35th Annual International Conference of the IEEE EMBS

Osaka, Japan, 3 - 7 July, 2013

A Wideband Scalar Network Analyzer for

Biomedical Dehydration Measurements
M. Hofmann, J. Nehring, R. Weigel, G. Fischer, and D. Kissinger

Institute for Electronics Engineering, University of Erlangen-Nuremberg,

Cauerstr. 9, 91058 Erlangen, Germany
Email: {hofmann, nehring, weigel, fischer, kissinger}@lte.eei.uni-erlangen.de

Abstract—This paper presents a wideband microwave ap- through the MUT. It is proposed to use this effect to determine
proach towards biomedical dehydration monitoring. The intro- the water content, which can be used for dehydration detection.
duced concept is verified via invasive measurements on several
blood samples. A microwave measurement circuit, based of a
two-port scalar vector network analyzer is presented. The circuit
III. I NVASIVE V ERIFICATION AND M ODELING
operates between 5 GHz and 20 GHz using a planar permittivity The proposed influence of water content in blood samples
sensor. Measurements of all subcomponents are shown together
with measurements of a Water-NaCl-Glycerol solution.
has been proven by invasive experiments. The measurement
setup for this experiment is illustrated in Fig. 1. The permittiv-
I. I NTRODUCTION
Dehydration is a leading cause of hospitalization among
elderly people and plays an important role in an aging society. Temperature control
By dehydration a loss of autonomy and cognitive functions
is indispensable for humans [1]. Thus it is of major interest
to accurately monitor the hydration status. Furthermore this
application will not be limited to elderly people only. Another Three Way Stop Cock
important application could help athletics that might suffer
from dehydration in sport. In this paper a microwave based
Forearm
concept for precise measurement of water content in aqueous
solutions is proposed.
II. C ONCEPT
The dielectric relaxation process describes the properties of
a material under test (MUT) which is excited with an external
electrical field. The dielectric relaxation can thus be used to
characterize the condition of a certain material, solution or Indwelling Venous Cannula
emulsion. A Debye relaxation process can be described as
Test-tube to VNA coaxial probe
denoted in Eq. (1) for aqueous solutions [2]. Where ǫ∞ is
the permittivity at high frequencies, ǫstat is the static, low
frequency permittivity, and τ is the characteristic relaxation
time of the MUT. Fig. 1. Invasive measurements on blood samples of 2 ml with temperature
controlled equipment with insert indwelling cannula and three-way stop cock
ǫstat − ǫ∞ (approval from the Ethical Review Committee (ERC) has been obtained).
ǫr (ω) = ǫ∞ + (1)
1 + jωτ
The permittivity can be measured using different techniques ity of different blood samples has been measured over different
such as coaxial or transmission based sensors. Both methods water concentrations. The water content has been changed by
make use of fringing electric fields that penetrate a MUT. By a manual dispensing. Fig. 2 shows the influence of an increasing
change in water content of a MUT, a variation of the fringing water content on the permittivity of blood. Since invasive
fields can be measured in magnitude and phase. In this paper, measurements with human blood samples are highly extensive,
a microwave based approach is proposed to determine the a substitute solution for further measurements was evaluated. It
water content within a MUT by measuring the attenuation over is shown in Fig. 3 that a NaCl-Water solutions with a variable
concentration. The amount of water in an emulsion or aqueous Glycerol concentration has comparable dielectric properties
solution, such as blood, will increases the transmission losses and could replace blood samples in further experiments.

978-1-4577-0216-7/13/$26.00 ©2013 IEEE 4050

Authorized licensed use limited to: M S RAMAIAH INSTITUTE OF TECHNOLOGY. Downloaded on June 15,2022 at 08:13:22 UTC from IEEE Xplore. Restrictions apply.
 approximately -15 dB is coupled out and used as a reference
80
signal. The transmission and reflection at the MUT is fed to
Water
a power detector directly and through a directional coupler,
70 Blood respectively. The power detector consists of a passive multi
+20 drops section matching network and a zero biased detector diode.
60
It is assumed that the transmission coefficient |S ∗21 | of the
Re(Permittivity)

+30 drops

+50 drops
50
+70 drops
Analog to Digital to Matlab
40
VReference Vreflection Vtransmission
30 RF to DC RF to DC

20 Matching
Matching Matching
Network Network Network
10
10 20 30 40 Reference Reflection Transmission
Frequency (GHz) a1 (fIN) b1 (fIN,χ) b2 (fIN,χ)
Reflectometer

Fig. 2. Real part of the measured permittivity of the 2 ml blood sample, Z0 Z0

obtained with a conventional vector network analyzer (Agilent PNA-X with
coaxial probe) and a drop volume of approximately 20 µl.

Wideband Permittivity sensor with χ being

80 ~ RF-Source
PIN< 3 dBm
the Glycerol concentration
in a H2O-NaCl solution
82% water con. 5 GHz < fIN < 20 GHz
70 84% water con.
90% water con. Fig. 4. Circuit concept for biomedical concentration measurements based on
60 94% water con. a microwave two port test set.
Re(Permittivity)

96% water con.

50
100% water con. sensor will change with the water concentration χ in a test
40 solution. An increasing water content will be followed by
higher transmission losses which can be obtained according
30 to Eq. (2).
20 |b2 (f, χ)| VT ransmission
|S ∗21 | (f, χ) = = (2)
|a1 (f )| VRef erence
10
10 20 30 40 The transmission coefficients are subsequently determined in
Frequency (GHz) Matlab after a calibration procedure. The calibration function
given in Eq. (3) is based on measuring four well known
Fig. 3. Real part of the measured permittivity of the NaCl-water-glycerol
mixture, obtained with a conventional vector network analyzer (Agilent PNA-X standards open, short, match, thru (OSMT).
with coaxial probe) over several concentrations.
|S 21 | (f, χ) = Cal(|S ∗21 | (f, χ)) (3)
For a suitable determination of the measured effect, all trans-
IV. C IRCUIT D ESIGN mission coefficients are normalized to the lowest water con-
centration as denoted in Eq. (4) and summed over a frequency
Concentration measurements using transmission based sen- range between fmin = 5 GHz and fmax = 20 GHz.
sors for biomedical application were introduced in [3] and [4].
Those methods are based on the wideband determination of |S 21 | (f, χ)
∆ |S 21 | (f, χ) = (4)
complex scattering parameters of the sensor or transmission |S 21 | (f, χ = 80%)
line, respectively. This work shows that a compact scalar The result of this integration is frequency independent and
network analyzer could be used for dehydration detection in shown in Eq. (5).
biomedical applications or evaluation of water content in liq-
uids. The circuit concept is illustrated in Fig. 4, consisting of a X
f =f max

STC(χ) = ∆ |S 21 | (f, χ) (5)

permittivity sensor [4], two directional couplers [5] (fabricated
in a reflectometer configuration) and three wideband power
detectors, feeding the analog to digital converter (ADC). A
wideband RF source feeds the material under test (MUT)
with an incident wave between 5 GHz and 20 GHz with an
output power less than 3 dBm. By the first directional coupler,
f =fmin
V. S YSTEM I MPLEMENTATION AND M EASUREMENTS
All subcomponents of the introduced measurement system
were fabricated and measured separately. One of the two
directional couplers is shown in Fig. 5 with a cross section

4051

Authorized licensed use limited to: M S RAMAIAH INSTITUTE OF TECHNOLOGY. Downloaded on June 15,2022 at 08:13:22 UTC from IEEE Xplore. Restrictions apply.
 illustration (right). The reflectometer is based on two broadside
coupled wideband directional coupler [5] in a four layer stack 0
up. The measured properties of a single coupler are illustrated
−5

-10
-15

S-Parameter (dB)
-20
RO4350B 168µm
-25

RO4450F 65µm 100µm -30

-35
168µm
RO4350B
-40 Coupl (meas)
Zoom: Top-Layer Feed Cross-Section 4 Layer Coupl (sim)
−45 Isol. (meas)
Isol. (sim.)
-50
0 5 10 15 20 25 30 35
Fig. 5. Fabricated directional coupler on four layer Rogers RO4350,4450 Frequency (GHz)
substrate.
Fig. 7. Simulated and measured scattering parameters (coupling and
in Figs. 6 and 7. One can obtain a return loss of approximately isolation) of a single directional coupler.
20 dB and an insertion loss of approximately 5 dB over the
entire frequency range. Further, a coupling and isolation of
approximately 15 dB has been simulated and measured. For
Matching Network

0
DC-Output
−5 RF-Signal Input Detector Circuit

-10
-15 Fig. 8. Fabricated wideband power detector for applications up to 35 GHz.
S-Parameter (dB)

-20
-25
-30 VI. E RROR D ISCUSSION
-35
-40 Thru (meas)
As one can observe by the presented results, a high precision
−45
Thru (sim) has been achieved, however minor errors can be read out.
Match. (meas)
Match. (sim.) Those error could be based on time-invariant transmission
-50
0 5 10 15 20 25 30 35 properties of the permittivity sensor. Since the sensor suffers
Frequency (GHz)
from minor water absorption.
Fig. 6. Simulated and measured scattering parameters (matching and
transmission) of a single directional coupler.

the measurement and reference channels, three power detectors 0

were designed, fabricated and measured in the entire frequency
range. A fabricated power detector is shown in Fig. 8. Two -10
Input matching (dB)

key parameter of the power detector are return loss and output
voltage versus RF-Input power. Both are shown in Figs. 9 -20
and 10 for simulation and measurement.
Fig. 11 shows the effect on Eq. 4 were one can obtain -30 Meas. @ PIN = −5 dBm

a clear relation in Fig.12 by the integration of Fig. 11 over Meas. @ P = −20 dBm
IN

the frequency as denoted in Eq. 5. The illustrated parameters -40 Meas. @ PIN = −30 dBm

are plotted in dB. As expected by theoretical investigations, Sim. @ PIN < −30 dBm

the transmission coefficient decreases with an increasing water -50

concentration due to the dielectric losses. This effect can be 0 10 20 30
Frequency (GHz)
clarified by considering the sum of all determined transmission
coefficients, which can be interpreted as the area above the Fig. 9. Simulated and measured wideband RF power detector matching from
curves of Fig. 11 and the frequency axis. 5 GHz to 35 GHz.

4052

Authorized licensed use limited to: M S RAMAIAH INSTITUTE OF TECHNOLOGY. Downloaded on June 15,2022 at 08:13:22 UTC from IEEE Xplore. Restrictions apply.
 VII. C ONCLUSION
0 In this paper a wideband microwave approach towards
10
biomedical dehydration monitoring has been presented. The
concept has been verified via invasive measurements on several
Measurement
blood samples. Further, a microwave measurement circuit,
Output voltage (V)

−2
10 based of a two-port scalar vector network analyzer was pre-
sented. Measurements of all subcomponents of the device
were shown together with measurements of a Water-NaCl-
−4 Glycerol solution. The overall measurement concept has been
10
successfully shown.
Simulation ACKNOWLEDGMENT
−6 This work has been supported by the German Federal
10
-60 -50 -40 -30 -20 -10 0 10 Ministry of Education and Research (BMBF) through the
Input power (dBm) research project SmartSensors within the excellence cluster
Medical Valley EMN e.V. under contract number 01EX1014T.
Fig. 10. Simulated and measured wideband RF power detector characteristic
from 5 GHz to 35 GHz. R EFERENCES
[1] M. Ferry, “Strategies for ensuring good hydration in the elderly.” Nutrition
Reviews, no. 63, pp. 22–9, 2005.
[2] M. Hofmann, F. Trenz, R. Weigel, G. Fischer, and D. Kissinger, “A
0 microwave sensing system for aqueous concentration measurements based
on a microwave reflectometer,” in IEEE MTT-S Int. Microw. Symp. Dig.,
Montreal, Canada, June 2012, pp. 1–3.
[3] K. Grenier, D. Dubuc, P.-E. Poleni, M. Kumemura, H. Toshiyoshi,
−2 82% water con. T. Fujii, and H. Fujita, “Integrated broadband microwave and microfluidic
∆ |S 21 | (f, χ) (dB)

84% water con. sensor dedicated to bioengineering,” Microwave Theory and Techniques,
86% water con. IEEE Transactions on, vol. 57, no. 12, pp. 3246 –3253, Dec. 2009.
−4 88% water con.
[4] M. Hofmann, M. Bloss, R. Weigel, G. Fischer, and D. Kissinger, “Non-
invasive glucose monitoring using open electromagnetic waveguides,” in
90% water con.
Proc. Eur. Microw. Conf., Amsterdam, The Netherlands, Oct. 2012, pp.
92% water con. 546–549.
−6 94% water con. [5] J. Nehring, M. Hofmann, C. Muenker, R. Weigel, G. Fischer, and
96% water con. D. Kissinger, “A universal method for the design and synthesis of
98% water con. wideband directional couplers using non-uniform coupled transmission
lines,” in Proc. Eur. Microw. Conf., Amsterdam, The Netherlands, Oct.
8.0 2012, pp. 475 – 478.
5 10 15 20
Frequency (GHz)

Fig. 11. Transmission parameter S21 over different water glycerol concen-
trations, normalized to the lowest glycerol concentration, obtained with the
introduced scalar network analyzer.

5
Sum of normalized S21

0
STC(χ) (dB)

−5

-10
80 85 90 95 100
χ Concentration (%)

Fig. 12. Sum of the normalized transmission coefficients (STC) over the
frequency range.

4053

Authorized licensed use limited to: M S RAMAIAH INSTITUTE OF TECHNOLOGY. Downloaded on June 15,2022 at 08:13:22 UTC from IEEE Xplore. Restrictions apply.