A Novel Pressure Sensing Circuit for Non-invasive

RF/Microwave Blood Glucose Sensors

Volkan Turgul and Izzet Kale

Applied DSP and VLSI Research Group (ADVRG), Department of Engineering
Faculty of Science and Technology, University of Westminster
London, W1W 6UW, United Kingdom
E-mail: [email protected], [email protected]

Abstract—A novel pressure sensing circuit for non-invasive variation in the dielectric properties of the media will induce a
RF/microwave blood glucose sensors is presented in this paper. change in the sensor’s frequency response and the BGL can be
RF sensors are of interest to researchers for measuring blood estimated by observing this. This response is a function of
glucose levels non-invasively. For the measurements, the finger is multiple factors, such as; the electrical properties of biological
a popular site that has a good amount of blood supply. When a tissue layers, constituents found in the blood and the position of
finger is placed on top of the RF sensor, the electromagnetic the sample (finger) on the sensor [4]. Another important factor
fields radiating from the sensor interact with the blood in the is the pressure exerted onto the sensor by the finger. These
finger and the resulting sensor response depends on the factors are explained in the following sections.
permittivity of the blood. The varying glucose level in the blood
results in a permittivity change causing a shift in the sensor’s II. BACKGROUND THEORY
response. Therefore, by observing the sensor’s frequency
response it may be possible to predict the blood glucose level. When a sample is placed on top of a transmission line, the
However, there are two crucial points in taking and subsequently fringing fields propagate through the sample, which acts as a
predicting the blood glucose level. These points are; the position superstrate as shown in Fig.1. (ε denotes the dielectric
of the finger on the sensor and the pressure applied onto the constant).
sensor. A variation in the glucose level causes a very small
frequency shift. However, finger positioning and applying
inconsistent pressure have more pronounced effect on the sensor
response. For this reason, it may not be possible to take a correct
reading if these effects are not considered carefully. Two novel
pressure sensing circuits are proposed and presented in this
paper to accurately monitor the pressure applied.

Keywords—non-invasive; glucose; sensing; RF; microwave;

pressure; sensor; permittivity

I. INTRODUCTION
Fig.1. Interaction of fringing fields with the sample
Diabetes Mellitus is affecting over 400 million people
worldwide, a well-known disease associated with high Blood In the case of using a fingertip as the sample, the fringing
Glucose Levels (BGL) [1]. This arises from the insufficient fields exist in several biological tissue layers. These layers are;
production of the insulin hormone or its ineffectiveness in the
skin, fat, blood and bone. The penetration depth depends on
patients’ body. In this case, the BGL can be controlled through
medication or by administering insulin externally [2]. the frequency of operation of the sensor and in a good sensor
However, the BGL must be known prior to taking action. design, penetration beyond the blood layer should be as
Conventional methods require a drop of blood to be taken from limited as possible. In any case, to measure the BGL the fields
the patient, usually from a pricked fingertip, that is applied need to penetrate the skin and the fat layers. The blood is
onto a test strip, which is then inserted into the measurement found in a complex network of capillaries within the
device for a BGL reading. Some diabetes patients may need up subcutaneous tissues beneath the fat layer. The BGL
to 10 measurements a day; hence, the process gets information is contained in the response resulting from this
uncomfortable. For this reason, researchers have been layer. Any response from other layers is not of interest, which
investigating various methods where the use of Radio is why the penetration depth should be limited to the blood
Frequency (RF)/microwave sensors is one of them for layer. The 4-layer fingertip model was realized in CST
measuring the BGL non-invasively [3]. The measurement Microwave Studio (MWS) electromagnetic simulation
principle lies in the detection of the permittivity of the software to validate the theory. The model is shown in Fig.2
measured media that is placed on top of the sensor. Any with the permittivity and thickness values of the layers.

978-1-5090-2586-2/16/$31.00 ©2016 IEEE

Authorized licensed use limited to: Indian Institute of Technology (ISM) Dhanbad. Downloaded on May 02,2025 at 16:23:44 UTC from IEEE Xplore. Restrictions apply.
 Fig. 2. The 4-layer fingertip model used for simulations

The complex permittivity of the layers was calculated

using the Cole-Cole relaxation model with the parameters
found in the open literature [5], [6].
Fig. 4. Measured and simulated return loss (S11) of the sensor with and
without finger
III. THE DESIGNED RF SENSOR
A compact non-invasive RF glucose sensor was designed, There is a good agreement between both when the sensor
fabricated and used in this study [7]. The designed sensor is is unloaded. The simulated response with the fingertip model
based on a one-port microwave resonator, which acts as an and the measured response with an actual finger show a
antenna used for near-field radiation. The resonance frequency resonant peak at the same frequency; however, it can be
of the unloaded sensor is 4.8GHz. When the fingertip is placed observed that the Q-factor of the sensor is lower in the
on top of the sensor, the resonance frequency shifts down to measured case leading to a wideband resonance. This is due to
about 3.25GHz depending on the pressure applied. The sensor the simplified nature of the 4-layer fingertip model used in
was fabricated on a Rogers RO3006 laminate with a thickness CST MWS and a more accurate model is difficult to achieve
of 32mil and a dielectric constant of 6.5. The sensor has a due to the complex structure of the tissues.
main sensing area that resembles an interdigital capacitor for IV. THE FINGER POSITION AND PRESSURE PROBLEMS
achieving high capacitance to maximize the electric field for
increased sensitivity to permittivity change. The 50Ω feed line The importance of the positioning of the finger on the
is located at the back of the substrate as a coplanar waveguide sensor and the pressure applied as well as their effects are
and is connected to the sensor through a plated via hole. This explained in this section.
is to ensure that the finger does not come into contact with the A. Finger Position on the Sensor
feed line, possibly altering the sensor response. The aim was
The fingertip should cover the whole sensing area and
to design a compact sensor so that the whole structure is
should be positioned in the same place every time for a
covered with the fingertip; therefore, the fringing fields are
repeatable measurement. Otherwise, different proportions of
contained within the fingertip as much as possible. The layout
fringing fields may exist in air and tissue layers with unknown
of the designed sensor is shown in Fig. 3.
fractions and it results in a shift in the sensor’s response [8].
.
This shift is greater than the shift caused by the BGL
variation. The findings will be presented in a future
publication in more detail. However, the position can easily be
controlled by placing the sensor in a housing that
accommodates a finger. Such a housing was custom 3D
printed and used as shown in Fig.5.

Fig. 3. Layout of the desgined sensor

The sensor was simulated in CST MWS and measured in

air for the unloaded response using a VNA. It was also
simulated with the 4-layer fingertip model and measured with
an actual fingertip. The simulated and measured responses are
shown in Fig. 4. Fig. 5. 3D printed sensor housing

Authorized licensed use limited to: Indian Institute of Technology (ISM) Dhanbad. Downloaded on May 02,2025 at 16:23:44 UTC from IEEE Xplore. Restrictions apply.
B. Pressure Applied onto the Sensor The pressure data is continuously sent to the computer by the
The sensor’s frequency response highly depends on the MCU. The circuit block diagrams are shown in Fig.6 for the
pressure exerted by the fingertip onto the sensor. The effect of proposed circuits.
pressure was briefly discussed in [8], where a pressure sensor
was used under the RF glucose sensor. Our experiments with
our proposed sensor showed that a measurement with a
fingertip is extremely sensitive to the pressure applied. The
frequency difference in the resonant peaks between a light and
a firm pressure on the sensor may be as high as 240 MHz
whereas the expected shift due to blood glucose variation is
well under 10 MHz [7]. This shows that if the pressure is not
monitored precisely, the shift due to glucose variation will be
masked under the shift resulting from the pressure variation
causing measurement error.

However, the sensitivity to the pressure is expected for the

following reasons:
o The effective thickness of the soft tissue layers change
depending on the pressure applied
o The electrical coupling between the sensor and the
fingertip varies depending on the pressure applied
Fig. 6. Diagrams for: a) Circuit Ver.1 b) Circuit Ver.2
o The blood is pushed away from the fingertip
proportional to the pressure applied, altering the
amount present for measurement The photographs of the designed and realized circuits are
shown in Fig. 7.
To the author’s best knowledge, there is no work in the
open literature discussing the importance of pressure applied
onto the sensor or proposing a precision pressure sensing
circuit. Using a pressure sensor such as a resistive force sensor
under the RF glucose sensor does not provide the precision
needed for repeatable results. For this reason, two novel
pressure sensing circuits were proposed and are reported in
this paper.
V. THE PROPOSED NOVEL PRESSURE SENSING CIRCUITS
The proposed circuit topology uses a 24-bit Capacitance to
Digital Converter (CDC) chip and acts as an interface between
the RF glucose sensor and the Vector Network Analyzer
(VNA) utilizing the RF glucose sensor as a capacitive pressure
sensor. The CDC uses a 16 kHz excitation signal and the RF
sensor provides a good sensitivity to proximity and pressure
due to its highly capacitive nature. However, the pressure
sensing circuit can be paired with any 1-port RF sensor.
Two variations of the topology are proposed. The first
circuit (Ver.1) incorporates an RF switch between the RF
sensor, CDC and the VNA. A Microcontroller Unit (MCU) is
used to control the RF switch and the CDC, generate a trigger
signal to trigger the VNA for a measurement and send Fig. 7. Photographs of: a) Circuit Ver.1 b) Circuit Ver.2
pressure information to a computer. A target pressure value is Three experiments were carried out to evaluate the
preset in the MCU and the RF sensor is connected to the CDC performance of the proposed circuits. First, a blind test was
via the RF switch. When the target pressure level is reached, performed with the sensor. The sensor was connected to the
the RF switch connects the RF sensor to the VNA and the VNA directly, a test subject was asked to place a finger on it
MCU triggers the VNA for a measurement. The second circuit and the resonant frequency was recorded. This was repeated 20
(Ver.2) allows for continuous measurement eliminating the RF times re-placing the finger where the subject was asked to
switch and using a high-pass filter (HPF) on the VNA line to apply approximately the same pressure each time without
block the excitation signal in that branch and an RF choke knowing the actual sensor response to avoid positive
(RFC) on the CDC line to block the RF signal in that branch. reinforcement. The results are shown in Fig. 8.

Authorized licensed use limited to: Indian Institute of Technology (ISM) Dhanbad. Downloaded on May 02,2025 at 16:23:44 UTC from IEEE Xplore. Restrictions apply.
 Fig. 8. Resonant frequency recorded in 20 tests (Sensor directly connected to Fig. 10. Resonant frequency vs. pressure applied (Sensor connected to the
the VNA) VNA through circuit Ver.2)

It can be seen from the results that the resonant frequency VI. CONCLUSION
varies substantially (standard deviation= 21.73 MHz) while
trying to apply the same pressure each time. The pressure applied onto the RF glucose sensor is critical
Next, the circuit Ver.1 was tested the same way where a to get repeatable results, independent of the type of sensor
preset pressure value was programmed into the MCU. The test used. The effective thickness of the biological layers changes
subject is guided on the pressure level via onboard LED lights depending on the pressure applied; hence, the sensor’s
and the VNA is triggered at the desired pressure level. The frequency response also varies accordingly. For this reason, the
results for 20 measurements are shown in Fig. 9. pressure needs to be accurately monitored while taking blood
glucose readings with an RF sensor. However, conventional
pressure sensors require a lot of force to be applied which
results in a crude resolution and do not provide enough
sensitivity. Therefore, the glucose measurement is hindered by
the effects of inconsistent pressure applied. To solve this
problem two novel pressure sensing circuits were proposed
making use of the already existing RF glucose sensor a
pressure sensor too. A good accuracy and repeatability were
achieved as presented.
REFERENCES

[1] “Diabetes Atlas, 6th ed.,” International Diabetes Federation (IDF), 2015.
[Online]. [Accessed 05 2016].

Fig. 9. Resonant frequency recorded in 20 tests (Sensor connected to the VNA [2] R. Holt, C. Cockram, A. Flyvbjerg and B. Goldstein, Textbook of
Diabetes, New York, NY: John Wiley & Sons, 2011.
through circuit Ver.1)
[3] S. Vashist, “Continuous Glucose Monitoring Systems: A Review,”
It can be clearly observed that the variation between Diagnostics, vol. 3, no. 4, pp. 385-412,, 2013.
measurements is very little with a standard deviation of [4] A. V. Vorst, A. Rosen and Y. Kotsuka, RF/microwave interaction with
0.366 MHz, significantly improving the measurement accuracy. biological tissues, Hoboken, NJ: John Wiley & Sons, 2006.
The accuracy can be further improved by averaging a number [5] K. Cole and R. Cole, “Dispersion and Absorption in Dielectrics I.
of measurements. Alternating Current Characteristics,” The Journal of Chemical Physics,
Finally, the relationship between the resonance frequency vol. 9, no. 4, p. 341, 1941.
and the pressure applied was established using the circuit [6] S. Gabriel, R. Lau and C. Gabriel, “The dielectric properties of biological
Ver.2. About 300 measurements were taken continuously while tissues: III. Parametric models for the dielectric spectrum of tissues,”
Physics in Medicine and Biology, vol. 41, no. 11, pp. 2271-2293, 1996.
varying the pressure applied onto the RF sensor. Pressure was
varied randomly between light and firm levels and result were [7] V. Turgul and I. Kale, “Characterization of the Complex Permittivity of
Glucose/Water Solutions for Noninvasive RF/Microwave Blood Glucose
plotted real-time in MATLAB with the frequency response Sensing,” in IEEE International Instrumentation and Measurment
gathered from the VNA and the pressure data received from the Technology Conference (I2MTC), Taipei, 2016.
proposed circuit. The obtained graph is shown in Fig. 10. The
graph shows that the resonance frequency is almost a linear [8] J. T. Bernhard and C. J. Tousignant, “Resonant frequencies of rectangular
microstrip antennas with flush and spaced dielectric superstrates,” IEEE
function of the pressure applied and a good repeatability can be Transactions on Antennas and Propagation, vol. 47, no. 2, 1999.
achieved for a specific pressure level. [9] B. Jean, E. Green and M. McClung, “A microwave frequency sensor for
non-invasive blood-glucose measurement,” in IEEE Sensors Applications
Symposium, Atlanta, GA, 2008.

Authorized licensed use limited to: Indian Institute of Technology (ISM) Dhanbad. Downloaded on May 02,2025 at 16:23:44 UTC from IEEE Xplore. Restrictions apply.