WBAN BASED PROTOTYPE FOR ACTIVE BODY CLIMATE
CONTROL BASED ON ENVIRONMENTAL AND INDIVIDUAL
SENSOR DATA
A. Gharbi, M. Breuel, W. Darmoul, W. Stork, K. D. Mueller-Glaser
Institute for Information Processing Technology, Karlsruhe Institute of Technology (KIT), Karlsruhe, Germany
S. Heuer
FZI (Research Center for Information Technology), Karlsruhe, Germany
S. Haertel
Institute of Sports and Sport Science, Karlsruhe Institute of Technology (KIT), Karlsruhe, Germany
Keywords:
Wireless body area network, Vital parameters monitoring, Active body climate control, Energy expenditure,
Regulation algorithm.
Abstract:
In this paper, a new textile integrated WBAN based prototype for active body climate control is presented.
The design of this prototype resulted from a previous evaluation system, which has been tested in a field study.
In addition, new algorithms for determining the metabolic activity from measured vital parameters (heart rate
and activity) and thus controlling the cooling mechanism by setting the necessary ventilation airflow have been
conceived. A field study involving nine test persons has been conducted in order to test the new prototype and
validate the conceived algorithms.
1 INTRODUCTION
Nowadays, the need for a local textile integrated ac-
tive body climate control is getting more and more im-
portant and essential. In fact, a system that offers such
functionality could relieve staff like rescue personnel
and special police units that have to wear protective
clothing (in this use case ballistic garments). This
type of clothing leads, even undesired, to a thermal
isolation, which prevents the wearer from dissipating
the generated body heat that varies between 80% and
100% of the total metabolic activity of the body (Sil-
bernagl and Despopoulos, 2003; Fiala et al., 1999).
In this way, the thermal load increases very fast with
an increase in the metabolic activity and the body can
overheat. Consequently, this leads to a decrease in
physical performance and thermal comfort and thus
to a limitation of the working time.
Another target group, which could benefit from
such a system, are the elderly, who have cardiovas-
cular problems. These people cannot stand excessive
heat that could lead to a collapse of their cardiovas-
cular system, if their physiological thermoregulation
mechanisms do not get active support by adjusting the
climate of their surroundings. This could be achieved
by either controlling the room temperature by air con-
ditioners or by varying the amount of air circulating in
the vicinity of their skin surface, which could be real-
ized by a textile integrated active body climate control
system.
The latter is primarily a cooling system that is
meant to support the heat exchange mechanisms of
the body by improving the convection and especially
the evaporation. For instance, the evaporation of 1
liter of sweat could take around 2400 kJ (equivalent
to 667 Wh) away from the body. This approximately
corresponds to a physical power dissipation of 100 W
(in form of body heat) during 6.7 hours. An airflow of
100 L/min having a temperature of 30
◦
C and a rela-
tive humidity of 20% (thus able to transport 24 g/m
3
of water) could theoretically transport the generated
heat away from the body.
The climate control system should therefore get an
input from several vital and environmental parameters
in order to identify the cooling needs of the user.
Another aspect of utmost importance is that such
276
Gharbi A., Breuel M., Darmoul W., Stork W., D. Mueller-Glaser K., Heuer S. and Haertel S..
WBAN BASED PROTOTYPE FOR ACTIVE BODY CLIMATE CONTROL BASED ON ENVIRONMENTAL AND INDIVIDUAL SENSOR DATA.
DOI: 10.5220/0003168202760284
In Proceedings of the International Conference on Bio-inspired Systems and Signal Processing (BIOSIGNALS-2011), pages 276-284
ISBN: 978-989-8425-35-5
Copyright
c
2011 SCITEPRESS (Science and Technology Publications, Lda.)
a system could play a major role in energy saving.
In fact, the active body climate control system not
only offers the advantage of portability but also needs
only about 2 to 3 W of electrical power (primarily
for the integrated ventilators), which corresponds to
about 0.1% of the energy consumed by a commercial
room air conditioner.
2 METHODS AND MATERIALS
After the implementation of a first prototype for an ac-
tive body climate control system and the conduction
of a field study (Gharbi et al., 2010), several enhance-
ments concerning system concept, design issues, tex-
tile integration and body climate regulation algorithm
have been carried out. Consequently, a second gener-
ation prototype has been realized and a second field
study, similar to the first one, was conducted in order
to collect feedback information from the users and to
validate the reliability of the sensor data and the con-
ceived algorithms.
2.1 Prototype
In this section, the realized prototype is described. As
Figure 1 shows, the prototype comprises three main
components; a sensor shirt, a cooling vest (worn on
the sensor shirt) and a user feedback terminal (in this
case a PDA). These components build up a wireless
body area network (WBAN) by means of two wire-
less technologies. The first wireless technology uses
the low power radio modules iM221A from IMST for
the 2.4 GHz ISM band (IMST GmbH, 2009). These
assure the data exchange between the sensor shirt
and the cooling vest. In order to exchange data be-
tween the cooling vest and the PDA, a Bluetooth mod-
ule (MITSUMI WML-C46) (MITSUMI ELECTRIC
CO., LTD., 2010) has been integrated into the con-
trol board of the cooling vest. Through that channel,
data can be bidirectionally transmitted for different
purposes like online visualization of the sensor data
and the calculated body heat and ventilation level,
input of user specific data like weight, height and
age into the system, and user feedback that adjusts
the determined ventilation level in the cooling vest.
Both control boards of the sensor shirt and the cool-
ing vest integrate a 16 bit RISC microcontroller from
the PIC24FJ256GB110 family (Microchip Technol-
ogy Inc., 2010).
Battery
Control board of
the sensor shirt
Front pocket
Control board of
the cooling vest
Combined
temperature &
humidity sensors
Ventilator
PDA
Figure 1: The new prototype for active body climate con-
trol.
Combined
temperature &
humidity sensor
Control board
Electrodes
Combined
temperature &
humidity sensor
Figure 2: The sensor shirt component.
2.1.1 Sensor Shirt
The sensor shirt, as shown in Figure 3, is made of an
elastic material permeable to moisture, which fits the
body form and lets the generated evaporation heat of
the body diffuse towards the ventilation layer of the
cooling vest. It also integrates several sensors and a
control board in order to determine the actual body
climate and its energy expenditure due to physical ac-
tivity. These modules can be attached to the textile
via snap fasteners.
The sensor shirt includes the following vital pa-
rameter sensors:
Two combined sensors for measuring temperature
and relative humidity (temperature compensated) are
used in the area of the chest and the upper back due to
the non uniform distribution of the skin temperature
(Fiala et al., 1999; Olesen and Fanger, 1973; Craw-
shaw et al., 1975). The relative humidity at the skin
surface is besides a good index for determining the
WBAN BASED PROTOTYPE FOR ACTIVE BODY CLIMATE CONTROL BASED ON ENVIRONMENTAL AND
INDIVIDUAL SENSOR DATA
277
Combined
temperature &
humidity sensor
Control board
Electrodes
Combined
temperature &
humidity sensor
Figure 3: The sensor shirt component.
amount of heat losses in form of evaporation. The lat-
ter represents the largest part of the body heat given up
to the surroundings and especially at high surround-
ings temperature and/or metabolic activity.
In order to detect the heart rate, two conductive
rubber based ECG electrodes are integrated into the
shirt. An analog front-end module, which is part
of the control board, detects the QRS complexes in
the ECG signal and generates digital square pulses
at each QRS event. This digital signal is connected
to an input capture module of the microcontroller,
which enables the calculation of the heart rate. To
get an ECG signal of a good quality, the crossover
impedance between the skin surface and the elec-
trodes has to be very low. In the case of sport activity,
the fast formation of a sweat film under the electrodes
decreases the crossover impedance. In the conducted
field study, a cream for ECG electrodes is applied.
Additionally, a 3D acceleration sensor is inte-
grated into the control board. From the measured
acceleration data, the Activity Equivalent ACcelera-
tion (AEAC) can be calculated according to the equa-
tions 1 and 2 (Gharbi et al., 2010). Thereby, EEAC
stands for Energy Equivalent ACceleration and rep-
resents the signal energy from all three axes together
averaged in a time period of one second (Jatob et al.,
2007). f
s
denotes the sampling rate of the accelera-
tion data (in the current system 20 Hz).
AEAC = |(EEAC − 1)| (1)
EEAC =
1
f
s
f
s
X
i=1
q
a
2
xi
+ a
2
yi
+ a
2
zi
(2)
2.1.2 Cooling Vest
The cooling vest is worn over the sensor shirt and in-
tegrates a 7 mm thick space holder material, through
which fresh air can circulate, and has a total weight of
approximately 680 g. It has two different ventilation
parts for separate control of the body climate at the
chest and the back area. These are assembled to each
other via velcro fastener. As Figure 4 shows, each
ventilation part has an air inlet in the top left corner
and an air outlet in the opposite bottom right corner,
where a ventilator is placed in order to aspirate the air
and thus lead to the circulation of the air over the up-
per part of the body. In this way, the air blown out at
the outlet becomes warmer and more humid due to the
released body heat via evaporation and convection.
Air inlet
Air outlet
Space holder
material
Figure 4: Design of the ventilation area of the cooling vest.
Thereby, the inner textile separation layer is, un-
like the outer one, permeable to the moisture. In ad-
dition, each ventilation part of the cooling vest inte-
grates two combined sensors for measuring temper-
ature and relative humidity at both the inlet and the
outlet. These latter can be used to monitor the ambi-
ent air (from the sensor at the inlet) and the thermal
exchange between the body surface under the vest and
the circulating air in the vest (from the sensor at the
outlet). Each ventilation part of the cooling vest also
has a pocket (see Figure 1), where a radial ventilator
that allows an airflow up to 50 L/min and a recharge-
able lithium-polymer battery (7.4 volt and a capac-
ity of 1100 mAh), which allows an operating time
of more than 7 hours for the ventilator at full perfor-
mance, are placed. The applied voltage to both of the
ventilators is controlled by a PWM modulator on the
control board, which is placed in the front pocket (see
Figure 1). As a result, the ventilation level in the vest
can be changed according to the current level of the
regulation algorithm and a maximal total airflow of
around 100 L/min can be set.
This control board represents the main node of
the system, where the management of the wireless
communication with the other slaves (sensor shirt and
PDA), the signal processing, the climate regulation al-
gorithm and the ventilation control take place. It also
integrates an SDHC memory card in order to save all
sensor and regulation algorithm data for offline analy-
sis.
BIOSIGNALS 2011 - International Conference on Bio-inspired Systems and Signal Processing
278
2.1.3 Feedback Interface: PDA
The PDA communicates with the control board of the
cooling vest via Bluetooth. On the one hand, it gives
the user the possibility to change the ventilation level
according to his individual preferences. On the other
hand, it allows the online visualization of all vital and
system parameters.
2.1.4 Textile Integration
All electronic modules of the different system com-
ponents, whether small sensor modules (for instance
the modules of the combined temperature and humid-
ity sensors) or control boards of both the sensor shirt
and the cooling vest, have snap fastener (like Figure
5 shows) and can be thus detached from the textile
material, when the latter needs to be washed.
(a)
(b)
Figure 5: Implemented control modules of a) the sensor
shirt, b) the cooling vest.
The snap fasteners on the textile material are con-
nected to each other via isolated conductive yarns.
In this prototype, ELITEX yarn (TITV, 2008), which
comprises many silver coated polyamide filaments in-
sulated by a polyurethane film, was used. It was sewn
meander-shaped on the textile material and soldered
at its ends to the snap fasteners before crimping these
into the textile like in (Lamparth et al., 2009) (see Fig-
ure 3).
2.2 Regulation Algorithm
The conceived regulation algorithm is based on the
measured vital and surroundings parameters and is
shown in Figure 6.
Vent
Level
hrel
Tsk
Tair
E
out
Wout
W
out_torso
W
out_fiala
W
out_fiala_torso
µ
Regulation
Fiala model
Prototype
Personal parameter:
•Weight
•Height
•Clothing
W
out_fiala_torso
W
out_fiala
%torso
Data fusion
HR
AEAC
Figure 6: Knowledge based regulation algorithm.
In a first stage, the heart rate (HR) and the activ-
ity equivalent for the acceleration (AEAC), described
in section 2.1.1, are fused in order to approximate the
actual metabolic activity of the body.
Therefore, the sensor data is beeing post processed in
order to avoid a strong fluctuation and thus an instabil-
ity of the regulation algorithm. The HR values, which
have been determined event-driven, are first filtered
by a median filter having the width of 11 samples
and then averaged over a moving window having the
width of 5 samples. The AEAC values, which have
been measured with a sample rate of 20 Hz, are only
averaged over a moving window having the width of
30 samples. These settings have been determined af-
ter conducting two tests and have shown a good be-
havior with a minimal fluctuation and without caus-
ing an inacceptable reaction time of the system due to
the resulting delays in the signals.
The data fusion model is shown in Figure 7.
HR
AEAC
activity
classification
late fusion
regression models
E
AEAC
individual
parameter (BMI)
E
HR
E
out
error detection
weight, height
Figure 7: Data fusion model for approximating the
metabolic activity of the body.
The error detection component filters outliers by
means of a confidence interval analysis (for both HR
and AEAC). In this way, faulty measurements, which
WBAN BASED PROTOTYPE FOR ACTIVE BODY CLIMATE CONTROL BASED ON ENVIRONMENTAL AND
INDIVIDUAL SENSOR DATA
279
can for instance be caused by an ECG electrode loos-
ing contact with the skin due to a high activity dy-
namic, and a malfunction of a sensor can be compen-
sated.
The data fusion model also integrates a decision
tree based activity classification that classifies activity
into two categories: low activity (e.g. laying, standing
or walking slowly) and middle to high activity (e.g.
walking fast or running). This classification is based
on two features related to the AEAC signal; the value
of AEAC itself and its variance. Since the classifica-
tion algorithm is beeing implemented on a microcon-
troller, which has limited resources, the variance was
approximated with an integral over the last 30 AEAC
variations (∆AEAC) according to formula 3.
AEACint30 =
t
X
t−30
∆AEAC(t). (3)
The results of the action analysis component
are then used by activity based regression models
that map the corresponding vital parameter (HR and
AEAC) to a reference measurement of the metabolic
activity (registered by a spirometer) by means of the
robust least mean square method as adopted in (Ren-
nie et al., 2000). At the same time, these regression
models take into consideration the body mass index
BMI (see equation 4) in order to minimize the ap-
proximation failure from one person to another.
BMI =
weight
body
[kg]
(height
body
[m])
2
(4)
In this way, the polynomial equations 5 to 8,
which approximate the energy expenditure of the
body respectively from HR and AEAC for both states,
low and high activities, have been determined.
E
HR,low
= α
HR,low
· HR + β
HR,low
+ γ
HR,low
· BMI (5)
E
HR,high
= α
HR,high
· HR + β
HR,high
+ γ
HR,high
· BMI (6)
E
AEAC,low
= α
AEAC,low
· AEAC + β
AEAC,low
+γ
AEAC,low
· BMI (7)
E
AEAC,high
= α
AEAC,high
· AEAC
3
+ β
AEAC,high
· AEAC
2
+γ
AEAC,high
· AEAC + δ
AEAC,high
+ φ
AEAC,high
· BMI (8)
The determined metabolic activities from HR and
AEAC are then fused in order to approximate the ac-
tual metabolic activity of the body and thus get E
out
.
Two fusion methods have been implemented: simple
mean building and kalman filter.
After multiplying the determined metabolic acti-
vity E
out
with the efficiency factor of approximately
20% (Fiala et al., 2001), the total amount of the ge-
nerated body heat W
out
that has to be released can be
calculated.
Besides, the measured temperature and relative
humidity of the skin, the aspirated air in the cooling
vest and the person specific parameter (weight, height
and clothing) are applied to a knowledge based model
according to the fiala model (Fiala et al., 1999; Fi-
ala et al., 2001). Thereby, the percentage of the body
heat, which is released by the body under the cooling
vest W
out torso
(torso area), can be approximated.
The actuating variable for the active climate
control is the ventilation level Vent
Level
. This latter is
proportional to the rotation speed of the ventilators at
the outlets of the cooling vest and is thus responsible
for varying the airflow. The operating range of the
used ventilators is linearly split up into 11 discrete
levels:
Level 0 = ventilators are inactive
Level 1-9 = ventilators are active
Level 10 = maximal ventilation (corresponds to 100
L/min)
The calculated ventilation level (Vent
Level
) com-
prises two components: the ventilation level calcu-
lated from the measured sensor data and the deter-
mined body heat W
out torso
(Alg
Level
), and the user
level (U ser
Level
), which is set online by the user via
the PDA and thus gives him the possibility to fine
tune the ventilation level according to his individual
comfort feeling and even switch it off, if desired (see
formula 9).
Vent
Level
= Alg
Level
+ U ser
Level
(9)
with 0 ≤ Vent
Level
≤ 10
0 ≤ Alg
Level
≤ 10
−Alg
Level
≤ U ser
Level
≤ 10 − Alg
Level
The concept of the regulation component is shown
in Figure 8. It uses the determined body heat
W
out torso
and a knowledge based regression model
from the data of the first field study. This way, the
necessary airflow and thus the corresponding ventila-
tion level can be calculated.
Since the metabolic activity, and especially the
basal metabolic rate, varies from one person to an-
other, a calibration phase has been implemented (as
shown in Figure 8) in order to compensate the base-
line body heat W
baseline
corresponding to the basal
metabolic rate. This calibration phase takes place in
the first 30 seconds of the measurement cycle, where
BIOSIGNALS 2011 - International Conference on Bio-inspired Systems and Signal Processing
280
calib?
W
out_torso
Vent
Level
= 1
calibrate W
baseline
W
out_torso
-W
baseline
Calc Vent
Level
regression model
Vent
Level
yes
no
Figure 8: Calibration based regulation component.
the test person is inactive. W
baseline
is calculated and
averaged over the 30 seconds. After that, the offset
of W
baseline
is subtracted from the calculated released
body heat W
out torso
.
2.3 Field Study
The conducted field study aimed at collecting detailed
information about the physiological interdependen-
cies and their correlation with the workload. As a ref-
erence, the consumption of oxygen is saved by means
of a spirometer during the load test. An offline analy-
sis later on allows for validating the sensor data of the
new prototype and adjusting the existing regression
models for the calculation of the energy expenditure
or metabolic activity of the body from the sensor data
(Hollmann, 2006) according to the new design of the
sensor shirt, in which the position of the acceleration
sensor has changed and a new material of the ECG
electrodes is used.
Besides, a verification of the suitability of the new
prototype for the daily use and a validation of the im-
plemented algorithm for active body climate control
can be conducted.
Nine test persons (high school students, males)
participated in the field study (see Table 1).
Table 1: Physiological data of the nine test persons.
Mean value Standard deviation
Age [years] 26,5 ± 3,5
Height [cm] 182 ± 8
Weight [kg] 81,25 ± 12,25
Field Study Procedure. Each test person under-
took a load test on a treadmill ergometer wearing the
implemented prototype and a spirometer. The test
procedure had to be conceived in a way that it makes
it possible to compare the results of the load tests
related to different test persons. Therefore, a low
workload level in the beginning of the test and short
workload phases, which do not exclude the reaching
of a bio-physiological steady state status, needed
to be considered. According to (Wahlund, 1948),
an absolute steady state status for light to middle
workload is reached after 6 minutes. Investigations of
the sports university in Koeln/Germany showed that
90 to 95% of the steady state status can be registered
already after 3 minutes of the beginning of a constant
workload (Knipping et al., 1953). For the conducted
load test, a tradeoff was met analog to (Hollmann,
1963). After a baseline phase of 5 minutes, an
intensity level of 6 km/h was set at the treadmill
ergometer for the period of 5 minutes followed by
an increase in the intensity level with 1 km/h. After
the load test phase with the maximum intensity level
of 10 km/h, the test person rested for 10 minutes.
This baseline phase after the end of the load test
enables the observation and the understanding of the
relaxation behavior of the body.
Test Person Feedback Form. During the study,
subjective feedback from the test persons was col-
lected by means of a test person feedback form. This
latter includes general questions before the beginning
of the load test dealing with the performance anamne-
sis (general fitness, physical and mental state on the
day of testing, feeling of the room climate, etc.) and
recording the individual physiological data of each
test person (e.g. age, body height, weight, etc.) and
the climate in the room of the study (temperature,
moisture, etc.). During the load test, the actual wear-
ing comfort of the vest and the local thermal comfort
at the chest and the back were registered in the mid-
dle of every phase of the test. After the end of the
test, the test persons gave their feedback related to the
absolved test and to the test settings (for instance the
new prototype and the efficiency of its active climate
control).
3 RESULTS AND DISCUSSION
3.1 Data Fusion
In order to assess the implemented data fusion algo-
rithm, the Mean Absolute Percentage Error (MAPE)
has been calculated according to equation 10 for all
nine test persons, who took part in the field study.
MAPE =
1
n
n
X
t=1
y
t
− x
t
y
t
(10)
Thereby, y
t
represents the reference value corre-
sponding to the energy expenditure calculated from
the spirometer data and x
t
represents the estimated
value from the sensor data (HR and AEAC) in case
WBAN BASED PROTOTYPE FOR ACTIVE BODY CLIMATE CONTROL BASED ON ENVIRONMENTAL AND
INDIVIDUAL SENSOR DATA
281
Table 2: MAPE for the nine test persons.
Test person MAPE [%]
1 18.41
2 28.80
3 22.68
4 27.42
5 12.74
6 27.93
7 23.77
8 41.29
9 19.47
∅ 24.72
Figure 9: Comparison of the approximated metabolic acti-
vity with the spirometer data for test person №5.
of fusion by the mean building method.
According to Table 2, the approximation of the
metabolic activity for test persons №1, 5 and 9 was
good, Figure 9 shows that for test person №5 ex-
emplarily. In fact, the calculated metabolic activity
(dashed red line) follows the reference measurement
of the spirometer (magenta line) and has a small mean
deviation of 12.74%. The fusion model, which uses
kalman filter for the fusion of the calculated metabolic
activities, (black line) has shown similar good results
like the mean building method, except that it has a
longer reaction time and thus a bigger time delay.
Since the mean building method additionally needs
less calculational resources than the kalman filter, it
is more appropriate for online implementation on the
microcontroller.
As for the other test persons, a stronger deviation
from the spirometer data can be obseved. This is due
to incorrect activity classification in the data fusion
model. In fact, the majority of the test persons were
walking fast in the first load test phase between the
5
th
and the 10
th
minute (see Figure 10). But this was
classified as a low activity rather than high activity
and thus the metabolic activity was underestimated.
(a)
(b)
Figure 10: Comparison of the approximated metabolic acti-
vity with the spirometer data for a) test person №9, b) test
person №8.
Test person №8 was walking fast even in the second
load phase corresponding to 7 km/h. In addition his
spirometer data (magenta line in Figure 10 b)) has
shown an instability that led to the corresponding bad
MAPE in Table 2. This can be explained by the fact
that the test person lifted the mask of the spirometer
during the load test in order to adjust it, which had an
impact on the initial calibration of the spirometer and
led to its mulfunction.
3.2 Regulation Algorithm
In order to validate the implemented regulation algo-
rithm, a comparison of the calculated ventilation level
(Vent
Level
) and the ventilation level calculated from
the measured sensor data and the determined body
heat W
out torso
(Alg
Level
or Vent
Level theoretical
) is nec-
essary. Figure 11 shows the observed trend for the
majority of the test persons.
BIOSIGNALS 2011 - International Conference on Bio-inspired Systems and Signal Processing
282
Figure 11: Comparison of the theoretical and the real ven-
tilation level for a test person during the load test for test
person №9.
One can see that the deviation is small during
the load test until the load phase of 8 km/h. Af-
ter that, the calculated ventilation level exceeds the
maximal ventilation level, which corresponds to the
maximal airflow in the cooling vest of around 100
L/min. This implies that the ventilation resources
are not sufficient for relatively high loads. At the end
of the load test during the second baseline phase (af-
ter the 30
st
minute), the theoretical ventilation level
decreased rapidly. But the test person wanted more
ventilation and has compensated it via the feedback
channel of the PDA. This can be explained by the fact
that the calculated metabolic activity has decreased
rapidly at the beginning of the second baseline phase
since the AEAC parameter decreased rapidly too. In
other words, there is a need for taking the relaxation
phase of the body after an important and lasting phys-
ical activity into consideration.
Figure 12 shows that the measured body heat
W
out vest total
(continuous red line), which is trans-
ported away by means of the airflow circulating
within the cooling vest, is basically an evaporation
heat (W
out evap
) and the convection part (W
out conv
) is
very negligible. In addition, a comparison of the cal-
culated released body heat at the torso area W
out torso
(magenta line) and the measured heat W
out vest total
(continuous red line) shows a gap between the two
lines, which can reach 150 W. This can be ex-
plained by the fact that the actual ventilation resources
are below the theoretically needed airflow. In addi-
tion, the absorption capacity of the textile material of
the sensor shirt prevents the total amount of the re-
leased evaporation heat from getting into the ventila-
tion layer.
Wgap
time (s)
W
out
(W) / Airflow (l/min)
W
out _evap.
W
out _con v.
W
out _vest_total
Airflow
W
out _torso
Figure 12: Comparison of the released body heat at the torso
and the heat transported away within the cooling vest for
test person №3.
4 CONCLUSIONS AND FUTURE
WORK
The new prototype for active body climate control
has shown better results than the previous design de-
scribed in (Gharbi et al., 2010). Thereby, the differ-
ent components of the implemented regulation algo-
rithm have been validated with the data of the con-
ducted field study. In addition, the test persons, who
took part in the conducted field study, gave an over-
all positive feedback regarding the suitability of the
new prototype for the daily use. They also confirmed
the added value of the cooling system that actively
supports their body thermoregulation. Nevertheless,
some limitations of the current system and its algo-
rithms could be identified and several enhancements
need to be carried out:
The fusion model, especially the activity classifi-
cation component, needs to be further optimized in or-
der to get a more accurate estimation of the metabolic
rate.
As for the regulation algorithm, the relaxation
phase of the body after an important and lasting phys-
ical activity needs to be modeled, so that an appropri-
ate ventilation level close to the subjective optimum
of the wearer of the system can always be guaranteed.
Concerning the presented prototype for active
body climate control, the design of the ventilation
area of the cooling vest needs to be improved, so that
the ventilation resources would be sufficient for high
workloads. For that, new designs of the space holder
material need to investigated in order to reduce the air
resistance inside the cooling vest. Another alterna-
tive is the integration of more than two ventilators in
the cooling vest and/or the selection of more powerful
ventilators.
WBAN BASED PROTOTYPE FOR ACTIVE BODY CLIMATE CONTROL BASED ON ENVIRONMENTAL AND
INDIVIDUAL SENSOR DATA
283
ACKNOWLEDGEMENTS
This work is done in the framework of the joint re-
search project ”KlimaJack”, which is financed by the
German ministry for education and research. Besides
the Institute for Information Processing Technology
(KIT), other German industrial parties (Beurer, IMST,
Lodenfrey, MVS and Ruhlamat) have contributed to
the conducted research work.
REFERENCES
Crawshaw, L. I., Nadel, E. R., Stolwijk, J. A. J., and Stam-
ford, B. (1975). Effect of local cooling on sweating
rate and cold sensation. Pflgers Archiv: European
Journal of Physiology, 354:19–27.
Fiala, D., Lomas, K., and Stohrer, M. (1999). A computer
model of human thermoregulation for a wide range of
environmental conditions: The passive system. Jour-
nal of Applied Physiology, 87:1957–1972.
Fiala, D., Lomas, K., and Stohrer, M. (2001). Com-
puter prediction of human thermoregulatory and tem-
perature responses to a wide range of environmental
conditions. International Journal of Biometeorology,
45:143–159.
Gharbi, A., Kraehling, C., Stork, C., and Mueller-Glaser,
K. D. (2010). Evaluation system for monitoring of
vital parameters and active body climate control. In
Biosignals: 3rd International Conference on Bio-
inspired Systems and Signal Processing. INSTICC
Press.
Hollmann, W. (1963). Hoechst- und Dauerleistungsfae-
higkeit des Sportlers. Barth, Munich.
Hollmann, W. (2006). Spiroergometrie : kardiopulmonale
Leistungsdiagnostik des Gesunden und Kranken.
Schattauer, Stuttgart.
IMST GmbH (2009). WiMOD - iM221A Datasheet: low
power radio module for the 2.4 GHz ISM band.
Jatob, L., Gromann, U., Ottenbacher, J., Haertel, S., von
Haaren, B., Stork, W., Mueller-Glaser, K., and Boes,
K. (2007). Obtaining energy expenditure and physical
activity from acceleration signals for context-aware
evaluation of cardiovascular parameters. In CLAIB.
Knipping, H., Albrecht, H., Valentin, H., and Venrath, H.
(1953). Ueber die atmung und das herzminutenvol-
umen bei arbeit und sport, sowie die herzleistung.
Zeitschrift fiir die gesamte experimentelle Medizin,
122:356–368.
Lamparth, S., Fuhrhop, S., Kirst, M., Wagner, G. v., and
Ottenbacher, J. (2009). A mobile device for textile-
integrated long-term ecg monitoring. In IFMBE Pro-
ceedings: World Congress on Medical Physics and
Biomedical Engineering, Munich, Germany.
Microchip Technology Inc. (2010). PIC24FJ256GB110
family Datasheet: 16 bit RISC microcontroller.
MITSUMI ELECTRIC CO., LTD. (2010). WML-C46
Datasheet: Bluetooth module, class 2.
Olesen, B. W. and Fanger, P. O. (1973). The skin temper-
ature distribution for resting man in comfort. Pflgers
Archiv: European Journal of Physiology, 27:385–393.
Rennie, K., Rowsell, T., Jebb, S. A., and Holdurn, D.
(2000). A combined heart rate and movement sensor
proof of concept and preliminary testing study. Euro-
pean Journal of Clinical Nutritiun, 54:409–414.
Silbernagl, S. and Despopoulos, A. (2003). Taschenat-
las Physiologie. Thieme ElectronicBook Library.
Thieme, Stuttgart [u.a.], 6., vollst. berarb. und erw.
aufl. edition.
TITV (2008). ELITEX Datasheet: conductive yarn.
Wahlund, H. (1948). Determination of the physical working
capacity. Acta Medica Scandinavica, 132:1–78.
BIOSIGNALS 2011 - International Conference on Bio-inspired Systems and Signal Processing
284