sensors

Review
Wearable Sensors for Remote Health Monitoring
Sumit Majumder 1 , Tapas Mondal 2 and M. Jamal Deen 1, *
1 Department of Electrical and Computer Engineering, McMaster University, Hamilton, ON L8S 4L8, Canada;
[email protected]
2 Department of Pediatrics, McMaster University, Hamilton, ON L8S 4L8, Canada; [email protected]
* Correspondence: [email protected]; Tel.: +1-905-525-9140 (ext. 27137)

Academic Editors: Stephane Evoy and Baris Fidan

Received: 30 September 2016; Accepted: 21 December 2016; Published: 12 January 2017

Abstract: Life expectancy in most countries has been increasing continually over the several few
decades thanks to significant improvements in medicine, public health, as well as personal and
environmental hygiene. However, increased life expectancy combined with falling birth rates are
expected to engender a large aging demographic in the near future that would impose significant
burdens on the socio-economic structure of these countries. Therefore, it is essential to develop
cost-effective, easy-to-use systems for the sake of elderly healthcare and well-being. Remote health
monitoring, based on non-invasive and wearable sensors, actuators and modern communication
and information technologies offers an efficient and cost-effective solution that allows the elderly
to continue to live in their comfortable home environment instead of expensive healthcare facilities.
These systems will also allow healthcare personnel to monitor important physiological signs of
their patients in real time, assess health conditions and provide feedback from distant facilities. In
this paper, we have presented and compared several low-cost and non-invasive health and activity
monitoring systems that were reported in recent years. A survey on textile-based sensors that
can potentially be used in wearable systems is also presented. Finally, compatibility of several
communication technologies as well as future perspectives and research challenges in remote
monitoring systems will be discussed.

Keywords: wearable sensors; smart textile; remote health monitoring; body sensor network; vital
sign monitoring; ambulatory monitoring

1. Introduction
Life expectancy has been increasing worldwide due to significant improvements in healthcare,
and medicine, as well as due to growing consciousness about personal and environmental hygiene [1,2].
In addition, over the past several decades, there has been increasing interest in family planning [3],
thereby contributing to declining birth rates around the globe. According to the World Health
Organization (WHO), by 2017, the elderly population over 65 years or older are expected to outnumber
the children less than 5 years of age [4]. However, this enormous aging population would create a
significant impact on the socio-economic structure of society in terms of social welfare and healthcare
needs. Besides this, the cost associated with health care services continues to soar because of the
increasing price of prescription drugs, medical instruments, and hospital care [5]. Therefore, it is an
utmost necessity to develop and implement new strategies and technologies in order to provide better
health care services at an affordable price to the aging population or to the people of those areas having
limited access to healthcare while ensuring maximum comfort, independence, and participation among
the people.
Remote healthcare monitoring allows people to continue to stay at home rather than in
expensive healthcare facilities such as hospitals or nursing homes. It thus provides an efficient and

Sensors 2017, 17, 130; doi:10.3390/s17010130 www.mdpi.com/journal/sensors

Sensors 2017, 17, 130 2 of 45

cost-effective alternative to on-site clinical monitoring [6]. Such systems equipped with non-invasive
and unobtrusive wearable sensors can be viable diagnostic tools to the healthcare personnel for
monitoring important physiological signs and activities of the patients in real-time, from a distant
facility [6–8]. Therefore, it is understandable that wearable sensors play a critical role in such
monitoring systems that attracted the attention of many researchers, entrepreneurs, and tech giants in
recent years. A variety of application specific wearable sensors, physiological and activity monitoring
systems were proposed in the literature. Apart from that, various wearable commercial products such
as the biometric shirt (by Hexoskin® , Montreal, QC, Canada), fitness trackers (by Fitbit® , San Francisco,
CA, USA, Jawbone® , San Francisco, CA, USA, Striiv® , Redwood city, CA, USA and Garmin® , Olathe,
KS, USA) are now available on the market. A list of some commercial products and their principal
applications is presented in Table 1.

Table 1. Listing of some commercial products for monitoring physiological signs and activities.

Battery
Product Name Monitored Parameters Wireless Platform Type Life
Heart rate (HR), HR variability, respiratory
Hexoskin® rate, number of steps, distance traveled,
Bluetooth 6–7 days (standalone) 14+ h (multi-training)
Biometric® Shirt pace, maximal oxygen consumption, and
calories burned.
Sleep stages (REM, light and deep), HR,
Jawbone UP3™ food and liquid intake, number of steps, Bluetooth LE Li-ion poly 7 days
Fitness Tracker
distance traveled, running.
Striiv®Fusion Bio HR, number of steps, distance traveled,
Bluetooth LE Li-ion 5 days
Fitness Tracker calories burned, and sleep quality.
HR, calories burned, sleep quality, food, and
Microsoft® Band 2 liquid intake, number of steps, elevation, Bluetooth Li-poly 2 days
climbing, running, biking.
HR, calories burned, sleep quality, food, and
Fitbit Charge HR™
liquid intake, number of steps, elevation, Bluetooth LE Li-poly 5–7 days
Fitness Tracker
climbing, running.
Garmin vivosmart® HR, calories burned, sleep quality, number
Bluetooth LE, ANT+ Li-ion 5 days
HR Fitness Tracker of steps, climbing, running, swimming.

Wearable devices can monitor and record real-time information about one's physiological
condition and motion activities. Wearable sensor-based health monitoring systems may comprise
different types of flexible sensors that can be integrated into textile fiber, clothes, and elastic bands or
directly attached to the human body. The sensors are capable of measuring physiological signs such as
electrocardiogram (ECG), electromyogram (EMG), heart rate (HR), body temperature, electrodermal
activity (EDA), arterial oxygen saturation (SpO2 ), blood pressure (BP) and respiration rate (RR) [9,10].
In addition, micro-electro-mechanical system (MEMS) based miniature motion sensors such as
accelerometers, gyroscopes, and magnetic field sensors are widely used for measuring activity related
signals [6,11]. Continuous monitoring of physiological signals could help to detect and diagnose
several cardiovascular, neurological and pulmonary diseases at their early onset. Also, real-time
monitoring of an individual’s motion activities could be useful in fall detection, gait pattern and
posture analysis, or in sleep assessment. The wearable health monitoring systems are usually equipped
with a variety of electronic and MEMS sensors, actuators, wireless communication modules and signal
processing units. The measurements obtained by the sensors connected in a wireless Body Sensor
Network (BSN) [8,12–14] are transmitted to a nearby processing node using a suitable communication
protocol, preferably a low-power and short-range wireless medium, for example, Bluetooth [15,16],
ZigBee [15,17], ANT [15,18,19] Near Field Communications (NFC) [20,21]. The processing node, which
could be a Personal Digital Assistant (PDA), smartphone, computer or a custom made processing
module based on a microcontroller or a Field Programmable Gate Array (FPGA) runs advanced
processing, analysis, and decision algorithms and may also store and display the results to the user.
It transmits the measured data over the internet to the healthcare personnel, thus functioning as
the gateway to remote healthcare facilities. The general overview of the remote health monitoring
system is presented in Figure 1, although actual implemented system could differ depending on the
application requirements. For example, some systems can be designed with few numbers of sensors
where each of them can send data directly to the nearby gateway. In other systems, the sensors can
Sensors 2017, 17, 130 3 of 45

be connected through a body sensor network (BSN) and the central BSN node gathers data from the
sensors, performs limited processing before transmitting the data to the advanced processing platform.

Figure 1. General overview of the remote health monitoring system.

In order to be used for long-term monitoring purposes, wearable health monitoring systems
need to satisfy certain medical and ergonomic requirements. For example, the system needs to be
comfortable; the components should be flexible, small in dimensions and must be chemically inert,
and nontoxic, hypo-allergenic to the human body. In addition, limitation of hardware resources
is a major concern for a multi-sensor BSN system where the central node needs to handle a large
amount of data coming from different sensor nodes. It also causes significant impact on the system
power requirements that needs to be minimized in order to extend the battery life for long-term use.
The measured and processed physiological data are, eventually, transmitted to the remote healthcare
facility over the internet. Therefore, it is also necessary to use a secured communication channel in
order to safeguard the privacy of sensitive personal medical data. Strong encryption techniques such
as Public Key Infrastructure (PKI), Secure Sockets Layer (SSL) as well as appropriate authorization
and authentication algorithms [22,23] could be implemented for enhanced data security. Finally, the
system needs to be inexpensive and user-friendly in order to ensure its widespread acceptance among
the people for ubiquitous health monitoring. Therefore, the critical design challenge for wearable
health monitoring system is to integrate several electronic and MEMS components while ensuring
measurement accuracy, efficient data processing, information security, and low-power consumption as
well as user’s wearing comfort.
In this paper, we present a review on the current state of research and development in wearable
systems for health monitoring by summarizing and comparing the most significant contributions in
this field. Recent works on wearable sensor-based physiological parameters and activity monitoring
systems are studied in Section 2, which is followed by a review on textile based flexible sensors
(Section 3) that are vital for smart textile technologies. A brief discussion on wireless communication
standards for the wireless monitoring system is presented in Section 4. Finally, the paper is concluded
in Section 5 by illustrating some key challenges and future research directions in the field of remote
health monitoring.

2. Wearable Health Monitoring Systems

Non-invasive, non-intrusive sensors are indispensable elements of ambulatory and long-term
health monitoring systems [6,9]. Wearable sensors, being progressively more comfortable and less
obtrusive, are appropriate for monitoring an individual’s health or wellness without interrupting their
daily activities. The sensors can measure several physiological signals/parameters as well as activity
Sensors 2017, 17, 130 4 of 45

and movement of an individual by placing them at different locations of the body. The advancement
in low-power, compact wearables (sensors, actuators, antennas, smart textiles), inexpensive computing
and storage devices coupled with modern communication technologies pave the way for low-cost,
unobtrusive, and long-term health monitoring system.

2.1. Cardiovascular Monitoring System

Electrocardiograms (ECGs) represent a non-invasive approach for measuring and recording
the fluctuations of cardiac potential. This is the most widely used and effective diagnostic tool
that physicians have used for decades to identify heart-related problems such as different forms
of arrhythmias.

(a) (b)

(c)

Figure 2. Cardiovascular monitoring: (a) One cycle of a typical ECG signal (not scaled); (b) Electrode
placement in a standard 12 lead ECG system; (c) General architecture of ECG monitoring system.

Although many arrhythmias are not life-threatening, some results from weak or damaged heart
such as myocardial infarction (MI) that may lead to cardiac arrest, if not managed immediately [24–26].
After a heart attack, patients are required to receive immediate medical attention, which, otherwise,
may turn fatal. These complications can be avoided if any inconsistency in cardiac activity is detected
and treated in an early stage that calls for outpatient ambulatory monitoring of ECG. Some rare,
serious arrhythmias (e.g., Brugada Syndrome, Arrhythmogenic Right ventricular Cardiomyopathy,
Long QT syndrome, hypertrophic Cardiomyopathy) are infrequent and only detected on prolonged
monitoring. Figure 2a shows one cycle of a typical ECG signal. In a conventional 12 lead ECG system,
electrical activities of the heart along 12 particular spatial orientations are measured using ten Ag-AgCl
electrodes (hydrogel method/wet ECG), which are affixed to some specific parts of the body. Figure 2b
Sensors 2017, 17, 130 5 of 45

shows the placement of the electrodes in a standard of 12-lead ECG system. The electrodes contain
conducting gel in the middle of the pad that functions as a conduction medium between the skin
and the electrode. This conducting gel has potential toxic and irritant effects on the skin and is thus
not best suitable to use for long–term ambulatory monitoring system though currently, it is the only
system available [10,27]. However, only a few numbers of electrodes are used in ambulatory ECG
monitoring system at the cost of limited information (Figure 2c). A continuous ambulatory monitoring
device requires a wearable and portable system that could be used comfortably without affecting an
individual’s daily activities.
Andreoni et al. [28] designed a custom T-shirt and textile belts with embedded textile electrodes
for monitoring ECG, HR, and R-R interval. The electrodes were made from silver based conductive
yarns. Instead of using any conductive gel, the electrodes relied on body sweat, an electrolyte medium,
to improve the conductivity of the skin-electrode interface and signal quality. The device also included
a SpO2 sensor and a three-axis accelerometer for fall detection, and it could transmit the data over
low-power Bluetooth 4.0. An elastic fabric-made ECG vest was presented in [29] which accommodated
three electrodes, a data acquisition module and also supported robust contact of the electrodes with
the skin. The electrodes were fabricated from Ni/Cu coated compressed urethane polymer foam that
was enclosed by an Au-coated conductive taffeta fabric. The ECG measured by the proposed system
exhibited high correlation with the simulated signal, although measurements on real subjects were
not shown. Due to the conductive and flexible nature of the substrate, the skin-electrode impedance
was low and remained stable over a longer period of time, thus reducing the electrode motion noise.
Jeong et al. [30] proposed an ECG monitoring system where they used similar technology as [29] to
develop flexible capacitive electrodes and integrated them in a chest belt. In addition to that, they
used a very high bias resistor at the input of the pre-amplifier, which assisted further in reducing
electrode motion artifacts. A noise cancellation and peak detection algorithm was performed on the
raw ECG data to find out the QRS complex, and HR, although a detailed description of the algorithm
was not provided. The authors reported achieving high sensitivity and high accuracy in detecting
QRS complexes.
The electrodes used in the ECG systems presented in [28–30] were in direct contact with the skin.
Nemati et al. [10] embedded a small, low–power, wireless ECG monitoring system in a stretchable
belt where three capacitive electrodes were integrated into a cotton T-shirt, thus enabling ECG
measurements to be performed over the cloth. The cotton functioned as the dielectric material between
the electrode and the skin. The signal processing and communication modules were mounted on a
small two-layer PCB board. Power consumption was minimized by selecting low power electronic
components for the system, ANT protocol for wireless communication as well as by adopting idle
mode signal sampling technique. However, the rigid electrodes can be inconvenient to the users and
may induce motion artifacts in the signal.
In order to minimize the common mode interference, an additional driven right-leg (DRL)
electrode was used in [10,29,30]. This DRL electrode is usually placed at a distant site, far from
the measurement electrodes and thus requires a long wired connection, which may not be convenient
for wearable and long-term monitoring systems. Komensky et al. [31] proposed an ECG monitoring
system without the DRL circuit, where only two active capacitive sensors were embedded in an elastic
chest band. In order to increase and stabilize the input impedance, two anti-parallel connected diodes
were used for biasing that has advantages over resistors [30] of low thermal noise and fast recovery
time. The ECG measurements on stable subjects were reasonably well, although the P waves were
indistinguishable, which might be attributed to the electrodes’ position on the body or the absence of
the common electrode. On the other hand, the measurements during walking were greatly affected by
the motion artifacts, but the QRS complexes were still recognizable.
Many researchers have developed and made use of piezoelectric pressure sensors for measuring
the HR by sensing the arterial pulse wave generated by the periodic contraction and relaxation of the
heart. A wireless HR monitoring device was presented in [32] that could estimate HR from the pressure
Sensors 2017, 17, 130 6 of 45

variation in the ear’s canal surface. A piezoelectric film pressure was used to sense the in-ear pulse
waves (EPW) and convert it to an electric current. A knowledge-based algorithm was implemented
in a microcontroller that could detect the pulse peak in real time from the signal after performing
a morphological conversion. However, the pressure variance, and thereby the peak height of the
pressure waves can be affected by body movements that introduce error in HR estimation. In addition,
an ear-mounted device is inconvenient for long-term use. A similar system was proposed in [33] where
the authors developed a polymer-based flexible piezoresistive pressure sensor that can sense pressure
variation on the skin caused by the pulsation of arterial blood. They used carbon black/silicone rubber
nanocomposite as the flexible piezoresistive material. High sensitivity and linearity of the pressure
sensor was achieved by forming microstructures at the contact surface of two piezoresistive layers.
They also proposed a low-cost analog signal processing (ASP) system that could perform denoising,
data processing, and HR measurement.
Yoon et al. [34] designed a skin attachable piezoelectric pressure sensor and demonstrated its
usability in HR estimation by sensing the pulse wave in human artery. The pressure sensor was
fabricated on a polyimide substrate with a small window. A thermally evaporated silver electrode
was spin-coated with a polyvinylidene fluoride-trifluoroethylene (P(VDF-TrFE)) piezoelectric layer.
The pressure variation in the radial artery causes mechanical stress on the piezoelectric layer, resulting
in potential variation across the electrodes. Tajitsu et al. [35] embedded a piezoresistive pressure
sensor in a wristband for HR monitoring. The piezoresistive material was made from nonwoven
acrylate-modified polytetrafluoroethylene (PTFE) fabric that was fabricated using electro-spinning.
The PTFE was deposited on an aluminum electrode on a polyethyleneterephthalate (PET) film.
The pulse wave measured from the wrist by this sensor had similar pattern as the ECG signal and
showed high accuracy as well as less vulnerability to motion-induced noise.
Some researchers have exploited system-on-chip (SOC) technologies to integrate both analog and
digital signal processing units for on-chip ECG signal processing. Izumi et al. [36] developed a wearable
system that incorporated a near field communication (NFC) module, a three-axis accelerometer,
and an ECG processor chip. The chip was designed to perform data acquisition, process ECG
and accelerometer signals, and communicate with the smartphone. The R-peak detection and HR
estimation was performed by utilizing short-term autocorrelation (STAC) between a template signal
and the measured signal. The chip was fabricated using a standard 130-nm CMOS technology.
The system was reported to consume ~13.7 µA current and perform monitoring for about 24 days
using a 35 mAh battery. An ultra-low-power ASIC was designed for cardiovascular monitoring in [37],
which was fabricated using a standard 0.18 µm CMOS technology and encompassed a two-stage
Miller-compensated programmable gain amplifier (PGA), QRS and baseline amplifiers, DC voltage
generator and a comparator. The PGA offered wide dynamic range, self-biasing capability, and low
supply voltage requirement. The ‘QRS Amp’ and a ‘Baseline Amp’ filtered the signal and isolated
QRS signal from the baseline drift. A DC voltage was added to the baseline drift and R–peaks were
detected by comparing the QRS complex signal with the shifted baseline. The system required only
58 nW of power and can operate continuously for one year with a 0.7 mAh thin-film battery, thereby
making it suitable for long-term monitoring applications. Helleputte et al. [38] proposed a design
of a 3-channel bio-potential acquisition integrated circuit. Each channel measures ECG as well as
electrode-tissue impedance (ETI), which is found to be strongly correlated with the motion artifacts.
Motion artifacts are estimated in real-time using an adaptive LMS filter and subtracted from the ECG
signal before amplification.
The system was reported to consume low power and suppress slow varying (<10 Hz) motion
artifacts in ECG. Table 2 presents a comparison among the cardiovascular monitoring systems
discussed above.
Sensors 2017, 17, 130 7 of 45

Table 2. Comparison among cardiovascular monitoring systems.

Signal Acquisistion Module

Proposition Moni-Tored Electrode Electrode Attachment Wireless Accuracy Bat. Life,
Ref. Signs Type Active Material Connectivity Size Freq. A/D
Size Method Power
Sensorized T-shirt Dry textile Silver based
[28] ECG, HR Snap buttons Bluetooth LE - 512 Hz 24 bit -
and textile belt electrodes conductive yarns
99.51% correlation
Ni/Cu coated
Wearable mobile Dry foam compressed Bluetooth with prerecorded 33 h,
ECG, HR, 14 mm × 8 4 cm × 2.5 cm
[29] electro-cardiogram location mm × 8 mm v2.0, and ECG data, × 0.6 cm 512 Hz 12 bit 1100 mAh
monitoring system electrode urethane GSM QRS detection Li-ion battery
polymer foam accuracy ~98.14%
Capacitive
Wireless, portable 33 mm × 33 Woven under 45 mm × 60 mm × 15 h with
[10] capacitive ECG ECG, HR electrode with a stretchable ANT 500 Hz 10 bit 256 mAh 3 V
cotton mm × 2 mm 9 mm
sensor belt Li battery
insulator
Use of flexible Ni/Cu coated 300 mm × Upto 91.32% QRS
capacitive Flexible foam Integrated
capacitive 20 mm × detection at
[30] ECG, HR (polyolefincovered (1.1 ± 0.2 into a Bluetooth - 256 Hz - -
electrodes for chest belt 7 km/h
electrodes mm) walking speed
reducing MA by polyurethane)
Common Active
Electrode-FreeECG ECG, HR capacitive 5 cm × 3 cm Adhesive - - -
[31] Copper layer 2 kHz 24 bit
monitoring System tape
electrodes
HR monitoring
from pressure Piezo-electric 3.5 mm × Earpiece like Sensitivity 97.25%, Coin-cell
[32] HR 2.4 GHz RF 15 mm × 17 mm 100 Hz 12 bit
variance film sensor 3.5 mm device PPV 97.18%. battery
in ear canal
C black/silicone
rubber
Heart Rate Piezo-resistive nanocomposite 15 mm ×
Monitoring with HR pressure Embedded in - Accuracy > 97% - - - -
[33] 30 mm
pressure sensor sensing encapsulated in elastic belt
conductive
FCCL films
Sensors 2017, 17, 130 8 of 45

2.2. Activity Monitoring System

Monitoring an individual’s physical activities and locomotion can be useful in rehabilitation,
sports, early detection of musculoskeletal or cognitive diseases, fall and balance assessment. It has been
reported that an individual’s walking patterns are strongly associated with their health condition [39].
Walking involves several joints including spine, hip, knee, ankle, tarsal and metatarsal joints. It equally
involves several muscles, for example, muscles of back, around hip joints, thigh, calf muscles and
several small muscles of foot. A typical walking cycle is shown in Figure 3. Walking, especially
turning event requires good balance and coordination among different parts of the body, which is
controlled by the cerebellum. Therefore, any abnormality in walking patterns can be indicative of
possible musculoskeletal, central nervous system (CNS) or peripheral nervous system diseases.

Figure 3. A typical human gait cycle.

The walking patterns of ailing people tend to differ from that of normal healthy people.
For example, people at the early onset of neurodegenerative disorders such as Alzheimer’s or
Parkinson’s tend to exhibit different walking patterns [40,41]. One of the possible early signs of
Parkinson’s disease is small and shuffled walking steps. Besides that, a person at the early stage
of Parkinson’s may experience difficulties in starting, stopping and turning events while walking.
They may show loss of associated movements. On the other hand, elderly people, owing to their
declining motor control and muscle strength are usually more vulnerable to fall, which, if occurs, may
cause joint injuries, hip and bone fractures and traumatic brain injury. These injuries demand longer
recovery time, restrict physical movement and affect the daily activities of the individual. Indeed
there is a strong correlation with mortality and fall-related fractures [41]. Quantitative analysis and
assessment of the gait can be useful for early detection of several diseases, fall prediction as well as
during the rehabilitation period after an injury.
Home-based fixed position monitoring, for example, camera-based systems are useful tool for
activity monitoring [42,43]. These systems are capable of recognizing complex gait activities. However,
such systems restrict the movement of the user within a specific range. Apart from that, these systems
are complex and expensive. In recent years, use of wearable motion sensors such as accelerometers,
gyroscopes, and magnetometers are gaining in popularity for measuring human gait activities [6,11]
in real time. The sensors measure linear and angular motion of the body from which a number
of key features are extracted. Table 3 presents a list of key features that can be extracted from the
signals. These features are used to quantify and classify human gait events. A schematic of the activity
monitoring system based on accelerometers and gyroscopes is presented in Figure 4.
Sensors 2017, 17, 130 9 of 45

Table 3. Features extracted from motion signal.

Spatial Domain Temporal Domain Frequency Domain Statistical Domain

Step length Double support time Spectral power Correlation
Stride length Stance time Peak Frequency Mean
Step width Swing time Maximum spectral amplitude Standard deviation
RMS acceleration Step time Covariance
Walking speed Stride time Skewness
Cadence (steps/min) Kurtosis
Energy

Derawi et al. [44] implemented an activity and gait recognition system on a smartphone.
They measured the gait data with the help of the accelerometer in the smartphone. An application
software was developed for the smartphone that performed detection, normalization, averaging of
gate cycles, activity, and gait recognition. The algorithm used Manhattan distance metric to compare
the average gait cycle of a test sample to three different template gait cycles that correspond to three
different walking speeds. The authors exploited both statistical and machine learning approaches in
order to classify among three different walking speeds and achieved high accuracy from the support
vector machine (SVM) approach. However, these methods rely on local peak and valley detection,
which is sensitive to variations in walking speed and/or style. Debraj et al. [45] proposed an activity
recognition system that used multi-modal sensors to detect complex daily activities. Each measurement
unit, which was a smartphone in this case, comprised an accelerometer and a gyroscope for activity
measurement; temperature, humidity, and barometric pressure sensors for environment sensing; and
Bluetooth beacon for location assessment. Four such measurement units were attached to waist,
back, leg, and wrist. After extracting a set of suitable features from the preprocessed sensor data,
classification was performed separately on each unit using a modified conditional random field (CRF)
algorithm. Final recognition was performed by assessing the classifier decisions from each unit based
on their relevance to the body positions. The authors reported on classifying 19 in-home activities
including using refrigerator, cleaning utensils, cooking, sitting and eating, using the bathroom sink
along with normal daily activities with high accuracy. Further study and development are required in
order to realize it in a fully wearable system and implement it for remote monitoring of the elderly.
In addition, more investigations on machine learning classifiers are necessary to achieve a more robust
recognition algorithm.
Bertolotti et al. [46] designed a lightweight, wireless wearable device for assessing the balance
control abilities of the body by measuring the limb movements for a longer period of time using
an accelerometer, a magnetometer, and a gyroscope. Several units can also be connected in a body
sensor network (BSN) for achieving more detailed measurements. The reliability of the system was
validated by comparing the center-of-mass (CoM) displacement estimated from the measurement
with that obtained from a Wii Balance Board (WBB). Further research is necessary in order to extract a
suitable set of features from the measurement data and classify subjects in terms of fall vulnerability.
Panahandeh et al. [47] proposed a human activity and joint classification and gait analysis algorithm
based on continuous hidden Markov model (HMM). They devised a chest mounted activity
measurement system that performed inertial measurements using one tri-axial accelerometer and
gyroscope. They proposed a feature extraction method, which is based on calculating discrete Fourier
transform (DFT) coefficients from segments of the signal. The authors achieved high classification
accuracy among different walking events. Due to the absence of wireless connectivity, the subjects had
to carry the laptop during the experiment that could affect the normal human locomotive behavior.
Besides this, the feature extraction was performed in an ad-hoc basis. The system can be further
improved by incorporating wireless connectivity and developing better feature extraction algorithms.
Sensors 2017, 17, 130 10 of 45

Figure 4. Schematic representation of activity monitoring systems.

Chia Bejarano et al. [48] proposed an adaptive algorithm for real-time gait-event detection.
They attached both inertial motion and magnetic sensors on the shanks and measured the angular
velocity and flexion-extension angle for each leg. After initial calibration, the algorithm employed
a threshold-based state machine in order to detect three gait-events: initial contact, end contact and
mid-swing for each leg at different walking speeds. This algorithm exhibited high detection accuracy
as well as low detection delay, thus making it suitable for real-time applications. However, further
validation with a larger number of subjects is necessary. In addition, the measurement from the
magnetometer that was used by the Kalman filter can be affected in the presence of ferrous materials
which can cause false detection. Another inertial measurement unit (IMU)-based gait recognition
algorithm was presented in [49] where three sets of IMUs were placed on the center, left, and right of
the back using a waist belt. The step signals were first segmented from the likelihood of heel strike
computed using a scale-space technique. An algorithm based on inter-class relationships extracted
the feature vectors. It also performed tilt correction using the gyroscope on the IMUs and employed
an iterative matching algorithm to resolve the inconsistency associated with the sensor orientation.
The researchers carried out experiments on a large number of subjects and reported an average
classification accuracy of 93%.
An intensity-independent activity recognition method was proposed in [50] that utilized the
uncertainty among the clusters. They selected a fixed set of features from the accelerometer data
during the training phase and clustered the data into group of activities by employing K-means or the
Gaussian mixture model (GMM) algorithms. The mean and variance of each activity cluster were used
to form a stochastic activity model (SAM) matrix. In the recognition phase, the appropriate group for
the test sample was determined by employing the nearest-neighbor algorithm on the SAM matrix.
Test results showed that the system could discriminate walking and running across three different
intensity levels with a high degree of accuracy. Ghasemzadeh et al. [51] proposed a real-time feature
selection algorithm for wearable systems, which was designed to minimize energy requirements
during classification. This algorithm identified and discarded irrelevant and redundant features, thus
Sensors 2017, 17, 130 11 of 45

increasing the learning speed and optimizing the power requirement for the system. The redundancy
analysis was performed by using symmetric uncertainty among the features producing strongly
correlated features. A graph model was deduced from the correlation analysis, which was used by
an integer programming, and a greedy approximation based optimization algorithm to find out the
optimal features. They used six sets of IMUs for movement monitoring and reported to obtain 30% of
energy savings while achieving 96.7% classification accuracy among 14 sets of movements. Further
development of algorithms for dynamic feature selection as well as sensor unit activation may result
in an improved power-optimized system.
A walking-phase dependent parameter optimization algorithm has been presented in [52] for
efficient classification of locomotion modes. The walking phase was detected from the signals of the
pressure sensors embedded in insoles. The gait signal measured by two IMUs and pressure insoles
were segmented into four walking phases. The authors, then optimized the feature vector, classification
algorithm, and window size separately for each phase. This algorithm exhibited high recognition
accuracy (~96.5%) and fast computation time when the performance of the complete gait cycles is
considered for optimization as compared to the conventional methods of evaluating each phase of
a gait cycle separately. Cristiani et al. [53] devised an electronic insole for long-term monitoring of
motor activities. The insole housed a tri-axis accelerometer, humidity and temperature sensors, and
four pressure sensors. The sensors were connected to a microcontroller having an integrated 2.4 GHz
transceiver in it. The onboard flash memory allows storing up to 10 h of measurement data. The insole
was comfortable to use in regular motor activities and thus can be used to monitor a subject’s movement
for longer periods of time. However, the insole requires initial calibration for each individual in order
to make an inclination correction. Integrating a gyroscope in the insole for inclination measurement
might be useful to avoid this initial calibration phase. Another shoe-based activity monitoring system
was presented in [54] where recognition is performed by rejecting unreliable data while employing the
classifiers. A set of nine features were calculated from each measurement obtained from each of five
insole pressure sensors as well as from the accelerometer, which was attached to the heel of the shoe.
The authors tested both SVM and multilayer perception (MLP) classifier and rejected those data points
residing near the cluster boundaries. Very high recognition accuracy (99.8% ± 0.1%) was achieved by
applying MLP on raw measurement data. However, the tests were performed only on nine subjects.
Besides this, the rejection threshold was determined in a heuristic manner, whereas an adaptive way
of calculating the threshold could be more useful for practical applications.
Friedman et al. [55] designed a wearable wrist and finger joint monitoring system using magnetic
sensing technology. The system comprised of a neodymium ring, a sensing and data storage unit.
The ring worn on the index finger generated a magnetic field, which was measured by two tri-axial
magnetometers mounted on the wrist-worn sensing unit. A radial basis function network estimated
the angles of the wrist and finger joints from the measurements. The authors reported a highly accurate
estimation of angular distance for wrist joints. However, with a high standard deviation, finger
flexion-extension estimation was poor. This estimation can be improved by ensuring accurate initial
calibration for each user. An approach for human joint angle estimation by combining kinematic arm
models with the state space algorithm has been presented in [56]. Inertial measurements were obtained
from three sets of IMUs attached on the upper arm, forearm, and, wrist of a robot arm. The state model
incorporated random drift models and zero-velocity updates that reduced the effect of sensor drifts.
The model also considered the physical constraints of joint movements in order to achieve higher
estimation accuracy. The authors estimated joint angles using the Unscented Kalman filter (UKF) [57],
an improved algorithm over the Extended Kalman filter (EKF) especially developed for systems with
higher nonlinearities. They achieved a high degree of estimation accuracy at different intensities of
arm motion. Nevertheless, validation of this algorithm on human subjects is necessary. It should be
noted that Kalman-based solutions are computationally intensive and also need a high sampling rate
in human activity monitoring applications. Therefore, realizing the systems for real-time applications
can be challenging.
Sensors 2017, 17, 130 12 of 45

Table 4. Comparison among activity monitoring systems.

Classification Sensor
Ref. Proposition Feature Extraction Sensors Placement Com. Tech. Detection Accuracy Power Req.
Method
Support Vector
Activity and gait Accelerometer is
recognition system Machine (SVM),
[44] Fixed set of features embedded in Different walking speed >99%.
Bayes network, and
on a smartphone smartphone
Random Tree
Walk and run indoors, use
refrigerator, clean utensil,
Smartphones’ (Samsung cook, sit and eat, use
Galaxy S4) onboard bathroom sink, move from
In-home,
sensors (accelerometer, indoor to outdoor, move from
fine-grained gyroscope, barometer,
Waist, lower 19 in-home
activity recognition Conditional back, thigh, USB outdoor to indoor, walk
[45] Fixed feature set random field (CRF) upstairs, and walk activities with
temperature and, >80% accuracy
multimodal and wrist
wearable sensors humidity sensor), along downstairs, stand, lie on the
with Gimbal bed, sit on the bed, lie on the
Bluetooth beacons floor, sit on the floor, lie on
the sofa, sit on the sofa, and
sit on the toilet
Wearable device Accelerometer, Balance hazards, balance
Fixed set Streaming ~6 h
[46] based on a gyroscope, and Limb or trunk Bluetooth monitoring for High correlation
of features Logging > 16 h
9-DOF IMU magnetometer fall prediction
Walking, running, ascending
Algorithm Time-Frequncy Hidden Markov 3-axis accelerometer, upstairs, descending
[47] Chest USB ~95%
development domain analysis Model 3-axis gyroscope downstairs
and standing
Two inertial and Gait events: Initial Contact
A real-time,
adaptive algorithm magnetic sensors (IC), End Contact (EC) and
External part of Mid-Swing for both right and F1-scores 1(IC, EC), 0.998 (IC)
[48] (1 IMU = 1
for gait-event both shanks and 0.944 (EC) for stroke subjects
accelerometer, left leg while walking at three
detection 1 gyroscope) different speed
Recognition Support vector 3 IMUs (each IMU: Fixed at the back, Walking on flat ground,
Inter-class relation ~93% average
[49] method for similar machine, K-nearest 1 tri-axis accelerometer, left, and right up/down stairs, and
Ship 1 tri-axis gyro) waist
gait action neighbor up/down slope
Stochastic K–means and 3 intensity level of walking:
Belt-like strap
[50] approximation Fixed set of features Gaussian Mixture Accelerometer 93.8%; 3 intensity level of
Models around the waist
framework running 95.6%
Switching between stand and
sit, sit and lie, bend to grasp,
Power-aware Minimum cost 6 IMUs (each IMU has Waist, right
feature selection by rising from bending, kneeling 30% energy savings
feature selection for one three-axis wrist, left wrist,
[51] using a K-nearest neighbor BSN right, rising from kneeling, with 96.7%
minimum accelerometer and a right arm, left
processing energy two-axis gyroscope) look back and return, turn accuracy
redundancy graph thigh, right ankle
clockwise, step forward and
backward, jumping
Parameter
optimization IMUs on the
2 IMUs, 2 pressure Walking, up/down stairs,
strategy for Fixed set shank and the
[52] insoles (each having shoe, pressure and up/down slope, passive 88%–98%
phase-dependent of features
4 pressure sensors) sensors insole mode
locomotion mode
recognition
Sensors 2017, 17, 130 13 of 45

Table 4. Cont.

Classification Sensor
Ref. Proposition Feature Extraction Sensors Placement Com. Tech. Detection Accuracy Power Req.
Method
Electronic insole for Humidity and Foot accelerations,
wireless monitoring Fixed set temperature sensors, orientation in space,
[53] Insole ZigBee 10 h of data logging
of motor activities of features accelerometer and temperature and
and shoe comfort 4 pressure sensors moisture data
Support vector
Shoe-based activity Five pressure sensors PS on insole and Sit, stand, walk, ascend stairs,
monitoringsystem Fixed set machine, 99.8% ± 0.1% with
[54] multilayer (PS) and one 3-D accelerometer on descend stairs
of features heel of shoe and cycling MLP
(smartshoe) accelerometer
perception (MLP)
A wearable device
for monitoring Watch-like enclosure worn on the
Fixed set K-means 2 tri-axial wrist and a small neodymium ring 92%–98% with a 20.5 mA at 3.3 V
[55] Finger and wrist movement
daily use of the of features magnetometers 19%–28% STD
worn on the index finger
wrist and fingers
Shoulder internal/external
Combined Upper arm, rotation; flexion/extension of
kinematic models shoulder, elbow, Average RMS angle
[56] Unscented Kalman filter 3 IMUs forearm,
to estimate human and wrist, error ~3◦
joint angles and wrist supination/pronation of
forearm, wrist twist
Wearable device
with automatic gait
3 IMUs (each IMU has a On feet for gait
and balance Gait parameters
3-d accelerometer, a analysis on waist
[58] analyzing Fixed set of features 30 mA at 3.7 V
uni-axial gyroscope, and for balance and balance
algorithms for
a biaxial gyroscope analysis
Alzheimer
patients (AP)
Backward fall, forward fall, Accuracy:
Accelerometer, 90.37%–100% 15 mA–34 mA
IMU based fall Waist lateral left fall, lateral right
[59] Madgwick orientation filter gyroscope, and Bluetooth Sensitivity: using 3.7 V
Detection system magnetometer fall, syncope 80.74%–100%
Sensors 2017, 17, 130 14 of 45

Hsu et al. [58] implemented an algorithm for gait and balance analysis in an IMU based wearable
system. For gait measurements, two IMUs were attached to the top of each shoe, and one IMU was
attached to the back of the waist for balance analysis. After the detection of strides, the gait cycles were
further decomposed by using the toe-off and heel-strike points obtained from the pitch signals of the
gyroscope. A number of gait parameters corresponding to walking speed, rhythm, and, variation were
calculated from the decomposed signal. The balance was measured by calculating the sway speed
from the center of mass (COM) analysis. A wearable fall detection system was proposed in [59] that
could determine fall events by employing acceleration and orientation thresholds. The acceleration
thresholds were obtained at the training phase from SVM, and the postural orientation thresholds
were determined from the subject’s tilt angle. The system used Madgwick’s orientation filter for
reducing magnetic distortion and gyroscope drift, resulting in high estimation accuracy. The IMU
was placed on the waist and could communicate over Bluetooth. The system analyzed the RMS data
obtained from the accelerometer and the orientation filter and could detect fall events using a threshold
based algorithm. This allows implementing the algorithm for real-time applications in a low profile
microprocessor. The algorithm was reported to achieve a high degree of accuracy and sensitivity.
Table 4 presents a comparison of the key features and performance characteristics among the activity
monitoring systems discussed above.

2.3. Body Temperature Monitoring System

Body temperature is one of the vital signs that can reflect health conditions. Body temperature
increases in infections, malignancy and many inflammatory conditions. Only serial temperature
measurements over a long period of time rather than spot checks may prompt the diagnosis. It was
reported that core body temperature (CBT) has strong influence on different physiologic conditions as
well. Disruptions in the body temperature rhythm are reported to be associated with different types of
insomnia [60]. For example, patients suffering from delayed sleep phase insomnia have ~2 h of delay
in reaching their minimum CBT compared to the group of good sleepers [60]. Besides this, variation
in body temperature rhythm with menstruation cycle was also observed in some studies [61,62].
Researchers [63] also reported observing a correlation between body temperature and initial stroke
severity, infarct size, mortality among stroke patients. It was observed that the infarct size worsens
by ~15 mm with 1 ◦ C increase in body temperature. In addition, a strong correlation between body
temperature and cognitive functions was also reported in the literature [64,65].
Various noninvasive approaches for continuous body temperature monitoring were reported in
the literature. Buller et al. [66] proposed a Kalman filter-based model that is inspired from the work
reported in [67] where the authors estimated core body temperature from the heart rate. Although the
model is validated with only a few subjects and accumulates error with time, it has the potential to be
embedded in wearable ECG monitoring systems. Further study and development are also necessary in
order to validate the estimation accuracy in the presence of rapid physical movement of the subject, for
example, running, exercising etc. A noninvasive, dual-channel body temperature measurement system
was reported in [68] that can measure temperature with an accuracy of ±0.1 ◦ C within the range
of 16–42 ◦ C and also transmit data over a Bluetooth communication platform. The device has two
temperature probes, each containing a digital temperature sensor. The probes measure temperature
from two ear canals simultaneously from which the mean temperature is calculated. However, because
of the ear-canal probes, the system may not be feasible for long-term monitoring.
Boano et al. [69] demonstrated an unobtrusive, wireless body temperature monitoring system that
can be worn by an individual for a long period of time. Two sensor units were attached to the skin that
can measure and send data to a more powerful body-worn central unit, thus forming a star-type body
sensor network (BSN). The central unit sends the data to a computer that can communicate with health
care facilities over the internet. The authors achieved an accuracy of 0.02 ◦ C over the temperature
range of 16–42 ◦ C. They also demonstrated the detection of circadian rhythm using this system.
The authors later developed a wireless core body temperature monitoring system for marathon runners
Sensors 2017, 17, 130 15 of 45

that measures the tympanic temperature [70]. A non-invasive, wearable temperature monitoring
systems was designed for neonates in [71]. They integrated a negative temperature coefficient (NTC)
resistor in a belt made of soft bamboo fabric. Instead of hard wires, the authors used soft and flexible
silver plated nylon yarns as the conductive medium woven in the belt. The system exhibits an accuracy
of 0.1 ◦ C when compared with the measurement obtained from a standard thermometer. A reliable
and stable connection between fabric wires and the sensor could is critical and was not investigated
in this research. However, the designed system can be integrated in a smart jacket or smart belt.
Thus, it has a great potential to be used in non-invasive, long-term monitoring, even for adults.
Mansor et al. [72] demonstrated the implementation of a wireless body temperature monitoring system
using comercial sensors. The temperature sensor, which comes with an integrated ZigBee wireless
node, measures and transmits data to a microcontroller. The microcontroller sends data to a remote
server over a wireless local area network (WLAN). The authors used Arduino Ethernet shield for
developing the prototype. Similar temperature and heart rate monitoring systems were reported
in [73,74].
Some researchers exploited RFID technology for body temperature monitoring systems.
Vaz et al. [75] designed and fabricated a low power RFID temperature sensor chip in 0.35-µm CMOS
process. The chip along with a matched impedance dipole antenna measures the temperature with an
accuracy of ~±0.1 ◦ C within the typical range of human body temperature and can communicate with
a 2 W ERP (effective radiated power) output reader at a frequency of 868 MHz from a distance as far
as 2 m. A RFID based real-time, continuous body temperature measurement system that is fabricated
on a poly ε-caprolactone (PCL) membrane was developed by Milici et al. [76]. The system comprises a
wearable RFID tag operating in the ultra-high frequency (UHF) band and a RFID microchip, EM4325
that has the capability of measuring temperature with a resolution of 0.25 ◦ C as well as functioning
as a regular RFID transceiver. The measured temperature was read by a short range RFID reader,
connected with a linear-polarized antenna. Although the linear-polarized antenna offers longer read
range, they are very sensitive to tag orientation. A circular-polarized read antenna could circumvent
this problem.
In order to measure the core body temperature, Sim et al. [77] designed a system by embedding a
dual-heat-flux probe and two double-sensor thermometers [78] into a neck pillow. Since the jugular
vein passes by the neck, the skin temperature over it would have a strong correlation with the core
body temperature (CBT). The temperature estimated by this system was found to be similar with that
measured by an IR thermometer from the tympanic membrane. The temperature curves obtained
from three sensors are distinct and also vary with different sleeping positions which may provide
sleep-related information. The authors also proposed a curve-fitting method in order to improve the
inherent slow response time of the dual-heat-flux thermometer. Although the system is designed
for patients to use in beds, embedding the system in the collar of the shirt or textile band is also
possible. Kitamura et al. [79] developed a temperature sensor probe that can measure CBT from the
surface of the skin. The circular metal probe comprises two heat-flow channels for two different
thermal resistances and each channel has a pair of temperature transducers attached at both ends.
The presented system demonstrates a long initial response time but higher accuracy (97% correlation
with the measurement from zero-heat-flow thermometer) once it reaches the equilibrium temperature.
The response time can be improved by using an Al probe instead of the Cu probe. Table 5 presents the
comparison among the body temperature monitoring systems discussed above.
Sensors 2017, 17, 130 16 of 45

Table 5. Body temperature monitoring systems.

Used Device for Location Wireless Accuracy

Ref. Proposed Device Principle Measured Parameters Measurement Connectivity Performance Evaluation
Compared with data
Kalman filter based body temp. Temperature variation Ag/AgCl gel
[66] HR, skin temperature Chest - from ingestible RMSE: 0.40 ◦ C
estimation model with HR electrodes
temperature capsule
Mean of
Wireless, dual channel body Digital temp ±0.1 ◦ C
[68] temp measurement system measurements from Core body temp sensor DS18B20 Ear canal Bluetooth
two ear canals
RF (Tyndall node)
Wearable wireless temperature Circadian rhythms, MF51E NTC Compared with data from
[69] Two-point calibration thermistor Skin over Body sensor 0.02 ◦ C
monitoring Skin temp a thermometer
network (BSN)
Embedded NTC temperature NTC Mon-A-Therm Compared with data from the
[71] sensor and conductive ECG, skin temperate 90045 and Shieldex® Skin - ±0.1 ◦ C
textile wires in a belt made with Silver Plated NICU sensor connected to the
soft bamboo Nylon yarn Solar® 8000M patient monitor
Wireless body LM35 ±0.25 ◦ C
[72] Skin temperature Hand ZigBbee and WLAN
temperature monitoring
Temperature
Tag and reader
[75]
RFID sensor chip in 0.35-µm dependence of the Measurement was performed in ~±0.1 ◦ C Resolution:
communicate 0.035 ◦ C
CMOS standard process frequency of at 868 MHz a climate chamber
ring oscillator
Epidermal-like RFID Tag and reader
Re-tunable epidermal communicate within a Compared with data from
[76] tag made on a Poli-caprolactone tag Skin temperature EM4325 Abdomen ±0.25 ◦ C
band PT104 thermocouple
membrane of 780–950 MHz
Deep body temperature Embedding 1
Compared with data from
Dual-heat-flux, Core body - -
[77] measurement system embedded Around neck infrared thermometer
2 double-sensor in temperature
in a neck pillow (thermoscan IRT 4520)
neck pillow
Heater-less deep body Dual-heat-flux Core body Compared with data from
[79] Forehead - Correlation: 97%
temperature probe method temperature zero-heat-flow thermometers
Sensors 2017, 17, 130 17 of 45

2.4. Galvanic Skin Response (GSR) Monitoring System

The autonomic nervous system (ANS) controls and regulates the response of the body to
internal or external stimuli by balancing the activities within its two subdivisions: sympathetic and
parasympathetic nervous systems [80]. The parasympathetic system, which is also termed as “rest and
digest" system conserves and restores energy of the body. On the other hand, the sympathetic system
triggers what is often referred to as fight or flight response by increasing metabolic output to deal
with the external stimuli. Increased activity of the sympathetic system accelerates heart rate, increases
blood pressure, and sweat secretion, as well as prepares the body for motor action by pumping more
blood to muscles, lungs, and brain.
Although currently, this information is of no significant clinical use, there is growing interest in
many conditions including dysautonomia, Postural orthostatic tachycardia syndrome (POTS), and
Inappropriate tachycardia syndrome. Increased sweat secretion from eccrine glands fills the sweat
ducts. Sweat, being a weak electrolyte, increases the conductance of the skin with increased secretion.
Therefore, variation in skin conductance (Figure 5), which is also referred to as electro-dermal activity
(EDA) or galvanic skin response (GSR), reflects the activity of the sympathetic nervous system and
provides a simple, sensitive and reliable parameter for assessing the sympathetic nervous activities
associated with stress and emotion [81,82].
Skin Conductance S)

Figure 5. Typical galvanic skin response (GSR) signal (not to scale).

Usually, GSR is measured from the part of the skin having a large number of sweat glands such
as the palm, fingers, or soles of the feet. In active measurement, a DC voltage is applied across two
on-body electrodes and the skin conductance is obtained from Ohm’s law by measuring the current.
The earlier researches in this field mostly focused on time-limited GSR measurement systems used in
laboratories and health care facilities. The development of low-power and wearable technologies opens
up a new window for unobtrusive GSR monitoring that can be worn for a longer period of time [83].
Long-term monitoring of GSR allows to observe and assess the response of the sympathetic nervous
system for a longer period of time and can potentially unfold important physiological information
that cannot be obtained by limited time monitoring. Besides this, wearable GSR monitoring system
allows patients to monitor GSR level in their home environment to provide a better assessment of their
psychophysiological condition than the evaluation made at laboratories or hospitals from short-term
measurement [83,84]. A schematic diagram of a wearable GSR monitoring system is presented in
Figure 6.
A low-cost GSR sensor has been presented in [85] that was embedded in a wristband. The sensor
module is small in size (20 mm × 30 mm × 0.8 mm) and accommodates measurement, processing,
and communication functionalities. The wrist-worn sensor measures GSR from the dorsal forearm and
transmits the data with a Gaussian Frequency Shift Keying (GFSK) transceiver. The elastic nature of
the wristband provides stable, consistent and undisturbed electrode-skin interface, thus minimizing
motion artifacts in the GSR measurements. Sugathan et al. [86] integrated a set of non-invasive sensors
in a shirt that facilitates real-time measurement of HR, GSR, and body temperature. They used an
Arduino-based wearable computing device (LilyPad) as the primary computing platform. Integration
Sensors 2017, 17, 130 18 of 45

of storage and wireless communication modules could make the system more feasible for long-term
and remote health monitoring applications.
A wearable GSR sensor has been reported in [87] that can conduct measurement from the
back of the body and transmit data over the Bluetooth platform. The sensor was fabricated on a
flexible PCB, which was covered with silicon providing stable contact with the curved body surface.
Here [87] a dry conductive polymer foam was used as the sensing material for the flexible electrodes.
The flexible nature of the electrodes offers stable and reliable skin-electrode interface as well as wearing
comfort to the user. The GSR measured by the reported system had good correlation (average ~0.768)
with the reference GSR system, although the validation was performed on a very small number of
subjects. Garbarino et al. [88] developed a multi-sensor wristband (Empatica E3) that included GSR,
PPG, temperature, and motion sensors. The GSR sensor offered high dynamic range measurement
between 0.01 µS to 100 µS at 900 pS resolution. The data acquisition device has a dimension
of 4 cm × 4 cm and was embedded in a wristband. It was capable of measuring and logging data
from all the four sensors continuously for 38 h. The system could also stream data in real time
using Bluetooth LE wireless medium. The long battery life, low power wireless connectivity, and
multi-parameter monitoring capability are key features of the proposed system.

Figure 6. Schematic diagram of the GSR monitoring system.

Guo et al. [89] presented a GSR monitoring system and developed an algorithm to identify
human emotions. A 20-dimensional feature vector was extracted from the pre-processed GSR signal.
The authors then employed Sequential Floating Forward Selection (SFFS) algorithm on the feature
vector for classifying the data into four sets of emotions: amusement, fear, sadness, and relaxation.
However, the method was validated only on a few subjects. An investigation on GSR data was
performed by Setz et al. [90] in an attempt to distinguish cognitive load and stress. The authors
separately measured GSR from the fingers of a number of subjects working in two artificially created
psychological conditions mimicking cognitive load and stress. A wearable monitoring system was
used that measured and transmitted GSR signal via a Bluetooth wireless link. It was observed that
among 16 selected features of GSR signal distributions, the GSR peak height and the instantaneous
peak rate provided better result in differentiating cognitive load from stress. Crifaci et al. [91] studied
the feasibility of wearable, wireless HR and GSR monitoring system in assessing the stress level and
the function of an autonomic system in ambulatory condition. ECG and GSR measurements were
taken from two groups of young adolescents. The authors performed statistical analysis on the features
extracted from the ECG and GSR. The group of healthy people exhibited significantly higher variance
and standard deviation in their GSR level compared to the group of people affected by the eating
disorder anorexia nervosa. Subramanya et al. [92] performed an investigation on the correlation
between blood pressure (BP) indexes and GSR exists. The authors implemented a wearable GSR
monitoring system based on the design presented in [93] and attempted to correlate the measured GSR
data with four BP indexes: systolic BP (SBP), diastolic BP (DBP), mean arterial pressure (MAP), and
pulse pressure (PP). The study showed that among the four indexes PP is highly correlated with GSR.
Research [94] showed that PP is a vital indicator of cardiovascular risks and a 10 mm Hg increase in
PP causes approximately 20% increase in cardiovascular death. Therefore, the correlation observed
in [92] between PP and GSR could be significant for cardiovascular assessment. Table 6 presents the
comparison among the GSR monitoring systems discussed here.
Sensors 2017, 17, 130 19 of 45

Table 6. GSR monitoring systems.

Electrode Measurement Wireless Battery

Ref. Proposition Location Size Sampling Rate A/D Life/Power Req. Evaluation Accuracy
Type/Device Connectivity
A small wristband for Measurement
unobtrusive and 2.4 GHz
Ag/AgCl transceiver 70 mm × 70 mm compared with overall
continuous EDA 32 Hz 1199 mAh,
[85] Dorsal forearms module × 20 mm 12 bit
3.7 V LiPo
correlation:
measurements during electrodes commerecial 93%–99%
(nRF2401) system.
everyday activities
An ambulatory device Arduino based e-textile
lilypad platform (SHT15 Not Supply voltage:
[86] for measuring HR, GSR, implemented 2 V to 5 V
and skin temperature for T measurement)
compared
Highly wearable and flexible dry average
42.5 mm × thesignal with
[87] reliable galvanic skin polymer foam Back Bluetooth 38.5 mm 10 bit Correlation:
a finger 76.8%
response (GSR) sensor Ni/Cu
reference GSR
Wearable multi-sensor
device for real-time resolution 900 pS
[88] Ag electrodes Bluetooth LE 4 cm × 4 cm 4 Hz 38 h of operation
biofeedback and between 0.01 µS and 100 µS
data acquisition
A pervasive and
Commercial Classification of 4 emotions with ~80%
unobtrusive system 65 mm × 32 mm 450 mAh Li-ion
[89] Shimmer GSR Finger Bluetooth 10 Hz of accuracy (amusement,
for sensing sensor
× 12 mm battery
fear, sadness, and relaxation)
human emotions
Investigated 6
Distinguishing stress
Power classifiers to Accuracy 82.8%
from cognitive load in an Dry Ag/AgCl left index and 41 mm ×
[90] Bluetooth 67 mm 16 Hz consumption: discriminate (max), achieved
office environment by electrodes middle fingers 182 mW cognitive load by LDA
EDA
from stress
Use of wearable sensors
and wireless technology Ag/AgCl GSR
Palm of preconditioning
[91] to measure the electrodes in non-dominant Bluetooth 30 Hz
autonomic function and Shimmer circuit consumes
hand 60 µA
stress level in the Platform
ambulatory setting
A wearable device for Fingers or 1280 Hz, averaged
predicting blood 10 h with 9 V correlation with
Ag/AgCl R2 value for PP:
[93] opposite sides of Bluetooth over 32 samples: 10 bit battery, 220 mA pulse pressure
pressure (BP) and electrodes 0.923, SBP: 0.801
palm results 40 Hz with Bluetooth with GSR
cardiovascular dynamics
Sensors 2017, 17, 130 20 of 45

2.5. Blood Oxygen Saturation (SpO2 ) Monitoring Systems

Peripheral capillary oxygen saturation (SpO2 ) is a measure of the amount of oxygenated
hemoglobin in the blood. The oxygen level in blood can be decreased due to health conditions
such as cardiovascular diseases, pulmonary diseases, anemia and sleep apnea. It can also be reduced
following excessive physical activities. It is essential to maintain an adequate amount of oxygen (>94%)
in the blood to ensure proper functioning of cells and tissues [95,96]. Therefore, it is important to
monitor SpO2 continuously, especially for persons having respiratory and heart-related diseases.
Pulse oximeters are widely used as a fast, non-invasive mean to measure the oxygen level in
blood. The estimate of SpO2 is from the absorption characteristics of blood in response to red (660 nm)
and infrared (940 nm) light. When hemoglobin becomes oxygenated, its color changes from dark red
to bright red that reduces the absorption of red light. The light absorption in blood also varies with the
change of arterial blood volume during systolic and diastolic phase of the heart, resulting in a time
varying signal called as photo-plethysmograph (PPG). The schematic representation of arterial blood
flow and its corresponding PPG signal is shown in Figure 7.

Figure 7. Arterial blood flow and corresponding PPG signal (not scaled).

The intensity of the transmitted light ( I ) can be determined by the well known Beer-Lambert law
that states that the intensity of transmitted light decreases logarithmically with the concentration of
,
oxygenated (CO ) and deoxygenated (Cd ) hemoglobin, the absorption coefficients of both (αO , αd ) at a
particular wavelength, and on the thickness (l ) of the arteries according to:

I = I0 exp(−(αO CO + αd Cd )l ) (1)

The SpO2 is calibrated from the PPG signal by measuring and comparing the intensity of
transmitted light at two wavelengths. The PPG signal can also be used to determine the respiratory
rate, pulse and heart rate [97,98]. It can also be used along with the ECG signal for estimating BP from
the pulse transit time [99,100].
The pulse oximeter usually utilizes red and infrared light emitting diodes (LED) as the light
sources. The residue lights after absorption are detected by the photodetector (PD). PPG or SpO2
sensors can be classified into two categories based on the working principles: transmittance and
reflectance oximetry (Figure 8a). In transmittance oximetry, the LEDs and PD are placed on opposite
sides of a transparent section of the body such as an earlobe, a fingertip or on the palms or soles
of small babies. Light transmitted through this section is collected by the PD. Currently, fingertip
based transmittance pulse oximeters are widely used for PPG signal measurements [101]. However,
this method is inconvenient for long-term monitoring. In reflectance oximetry, the LEDs and PD are
placed side by side on the same body surface and intensity of the reflected light is measured by the
PD. It offers flexibility for measuring PPG signal from different locations on the body that makes it
more suitable for the non-invasive wearable platform. The schematic diagram of the SpO2 monitoring
system has been shown in Figure 8b.
Sensors 2017, 17, 130 21 of 45

(a) (b)

Figure 8. Photoplethysmography (PPG): (a) Different approaches for measuring PPG; (b) Schematic
diagram of the SpO2 monitoring system.

A transmittance SpO2 sensor probe has been proposed in [102] that could be embedded in a
finger ring, unlike the conventional fingertip probes. A novel distribution of optical sensors and LEDs
around the phalanx and mounted them on a flexible PCB was proposed. Experiments were carried out
on 10 subjects and the results were in good agreement with that measured by commercial finger-tip
oximeters. This ring-type SpO2 sensor probe could be useful for monitoring arterial oxygen level and
heart rate for a long period of time. Guo et al. [100] integrated a vital sign monitoring system in a
chest band, embedded with micro-machined Pt electrodes. The band contains a miniaturized PCB that
comprises an ECG analog front-end, a driver circuit for an ear-worn PPG probe and UART/wireless
transceiver. The system measures ECG, and PPG signals at a rate of 200 Hz and transmits the data
using BSN (body sensor network) node to a hand-held device (e.g., PDA) where motion induced noise
are removed by using a wavelet de-noising process. The PDA calculates and displays heart rate, BP
and SpO2 and can transmit the data through GSM to healthcare personnel, if necessary.
A reflectance probe has been designed by Cai et al. [103] for measuring PPG signal that can be
worn as a wristband. The wristband is also equipped with an RF transmitting module that facilitates
wireless communication between the measurement system and the health care center. Experimental
results show that the system can detect the change in oxygen level in blood effectively and thus could
be useful in non-invasive, continuous and remote monitoring systems. Chen et al. [98] demonstrated
a non-invasive oxygen saturation monitoring system for newborn babies. The reflectance sensor is
embedded in soft fabrics that makes it suitable for a wearable, long-term monitoring system. The
system measures HR, SpO2 and can transmit the data by an RF transceiver. Experimental result showed
that the measured data closely follows the measurement obtained from the commercial monitoring
system. However, further improvement in the design is needed in order to minimize the impact of
motion artifacts that causes false reading. A small, low-cost wearable reflectance pulse oximeter was
proposed in [99] that can provide quality PPG signal without using any filter circuit. The PPG signal
is sampled at a rate of 240 Hz which is then transmitted to a host computer by a ZigBee transceiver
module or by a mini-USB. Further processing is done in computer in order to remove the ambient
noise and slowly varying motion artifacts. This system can be used to monitor HR, respiration rate,
SpO2 , and BP.
Huang et al. [104] performed the Monte-Carlo simulation of optical interaction with human tissue
in order to investigate the feasibility of a ring-type pulse oximeter with multiple detectors [105,106].
The optimum positions of the light source and the detectors are determined from the simulation results.
It has also been reported that multi-detector sensor improves the stability and the light gathering
efficiency. The authors developed a wearable wireless sensor based on the simulation results and
compared the SpO2 measurement result with the measurement from the commercial fingertip-type
pulse oximeter that showed a high degree of correlation (~98.26%) between the measurements.
Sensors 2017, 17, 130 22 of 45

Table 7. SpO2 monitoring systems.

Proposition Measured Size Power/Current Wireless

Ref. Principle Sampling Rate Performance Evaluation
Parameters Req. Connectivity
Ring shaped Quantum eff. = 62% Reverse current
Transmittance Temperature, Radius = 3.68 mm
[97] backside silicon 8 kHz <10 mA density = 55 nA/cm2 Forward
oximetry Pulse, SpO2 width = 0.78 mm
p-n photodiode saturation current = 0.14 nA/cm2
Measurement compared
Sensors embedded Reflectance HR, SpO2 graphically with
[98]
in soft fabrics oximetry commercial oximeter
measurements
Reflectance HR, RR, SNR of IR = 8
[99] Wireless oximeter 240 Hz 41mm × 36 mm <150 mA ZigBee
oximetry SpO2 , PPT SNR of red =3
Micro-machined Pt Transmittance ECG, HR, SpO2
[100] 200 Hz <35 mA ZigBee
electrodes oximetry and SBP
Ring probe, Measurement compared
novel distribution Transmittance Diameter of graphically with
[102] HR, SpO2
of optical sensors oximetry the finger commercial oximeter
around the phalanx measurements
Reflectance Ratio of change rates of reflected light intensity in
[103] Wrist band Sensor HR, SpO2 CC2500 RF TRX
oximetry two wavelengths (660 nm and 900 nm)
Correlation between SpO2
Ring-type pulse Reflectance HR, RR, values measured by the
[104] Bluetooth 98.26%
oximeters oximetry SpO2 , PPT proposed and commercial
oximeter
Measurement compared
Analog single-chip Mean diff. ~−1.2%
[107] SpO2 2.2 mm × 2.2 mm 4.8 mW with commercial
pulse oximeter SD = 1.5%
oximeter measurements
Forehead mounted Reflectance HR, SpO2 WiFi Measurement compared with commercial
[108]
sensor oximetry oximeter measurements
Electronic Patch PPG is measured using Datex pulse oximeter. SpO2
with an optical Reflectance 88 mm × 60 mm I < 33 mA
[109] PPG, HR, RR 125 Hz (× 5 mm) P < 99 mW is calculated and plotted against optical ratio for
oximetry
biomedical sensor calibration, MSE ~2.6%
Sensors 2017, 17, 130 23 of 45

Some researchers also worked on designing on-chip oximeters for SpO2 measurement.
Tavakoli et al. [107] developed a low-power analog single-chip pulse oximeter fabricated
in a 1.5 µm AMI BiCMOS n-well process. They used a conventional fingertip probe to measure
the SpO2 level in blood of 11 subjects and compared the results with that of a commercial pulse
oximeter. It has been observed that the result closely follows the measurements obtained from the
commercial pulse oximeter having a standard deviation of ~1.5%. The chip has a dimension of
2.2 × 2.2 mm2 and only requires 4.8 mW of power, which according to the authors is more than
an order-of-magnitude less than the power consumption of the best pulse oximeters available in
the market.
A multi-sensor chip was designed by Dunn et al. [97] that can be integrated into an electronic
patch for measuring temperature, pulse and SpO2 . The authors proposed a ring-shaped photodiode
(PD) for reflectance oximetry that maximizes the gathering of backscattered light from the tissue, thus
enabling much lower driving current for LEDs. The PD also contains an Au thermistor which facilitates
measurement of skin temperature. With a current requirement of less than 10 mA, this multi-sensor
oximeter chip is suitable for long-term monitoring purposes.
Other SpO2 monitoring systems such as forehead mounted sensor [108], electronic patch
sensor [109] and phone camera based measurement systems [110] can be found in the literature.
Table 7 presents the comparison of SpO2 monitoring systems discussed above.

2.6. Multi-Sensor Monitoring System

As discussed above, most of the systems were developed aiming to measure or monitor only
a specific bio-signal or parameter, for example, only ECG and HR were monitored in [10,28,30–33].
However, it is necessary to monitor a set of physiological signs such as HR or pulse, BP, respiratory
rate, and body temperature; often referred together as vital signs as well as oxygen saturation level
in blood and GSR level in order to perform a better assessment of an individual’s health condition.
Using parameter specific monitoring systems for each parameter is neither practical nor ergonomically
sound for continuous and ambulatory monitoring. A network of multiple on-body sensors embedded
in a wearable platform along with an on-body data acquisition and transceiver module can be a viable
solution for multi-parameter monitoring.
From the discussion presented above, it is noted that a set of important physiological parameters
can be measured and monitored by using four sensors: ECG, PPG, GSR, and temperature sensor.
An ECG sensor measures ECG signal, HR, and HRV while GSR and temperature sensor measures skin
conductivity and body temperature, respectively. PPG signal is generally used to determine the arterial
oxygen saturation (SpO2 ) level. Many researchers exploited Moens–Korteweg or Bramwell–Hill
relationship in order to estimate the BP from the pulse transit time (PTT) (Figure 9a), the time interval
between the ECG and PPG signal peaks [111–115]. Furthermore, ECG and PPG signal can also be
used to determine respiration rate (RR) [116–119] by employing signal decomposition techniques such
as empirical mode decomposition (EMD) [117,118], principal component analysis (PCA) [118,119] or
wavelet transform [120,121]. Figure 9b presents the concept of a four sensor monitoring system.

(a) (b)

Figure 9. (a) Pulse transit time (PTT); (b) Four sensor health monitoring system.
Sensors 2017, 17, 130 24 of 45

3. Textile-Based Wearable Sensors

Smart textiles associated with healthcare include sensors, actuators, communication, computing,
and electronic systems that are made of textile or are suitable for embedding into textiles thus enabling
unobtrusive and comfortable means of monitoring physiological signals of the individuals [122,123].
It makes use of conventional fabric manufacturing techniques such as weaving, knitting, embroidery,
and stitching to realize or integrate sensing materials in clothes. Advanced fabrication methods,
for example, inkjet-printing, coating, lithography, chemical vapor deposition (CVD) are also used
in order to achieve high performance in terms of noise and sensitivity. Figure 10 shows different
conventional textile manufacturing technologies. The active or sensing material is usually built on
a substrate and can either be in direct contact with the body surface or remain encapsulated in a
fabric-based layer [124–126].

(a) (b) (c)

(d) (e) (f)

Figure 10. Different textile/fabric manufacturing technologies. (a) Embroidery; (b) Stitching;
(c) Weaving; (d) Knitting; (e) Spinning; (f) Printing. Image source: https://pixabay.com under Creative
Commons CC0.

3.1. Textile Electrodes

Smart textiles can be used to develop wearable on-body electrodes in order to measure
electro-physiological signals such as ECG, electroencephalography (EEG), GSR, and electromyography
(EMG). Textile based electrodes were reported to be as reliable as the traditional wet gel Ag-AgCl
electrodes [127,128]. Textile electrodes can be classified into two basic categories: active and passive.
Passive textile electrodes sense electrical properties from the skin surface. It can be used to monitor
cardiac or muscle activities by sensing potential fluctuations caused by the heart or muscle. They also
have applications in GSR measurement where the change in the skin conductivity due to sweating
is detected by attaching electrodes on the body surface. Traditional electrodes use adhesive and
conductive gel to affix them to the skin. It requires skin preparation such as shaving and cleaning
the attachment site. In addition, the conductive gel may cause irritation, allergic reactions, or
inflammation [10,129,130]. Furthermore, the gel dries out with time causing degraded signal quality.
Although wet electrodes provide superior signal quality, they are not suitable for wearable and
long-term monitoring system [10].
On the other hand, dry electrodes do not use adhesive or conductive gel and are usually
biocompatible. Owing to their “dry” nature, they are more suitable for long-term monitoring and are
Sensors 2017, 17, 130 25 of 45

being used extensively in textile-based health monitoring systems [129–135]. However, dry electrodes
suffer from very high electrode-skin impedance and thus are more vulnerable to noise and motion
artifacts compared to the wet electrodes. Active electrodes often incorporate a high input impedance
preamplifier that reduces the impact of noise and motion artifacts by reducing the electrode-skin
impedance [131,132,136]. This also helps to reduce the impedance mismatch between the electrodes
resulting in lower differential common mode voltage, which may otherwise cause signal saturation.
Some active textile electrodes can stimulate muscle or nerve cells by applying an electrical current
to the underlying tissues. This technique of muscle and nerve activation, commonly referred to
transcutaneous electrical nerve stimulation (TENS) is widely applied in rehabilitation and therapeutic
applications [133–135] such as chronic and postoperative pain management.
Textile electrodes can be realized by integrating prefabricated electrodes into finished garments
by simply stitching them at suitable locations on clothes. They can also be developed by directly
depositing conductive layers on the fabric. The conductive layers can be formed on the surface of the
fabric by depositing nano-fibers [137–139] using electrodeposition method or by applying a conductive
layer with the help of screen printing [140], sputtering, carbonizing and evaporation [141].
Although conductive coating on the surface of the fabric results in superior conductivity,
the performance may deteriorate with time, especially after a number of wash cycles. Another
attractive technique of textile electrode fabrication is weaving or knitting [136,140,141] garment fabrics
using conductive yarn. The conductive yarn can be made of metal filaments [141–143], conductive
nano-filaments [144] or produced by applying a metal coating on fibers such as cotton [136,137,140],
nylon [141], Kevlar or polyester [136,142–145]. Nano-fibers can be grown by the electrospinning
method [138] whereas metal coating on the fiber is formed by employing chemical deposition process
such as polymerization [146], electroless plating [141,145], electroplating [147] and sputtering [141,144].
Table 8 presents the summary of several textile electrodes reported in the literature.

3.2. Textile Based Temperature Sensors

Rectal thermometer is usually the most accurate method of measuring body temperature.
However, this method is invasive, time-consuming and require private arrangements, thus making it
unsuited for continuous monitoring purposes. A tympanic thermometer measures temperature from
the tympanic membrane of the ear by placing a probe in the auditory canal which is not convenient
for long-term use. Besides this, the ear canal needs to be properly cleaned as earwax can decrease the
measurement accuracy [148]. Oral thermometers are also known to provide reliable readings when
measurement is taken from the sublingual pocket. Axillary (armpit) temperature measurement is
more convenient compared to the above-mentioned methods. This method is sometimes considered
unreliable for core body temperature estimation owing to the absence of any primary blood vessel
at this site [149]. The measurement can also be affected by the environmental temperature and
perspiration. However, both oral and axillary measurement techniques restrict movement of certain
body parts and thus, are not suitable for continuous monitoring.
The normal core body temp of a person remains within a narrow range of 36 ◦ C–37.5 ◦ C, although
it can vary by few degrees due to environmental influence, diseases or physical activities. The skin
temperature differs from the core body temperature by as much as 2.5 ◦ C [150]. Therefore, the
temperature sensors used for body temperature measurement need to cover a temperature range
from 35 ◦ C–40 ◦ C. It is also important to ensure measurement accuracy that, otherwise can impact
the diagnosis and treatment. Typically, an accuracy of 0.1 ◦ C is desired [151]. Textile based flexible
temperature sensors can be useful for measuring human body temperature in a wearable platform.
They can be fabricated from fiber or single yarn using conventional textile manufacturing technologies
such as weaving, knitting, embroidery, and printing. Temperature sensors fabricated on flexible
substrates can also be incorporated into textiles. Whatever be the case, higher accuracy, linearity,
sensitivity, and fast response time within the range of 35 ◦ C–40 ◦ C are critical for the temperature
sensors. Apart from that, selection of proper encapsulation material is also important in order to
protect it from external mechanical and environmental impacts.
Sensors 2017, 17, 130 26 of 45

Table 8. Summary of textile electrodes.

Ref. Proposition Electrode Type Size Base Material Conductive Material Technology Performance Contact Resistance
20 × 13 mm2 Screen printing, *PSDs for sitting and jogging
Direct attach and (direct-attach) Nonwoven Evolon are close to Ag/AgCl
[131] Active electrode Conductive ink (CMI 112-15) stenciling, curing, and
Interposer electrode 11.6 × 11.6 mm2 fabrics electrodes*Durable
encapsulation
(Interposer) upto 5 washing cycles
Woven textile The printed active and
Active electrodes on 28 mm × 23 mm (skin composed of cotton, Silver polymer paste (Fabinks Screen and Ag/AgCl electrodes had very
[136] Active electrode
woven textiles contact area) polyester and TC-C-4001) stencil printing similar rms levels
Lycra fibers after filtering
Poly
2 textile nanofiber Electrospinning-vapor Tested ECG is 95% similar to
[138] Dry electrode 9 mm diameter PVDF Nanofiber Web (3,4-ethylene-dioxythiophene) ~1000 Ω
web electrodes phase polymerization Ag/AgCl electrodes
(PEDOT)
Tested ECG is ~92% similar to ~100 Ω
PVDF Nano fiber Web Silver Silver mirror reaction
Ag/AgCl electrodes
Nano copper loaded
Polypropylene Copper nanoparticles Multiple dip chemical Max conductivity:
[139] poly-propylene based Dry electrode 4 cm × 6 cm
nonwoven fabric on fabric processes 142.8 kΩ·m
textile electrode
32 kΩ at 120 Hz
8 types of Inclusion of one (for 2 Ag strand based
[143]
electro-thread
Dry fabric electrode 2 × 2 cm2 , 2 × 5 cm2 Polyester 75 denier Silver thread strand or two strands 1300TM polyester
of 50 µm silver thread
fabric)
5.7 Ω (PU laminated
Several textile-based 1.5 cm × 3 cm PU laminated or Copper coating Sputtering nylon), 10.26 Ω (PU
[141] Dry fabric electrode
electrodes dry- coated nylon
dry-coated nylon).
0.23 Ω/sq (Ripstop),
Ripstop, Mesh fabric Cu/Ni coating Electroless Plating
0.29 Ω/sq (Mesh)
Stainless Steel Embroidering or R peak detection accuracy: 32.55 Ω/m
5 cm × 5 cm Cotton, Steel/cotton
Filament Yarn Knitting 58.8% and 64.2% (linear resistance)
FFT response of the
Silver coated nylon, stainless
Knitted fabric 20 mm × 20 mm Knitting multifilament electrodes
[142] electrodes Dry electrodes Wool and polyester steel yarn, and silver coated retains ECG
copper spectralcomponents
Wet, moisturized by Polyethylene Similar signal quality and
[144] Embroidered textile water vapor using the 2 cm × 7 cm terephthalate yarn of Silver and ultra-thin titanium Coating by plasma signal strength after 1 h as
electrode sputtering
polyester wetting pad. 50 µm diameter after 72 h of use
Sensors 2017, 17, 130 27 of 45

Based on their operation principles, textile-based temperature sensors can be categorized as

thermocouples and resistance temperature detectors (RTD). Thermocouples exploit the Seebeck effect
that develops a corresponding potential difference across the junctions of two dissimilar metals due to
the temperature difference between the junctions. The sensitivity of thermocouples usually ranges from
~10 µV/◦ C to 70 µV/◦ C [152,153]. Various textile thermocouple structures using different conductive
and metal coated non-conductive threads were reported in [154]. Among different pairs of materials,
thermocouple developed with steel knitted fabric-constantan wire pair exhibited the highest sensitivity
(41.4 µV/◦ C). However, thermocouples exhibit non-linear relationship between potential variation
and temperature as well as very low sensitivity that is not suitable for human body temperature
measurements [155,156].
The RTDs exploit the materials’ temperature dependence of electrical resistivity to determine the
temperature. RTDs offer higher accuracy, sensitivity, and shorter response time as well as linearity
with temperature. Textile based RTDs can be realized by simply embedding metal or conductive
threads with a strong TCR (temperature coefficient of resistance) into fabric using conventional
textile manufacturing technology such as knitting [157], weaving [158], and embroidery [159]. Many
researchers are also working to develop flexible temperature sensors for textile applications. Such
sensors were fabricated by depositing sensing materials on flexible polymer substrates such as
Kapton [158,160–162], polydimethylsiloxane (PDMS) [163] or paper substrates [164] using screen
printing [160,162,165], inject printing [161,164], CVD [163] techniques. Metals such as silver, platinum,
nickel or metal alloys were used widely as the sensing material, whereas some researchers exploited
nano-materials such as carbon nanotube [162], graphene nano-wall [163] as the temperature sensing
element. Hongqiang et al. [166] presented an optical sensing approach for body temperature
measurement. They used a distributed Bragg reflector (FBG) that reflects light of specific wavelengths
and transmits at other wavelengths. The FBG was encapsulated with a polymer substance which was
later weaved into the textile. The authors mathematically analyzed the transmission of heat from skin
to the environment via FBG sensors and employed a weighted coefficients model to estimate the body
temperature. The authors reported achieving high measurement accuracy (~±0.18 ◦ C) within the range
of 33 ◦ C–42 ◦ C. Table 9 presents the summary of textile based temperature sensors discussed above.

3.3.Textile Sensors for Activity Measurement

Most researchers have used MEMS based inertial sensors such as accelerometers, gyroscopes
or magnetic field sensors or their combinations in order to measure the signal corresponding to
human locomotion. They are mounted on small PCB boards, which are usually embedded in belts,
elastic bands, and Velcro straps. MEMS-based motion sensors are cheap and small in size. Having
good sensitivity, accuracy, and low power features, they are suitable for long-term and real-time
activity monitoring systems. However, the rigid PCB boards may feel uncomfortable to some users.
Rajdi et al. [167] fabricated a MEMS accelerometer on cotton cloth for measuring pelvic tilt angle.
The accelerometer exploited the piezoresistive effect of the conductive Ag nanoparticles that was
patterned on the textile by stamping and ironing. When the tilt angle varies, the cantilever beam
structure of the accelerometer experiences mechanical stress that eventually changes the resistance of
the conductive material in a linear fashion. The authors reported a strong positive correlation between
relative resistance change and the strain applied on the textile-based accelerometer.
Flexible and stretchable strain sensors were also used by many researchers in textile-based activity
monitoring systems. Strain sensors measure the physical deformation by changing its electrical
characteristics such as resistance and capacitance in response to mechanical stress. The strain sensors
for textile applications need to be highly flexible, stretchable, and durable. In addition, high sensitivity,
and fast response/recovery time are critical for real-time activity detection [152,153].
Sensors 2017, 17, 130 28 of 45

Table 9. Summary of textile based temperature sensors.

Fabrication Temperature Substrate/Embedding Nominal

Ref. Proposition Type Range Sensing Material Sensitivity Size Performance
Method Platform Resistance
High flexibility, linear
Screen printing Carbon polymer characteristic,
[160] Polymer sensor Thermistor Polyamide foil Kapton
paste high thermal
resistance change
Encapsulated with
polymer
Optical fiber Grinding, (copolymerization of
33 ◦ C to 42 ◦ C Fiber Bragg Accuracy ~± 0.18 ◦ C
[166] Bragg grating Optical grating 0.15 nm/◦ C unsaturated Methyl
polishing
based sensor Ethyl Ketone Peroxide
(MEKP) and cobalt
naphthenate) filled strip.
Printed sensors
RTD Screen printing 20 ◦ C to 80 ◦ C PTC and NTC 0.025 V/◦ C at 320 mm × Poly Ethylene
[165] on flexible
resistive pastes 37 ◦ C 380 mm Naphtalate (PEN)
substrate
Good linearity
Inkjet printed
20 ◦ C to 60 ◦ C 4.5 Ω/◦ C at 2.85 cm × Polyimide substrate (coefficient of linearity 2.032 kΩ
[161] Thermistor Inkjet printing Silver
flexible sensor 38.5 ◦ C 2.26 cm (Kapton HN) ~0.9998) Hysteresis at 38.5 ◦ C
less than 5%
Electron beam
Arrays of single evaporation Meander shaped Kapton E foils, The sensors damage at strong
67.5 mm ×
[158] sensors on a RTD 25◦ C to 90◦ C structures of 1.52 Ω/◦ C 67.5mm Integrated into textile bending of around 11% due to cracking
followed by
flexible substrate platinum using weaving of the sensing lines
photolitho-graphy
Good linearity with a
Sensors on Nano-porous oxide film TCR of 0.0011/◦ C,
RTD −20 ◦C to 60 ◦C Silver 16 mm × 16 mm 740 Ω with
[164] paper substrate Inkjet printing with perylene coating
nano-particles coated paper perylene coating
linearity, is 0.9999,
resistivity 30 µΩ·cm
Conductive yarn
Embroidered made of Embroidered on a Good resistance
[159] RTD Embroidery 20 ◦ C to 100 ◦ C austenitic Cr-Ni 2.68 Ω/◦ C 90 mm × 90 mm
sensors textile substrate against washing cycles
stainless steel
wires
Mixer of carbon Good stability,
[162] Printed wearable RTD Shadow mask 22 ◦ C to 50 ◦ C nanotube and 0.6 %/◦ C SiO2 -coated Kapton
sensor printing PEDOT:PSS highly sensitive
TCR = 0.214/◦ C,
PECVD and
Ultrasensitive Grapheme Polydimethylsiloxane response time 1.6 s
[163] RTD polymer-assisted 35◦ C to 45◦ C 20 mm × 10 mm 706.2 Ω at 25 ◦ C
wearable sensor nanowalls (PDMS) and recovery
transfer method time 8.52 s
Ni microparticle-
filled binary
RTD with polymer
Flexible wireless integrated 35 ◦ C to 42 ◦ C (polyethylene Accuracy ~± 2.7 ◦ C
[153] 0.1 to 0.3 V/◦ C
sensors passive RFID (PE) and
antenna polyethylene
oxide (PEO))
composites
Metal wire inlaid
Temperature Platinum wire, Coefficient linearity in
RTD in the middle of a 20 ◦ C to 50 ◦ C 8 cm × 8 cm 3 Ω to 130 Ω
[157] Diameter Polyester fabric
sensing fabric rib knitted < 25 mm the range of 0.99–0.999
structure
Sensors 2017, 17, 130 29 of 45

Amjadi et al. [168] presented highly flexible strain sensors based on the nanocomposite of Ag
nanowire mesh, which was embedded between PDMS layers. The piezoresistive strain sensors
exhibited high sensitivity with a tunable gauge factors (GF) in the ranges of 2 to 14, stretchability up
to 70%, and fast response. The authors demonstrated the usability of their designed strain sensors
by detecting the motion of the fingers in real-time using a glove that had five sensors embedded
in its five fingers. Similarly, Lee et al. [169], developed a highly sensitive, stretchable, and durable
strain sensor using a thin film of Ag nano-particles. The film was printed on a PDMS membrane by
the direct transfer process. When a mechanical stress is applied, it creates micro-cracks in the film,
causing a change in the resistance of the sensor. The sensor showed fast response (~1 s) and recovery
(~0.5 s) times. The devices were embedded in a glove to detect the activity (bending/relaxation)
of the finger joints. Shyr et al. [170] developed a textile strain sensor that was implemented in a
gesture sensing device to measure the flexion angle of elbow and knee movements. The strain sensor
was fabricated from elastic and conductive webbing. The conductive yarns were made from carbon
particles coated polyamide fiber twisted with polyester yarn whereas the elastic yarns were made
from Lycra fiber wrapped with two polyester yarns. The elastic conductive webbing exhibited good
linearity in resistance with flexion angles.
Zhang et al. [171] developed a textile-structured flexible strain sensor by using conductive fiber.
The authors exploited the variation in fiber-fiber, yarn-yarn loop, and fabric-fabric contact resistances
with strain in a textile type structure. They used metal and carbon fibers as the sensing material due
to their good physical and electrical properties. The sensors exhibited good linearity within a large
strain range and also achieved high gauge factor and sensitivity. However, the sensitivity of the sensor
can be affected by the shape, width, and density of the fibers. The strain range can be improved
by including elastic fibers in the structure. Another interlock based textile structure was presented
in [172] where the authors knitted a strain sensing fabric from elastomeric yarns. A series of loops
were embedded in this fabric by knitting a silver coated polymeric yarn in it. The interlock structure
has the advantage of having higher dimensional stability, and enhanced sensor repeatability. The loop
structure of the sensing conductive yarn also helps to minimize structural deformation over long term
use. The authors used three different sensors fabric yarns with different density, tightness factor, and
input tension and observed different linear range and gauge factor for different sensors. However,
knitting with metallic fibers or yarns may cause structural damages to the textile due to excessive
friction [173]. The knitted fabric also can be uncomfortable due to added stiffness of the metallic fibers
or yarns. An electrically conductive and flexible all-polymeric fiber was presented in [174]. Polymer
fibers offer high flexibility and less friction; and thus, they increase the lifetime of the structure. The
fiber was developed from PU/PEDOT:PSS by using the wet-spinning method. Up to four such fibers
were knitted with commercial Spandex yarn. It was observed that the sensitivity of the strain sensor
increased with the number of PU/PEDOT:PSS fibers. The sensor exhibited stable sensing performance
upto 160% of applied strain.
Sensors 2017, 17, 130 30 of 45

Table 10. Summary of textile-based strain sensors.

Proposition Gauge Factor Demonstrated/Potential

Ref. Sensing Mechanism Structure/Base Sensing Material Stable Strain Range Applications
Textile-structured flexible Contact resistance of 10–200 depending
[171] Single warp fabric Carbon fiber Max 200% Wearable strain sensor
strain sensor fiber/yarn/fabric on fiber length
Contact resistance of Fabric with elastomeric Silver coated polymeric
[172] Textile-based strain sensor 0.75 40% Wearable strain sensor
conductive fiber loops yarns yarn made loops
Stretchable and Sensitive Piezoresistive PDMS 2 to 14 70% Finger movements
[168] Ag nano-walls thin film
Strain Sensor
Textile-based strain sensor for Elastic yarns made from Carbon particles coated
monitoring the elbow Flexion angle of elbow
[170] Piezoresistive Lycra fiber wrapped with polyamide fiber twisted ~0.3 30%
and knee movements and knee movements
two polyester yarns. with polyester yarn
Stretchable strain sensor based
[169] on a metal nanoparticle thin film Piezoresistive PDMS Silver nanoparticle 2.5 20% Finger movements
for human motion detection
Knee’s kinematic monitoring Polymer encapsulated Knee, finger movements,
[175] Fiber Bragg grating Optical Fiber ~0.8 0.04%
using single optical FBG sensor FBG sensor HR, RR
Force sensors based
on light pipes in the form of Loss of light due to
Copolymers containing
[176] deflection of the Multimodal optical fiber Force sensing
multimode optical fibers silicon and polyurethane
fiber with force
made of copolymers.
Textile-based MEMS Piezoresistive 7.796 ± 2.835 Motion sensing
[167] Cotton fiber Silver nanoparticles
accelerometer
All-polymeric knitted Commercial Knee bending
[174] Piezoresistive PU/PEDOT:PSS fibers 0.2 to 1 160%
textile strain sensor Spandex yarn movements
Sensors 2017, 17, 130 31 of 45

A few researchers investigated textile-based optical sensors for activity monitoring. An optical
Fiber Bragg Grating (FBGs) based knee monitoring system was proposed in [175]. A single optical
Fiber Bragg Grating (FBGs) sensor encapsulated with a polymer foil was integrated into an elastic knee
band. The flexion-extension movement of the knee causes strain variations resulting deflection in the
resonance wavelength of the FBG. They also integrated the FBG sensors in gloves and Velcro straps in
order to demonstrate the usability of FBG sensors in detecting finger movements and heart, respiration
rate. Instead of optical power, the FBG sensor performed measurement based on the wavelength thus
was less sensitive to external noise and fluctuations in the optical source. The system exhibited high
sensitivity, stability, and measurement accuracy. Krehel et al. [176] designed an optical fiber based
flexible force sensor that could be potentially be integrated into textiles. The optical fiber was fabricated
from a flexible copolymer containing silicon and polyurethane. In the presence of an external force, the
optical fiber experiences an elliptical deformation along the plane of its cross section. This deformation
causes increased deflection of light within the fiber resulting reduced intensity of light at the output.
The force, therefore, can be estimated from the output light intensity. The sensor is flexible and can
be integrated into textile for detecting moderate to strong forces corresponding to, for example, limb
motion, and respiratory rate. The sensor is sensitive to strain, bends, and temperature which cause
inaccuracies in the measured data. Table 10 presents the summary of textile-based strain sensors.

4. Communication Technologies for Wearable Systems

The physiological signals measured by the on-body sensors need a two-stage communication
to transmit the data to the remote healthcare server. In the first stage, a short-range communication
protocol is employed to transmit the measured data to a nearest gateway node such as PDA,
smartphone, computer, custom-designed FPGA, or a microcontroller-based processing board. The
gateway is responsible for advanced data processing, display, and the next long range communication
stage, where the processed signal is transmitted to a distant server placed in a healthcare facility.
The data can be transmitted over the internet or cellular communication network. Currently,
most cellular networks offer seamless access to the internet through General Packet Radio Service
(GPRS), Enhanced Data GSM Environment (EDGE), 3G, High Speed Packet Access (HSPA), Long-Term
Evolution (LTE) services [177–179]. However, it is essential to implement strong encryption and
authentication technology in order to ensure a secure transmission channel over the long range
communication medium for safeguarding of personal medical information [180–182].
In the case of short-range communication, the sensors can communicate to the gateway directly
over a wireless medium. Alternatively, the sensors can form a body sensor network (BSN), a star
network topology and send data to the central BSN node. The BSN node can send data to the gateway
after performing some processing. The on-body sensors and the BSN node could communicate by
using wired or wireless medium. However, wired connections can hinder the users’ mobility and may
cause frequent failed connections. Thus, they are not suitable for wearable and long-term monitoring
systems. A good option is to use conductive fabric yarns as the alternative conductive medium. These
fabrics can be easily integrated into clothing to communicate with textile embedded sensors [183–185].
As discussed earlier, the conductive fabric can be fabricated using conventional textile technologies
such as weaving, stitching, embroidery, and printing. But conductive fibers do have a problem due to
their low durability and washability that may lead to poor or failed connectivity after a long period
of use. Therefore, wireless technology can be adopted as the most viable and reliable alternative for
short-range communication. In this section, we will present a brief overview on short range wireless
communication protocols that have been used in recent researches.
Bluetooth is a popular low power RF communication technology that has been widely used in
devices such as laptops, smartphones and fitness trackers for short range data communication [186,187].
It uses the 2.4 GHz frequency band in the industrial, scientific and medical (ISM) radio spectrum and
transmits signal over 79 designated channels using the Frequency Hopping Spread Spectrum (FHSS)
method. The FHSS method is less susceptible to noise and interference and also offers highly secured
Sensors 2017, 17, 130 32 of 45

data transmission. One master device can communicate with seven slave devices thus, forming a
star type network structure based on Bluetooth connectivity (Piconet). The master defines the clock
and hopping sequence for the whole Piconet. The Bluetooth technology can support a data rate
of ~3 Mbps depending on the modulation schemes, although the maximum throughput may only
reach ~2.1 Mbps. For general applications, the transmission distance typically ranges from 1 m to 10 m.
An ultralow-power version of Bluetooth technology, named as Bluetooth low energy (BLE) or Bluetooth
V4 was later introduced for portable and wearable devices with limited battery capacity [15,187].
BLE uses the same frequency band as classical Bluetooth technology but hops over 40 channels with
each channel having a bandwidth of 2 MHz. BLE, as the name indicates, offers low power (~10 mW)
wireless connectivity and, thereby is a strong candidate for short range communication in long-term
monitoring systems.
Another popular and open wireless standard for low power and low-cost communication within
short range is ZigBee [15,17]. It operates in the unlicensed 2.4 GHz (worldwide), 915 MHz (Americas
and Australia) and 868 MHz (Europe) frequency bands of the ISM spectrum and transmits data
over sixteen, ten and one channels, respectively. The 868 and 915 MHz bands use the binary
phase-shift keying (BPSK) modulation whereas offset quadrature phase-shift keying (OQPSK) is used at
the 2.4 GHz band. Unlike Bluetooth and BLE that only supports peer-to-peer (P2P) and star topologies,
ZigBee devices can be connected using P2P, star, tree and mesh network topologies. Prior to
transmitting a packet, the ZigBee protocol first assesses the communication link by using CSMA/CA
(carrier sense multiple access with collision avoidance) protocol or by sending beacons to other nodes
in the network. The transmission range of ZigBee standard is limited to within 10–20 m for indoor
applications mostly because of its low output power and also the presence of high dielectric materials.
However, the range can increase up to 1500 m with no obstacles in the line of sight. The data rate is
much lower compared to the Bluetooth technology and can reach maximum 250 kbps for the 2.4 GHz
band. However, the low power requirement of ZigBee standard leads to extended battery life that is
advantageous for long-term health monitoring applications, although the lower data rate may impose
limitations on the number of sensors, number of simultaneous measurements, and data buffering in a
multi-sensor network. Faster RAM along with efficient first-in-first-out (FIFO) and data compression
algorithm need to be implemented in the central BSN processing hardware.
ANT is a proprietary protocol stack designed for ultra-low-power, short-range wireless
communications in sensor networks, especially for health and fitness monitoring systems [19,188].
It ensures low power consumption by using low data rate, shorter delay cycles, and deep-sleep mode
and can operate for longer periods of time, for example, it can run a year on a 250 mAh coin cell
battery. Similar to other wireless protocols presented above, it also operates in the 2.4 GHz ISM
band. It uses TDMA (time division multiple access) to communicate with multiple nodes over a
single 1 MHz channel. It can switch channels if any interference occurs. ANT can be distinguished
from other wireless protocols by its unique feature in which it acts as a master for one channel while
simultaneously serving as a slave for another channel. Like ZigBee, ANT supports multiple network
topologies and also ensures coexistence with neighboring ANT nodes using adaptive isochronous
network technology. The maximum data rate achieved by ANT systems ranges from 20–60 kbps
and there is a trade-off between data rate and low power consumption. A recent advancement in
ANT protocol, ANT+, uses application specific 'device profiles’ to communicate between two devices.
For example, the applications can be one of vital signs monitoring, walking speed and distance
monitoring, or fitness monitoring and “device profiles” are the set of network rules, parameters, data
format specific to a particular application. Furthermore, the ANT+ protocol has the advantage from
interoperability with other ANT+ devices having the same device profile.
The medical implant communication service or MICS, is a short-range, ultra-low power wireless
technology that was developed to communicate with implanted medical devices such as cardiac
pacemakers, defibrillator, and neuro-stimulators [189,190]. It operates within the frequency band
of 402–405 MHz with 300 kHz channels. This frequency band offers good signal propagation
Sensors 2017, 17, 130 33 of 45

characteristics in the human body that makes it suitable for implantable devices. MICS uses the
listen-before-talk (LBT) protocol to assess the link before starting transmission. In the case of any
interference, MICS switches to a different radio channel and listens again. The MICS system has a
typical transmission range of ~2 m and consumes as little as 25 µW of power. However, due to limited
availability of commercial MICS devices along with some networking constraints [191], this technology
has not been used much in wearable systems.
There are a few other wireless technologies for short-range communication such as Infrared
Data Association (IrDA), Ultra-Wide Band (UWB), Radio Frequency Identification (RFID), Near Field
Communication (NFC) and WiFi are also available. IrDA was one of the most popular wireless
technologies for very short-range communication (<10 cm) because of their high data rate. However,
the communicating IrDA devices need to maintain line-of-sight for transmission that makes it infeasible
for wearable monitoring systems. The UWB operates in the wide frequency spectrum of 3.1–10.6 GHz
and uses short Gaussian impulses or multi-band orthogonal frequency-division multiplex (OFDM)
signal for communication. It offers very high data rate at a very low power spectral density, which
protects it from possible interference with other radio waves. Although UWB technology has promising
applications in medical monitoring and imaging [192,193], the complexity and limited availability of
commercial UWB systems makes it impractical for wearable systems. RFID is another popular wireless
technology that is widely used primarily for tracking and identification purposes. RFID technology
uses different frequency bands including the ISM band. A reader or interrogator sends a signal to a tag
or label that is attached to an object to be identified [194,195]. On the other side, NFC is a low cost, low
power wireless technology with a communication range of ~20 cm. It operates in the frequency band of
13.56 MHz and is compatible with passive RFID technology that comes at the price of increased power
consumption. NFC only supports P2P communication between two devices, so it is not appropriate
for wearable BSN systems [196,197].
Finally, WiFi, due to its extremely high power consumption and complex configurations, is
inefficient for long-term monitoring systems where longer battery life is indispensable. Table 11
presents the key features of currently available wireless technologies and their usage in current
wearable health monitoring systems.
Sensors 2017, 17, 130 34 of 45

Table 11. Key features of currently available wireless technologies.

Wireless Power Maximum Number of Supported Network

Frequency Band Range Data Rate Consumption Security Modulation Reference
Technology Nodes Supported Topologies
RFID 13.56 MHz 860–960 MHz 0-3 m 640 kbps 200 mW 1 at a time P2P (passive) N/A ASK, PSK, FSK [73,74,196]
Bluetooth 2.4–2.5 GHz 1–100 m 1–3 Mbps 2.5–100 mW 1 master + 7 slave P2P, star 56–128 bit key GFSK [29,30,45,59,66,87,90–93,104]
BLE 2.4–2.5 GHz 1–100 m 1 Mbps 10 mW 1 master + 7 slave P2P, star 128-bit AES GFSK [28,88]
ZigBee 2.4–2.5 GHz 10–100 m 250 kbps 35 mW 65,533 P2P, star, tree and mesh 128-bit AES OQPSK, BPSK [70,99,100]
WiFi 2.4–2.5 GHz 150–200 m 54 Mbps 1W 255 P2P, star WEP, WPA, WPA2 BPSK, QPSK, QAM [108]
UWB 3.1–10.6 GHz 3–10 m 53–480 Mbps 250 mW 1 master + 7 slave P2P, star BPPM, FSK
ANT 2.4–2.5 GHz 30 m 20–60 kbps 0.01–1 mW 65,533 in one channel P2P, star, tree and mesh 64-bit key GFSK [10,19,188]
MICS 402–405 MHz 2m 200–800 kbps 25 µW P2P, star FSK
IrDA 38 kHz 10 cm 1 Gbps 1 at a time P2P
NFC 13.56 MHz 5 cm 424 kbps 15 mW 1 at a time P2P AES ASK
Sensors 2017, 17, 130 35 of 45

5. Conclusions and Research Challenges

In this paper, we have presented a state-of-the-art survey on physiological parameters and activity
monitoring systems developed in a wearable platform. The primary purpose of a wearable health
monitoring system is to allow people to lead independent and active lives in their familiar home
environment while ensuring continuous, non-invasive, non-intrusive, and seamless surveillance
of their health and physical well-being. The enormous development of technology in the past
few decades leads to manufacturing and use of miniature, low-power, low-cost sensors, actuators,
electronic components, and powerful computers that paves the way to non-invasive, non-intrusive,
and continuous monitoring of an individual’s health condition at a very low-cost.
Continuous monitoring of health status can provide comprehensive information about individuals’
health status over a period of time. The wearable sensors and actuators, coupled with the advanced
information and communications technologies have opened the window to a new era of cost-effective
remote healthcare services. The systems can include monitoring and data analysis as well as
predictive algorithms, which can potentially make the prognosis of certain diseases with a higher
degree of confidence, thereby leading to early diagnosis and treatment. In the case of any potential
health issue is detected, the system can raise an alarm and notify the persons concerned or the
healthcare services via secure wireless media such as the internet and the cellular network so that
immediate medical intervention can be initiated. Incorporation of smart textiles technologies such
as textile-based interconnections for sensors in wearable healthcare systems could lead to more
comfortable, non-intrusive platforms for health monitoring.
In addition to the significant advances in the past decade, there are challenges that require further
research and development in order to improve the performance of the heart monitoring system.
First, the bio-potential (ECG, EDA) measurement systems very often suffer from low
signal-to-noise ratio (SNR) that primarily evolves from the noise induced by the movement of the user.
Motion artifact (MA) can be minimized by employing flexible electrodes or high input impedance
front end amplifiers. It is also possible to improve the SNR by exploiting signal processing techniques
such as adaptive filtering, empirical mode decomposition (EMD), independent component analysis
(ICA) or time-frequency analysis.
Second, the hardware and computation resource for the on-body central node of a multi-sensor
BSN system can be a limiting factor for seamless connectivity and data handling. The central processing
node of the BSN network exchanges data with the on-body sensors as well as the home gateway, and
sometimes performs limited processing. Therefore, a robust and efficient algorithm is required for
the central BSN node to optimize its performance. In addition to that, an efficient data compression
algorithm needs to be implemented in the central node in order to deal with a large volume of data
and transmit them to the nearest gateway.
Third, a key concern for the wearable healthcare system is associated with the privacy and
security of the sensitive medical information of the user. More efforts are needed in order to develop
algorithms to ensure highly secured communication channels in existing low power, short range
wireless platforms.
Fourth, low power consumption and high energy efficiency are critical for long-term monitoring
systems. Power requirement of the system can be satisfied by using low power components, more
efficient batteries or by employing energy harvesting techniques. Battery lifetime can also be improved
by ensuring ‘sleep and wake up’ of the sensors in a timely fashion without disrupting the desired
measurement frequency.
Fifth, in order to achieve widespread acceptance among the people, the systems need to
be affordable, easy-to-use, un-obtrusive, and inter-operable among various computing platforms.
A minimum numbers of electrodes, sensors need to be used yet not to lose the most important
clinical information. Therefore, more research and development efforts are needed to enhance the
systems’ ease-of-use.
Sensors 2017, 17, 130 36 of 45

Sixth, we have also presented a brief review on textile-based sensors highlighting their
applications in sensing physiological signs. They can be fabricated by using conventional textile
technologies such as knitting, weaving, embroidery or printing. Textile based sensors, that is smart
textiles, have great potential in wearable monitoring systems. For example, textile-based electrodes and
temperature sensors can be used for physiological measurements, whereas textile-based strain sensors
can be exploited for monitoring HR, respiration rate, pulse as well as human activities. However,
ensuring high signal accuracy, sensitivity, SNR and stability in a textile-based platform are the key
design challenges. Further, more work is needed for the proper selection of sensing materials and
embedding technique as well as stable sensor-skin interface to ensure superior sensor performance.
In addition, durability and signal integrity of the sensors with time and washing cycles should also be
improved while fabricating smart textiles for long-term health monitoring.
We have discussed and compared various wireless technologies and assessed their feasibility in
wearable health monitoring systems. Generally, the systems measure several physiological parameters
from the human body and transmit them to a central node or main gateway. The gateway node
processes and transmits the data to a healthcare personnel in a remote facility. However, more research
and technology development is needed to ensure information privacy and data security, robust data
compression algorithms, reliable communication link, and energy efficiency.

Acknowledgments: This research is supported by a Discovery Grant from the Natural Science and Engineering
Research Council of Canada (NSERC), an infrastructure grant from the Canada Foundation for Innovation,
an Ontario Research Fund for Research Excellence Funding Grant, a FedDev of Southern Ontario Grant, and the
Canada Research Chair Program.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Centers for Disease Control and Prevention. The State of Aging and Health in America 2013; Centers for Disease
Control and Prevention, US Department of Health and Human Services: Atlanta, GA, USA, 2013.
2. Global Age Watch Index 2015. Available online: http://www.helpage.org/global-agewatch/ (accessed on
20 June 2016).
3. World Health Organization. Family Planning/Contraception. 2015. Available online: http://www.who.int/
mediacentre/factsheets/fs351/en/ (accessed on 20 June 2016).
4. World Health Organization. Are You Ready? What You Need to Know about Ageing. World Health Day.
2012. Available online: http://www.who.int/world-health-day/2012/toolkit/background/en/ (accessed
on 20 June 2016).
5. U.S. Health Care Costs Rise Faster Than Inflation. Available online: http://www.forbes.com/
sites/mikepatton/2015/06/29/u-s-health-care-costs-rise-faster-than-inflation/#1384765c6ad2 (accessed on
20 June 2016).
6. Deen, M.J. Information and communications technologies for elderly ubiquitous healthcare in a smart home.
Pers. Ubiquitous Comput. 2015, 19, 573–599. [CrossRef]
7. Agoulmine, N.; Deen, M.; Lee, J.-S.; Meyyappan, M. U-Health Smart Home. IEEE Nanotechnol. Mag. 2011, 5,
6–11. [CrossRef]
8. Wang, H.; Choi, H.-S.; Agoulmine, N.; Deen, M.J.; Hong, J.W.-K. Information-based sensor tasking wireless
body area networks in U-health systems. In Proceedings of the 2010 International Conference on Network
and Service Management, Niagara Falls, ON, Canada, 25–29 October 2010; pp. 517–522.
9. Pantelopoulos, A.; Bourbakis, N. A Survey on Wearable Sensor-Based Systems for Health Monitoring and
Prognosis. IEEE Trans. Syst. Man Cybern. C 2010, 40, 1–12. [CrossRef]
10. Nemati, E.; Deen, M.; Mondal, T. A wireless wearable ECG sensor for long-term applications. IEEE Commun.
Mag. 2012, 50, 36–43. [CrossRef]
11. Hong, Y.; Kim, I.; Ahn, S.; Kim, H. Mobile health monitoring system based on activity recognition using
accelerometer. Simul. Model. Pract. Theory 2010, 18, 446–455. [CrossRef]
Sensors 2017, 17, 130 37 of 45

12. Ullah, S.; Higgins, H.; Braem, B.; Latre, B.; Blondia, C.; Moerman, I.; Saleem, S.; Rahman, Z.; Kwak, K.
A Comprehensive Survey of Wireless Body Area Networks. J. Med. Syst. 2012, 36, 1065–1094. [CrossRef]
[PubMed]
13. Al Ameen, M.; Liu, J.; Kwak, K. Security and Privacy Issues in Wireless Sensor Networks for Healthcare
Applications. J. Med. Syst. 2012, 36, 93–101. [CrossRef] [PubMed]
14. Castillejo, P.; Martinez, J.; Rodriguez-Molina, J.; Cuerva, A. Integration of wearable devices in a wireless
sensor network for an E-health application. IEEE Wirel. Commun. 2013, 20, 38–49. [CrossRef]
15. Dementyev, A.; Hodges, S.; Taylor, S.; Smith, J. Power consumption analysis of Bluetooth Low Energy, ZigBee
and ANT sensor nodes in a cyclic sleep scenario. In Proceedings of the 2013 IEEE International Wireless
Symposium (IWS), Beijing, China, 14–18 April 2013; pp. 1–4.
16. Suzuki, T.; Tanaka, H.; Minami, S.; Yamada, H.; Miyata, T. Wearable wireless vital monitoring technology
for smart health care. In Proceedings of the 2013 7th International Symposium on Medical Information and
Communication Technology (ISMICT), Tokyo, Japan, 6–8 March 2013; pp. 1–4.
17. Malhi, K.; Mukhopadhyay, S.; Schnepper, J.; Haefke, M.; Ewald, H. A Zigbee-Based Wearable Physiological
Parameters Monitoring System. IEEE Sens. J. 2012, 12, 423–430. [CrossRef]
18. Valchinov, E.; Antoniou, A.; Rotas, K.; Pallikarakis, N. Wearable ECG System for Health and Sports
Monitoring. In Proceedings of the 4th International Conference on Wireless Mobile Communication and
Healthcare—“Transforming Healthcare through Innovations in Mobile and Wireless Technologies”, Athens,
Greece, 3–5 November 2014; pp. 63–66.
19. Mehmood, N.Q.; Culmone, R. An ANT Protocol Based Health Care System. In Proceedings of the 2015
IEEE 29th International Conference on Advanced Information Networking and Applications Workshops,
Guwangiu, Korea, 24–27 March 2015; pp. 193–198.
20. Coskun, V.; Ozdenizci, B.; Ok, K. A Survey on Near Field Communication (NFC) Technology. Wirel. Pers.
Commun. 2013, 71, 2259–2294. [CrossRef]
21. Pang, Z.; Zheng, L.; Tian, J.; Kao-Walter, S.; Dubrova, E.; Chen, Q. Design of a terminal solution for integration
of in-home health care devices and services towards the Internet-of-Things. Enterp. Inf. Syst. 2013, 9, 86–116.
[CrossRef]
22. Corchado, J.; Bajo, J.; Abraham, A. GerAmi: Improving Healthcare Delivery in Geriatric Residences.
IEEE Intell. Syst. 2008, 23, 19–25. [CrossRef]
23. Stav, E.; Walderhaug, S.; Mikalsen, M.; Hanke, S.; Benc, I. Development and evaluation of SOA-based AAL
services in real-life environments: A case study and lessons learned. Int. J. Med. Inform. 2013, 82, e269–e293.
[CrossRef] [PubMed]
24. Vaishnav, S.; Stevenson, R.; Marchant, B.; Lagi, K.; Ranjadayalan, K.; Timmis, A.D. Relation between heart
rate variability early after acute myocardial infarction and long-term mortality. Am. J. Cardiol. 1994, 73,
653–657. [CrossRef]
25. Bigger, J.T.; Fleiss, J.L.; Kleiger, R.; Miller, J.P.; Rolnitzky, L.M. The relationships among ventricular
arrhythmias, left ventricular dysfunction, and mortality in the 2 years after myocardial infarction. Circulation
1984, 69, 250–258. [CrossRef] [PubMed]
26. Kleiger, R.E.; Miller, J.; Bigger, J.; Moss, A.J. Decreased heart rate variability and its association with increased
mortality after acute myocardial infarction. Am. J. Cardiol. 1987, 59, 256–262. [CrossRef]
27. Hadjem, M.; Salem, O.; Nait-Abdesselam, F. An ECG monitoring system for prediction of cardiac anomalies
using WBAN. In Proceedings of the 2014 IEEE 16th International Conference on e-Health Networking,
Applications and Services (Healthcom), Natal, Brazil, 15–18 October 2014.
28. Andreoni, G.; Perego, P.; Standoli, C. Wearable monitoring of elderly in an ecologic setting: The SMARTA
project. Available online: https://sciforum.net/conference/ecsa-2/paper/3192/download/pdf (accessed
on 5 January 2017).
29. Tseng, K.C.; Lin, B.-S.; Liao, L.-D.; Wang, Y.-T.; Wang, Y.-L. Development of a Wearable Mobile
Electrocardiogram Monitoring System by Using Novel Dry Foam Electrodes. IEEE Syst. J. 2014, 8, 900–906.
[CrossRef]
30. Lee, J.; Heo, J.; Lee, W.; Lim, Y.; Kim, Y.; Park, K. Flexible Capacitive Electrodes for Minimizing Motion
Artifacts in Ambulatory Electrocardiograms. Sensors 2014, 14, 14732–14743. [CrossRef] [PubMed]
Sensors 2017, 17, 130 38 of 45

31. Komensky, T.; Jurcisin, M.; Ruman, K.; Kovac, O.; Laqua, D.; Husar, P. Ultra-wearable capacitive coupled and
common electrode-free ECG monitoring system. In Proceedings of the 2012 Annual International Conference
of the IEEE Engineering in Medicine and Biology Society, San Diego, CA, USA, 28 August–1 September 2012;
pp. 1594–1597.
32. Park, J.-H.; Jang, D.-G.; Park, J.; Youm, S.-K. Wearable Sensing of In-Ear Pressure for Heart Rate Monitoring
with a Piezoelectric Sensor. Sensors 2015, 15, 23402–23417. [CrossRef] [PubMed]
33. Shu, Y.; Li, C.; Wang, Z.; Mi, W.; Li, Y.; Ren, T.-L. A Pressure sensing system for heart rate monitoring with
polymer-based pressure sensors and an anti-interference post processing circuit. Sensors 2015, 15, 3224–3235.
[CrossRef] [PubMed]
34. Yoon, S.; Cho, Y.-H. A Skin-attachable Flexible Piezoelectric Pulse Wave Energy Harvester. J. Phys. Conf. Ser.
2014, 557, 012026. [CrossRef]
35. Tajitsu, Y. Piezoelectret sensor made from an electro-spun fluoropolymer and its use in a wristband for
detecting heart-beat signals. IEEE Trans. Dielect. Electr. Insul. 2015, 22, 1355–1359. [CrossRef]
36. Izumi, S.; Yamashita, K.; Nakano, M.; Kawaguchi, H.; Kimura, H.; Marumoto, K.; Fuchikami, T.; Fujimori, Y.;
Nakajima, H.; Shiga, T.; et al. A Wearable Healthcare System With a 13.7 µA Noise Tolerant ECG Processor.
IEEE Trans. Biomed. Circuits Syst. 2015, 9, 733–742. [CrossRef] [PubMed]
37. He, D.D.; Sodini, C.G. A 58 nW ECG ASIC With Motion-Tolerant Heartbeat Timing Extraction for Wearable
Cardiovascular Monitoring. IEEE Trans. Biomed. Circuits Syst. 2015, 9, 370–376. [CrossRef] [PubMed]
38. Helleputte, N.V.; Kim, S.; Kim, H.; Kim, J.P.; Hoof, C.V.; Yazicioglu, R.F. A 160 µA biopotential acquisition
ASIC with fully integrated IA and motion-artifact suppression. In Proceedings of the 2012 IEEE International
Solid-State Circuits Conference, San Francisco, CA, USA, 19–23 February 2012; pp. 552–561.
39. Mulroy, S.; Gronley, J.; Weiss, W.; Newsam, C.; Perry, J. Use of cluster analysis for gait pattern classification
of patients in the early and late recovery phases following stroke. Gait Posture 2003, 18, 114–125. [CrossRef]
40. Snijders, A.H.; Warrenburg, B.P.V.D.; Giladi, N.; Bloem, B.R. Neurological gait disorders in elderly people:
Clinical approach and classification. Lancet Neurol. 2007, 6, 63–74. [CrossRef]
41. Coutinho, E.S.F.; Bloch, K.V.; Coeli, C.M. One-year mortality among elderly people after hospitalization due
to fall-related fractures: Comparison with a control group of matched elderly. Cadernos de Saúde Pública 2012,
28, 801–805. [CrossRef] [PubMed]
42. Zhou, Z.; Dai, W.; Eggert, J.; Giger, J.; Keller, J.; Rantz, M.; He, Z. A real-time system for in-home activity
monitoring of elders. In Proceedings of the 2009 Annual International Conference of the IEEE Engineering
in Medicine and Biology Society, Minneapolis, MN, USA, 3–6 September 2009; pp. 6115–6118.
43. Ni, B.; Wang, G.; Moulin, P. RGBD-HuDaAct: A Color-Depth Video Database for Human Daily Activity
Recognition. In Consumer Depth Cameras for Computer Vision; Springer: London, UK, 2013; pp. 193–208.
44. Derawi, M.; Bours, P. Gait and activity recognition using commercial phones. Comput. Secur. 2013, 39,
137–144. [CrossRef]
45. De, D.; Bharti, P.; Das, S.K.; Chellappan, S. Multimodal Wearable Sensing for Fine-Grained Activity
Recognition in Healthcare. IEEE Internet Comput. 2015, 19, 26–35. [CrossRef]
46. Bertolotti, G.M.; Cristiani, A.M.; Colagiorgio, P.; Romano, F.; Bassani, E.; Caramia, N.; Ramat, S. A Wearable
and Modular Inertial Unit for Measuring Limb Movements and Balance Control Abilities. IEEE Sens. J. 2016,
16, 790–797. [CrossRef]
47. Panahandeh, G.; Mohammadiha, N.; Leijon, A.; Handel, P. Continuous Hidden Markov Model for Pedestrian
Activity Classification and Gait Analysis. IEEE Trans. Instrum. Meas. 2013, 62, 1073–1083. [CrossRef]
48. Bejarano, N.C.; Ambrosini, E.; Pedrocchi, A.; Ferrigno, G.; Monticone, M.; Ferrante, S. A Novel Adaptive,
Real-Time Algorithm to Detect Gait Events from Wearable Sensors. IEEE Trans. Neural Syst. Rehabil. Eng.
2015, 23, 413–422. [CrossRef] [PubMed]
49. Ngo, T.T.; Makihara, Y.; Nagahara, H.; Mukaigawa, Y.; Yagi, Y. Similar gait action recognition using an
inertial sensor. Pattern Recognit. 2015, 48, 1289–1301. [CrossRef]
50. Alshurafa, N.; Xu, W.; Liu, J.J.; Huang, M.-C.; Mortazavi, B.; Roberts, C.K.; Sarrafzadeh, M. Designing a
Robust Activity Recognition Framework for Health and Exergaming Using Wearable Sensors. IEEE J. Biomed.
Health Inform. 2014, 18, 1636–1646. [CrossRef] [PubMed]
51. Ghasemzadeh, H.; Amini, N.; Saeedi, R.; Sarrafzadeh, M. Power-Aware Computing in Wearable Sensor
Networks: An Optimal Feature Selection. IEEE Trans. Mob. Comput. 2015, 14, 800–812. [CrossRef]
Sensors 2017, 17, 130 39 of 45

52. Chen, B.; Zheng, E.; Wang, Q.; Wang, L. A new strategy for parameter optimization to improve
phase-dependent locomotion mode recognition. Neurocomputing 2015, 149, 585–593. [CrossRef]
53. Cristiani, A.M.; Bertolotti, G.M.; Marenzi, E.; Ramat, S. An Instrumented Insole for Long Term Monitoring
Movement, Comfort, and Ergonomics. IEEE Sens. J. 2014, 14, 1564–1572. [CrossRef]
54. Tang, W.; Sazonov, E.S. Highly Accurate Recognition of Human Postures and Activities through Classification
with Rejection. IEEE J. Biomed. Health Inform. 2014, 18, 309–315. [CrossRef] [PubMed]
55. Friedman, N.; Rowe, J.B.; Reinkensmeyer, D.J.; Bachman, M. The Manumeter: A Wearable Device for
Monitoring Daily Use of the Wrist and Fingers. IEEE J. Biomed. Health Inform. 2014, 18, 1804–1812. [CrossRef]
[PubMed]
56. El-Gohary, M.; Mcnames, J. Human Joint Angle Estimation with Inertial Sensors and Validation with a Robot
Arm. IEEE Trans. Biomed. Eng. 2015, 62, 1759–1767. [CrossRef] [PubMed]
57. Wan, E.; Merwe, R.V.D. The unscented Kalman filter for nonlinear estimation. In Proceedings of the IEEE
2000 Adaptive Systems for Signal Processing, Communications, and Control Symposium (Cat. No. 00EX373),
Lake Louise, AB, Canada, 1–4 October 2000; pp. 153–158.
58. Hsu, Y.-L.; Chung, P.-C.; Wang, W.-H.; Pai, M.-C.; Wang, C.-Y.; Lin, C.-W.; Wu, H.-L.; Wang, J.-S. Gait and
Balance Analysis for Patients with Alzheimer’s Disease Using an Inertial-Sensor-Based Wearable Instrument.
IEEE J. Biomed. Health Inform. 2014, 18, 1822–1830. [CrossRef] [PubMed]
59. Pierleoni, P.; Belli, A.; Palma, L.; Pellegrini, M.; Pernini, L.; Valenti, S. A High Reliability Wearable Device for
Elderly Fall Detection. IEEE Sens. J. 2015, 15, 4544–4553. [CrossRef]
60. Lack, L.C.; Gradisar, M.; Someren, E.J.V.; Wright, H.R.; Lushington, K. The relationship between insomnia
and body temperatures. Sleep Med. Rev. 2008, 12, 307–317. [CrossRef] [PubMed]
61. Kräuchi, K.; Konieczka, K.; Roescheisen-Weich, C.; Gompper, B.; Hauenstein, D.; Schoetzau, A.; Fraenkl, S.;
Flammer, J. Diurnal and menstrual cycles in body temperature are regulated differently: A 28-day ambulatory
study in healthy women with thermal discomfort of cold extremities and controls. Chronobiol. Int. 2013, 31,
102–113. [PubMed]
62. Coyne, M.D.; Kesick, C.M.; Doherty, T.J.; Kolka, M.A.; Stephenson, L.A. Circadian rhythm changes in core
temperature over the menstrual cycle: Method for noninvasive monitoring. Am. J. Physiol. Regul. Integr.
Comp. Physiol. 2000, 279, R1316–R1320. [PubMed]
63. Reith, J.; Jorgensen, H.S.; Pedersen, P.M.; Nakamaya, H.; Jeppesen, L.L.; Olsen, T.S.; Raaschou, H.O. Body
temperature in acute stroke: Relation to stroke severity, infarct size, mortality, and outcome. Lancet 1996, 347,
422–425. [CrossRef]
64. Wright, K.P.; Hull, J.T.; Czeisler, C.A. Relationship between alertness, performance, and body temperature in
humans. Am. J. Physiol. Regul. Integr. Comp. Physiol. 2002, 283, R1370–R1377. [CrossRef] [PubMed]
65. Shibasaki, K.; Suzuki, M.; Mizuno, A.; Tominaga, M. Effects of Body Temperature on Neural Activity in
the Hippocampus: Regulation of Resting Membrane Potentials by Transient Receptor Potential Vanilloid 4.
J. Neurosci. 2007, 27, 1566–1575. [CrossRef] [PubMed]
66. Buller, M.J.; Tharion, W.J.; Cheuvront, S.N.; Montain, S.J.; Kenefick, R.W.; Castellani, J.; Latzka, W.A.;
Roberts, W.S.; Richter, M.; Jenkins, O.C.; et al. Estimation of human core temperature from sequential heart
rate observations. Physiol. Meas. 2013, 34, 781–798. [CrossRef] [PubMed]
67. Buller, M.J.; Tharion, W.J.; Hoyt, R.W.; Jenkins, O.C. Estimation of human internal temperature from
wearable physiological sensors. In Proceedings of the 22nd Conference on Innovative Applications of
Artificial Intelligence (IAAI), Atlanta, GA, USA, 11–15 July 2010; pp. 1763–1768.
68. Oguz, P.; Ertas, G. Wireless dual channel human body temperature measurement device. In Proceedings of
the 2013 International Conference on Electronics, Computer and Computation (ICECCO), Ankara, Turkey,
7–9 November 2013; pp. 52–55.
69. Boano, C.A.; Lasagni, M.; Romer, K.; Lange, T. Accurate Temperature Measurements for Medical Research
Using Body Sensor Networks. In Proceedings of the 2011 14th IEEE International Symposium on
Object/Component/Service-Oriented Real-Time Distributed Computing Workshops, Newport Beach, CA,
USA, 28–31 March 2011; pp. 189–198.
70. Boano, C.A.; Lasagni, M.; Romer, K. Non-invasive measurement of core body temperature in Marathon
runners. In Proceedings of the 2013 IEEE International Conference on Body Sensor Networks, Cambridge,
MA, USA, 6–9 May 2013; pp. 1–6.
Sensors 2017, 17, 130 40 of 45

71. Chen, W.; Dols, S.; Oetomo, S.B.; Feijs, L. Monitoring body temperature of newborn infants at neonatal
intensive care units using wearable sensors. In Proceedings of the Fifth International Conference on Body
Area Networks—BodyNets ′ 10, Corfu, Greece, 10–12 September 2010; pp. 188–194.
72. Mansor, H.; Shukor, M.H.A.; Meskam, S.S.; Rusli, N.Q.A.M.; Zamery, N.S. Body temperature measurement
for remote health monitoring system. In Proceedings of the 2013 IEEE International Conference on Smart
Instrumentation, Measurement and Applications (ICSIMA), Kuala Lumpur, Malaysia, 25–27 November
2013; pp. 1–5.
73. Rahman, M.A.; Barai, A.; Islam, M.A.; Hashem, M.A. Development of a device for remote monitoring of
heart rate and body temperature. In Proceedings of the 2012 15th International Conference on Computer
and Information Technology (ICCIT), Chittagong, Bangladesh, 22–24 December 2012; pp. 411–416.
74. Miah, M.A.; Kabir, M.H.; Tanveer, M.S.R.; Akhand, M.A.H. Continuous heart rate and body temperature
monitoring system using Arduino UNO and Android device. In Proceedings of the 2015 2nd International
Conference on Electrical Information and Communication Technologies (EICT), Khulna, Bangladesh,
10–12 December 2015; pp. 183–188.
75. Vaz, A.; Ubarretxena, A.; Zalbide, I.; Pardo, D.; Solar, H.; Garcia-Alonso, A.; Berenguer, R. Full Passive UHF
Tag With a Temperature Sensor Suitable for Human Body Temperature Monitoring. IEEE Trans. Circuits
Syst. II 2010, 57, 95–99. [CrossRef]
76. Milici, S.; Amendola, S.; Bianco, A.; Marrocco, G. Epidermal RFID passive sensor for body temperature
measurements. In Proceedings of the 2014 IEEE RFID Technology and Applications Conference (RFID-TA),
Tampere, Finland, 8–9 September 2014; pp. 140–144.
77. Sim, S.Y.; Lee, W.K.; Baek, H.J.; Park, K.S. A nonintrusive temperature measuring system for estimating
deep body temperature in bed. In Proceedings of the 2012 Annual International Conference of the
IEEE Engineering in Medicine and Biology Society, San Diego, CA, USA, 28 August–1 September 2012;
pp. 3460–3463.
78. Kimberger, O.; Thell, R.; Schuh, M.; Koch, J.; Sessler, D.I.; Kurz, A. Accuracy and precision of a novel
non-invasive core thermometer. Br. J. Anaesth. 2009, 103, 226–231. [CrossRef] [PubMed]
79. Kitamura, K.-I.; Zhu, X.; Chen, W.; Nemoto, T. Development of a new method for the noninvasive
measurement of deep body temperature without a heater. Med. Eng. Phys. 2010, 32, 1–6. [CrossRef]
[PubMed]
80. Jänig, W. Integrative Action of the Autonomic Nervous System: Neurobiology of Homeostasis; Cambridge University
Press: Cambridge, UK, 2008.
81. Critchley, H.D. Book Review: Electrodermal Responses: What Happens in the Brain? Neuroscientist 2002, 8,
132–142. [CrossRef]
82. Bakker, J.; Pechenizkiy, M.; Sidorova, N. What's your current stress level? Detection of stress patterns from
GSR sensor data. In 2011 IEEE 11th International Conference on Data Mining Workshops, Vancouver, BC,
Canada, 11–14 December 2011; pp. 573–580.
83. Bonato, P. Wearable sensors/systems and their impact on biomedical engineering. IEEE Eng. Med. Biol. Mag.
2003, 22, 18–20. [CrossRef] [PubMed]
84. Sano, A.; Picard, R.W. Stress Recognition Using Wearable Sensors and Mobile Phones. In Proceedings of
the 2013 Humaine Association Conference on Affective Computing and Intelligent Interaction, Geneva,
Switzerland, 2–5 September 2013; pp. 671–676.
85. Poh, M.-Z.; Swenson, N.C.; Picard, R.W. A Wearable Sensor for Unobtrusive, Long-Term Assessment of
Electrodermal Activity. IEEE Trans. Biomed. Eng. 2010, 57, 1243–1252. [PubMed]
86. Sugathan, A.; Roy, G.G.; Kirthyvijay, G.J.; Thomson, J. Application of arduino based platform for wearable
health monitoring system. In Proceedings of the 2013 IEEE 1st International Conference on Condition
Assessment Techniques in Electrical Systems (CATCON), Kolkata, India, 6–8 December 2013; pp. 1–5.
87. Kim, J.; Kwon, S.; Seo, S.; Park, K. Highly wearable galvanic skin response sensor using flexible and
conductive polymer foam. In Proceedings of the 2014 36th Annual International Conference of the IEEE
Engineering in Medicine and Biology Society, Chicago, IL, USA, 26–30 August 2014; pp. 6631–6634.
88. Garbarino, M.; Lai, M.; Tognetti, S.; Picard, R.; Bender, D. Empatica E3—A wearable wireless multi-sensor
device for real-time computerized biofeedback and data acquisition. In Proceedings of the 4th International
Conference on Wireless Mobile Communication and Healthcare—Transforming healthcare through
innovations in mobile and wireless technologies, Athens, Greece, 3–5 November 2014; pp. 39–42.
Sensors 2017, 17, 130 41 of 45

89. Guo, R.; Li, S.; He, L.; Gao, W.; Qi, H.; Owens, G. Pervasive and Unobtrusive Emotion Sensing for Human
Mental Health. In Proceedings of the 7th International Conference on Pervasive Computing Technologies
for Healthcare, Venice, Italy, 5–8 May 2013; pp. 436–439.
90. Setz, C.; Arnrich, B.; Schumm, J.; Marca, R.L.; Troster, G.; Ehlert, U. Discriminating Stress From Cognitive
Load Using a Wearable EDA Device. IEEE Trans. Inform. Technol. Biomed. 2010, 14, 410–417. [CrossRef]
[PubMed]
91. Crifaci, G.; Billeci, L.; Tartarisco, G.; Balocchi, R.; Pioggia, G.; Brunori, E.; Maestro, S.; Morales, M.A.
ECG and GSR measure and analysis using wearable systems: Application in anorexia nervosa adolescents.
In Proceedings of the 2013 8th International Symposium on Image and Signal Processing and Analysis
(ISPA), Trieste, Italy, 4–6 September 2013; pp. 499–504.
92. Subramanya, K.; Bhat, V.V.; Kamath, S. A wearable device for monitoring galvanic skin response to accurately
predict changes in blood pressure indexes and cardiovascular dynamics. In Proceedings of the 2013 Annual
IEEE India Conference (INDICON), Mumbai, India, 13–15 December 2013; pp. 1–4.
93. Yoon, Y.; Cho, J.H.; Yoon, G. Non-constrained Blood Pressure Monitoring Using ECG and PPG for Personal
Healthcare. J. Med. Syst. 2008, 33, 261–266. [CrossRef]
94. Blacher, J.; Staessen, J.A.; Girerd, X.; Gasowski, J.; Thijs, L.; Liu, L.; Wang, J.G.; Fagard, R.H.; Safar, M.E. Pulse
Pressure Not Mean Pressure Determines Cardiovascular Risk in Older Hypertensive Patients. Arch. Intern.
Med. 2000, 160, 1085–1089. [CrossRef] [PubMed]
95. Baker, C. Method and System for Controlled Maintenance of Hypoxia for Therapeutic or Diagnostic Purposes.
U.S. Patent No. US 11/241,062, 30 September 2005.
96. O’driscoll, B.R.; Howard, L.S.; Davison, A.G. BTS guideline for emergency oxygen use in adult patients.
Thorax 2008, 63, vi1–vi68. [CrossRef] [PubMed]
97. Duun, S.B.; Haahr, R.G.; Birkelund, K.; Thomsen, E.V. A Ring-Shaped Photodiode Designed for Use in
a Reflectance Pulse Oximetry Sensor in Wireless Health Monitoring Applications. IEEE Sens. J. 2010, 10,
261–268. [CrossRef]
98. Chen, W.; Ayoola, I.; Oetomo, S.B.; Feijs, L. Non-invasive blood oxygen saturation monitoring for neonates
using reflectance pulse oximeter. In Proceedings of the 2010 Design, Automation & Test in Europe Conference
& Exhibition (DATE 2010), Dresden, Germany, 8–12 March 2010; pp. 1530–1535.
99. Li, K.; Warren, S. A Wireless Reflectance Pulse Oximeter with Digital Baseline Control for Unfiltered
Photoplethysmograms. IEEE Trans. Biomed. Circuits Syst. 2012, 6, 269–278. [CrossRef] [PubMed]
100. Guo, D.; Tay, F.E.; Xu, L.; Yu, L.; Nyan, M.; Chong, F.; Yap, K.; Xu, B. A Long-term Wearable Vital Signs
Monitoring System using BSN. In Proceedings of the 2008 11th EUROMICRO Conference on Digital System
Design Architectures, Methods and Tools, Parma, Italy, 3–5 September 2008; pp. 825–830.
101. Petersen, C.; Chen, T.; Ansermino, J.; Dumont, G. Design and Evaluation of a Low-Cost Smartphone Pulse
Oximeter. Sensors 2013, 13, 16882–16893. [CrossRef] [PubMed]
102. Sola, J.; Castoldi, S.; Chetelat, O.; Correvon, M.; Dasen, S.; Droz, S.; Jacob, N.; Kormann, R.; Neumann, V.;
Perrenoud, A.; et al. SpO2 Sensor Embedded in a Finger Ring: Design and implementation. In Proceedings
of the 2006 International Conference of the IEEE Engineering in Medicine and Biology Society, New York,
NY, USA, 31 August–3 September 2006; pp. 4295–4298.
103. Cai, Q.; Sun, J.; Xia, L.; Zhao, X. Implementation of a wireless pulse oximeter based on wrist band sensor. In
Proceedings of the 2010 3rd International Conference on Biomedical Engineering and Informatics, Yantai,
China, 16–18 October 2010; pp. 1897–1900.
104. Huang, C.-Y.; Chan, M.-C.; Chen, C.-Y.; Lin, B.-S. Novel Wearable and Wireless Ring-Type Pulse Oximeter
with Multi-Detectors. Sensors 2014, 14, 17586–17599. [CrossRef] [PubMed]
105. Deen, M.J.; Basu, P.K. Silicon Photonics—Fundamentals and Devices; John Wiley and Sons: Chichester,
UK, 2012.
106. Palubiak, D.; El-Desouki, M.M.; Marinov, O.; Deen, M.J.; Fang, Q. High-speed, single-photon
avalanche-photodiode imager for biomedical applications. IEEE Sens. J. 2011, 11, 2401–2412. [CrossRef]
107. Tavakoli, M.; Turicchia, L.; Sarpeshkar, R. An Ultra-Low-Power Pulse Oximeter Implemented With an
Energy-Efficient Transimpedance Amplifier. IEEE Trans. Biomed. Circuits Syst. 2010, 4, 27–38. [CrossRef]
[PubMed]
Sensors 2017, 17, 130 42 of 45

108. Mendelson, Y.; Duckworth, R.J.; Comtois, G. A Wearable Reflectance Pulse Oximeter for Remote
Physiological Monitoring. In Proceedings of the 2006 International Conference of the IEEE Engineering in
Medicine and Biology Society, New York, NY, USA, 31 August–3 September 2006; pp. 912–915.
109. Haahr, R.G.; Duun, S.B.; Toft, M.H.; Belhage, B.; Larsen, J.; Birkelund, K.; Thomsen, E.V. An Electronic Patch
for Wearable Health Monitoring by Reflectance Pulse Oximetry. IEEE Trans. Biomed. Circuits Syst. 2012, 6,
45–53. [CrossRef] [PubMed]
110. Scully, C.G.; Lee, J.; Meyer, J.; Gorbach, A.M.; Granquist-Fraser, D.; Mendelson, Y.; Chon, K.H. Physiological
Parameter Monitoring from Optical Recordings With a Mobile Phone. IEEE Trans. Biomed. Eng. 2012, 59,
303–306. [CrossRef] [PubMed]
111. Buxi, D.; Redouté, J.M.; Yuce, M.R. A survey on signals and systems in ambulatory blood pressure monitoring
using pulse transit time. Physiol. Meas. 2015, 36, R1–R26. [CrossRef] [PubMed]
112. Mccombie, D.B.; Reisner, A.T.; Asada, H.H. Adaptive blood pressure estimation from wearable PPG sensors
using peripheral artery pulse wave velocity measurements and multi-channel blind identification of local
arterial dynamics. In Proceedings of the 2006 International Conference of the IEEE Engineering in Medicine
and Biology Society, New York, NY, USA, 31 August–3 September 2006; pp. 3521–3524.
113. Puke, S.; Suzuki, T.; Nakayama, K.; Tanaka, H.; Minami, S. Blood pressure estimation from pulse wave
velocity measured on the chest. In Proceedings of the 2013 35th Annual International Conference of the IEEE
Engineering in Medicine and Biology Society (EMBC), Osaka, Japan, 3–7 July 2013; pp. 6107–6110.
114. Zheng, Y.-L.; Yan, B.P.; Zhang, Y.-T.; Poon, C.C.Y. An Armband Wearable Device for Overnight and Cuff-Less
Blood Pressure Measurement. IEEE Trans. Biomed. Eng. 2014, 61, 2179–2186. [CrossRef] [PubMed]
115. Lin, H.; Xu, W.; Guan, N.; Ji, D.; Wei, Y.; Yi, W. Noninvasive and Continuous Blood Pressure Monitoring
Using Wearable Body Sensor Networks. IEEE Intell. Syst. 2015, 30, 38–48. [CrossRef]
116. Allen, J. Photoplethysmography and its application in clinical physiological measurement. Physiol. Meas.
2007, 28, R1–R39. [CrossRef] [PubMed]
117. Madhav, K.V.; Ram, M.R.; Krishna, E.H.; Komalla, N.R.; Reddy, K.A. Estimation of respiration rate from ECG,
BP and PPG signals using empirical mode decomposition. In Proceedings of the 2011 IEEE International
Instrumentation and Measurement Technology Conference, Hangzhou, China, 10–12 May 2011; pp. 1–4.
118. Prathyusha, B.; Rao, T.S.; Asha, D. Extraction of Respiratory Rate from Ppg Signals Using Pca and Emd.
Int. J. Res. Eng. Technol. 2012, 1, 164–184.
119. Madhav, K.V.; Raghuram, M.; Krishna, E.H.; Komalla, N.R.; Reddy, K.A. Use of multi scale PCA for extraction
of respiratory activity from photoplethysmographic signals. In Proceedings of the 2012 IEEE International
Instrumentation and Measurement Technology Conference, Graz, Austria, 13–16 May 2012; pp. 1784–1787.
120. Leonard, P.A.; Douglas, J.G.; Grubb, N.R.; Clifton, D.; Addison, P.S.; Watson, J.N. A Fully Automated
Algorithm for the Determination of Respiratory Rate from the Photoplethysmogram. J. Clin. Monit. Comput.
2006, 20, 33–36. [CrossRef] [PubMed]
121. Clifton, D.; Douglas, J.G.; Addison, P.S.; Watson, J.N. Measurement of Respiratory Rate from the
Photoplethysmogram in Chest Clinic Patients. J. Clin. Monit. Comput. 2006, 21, 55–61. [CrossRef] [PubMed]
122. Cherenack, K.; Pieterson, L.V. Smart textiles: Challenges and opportunities. J. Appl. Phys. 2012, 112, 091301.
[CrossRef]
123. Chan, M.; Estève, D.; Fourniols, J.-Y.; Escriba, C.; Campo, E. Smart wearable systems: Current status and
future challenges. Artif. Intell. Med. 2012, 56, 137–156. [CrossRef] [PubMed]
124. Sibinski, M.; Jakubowska, M.; Sloma, M. Flexible Temperature Sensors on Fibers. Sensors 2010, 10, 7934–7946.
[CrossRef] [PubMed]
125. Zeng, W.; Shu, L.; Li, Q.; Chen, S.; Wang, F.; Tao, X.-M. Fiber-based wearable electronics: A review of
materials, fabrication, devices, applications. Adv. Mater. 2014, 26, 5310–5336. [CrossRef] [PubMed]
126. Park, S.; Vosguerichian, M.; Bao, Z. A review of fabrication and applications of carbon nanotube film-based
flexible electronics. Nanoscale 2013, 5, 1727–1752. [CrossRef] [PubMed]
127. Rai, P.; Kumar, P.S.; Oh, S.; Kwon, H.; Mathur, G.N.; Varadan, V.K.; Agarwal, M.P. Smart healthcare
textile sensor system for unhindered-pervasive health monitoring. Nanosens. Biosens. Info-Tech Sens. Syst.
2012, 8344E.
128. Rai, P.; Oh, S.; Shyamkumar, P.; Ramasamy, M.; Harbaugh, R.E.; Varadan, V.K. Nano-Bio-Textile Sensors with
Mobile Wireless Platform for Wearable Health Monitoring of Neurological and Cardiovascular Disorders.
J. Electrochem. Soc. 2014, 161, B3116–B3150. [CrossRef]
Sensors 2017, 17, 130 43 of 45

129. Fuhrhop, S.; Lamparth, S.; Heuer, S. A textile integrated long-term ECG monitor with capacitively coupled
electrodes. In Proceedings of the 2009 IEEE Biomedical Circuits and Systems Conference, Beijing, China,
26–28 Novemeber 2009; pp. 21–24.
130. Ouwerkerk, M.; Pasveer, F.; Langereis, G. Unobtrusive Sensing of Psychophysiological Parameters.
In Probing Experience; Springer: Dordrecht, The Netherlands, 2008; pp. 163–193.
131. Merritt, C.; Nagle, H.; Grant, E. Fabric-Based Active Electrode Design and Fabrication for Health Monitoring
Clothing. IEEE Trans. Inform. Technol. Biomed. 2009, 13, 274–280. [CrossRef] [PubMed]
132. Fonseca, C.; Cunha, J.P.S.; Martins, R.E.; Ferreira, V.M.; Sa, J.P.M.D.; Barbosa, M.A.; Silva, A.M.D. A Novel
Dry Active Electrode for EEG Recording. IEEE Trans. Biomed. Eng. 2007, 54, 162–165. [CrossRef] [PubMed]
133. Keller, T.; Kuhn, A. Electrodes for transcutaneous (surface) electrical stimulation. J. Autom. Control 2008, 18,
35–45. [CrossRef]
134. Li, L.; Au, W.M.; Li, Y.; Wan, K.M.; Wan, S.H.; Wong, K.S. Design of Intelligent Garment with Transcutaneous
Electrical Nerve Stimulation Function Based on the Intarsia Knitting Technique. Text. Res. J. 2009, 80, 279–286.
[CrossRef]
135. Malešević, N.M.; Maneski, L.Z.; Ilić, V.; Jorgovanović, N.; Bijelić, G.; Keller, T.; Popović, D.B. A multi-pad
electrode based functional electrical stimulation system for restoration of grasp. J. Neuroeng. Rehabil. 2012, 9,
66. [CrossRef] [PubMed]
136. Paul, G.; Torah, R.; Beeby, S.; Tudor, J. Novel active electrodes for ECG monitoring on woven textiles
fabricated by screen and stencil printing. Sens. Actuators A Phys. 2015, 221, 60–66. [CrossRef]
137. Lim, Z.; Chia, Z.; Kevin, M.; Wong, A.; Ho, G. A facile approach towards ZnO nanorods conductive textile
for room temperature multifunctional sensors. Sens. Actuators B Chem. 2010, 151, 121–126. [CrossRef]
138. Inoh, T.; Yoon, S.; Kim, T.E.; Wi, H.; Kim, K.J.; Woo, E.J.; Sadleir, R.J. Nanofiber Web Textile Dry Electrodes for
Long-Term Biopotential Recording. IEEE Trans. Biomed. Circuits Syst. 2013, 7, 204–211. [CrossRef] [PubMed]
139. Shaikh, T.N.; Chaudhari, S.; Patel, B.H.; Patel, M. Study of conductivity behavior of nano copper loaded
nonwoven polypropylene based textile electrode for ECG. Int. J. Emerg. Sci. Eng. 2015, 3, 11–14.
140. Rattfält, L.; Björefors, F.; Nilsson, D.; Wang, X.; Norberg, P.; Ask, P. Properties of screen printed
electrocardiography smartware electrodes investigated in an electro-chemical cell. Biomed. Eng. Online 2013,
12, 64. [CrossRef] [PubMed]
141. Cho, G.; Jeong, K.; Paik, M.J.; Kwun, Y.; Sung, M. Performance Evaluation of Textile-Based Electrodes and
Motion Sensors for Smart Clothing. IEEE Sens. J. 2011, 11, 3183–3193. [CrossRef]
142. Mestrovic, M.A.; Helmer, R.J.N.; Kyratzis, L.; Kumar, D. Preliminary study of dry knitted fabric electrodes
for physiological monitoring. In Proceedings of the 2007 3rd International Conference on Intelligent Sensors,
Sensor Networks and Information, Melbourne, Australia, 3–6 December 2007; pp. 601–606.
143. Cho, H.; Lim, H.; Cho, S.; Lee, J.-W. Development of textile electrode for electrocardiogram measurement
based on conductive electrode configuration. Fibers Polym. 2015, 16, 2148–2157. [CrossRef]
144. Weder, M.; Hegemann, D.; Amberg, M.; Hess, M.; Boesel, L.; Abächerli, R.; Meyer, V.; Rossi, R. Embroidered
Electrode with Silver/Titanium Coating for Long-Term ECG Monitoring. Sensors 2015, 15, 1750–1759.
[CrossRef] [PubMed]
145. Jiang, S.Q.; Newton, E.; Yuen, C.W.M.; Kan, C.W. Chemical silver plating on polyester/cotton blended fabric.
J. Appl. Polym. Sci. 2006, 100, 4383–4387. [CrossRef]
146. Xue, P.; Tao, X.M. Morphological and electromechanical studies of fibers coated with electrically conductive
polymer. J. Appl. Polym. Sci. 2005, 98, 1844–1854. [CrossRef]
147. Brevnov, D.A. Electrodeposition of Porous Silver Films on Blanket and Patterned Aluminum—Copper Films.
J. Electrochem. Soc. 2006, 153, C249–C253. [CrossRef]
148. Doezema, D.; Lunt, M.; Tandberg, D. Cerumen Occlusion Lowers Infrared Tympanic Membrane Temperature
Measurement. Acad. Emerg. Med. 1995, 2, 17–19. [CrossRef] [PubMed]
149. Sund-Levander, M.; Grodzinsky, E. Time for a change to assess and evaluate body temperature in clinical
practice. Int. J. Nurs. Pract. 2009, 15, 241–249. [CrossRef] [PubMed]
150. Togawa, T. Body temperature measurement. Clin. Phys. Physiol. Meas. 1985, 6, 83–108. [CrossRef] [PubMed]
151. Niedermann, R.; Wyss, E.; Annaheim, S.; Psikuta, A.; Davey, S.; Rossi, R.M. Prediction of human core
body temperature using non-invasive measurement methods. Int. J. Biometeorol. 2014, 58, 7–15. [CrossRef]
[PubMed]
Sensors 2017, 17, 130 44 of 45

152. Imran, M.; Bhattacharyya, A. Thermal response of an on-chip assembly of RTD heaters, sputtered sample
and microthermocouples. Sens. Actuators A Phys. 2005, 121, 306–320. [CrossRef]
153. Jeon, J.; Lee, H.-B.-R.; Bao, Z. Flexible Wireless Temperature Sensors Based on Ni Microparticle-Filled Binary
Polymer Composites. Adv. Mater. 2012, 25, 850–855. [CrossRef] [PubMed]
154. Ziegler, S.; Frydrysiak, M. Initial research into the structure and working conditions of textile thermocouples.
Fibres Text. East. Eur. 2009, 17, 84–88.
155. Houdas, Y.; Ring, E.F.J. Human Body Temperature: Its Measurement and Regulation; Springer Science & Business
Media: New York, NY, USA, 1982.
156. Cooper, K.; Veale, W.; Malkinson, T. Measurements of Body Temperature. Methods Psychobiol. Adv. Lab.
Tech. Neuropsychol. 1977, 3, 149–187.
157. Husain, M.D.; Kennon, R.; Dias, T. Design and fabrication of Temperature Sensing Fabric. J. Ind. Text. 2014,
44, 398–417. [CrossRef]
158. Kinkeldei, T.; Zysset, C.; Cherenack, K.; Troester, G. Development and evaluation of temperature sensors
for textile integration. In Proceedings of IEEE Sensors 2009 Conference, Christchurch, New Zealand,
25–28 October 2009; pp. 1580–1583.
159. Soukup, R.; Hamacek, A.; Mracek, L.; Reboun, J. Textile based temperature and humidity sensor elements
for healthcare applications. In Proceedings of the 2014 37th International Spring Seminar on Electronics
Technology, Dresden, Germany, 7–11 May 2014; pp. 407–411.
160. Bielska, S.; Sibinski, M.; Lukasik, A. Polymer temperature sensor for textronic applications. Mater. Sci.
Eng. B 2009, 165, 50–52. [CrossRef]
161. Dankoco, M.; Tesfay, G.; Benevent, E.; Bendahan, M. Temperature sensor realized by inkjet printing process
on flexible substrate. Mater. Sci. Eng. B 2016, 205, 1–5. [CrossRef]
162. Honda, W.; Harada, S.; Arie, T.; Akita, S.; Takei, K. Printed wearable temperature sensor for health monitoring.
In Proceedings of the 2014 IEEE Sensors, Valencia, Spain, 2–5 November 2014; pp. 2227–2229.
163. Yang, J.; Wei, D.; Tang, L.; Song, X.; Luo, W.; Chu, J.; Gao, T.; Shi, H.; Du, C. Wearable temperature sensor
based on graphene nanowalls. RSC Adv. 2015, 5, 25609–25615. [CrossRef]
164. Courbat, J.; Kim, Y.; Briand, D.; Rooij, N.D. Inkjet printing on paper for the realization of humidity and
temperature sensors. In Proceedings of the 2011 16th International Solid-State Sensors, Actuators and
Microsystems Conference, Beijing, China, 5–9 June 2011; pp. 1356–1359.
165. Aliane, A.; Fischer, V.; Galliari, M.; Tournon, L.; Gwoziecki, R.; Serbutoviez, C.; Chartier, I.; Coppard, R.
Enhanced printed temperature sensors on flexible substrate. Microelectron. J. 2014, 45, 1621–1626. [CrossRef]
166. Li, H.; Yang, H.; Li, E.; Liu, Z.; Wei, K. Wearable sensors in intelligent clothing for measuring human body
temperature based on optical fiber Bragg grating. Opt. Express 2012, 20, 11740–11752. [CrossRef] [PubMed]
167. Rajdi, N.N.Z.M.; Bakira, A.A.; Saleh, S.M.; Wicaksono, D.H. Textile-based Micro Electro Mechanical System
(MEMS) Accelerometer for Pelvic Tilt Mesurement. Procedia Eng. 2012, 41, 532–537. [CrossRef]
168. Amjadi, M.; Pichitpajongkit, A.; Lee, S.; Ryu, S.; Park, I. Highly Stretchable and Sensitive Strain Sensor Based
on Silver Nanowire–Elastomer Nanocomposite. ACS Nano 2014, 8, 5154–5163. [CrossRef] [PubMed]
169. Lee, J.; Kim, S.; Lee, J.; Yang, D.; Park, B.C.; Ryu, S.; Park, I. A stretchable strain sensor based on a metal
nanoparticle thin film for human motion detection. Nanoscale 2014, 6, 11932–11939. [CrossRef] [PubMed]
170. Shyr, T.-W.; Shie, J.-W.; Jiang, C.-H.; Li, J.-J. A Textile-Based Wearable Sensing Device Designed for Monitoring
the Flexion Angle of Elbow and Knee Movements. Sensors 2014, 14, 4050–4059. [CrossRef] [PubMed]
171. Zhang, H. Flexible textile-based strain sensor induced by contacts. Meas. Sci. Technol. 2015, 26, 105102.
[CrossRef]
172. Atalay, O.; Kennon, W.; Husain, M. Textile-Based Weft Knitted Strain Sensors: Effect of Fabric Parameters on
Sensor Properties. Sensors 2013, 13, 11114–11127. [CrossRef] [PubMed]
173. Atalay, O.; Kennon, W. Knitted Strain Sensors: Impact of Design Parameters on Sensing Properties. Sensors
2014, 14, 4712–4730. [CrossRef] [PubMed]
174. Seyedin, S.; Razal, J.M.; Innis, P.C.; Jeiranikhameneh, A.; Beirne, S.; Wallace, G.G. Knitted Strain Sensor
Textiles of Highly Conductive All-Polymeric Fibers. ACS Appl. Mater. Interfaces 2015, 7, 21150–21158.
[CrossRef] [PubMed]
175. Silva, A.F.D.; Pedro, R.; Paulo, J.; Higino, J. Photonic Sensors Based on Flexible Materials with FBGs for Use
on Biomedical Applications. In Current Trends in Short- and Long-Period Fiber Gratings; InTech: Rijeka, Croatia,
2013; pp. 105–132.
Sensors 2017, 17, 130 45 of 45

176. Krehel, M.; Rossi, R.; Bona, G.-L.; Scherer, L. Characterization of Flexible Copolymer Optical Fibers for Force
Sensing Applications. Sensors 2013, 13, 11956–11968. [CrossRef] [PubMed]
177. Halonen, T.; Romero, J.; Melero, J. GSM, GPRS and EDGE Performance: Evolution towards 3G/UMTS;
John Wiley & Sons: Hoboken, NJ, USA, 2002.
178. Salkintzis, K. Mobile Internet: Enabling Technologies and Services; CRC Press: Boca Raton, FL, USA, 2004.
179. Dahlman, E.; Parkvall, S.; Skold, J.; Beming, P. 3G Evolution: HSPA and LTE for Mobile Broadband; Academic
Press: Cambridge, MA, USA, 2010.
180. Ren, Y.; Werner, R.; Pazzi, N.; Boukerche, A. Monitoring patients via a secure and mobile healthcare system.
IEEE Wirel. Commun. 2010, 17, 59–65. [CrossRef]
181. Jang, C.S.; Lee, D.G.; Han, J.-W.; Park, J.H. Hybrid security protocol for wireless body area networks.
Wirel. Commun. Mob. Comput. 2011, 11, 277–288. [CrossRef]
182. Agrawal, V. Security and Privacy Issues in Wireless Sensor Networks for Healthcare. Internet Things
User-Centric IoT 2015, 36, 223–228.
183. Shoshani, B.; David, R.B. Vertical Conductive Textile Traces and Methods of Knitting Thereof.
U.S. Patent No. US 14/646,971, 23 November 2013.
184. Matsuhisa, N.; Kaltenbrunner, M.; Yokota, T.; Jinno, H.; Kuribara, K.; Sekitani, T.; Someya, T. Printable
elastic conductors with a high conductivity for electronic textile applications. Nat. Commun. 2015, 6, 7461.
[CrossRef] [PubMed]
185. Parkova, I.; Vališevskis, A.; Ziemele, I.; Briedis, U.; Vilumsone, A. Improvements of Smart Garment Electronic
Contact System. Adv. Sci. Technol. 2013, 80, 90–95. [CrossRef]
186. Laine, T.H.; Lee, C.; Suk, H. Mobile Gateway for Ubiquitous Health Care System Using ZigBee and Bluetooth.
In Proceedings of the 2014 Eighth International Conference on Innovative Mobile and Internet Services in
Ubiquitous Computing, Birmingham, UK, 2–4 July 2014; pp. 139–145.
187. Touati, F.; Tabish, R. U-Healthcare System: State-of-the-Art Review and Challenges. J. Med. Syst. 2013, 37,
1–20. [CrossRef] [PubMed]
188. Johansson, A.; Shen, W.; Xu, Y. An ANT Based Wireless Body Sensor Biofeedback Network for Medical
E-Health Care. In Proceedings of the 2011 7th International Conference on Wireless Communications,
Networking and Mobile Computing, Wuhan, China, 23–25 September 2011; pp. 1–5.
189. Savci, H.; Sula, A.; Wang, Z.; Dogan, N.; Arvas, E. MICS Transceivers: Regulatory Standards and Applications.
In IEEE SoutheastCon Proceedings; IEEE: Piscataway, NJ, USA, 2005; pp. 179–182.
190. Fang, Q.; Lee, S.Y.; Permana, H.; Ghorbani, K.; Cosic, I. Developing a Wireless Implantable Body Sensor
Network in MICS Band. IEEE Trans. Inform. Technol. Biomed. 2011, 15, 567–576. [CrossRef] [PubMed]
191. Zhen, B.; Li, H.B.; Kohno, R. Networking issues in medical implant communications. Int. J. Multimed.
Ubiquitous Eng. 2009, 4, 23–38.
192. Lazaro, A.; Girbau, D.; Villarino, R. Analysis of Vital Signs Monitoring Using an Ir-Uwb Radar.
Prog. Electromagn. Res. 2010, 100, 265–284. [CrossRef]
193. Chu, Y.; Ganz, A. A UWB-based 3D location system for indoor environments. In Proceedings of the 2nd
International Conference on Broadband Networks, Boston, MA, USA, 3–7 October 2005; pp. 1147–1155.
194. Yoo, H.-J.; Cho, N. Body channel communication for low energy BSN/BAN. In Proceedings of the APCCAS
2008—2008 IEEE Asia Pacific Conference on Circuits and Systems, Macao, China, 30 November–3 December
2008; pp. 7–11.
195. Sole, M.; Musu, C.; Boi, F.; Giusto, D.; Popescu, V. RFID sensor network for workplace safety management.
In Proceedings of the 2013 IEEE 18th Conference on Emerging Technologies & Factory Automation (ETFA),
Cagliari, Italy, 10–13 September 2013; pp. 1–4.
196. Liu, Y. A Heart Rate Finger Ring and Its Smartphone APP through Customized NFC. Master’s Thesis,
Rose-Hulman Institute of Technology, Terre Haute, IN, USA, 2015.
197. Fontecha, J.; Hervas, R.; Bravo, J.; Villarreal, V. An NFC Approach for Nursing Care Training. In Proceedings
of the 2011 Third International Workshop on Near Field Communication, Hagenberg, Austria, 22–23 February
2011; pp. 38–43.

© 2017 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC-BY) license (http://creativecommons.org/licenses/by/4.0/).