bandwidth efficiency at a reasonable cost. From the results of the time domain and frequency domain analyses above, it is concluded that signal in passband transmission enjoys the benefit of lower attenuation and lower distortion than the baseband transmission counterpart. Therefore, we now evaluate and compare the performance of different modulation schemes with a carrier frequency of 500 kHz using the proposed IBC channel model. Moreover, the use of a carrier frequency at 500 kHz could be a balance between data rate and electromagnetic radiation.

Among the commonly available digital modulation techniques, FSK, PSK and QAM, high-ary modulation would not be considered due to either its high complexity in hardware, such as 16QAM, 16PSK, or due to its low bandwidth efficiency, such as 8FSK and 16FSK. PSK is generally superior than FSK in terms of bandwidth and power efficiency in the low-ary (≤8-ary) [

The data symbols are first digitally modulated using BPSK, QPSK or 8PSK and then are pulse-shaped by the square root raised cosine filter with roll-off factor α = 0.5 before sending to the IBC channel. The noise within the human body mainly comes from various bioelectric signals, the environment and contact impedance between the electrodes and the human body. As the frequency ranges of bioelectric signals such as ECG, EMG and EEG signal are all lower than 10 kHz [_{0}_{0} is found to be around −123 dBm/Hz. The transmit power for the transmitter was set to −25 dBm for a compromise between feasible calculation time and bit-error-rate (BER). The communication performance (BER) in simulation and theoretical calculation is compared. In the simulation, the channel characterized by transfer function (7) and AWGN noise with PSD −123 ^{7} bits) of random bit stream with different bit rate and different modulation scheme is applied into this channel, and output bit-stream (with possible error bits) is obtained at the receiver after demodulation. The BER can be calculated based on the comparison between input and received data bits. In the theoretical calculation, the equivalent signal to noise ratio (_{b}_{0}) can be calculated before demodulation at the receiver using the transmit power, the transfer function attenuation, _{0} and data symbol rates. The theoretical BER can be calculated by the following equation:
_{b}_{0} are presented in ^{−3} when _{b}_{0} is higher than 6.5 dB with the maximum bit rate of 70 kbps for both BPSK and QPSK modulation. For 8PSK, BER is lower than 0.01 when _{b}_{0} is higher than 7 dB with the bit rate lower than 30 kbps. Among the three modulation methods, 8PSK obtains the highest bandwidth efficiency; while QPSK is twice as high as that of BPSK. However, the 8PSK suffers high BER at the same bit rate. In other words, 8PSK trades bandwidth efficiency off BER while QPSK obtains lowest BER with higher bandwidth efficiency. Thus, considering the bandwidth efficiency and BER, QPSK modulation scheme obtains the highest bit rate with low BER and is the optimal modulation scheme among BPSK, QPSK and 8PSK.

In this paper, we have investigated the channel characteristics of a galvanic-type IBC based on the transfer function derived from our analytical quasi-static EM model. Both time and frequency domain experimental results evidence good agreements with the model calculations, which demonstrate the benefits of using an analytical transfer function in analyzing IBC channel characteristics. The high pass channel characteristic suggests that a galvanic-type IBC can be suitably operated at least in hundreds of kHz of carrier for lower attenuation in sub-MHz frequency region and maintain reasonably good data rates. Moreover, the harmonic distortion in the IBC channel time domain analyses is around 96% for baseband transmission with the 500 Hz input square wave; while it reduces to around 41.9% for passband transmission, close to the ideal case of considering equipment ability of four harmonics. The results suggest that passband transmission can achieve low distortion even without channel equalization. Considering power efficiency, bandwidth efficiency and hardware complexity through theoretical calculation and simulation, QPSK is the optimum modulation method among BPSK, QPSK and 8PSK modulation schemes for reasonable BER at the low transmitted power of −25 dBm in passband transmission. The calculation, experimental and simulation results in this work have provided useful information for transmission method and modulation scheme selection in the IBC system design. In the future, we will continue to research on the optimum IBC method in terms of power, bandwidth and cost.

The work was supported by The Science and Technology Development Fund of Macau (FDCT) under Grant 014/2007/A1, Grant 063/2009/A, and Grant 024/2009/A1; the Research Committee of the University of Macau, under Grant RG072/09-10S/MPU/FST; the National Natural Science Foundation of China under Grant 61201397; the Natural Science Foundation of Fujian Province under Grant 2011J05077; the Program of International S&T Cooperation under Grant S2013GR0188. The authors would like to express heartfelt thanks to the financial support from the Science and Technology Development Fund of Macau, the Research Committee of the University of Macau, the Funds of Fujian Provincial Department of Science and Technology, the Natural Science Foundation of Fujian Province and the National Natural Science Foundation of China.

Illustration of galvanic-type IBC geometry on the human limb.

Calculated and measured transfer function characteristic of (

Measured and calculated (with correction) transfer function characteristic of (

IBC channel baseband and passband transmission experiment set-ups: (

Input and Output of baseband IBC transmission for 500 Hz square wave: (

Input and Output (after demodulation) in passband IBC transmission for square wave: (

Performance of BPSK, QPSK and 8PSK

Performance of BPSK, QPSK and 8PSK