IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL.

71, 2022 9505207

Physically Flexible Ultralow-Power Wireless Sensor

Dung Nguyen , Colm Mc Caffrey , Member, IEEE, Olli Silvén , Senior Member, IEEE, and Martin Kögler
Abstract— The key challenges of local sensor networks are in possible, and should often be bendable or even stretchable.
supporting high sensor density, information security, physical This implies flexible electronics that have become a rapidly
size, and especially energy efficiency at a level that could eliminate evolving market because of the versatility of applications
the need for batteries or external power supplies. This article
presents a novel scheme that answers all issues at the cost with components on flexible polymeric foils [2]. Applications
of minor information losses in low data rate applications that using flexible designs in consumer electronics, medical and
tolerate latency. Experimental verification is made using a sensor healthcare, automotive, aerospace, and defense industry are
node implemented on a flexible electronics platform. Lightly estimated to grow with a CAGR of 7.4% from 2020 until
encrypted data are transmitted by embedding it into Bluetooth 2027 [3] to a global market size of around 42.48 billion USD
advertising packets, contributing to ultralow-energy wireless
power consumption, and theoretically enabling an unlimited by the end of 2027. At the same time, awareness of security
number of nodes in the local network. In the experiments, in IoT is rising with the necessity to add a strong level of data
the energy dissipation per transmitted 14-B information packet protection on the data collector side [4].
varied between 19.83 and 105.93 µW depending on the system A flexible printed circuit (FPC) is defined as “a patterned
configuration, while the data loss rates ranged from 7.4% to arrangement of printed circuitry and components that utilize
0.004%, respectively. As the flexible substrate can be attached
to various surfaces, the applications extend from wearable to flexible base material with or without flexible cover lay” [5].
industrial condition monitoring devices. Printed flexible circuits were first described in the literature in
1948 [6] with a suggestion that printed circuits could be made
Index Terms— Flexible printed circuits (FPCs), information
security, low-power electronics, wireless sensor networks (WSNs). on paper. Recently, there has been a resurgence of paper-based
flexible electronics utilizing both inkjet printing [7] and pen-
I. I NTRODUCTION to-paper [8] patterning of conductors.
The current industry standard employs polyimide as the

E NABLING high-density networks of ultralow-power

sensors is a persisting challenge. The sensors may mea-
sure, for instance, biosignals as wearable devices or monitor
preferred substrate where conductors consist largely of etched
rather than printed copper to allow for soldered components.
Kapton, one particular polyimide, is the industry standard in
mechanical components, such as intelligent strain gauges. flexible circuit fabrication, selected for its excellent durability,
In such applications, the designs need to be highly minia- stability, and thermal properties [9]. Kapton-based substrates
turized and mountable to surfaces of various shapes. These have been effectively used in the 3-D integration of sensor
call for technology combinations and solutions that challenge platforms using a folding technique [10].
the current state-of-the-art design practices. To answer these Advances in flexible circuit technology include flexible thin-
needs, in this article, we present a design that is not only film transistors [11]. In the current state of the art, flexible
flexible to configure and deploy but also physically flexible. electronics are viewed as a platform technology [12] exploit-
The hardware is intended to be a flexible platform that can be ing recent advances in graphene, nanostructures, and printed
adapted for use in a wide range of applications and deployed conductors that can be manufactured in roll-to-roll (R2R)
in a broad number of scenarios. Particular applications’ targets process [13]. A wide range of applications is described [12],
include wearables, skin contact and clothing integration, med- including healthcare, human interactivity, e-textiles, energy,
ical, well-being and sports, smart environment, and industrial display, and computation.
and environmental monitoring. Flexible hybrid electronics [14] is a new approach aiming to
The vision of the Internet of Things (IoT) is to sense combine the best of both flexible and conventional electronics
data in any application and location [1]. The devices needed unlocking significant opportunities while presenting major
must possess a small form factor, consume as little power as challenges. Hybrid electronics in the field of wearable technol-
ogy [15] is a particular area of research interest where physical
Manuscript received December 17, 2021; revised March 3, 2022; accepted
March 14, 2022. Date of publication April 1, 2022; date of current version (temperature, activity, pressure, flow, and light) and chemical
April 20, 2022. This work was supported by the European Union’s Hori- (sweat and glucose) parameters can be monitored. Flexible
zon 2020 Research and Innovation Program through the Project Smart and hybrid electronics is especially useful in “beyond wearable”
Flexible Energy Supply Platform for Wearable Electronics (Smart2Go) under
Grant 825143. The Associate Editor coordinating the review process was applications where lab-on-skin approaches are being applied
Dr. Zhengyu Peng. (Corresponding author: Dung Nguyen.) to measure biopotentials (ECG, EMG, and EEG), hydration,
Dung Nguyen and Martin Kögler are with VTT Technical Research Centre blood oxygenation, blood pressure, chemical components, and
of Finland, 90570 Oulu, Finland (e-mail: nguyenngocquocdung@gmail.com).
Colm Mc Caffrey is with the Device Concept Laboratory, Huawei wound monitoring [16].
Technologies Finland Oy, 00180 Helsinki, Finland. Along with flexible electronics, the advancement of flex-
Olli Silvén is with the Center for Machine Vision and Signal Analy- ible batteries is also notable. There are many electrode
sis, Faculty of Information Technology and Electrical Engineering, University
of Oulu, 90570 Oulu, Finland. designs each with their own advances and drawbacks [17].
Digital Object Identifier 10.1109/TIM.2022.3164157 According to recent research [18] a 3.6-V AgO-Zn recharge-
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
9505207 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 71, 2022

Fig. 1. FlexNode architecture.

able flexible battery can achieve up to 54 mAh/cm2 with

1-mA constant discharge current and 20-mA peak current, Fig. 2. FlexNode photograph.
enabling 70-mW momentary dissipation of the powered
device.
Cutting the power dissipation not only extends the life-
time of the power source but also allows using low current
output flexible batteries or alternative energy sources [19].
Ultralow-power wireless communication on such a flexible
sensor platform can be performed when the microcontroller
unit (MCU) is capable to wake up only when necessary
from sleep-mode or deep-sleep mode [20], [21]. With a low
power consumption design, a small energy harvester would
be enough to make the system self-powered [22], [23]. Also,
designing the embedded software to power down the unneces-
sary parts of hardware to further reduce energy consumption
can decrease the consumption to the µA level.
The functions of an IoT sensor device can be rather simple,
i.e., measuring the temperature, humidity, and so on, in a spe-
cific place over an extended time. In such cases, bidirectional
communications are not necessary since just the sensor data
are collected, providing for further opportunity to save energy.
Often, these simpler IoT devices are deployed as wireless
sensor networks (WSNs), which consists of a group of multiple Fig. 3. FlexNode modularity. (a) Full size. (b) NFC cut out. (c) Core, sensor,
embedded devices that employ sensors to record the physical and expansion port only.
conditions of the environment. The most popular architecture
is centralized WSN, which has one central unit and many sens- A. Flexible Substrate
ing units. In recent years, fog computing and edge computing
have gradually been applied to the domain IoT [24]. A fabricated Flexnode is shown in Fig. 2. To enable this
Wireless communication is responsible for the majority versatility, the key design goals were flexibility and modu-
of power dissipation of sensor nodes [25]. In the future, larity. Physical and mechanical flexibilities are achieved by
wake-up radio receivers (WURs) would enable turning on the implementing the hardware on a thin polyimide substrate.
transmitting system only when needed [26], saving substantial Modularity aims at the capability to scale the functionality
energy. However, currently, Bluetooth low energy (BLE) is of the design based on application needs. In Flexnode, this
available in most mobile devices and is one of the lowest translates into cutting out unnecessary modules, also reduc-
energy wireless communication approaches available for wire- ing the size of the device. Fig. 3(a) shows the full-sized
less sensor designs [27], [28]. (61 mm × 19 mm) design. This includes all the functional-
ities described above.
II. M ATERIALS Fig. 3(b) shows how the rightmost section of the design can
The objective of the developed hardware is to enable quick be cut out, eliminating the NFC receiver coil that connects to
turnaround of implementing IoT concepts for demonstra- either the nRF or NXP receiver. For applications in which NFC
tions and customization for “sensor to the cloud” solutions. communications and/or wireless charging of the battery are not
As described in Fig. 1, the architecture is divided into four needed, this gives a 39-mm-long device. A further reduction
subsections: sensors, processing and communications’ core, to 20 mm shown in Fig. 3(c) can be made if an external
power supply, and external interface. expansion includes compatible energy storage or harvester.
NGUYEN et al.: PHYSICALLY FLEXIBLE ULTRALOW-POWER WIRELESS SENSOR 9505207

TABLE I
F LEX N ODE C OMPONENT S PECIFICATION

Then, the remaining functionalities include the processing and connected via this expansion port. The interface (Expansion B)
communications’ core, sensors, and expansion ports. is on the bottom side of the Flexnode and is intended for
energy harvester add-ons.
B. Sensors
D. Energy Budget and Power Supply
The selection of the sensor components was motivated by
the aim at ultralow-energy dissipation, while the use cases The power supply subsection includes a wireless power
were anticipated to employ low sampling rates. This resulted transfer receiver and conversion using the NFC coil tuned to
in the preference for sensors with low-power modes. The series resonance at 13.56 MHz. The wireless power signal
key components are summarized in Table I, including their from the coil is half-wave rectified and converted for battery
functionalities, specifications, and current consumption from charging using an LTC4079.
manufacturers’ datasheet. Energy is currently stored on a 3.7-V Lithium polymer
The integrated sensor devices include LGA 12 encapsulated battery measuring 10 × 20 × 2 mm3 with a 22-mAh capacity.
LSM303 [29] which measures three-axis acceleration and The battery is monitored with a MAX17043 battery-level
a three-axis magnetometer. Also included is a temperature gauge reporting remaining capacity. The battery voltage is
sensor with an accuracy of 1 ◦ C. The other integrated low- regulated to a stable 2.5 V using an NCP170 linear dropout
power sensor device is 2.5 mm × 2.5 mm LGA packaged regulator.
Bosch BMA280 [30] environmental sensor, which combines III. F UNCTIONALITY
humidity, barometric pressure, and temperature measurements.
The FlexNode is intended for uses in which the main
need is reading the sensor data and sending it to further
C. Processing and Communications’ Core and Interfaces processing without being requested and without protocol-level
The core of Flexnode is the System-on-Chip Nordic handshakes.
Semiconductors nRF52832 with 3 × 3.2 × 0.4 mm3 phys-
ical dimensions [31]. It contains an ultralow-power ARM A. Communications
Cortex-M4F microcontroller, and 2.4-GHz Bluetooth 5 and The wireless communications between the Flexnode and
ANT/ANT+ radios. For storing functionality, a 64-Mb receiver (a BLE-enabled laptop) are based on 2.4-GHz BLE
Macronix MX25R6435F in a 4 × 4 package is included. radios. While BLE provides for substantial flexibility, in the
A PI configuration matching network is used to match the intended applications, there is no requirement for data stream-
RF output to a Johanson 2450AT18B100E chip antenna. The ing or the sensors establishing connections with the hubs.
nRF52832 additionally includes a 13.56-MHz NFA Type 1 Flexnode can utilize connection-based BLE protocols for
106-kb/s receiver, which enables “wake-on-NFC” and “touch- reliable communications; those are reserved for diagnostics
to-pair” capabilities. The NFC coil and NXP Semiconductors and software reconfiguration, and, in the future, for providing
NTAG NT3H1101 support energy harvesting through NFC, public encryption keys.
microcontroller interrupt for wake-on-NFC, and I2C interface In the energy minimizing communication scheme developed
for communication over NFC. Also, an EEPROM for passive for Flexnode, only the advertising part of the connection
data retention is provided. procedure is utilized for data transfers. When comparing
Two interface connectors are included for expansion, allow- the duration of radio communication between connection
ing the addition of further sensing capabilities, or power establishment and advertisement period every 1 or 2 s, the
supplies. The primary interface (Expansion A) is shown on energy efficiency of using advertising packets to transmit
the right edge of the device in Fig. 3. The interface has small amounts of data is obvious. Furthermore, in the connec-
18 signals that include both battery and regulated power tion data transmission scheme, if the connection is dropped,
supplies, ADC inputs, I2C, SPI, and undedicated GPIO pins. the effort to reconnect will significantly increase the power
For software development and testing, the platform can be consumption.
9505207 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 71, 2022

Fig. 5. Advertising packet structure.

TABLE II
B IT A RRAY OF BME 280 D ATA

TABLE III
Fig. 4. Advertisement current profile.
A DVERTISING PACKET E XAMPLE

To make its presence known to the receivers, a BLE device

periodically broadcasts advertising packets in three channels
37, 38, and 39 (2402, 2426, and 2480 MHz), while the
TABLE IV
remaining channels (0–36) are dedicated to connected proto-
D ATA E NCRYPTION
cols. For the communication hubs listening to the advertising
channels, this is a one-way discovery mechanism, and it
provides Flexnode, a very-low-power means to deliver data.
Fig. 4 shows the current drawn during an advertising
event that uses the three channels. The total duration of an
advertising event, including the MCU activity, is 3–4 ms, and humidity and air pressure sensor, and so forth, while the
the advertising interval can be from 20 to 2000 ms. In our space for data depends on the sensor, for example, 45 b
case, the peak power dissipation was at a 37.5-mW level. for temperature, humidity, and air pressure). The 200 b are
The power consumption of an advertising event is differed by successively filled in this manner and suffice for all the sensors
the advertising interval, antenna power, and advertising packet in the current implementation.
length. A more detailed analysis of the energy impacts of BLE An example of the sensor data packetizing is shown in
advertising is in [32], while Song et al. [33] have introduced Table II. 23.00 ◦ C temperature, 21.23% humidity, and 900-hPa
a method to optimize its power consumption. pressure readings are converted into a 45-b section of the
array under assembly for transmission. The initial 001 is the
identification code of the environmental sensor.
B. Data Packets and Security Although the sensor data are in a bit array with different
The BLE advertising packets comprise at minimum 1 and field lengths for each sensor type, it could still be easily
at most 31 B, as shown in Fig. 5. The first three bytes decoded by trial and error. To prevent the straightforward
contain the name of the device, and in our case, the first misuse of data through eavesdropping BLE receivers, a low-
byte is always the ASCII code “F” (for Flex). The two level energy-efficient information obfuscation approach is
next ones contain the device number, limiting the theoretical employed. Its intention is to slightly delay the access of
number of devices to 65 535 as device number 0 is not used. unauthorized parties to the measurements.
In the current application, however, the maximum is 255 to The current symmetric encryption scheme employs the 8-b
avoid excessive BLE advertising collisions. The extra currently advertising packet number as the key. XOR between each
unused device number byte has been considered to be useful packet number and the second byte of device ID generates a
for system expansions. Two bytes are for the manufacturer’s changing XOR cipher that is renewed for every new packet with
device ID. In total, 28 B have been left for the free use of the bits of the cipher used in alternating order depending on
the manufacturer. Flexnode uses 1 B for packet ID, and at the packet number. A simple example of this energy-efficient
most 25 B are available for sensing data. In our experiments, scheme is described in Table III.
only 14 B information packets were used for transmitting The key is calculated as 0 × 01 XOR 0 × D4 = 0 × D5,
temperature, humidity, air pressure, and battery charge level so the bit reversed key is 0 × AB. As the device ID is “1”
measurements. (F01), we XOR the first, third, fifth, and seventh bytes with
Since the bit space in the advertising packet is a signifi- 0 × D5 and the other bytes with 0 × AB. This encryption
cant constraint, the sensor data are transmitted as bit-array. procedure is described byte-by-byte in Table IV. The data
Currently, 3 b are used to identify each of the seven sensors buffer in the advertising packet becomes the encrypted byte
(000 stands for no data, 001 for the combined temperature, array 0 × F99E52A245E798D6.
NGUYEN et al.: PHYSICALLY FLEXIBLE ULTRALOW-POWER WIRELESS SENSOR 9505207

TABLE V
E VALUATE D EVICES S ETTINGS

Fig. 6. Database design.

While the approach provides for a fairly low level of

information security, it may suffice for noncritical applications.
However, through brute force effort, an attacker knowing the
bit allocation scheme can check if a deciphered result could
be correct.
Employing a more powerful public-key encryption scheme
is a future alternative. While the short bit length reserved for
data does not allow for high-level security, new keys could be
provided through BLE connections for each device.

IV. E XPERIMENTS AND R ESULTS

Fig. 7. Server flowchart.
A. Evaluation Procedure TABLE VI
To evaluate the Flexnode implementation, a Windows PC P OWER C ONSUMPTION S UMMARY
software application for Bluetooth scanning and database has
been employed. Fig. 6 shows that the data tables model
the network and the sensors. The information of the sensors
and nodes is in SensorInfo and NodeInfo, respectively, while
FlexStatistics contains all data collected from the sensors
and nodes, enabling to track the source and time of each
measurement. The settings of each device are given in Table V. The system
The multithreaded PC application is developed using the was consecutively run running for 48 days.
Microsoft .NET framework. Three threads run simultaneously:
Scanning and Data Collecting run always in the background, B. Results
while the Discover Thread is launched whenever the user Based on the experiment, the packet loss rates of nodes F01
wants to add new nodes into the network. The advertising and F02 were 5.8% and 7.4%, respectively, while F03 suffered
packets of the devices in the database are collected and queued only 0.004% loss. The battery voltages of the FlexNodes
by the Scanning Thread. Each packet is, in turn, decoded and dropped less than 200 mV during the 48-day test. The most
stored to the FlexStatistics by the Data Collecting Thread. The energy-hungry one, F03, experienced a 152-mV drop, while
internal organization of the application is shown in Fig. 7. F02 used its battery most conserving losing only 45 mV.
The experimental network used to determine the energy Table VI describes the voltage drop and power consumption
characteristics of the Flexnode design consists of three devices of the three nodes.
identified as F01, F02, and F03, a laptop PC collecting the sta- Based on the current waveform analyzer (CX3324A,
tistics, and running the application. The nodes were configured Keysight Technologies), the power on sleep-mode current need
to transmit advertising packets at different rates, going to sleep of Flexnode is 3.7 µA and, in the active sensing mode, about
between transmissions, to gain a concrete understanding of the 1.4 mA for a very short time, about 300 µs. The current
impact on the endurance. The nodes were powered by two consumption during transmission depends on the number of
Varta Ultra Lithium AA batteries having 2900-mAh capacity. advertising packets in advertising events.
9505207 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 71, 2022

Node F03 had the most power-consuming settings with TABLE VII
average current consumption during transmission at 100.3 µA. P OWER C ONSUMPTION C OMPARISON
Consequently, its average power dissipation is
Iactive ∗ tactive + Isleep ∗ tsleep + Iprocess ∗ tprocess
Iavg = (1)
Ttotal
where I active and t active are the BLE advertising current and
time; I sleep and t sleep are the current and time between two However, for the application that requires data transmission
advertising event; I process and t process are the current and through extended distances, the Long Range Wide Area Net-
time for data processing; and T total is the total time for one work (LoRaWAN) is the most compatible approach [35].
cycle [34]. The average current can be obtained by substituting The selection of the hardware components, including the
the measured values into 1 sensors and the communications scheme, plays an important
3.7 µA ∗ 30 s + 100.3 µA ∗ 20 s + 1400 µA ∗ 0.0003 s role in power dissipation minimization. The Nordic Semi-
50.0003 s conductor nRF52 series with the Arm Cortex M4 processor
= 42.35 µA. (2) is among the best low-power system chips that incorporate
2.4-GHz radio. A power consumption comparison between
From a 2000-mAh battery at 70% efficiency, the node will run our solution against the roughly equivalent ones cited in this
for over 30 000 h or 3.5 years. article is given in Table VII. In the scope of this article,
If the power supply module is an organic photovoltaic (PV) we demonstrate a simple but effective encryption method
harvester, the size of the cell is targeting to minimize power consumption, but the chip is
E electrical capable of using a stronger encryption algorithm with more
A= (3)
H ∗ R ∗ PVefficiency power and time needed.
The Flexnode platform has many extension pins that can be
where used to connect sensors, energy harvesters’ actuators, and indi-
A: area of PV Panel; cators. The future enhancements will include plug-and-play
E electrical : energy required; functionality for the extensions, e.g., through the utilization
H: annual average solar radiation on tilted panels of I2C protocol chip addresses. Moreover, as an I2C channel
−200 kWh/m2 /year; can support multiple sensors, the complexity of the platform
R: solar panel yield or efficiency 15%; design would be well controlled. Also, an energy monitoring
PVefficiency : performance ratio, coefficient for losses −0.75. scheme needs to be implemented to allow balancing the use
In theory, the system could become completely autonomous of sensors against the energy being harvested.
that would not require any battery replacement with an about
1.5-cm2 PV cell minimum size. However, with the current VI. C ONCLUSION
power supply hardware, it supports adjustable battery charging
current to operate from 10 to 250 mA. In many applications, the thickness, size, and flexibility of
the physical platform are important. The current contribution
demonstrates not only those characteristics but also wireless
V. D ISCUSSION
energy-efficient data communications in such a system. The
The aim of this contribution is to present an energy- platform can be expanded with additional sensors and energy
autonomous wireless sensor platform that can be used from harvesters to fit the requirements of relevant applications. With
industrial condition monitoring instruments to wearable human the capability of extending sensor and platform flexibility, this
biosignal measurements devices. The energy consumption has platform can be applied to wearable, bendable, rollable use
been minimized by employing BLE advertising packets for cases with extension platforms, such as sweat analytics and
data transmission. This not only reduces the energy consump- ECG/PPG measurement.
tion of the radio but eliminates almost all protocol complexity. Many monitoring needs do not require high data rates.
While the lack of protocol connection results in data losses, The demonstrated novel low data bandwidth communication
many applications can tolerate moderate ones. Furthermore, scheme is compatible with common standards, enabling to
if the nodes of the network are known, the server side is access sensor nodes even using a regular cell phone with BLE.
capable of detecting failed nodes or excessive data loss rates.
For a particular application, the physically flexible platform
that allows for cutting out redundant parts enables minimizing C ONFLICT OF I NTEREST
the size and weight of the sensor node. The experimental The authors declare to have no known competing financial
sensor nodes were tested continuously for more than a month or any conflict of interests that would have given influence to
and a half without hardware or software complications. the work presented in this article.
The system design suits applications that have low data rates
and benefit from light, small, and thin sensor nodes in low and R EFERENCES
medium ranges. These include, for example, monitoring the
minimum and maximum vibrations of industrial motors or the [1] M. Bhagat, D. Kumar, and S. M. Balgi, “Application of Internet of
Things in digital pedagogy,” in Computational Intelligence in Digital
temperature threshold of cylindrical objects, such as shafts. Pedagogy. Cham, Switzerland: Springer, 2021, pp. 219–234.
NGUYEN et al.: PHYSICALLY FLEXIBLE ULTRALOW-POWER WIRELESS SENSOR 9505207

[2] B. Rivkin et al., “Shape-controlled flexible microelectronics facilitated [28] K. Nair et al., “Optimizing power consumption in IoT based wireless
by integrated sensors and conductive polymer actuators,” Adv. Intell. sensor networks using Bluetooth low energy,” in Proc. Int. Conf. Green
Syst., vol. 3, no. 6, Jun. 2021, Art. no. 2000238. Comput. Internet Things (ICGCIoT), Oct. 2015, pp. 589–593.
[3] R. Humbare, S. Wankhede, and V. Kumar, “Flexible electronics [29] Ultra-Compact High-Performance eCompass Module: Ultra-Low-Power
market by component and application: Global opportunity analy- 3D Accelerometer 3D Magnetometer, ST Microelectronics, Geneva,
sis and industry forecast, 2020–2027,” Allied Market Res., Port- Switzerland, 2016.
land, OR, USA, Tech. Rep. A00850, Nov. 2020. [Online]. Available: [30] Combined Humidity Pressure Sensor, Bosch, Gerlingen, Germany, 2014.
https://www.alliedmarketresearch.com/flexible-electronics-market [31] nRF52832 Product Specification, Nordic Semiconductors, Oslo, Norway,
[4] M. B. M. Noor and W. H. Hassan, “Current research on Internet of 2017.
Things (IoT) security: A survey,” Comput. Netw., vol. 148, pp. 283–294, [32] C. Gomez, J. Oller, and J. Paradells, “Overview and evaluation of
Jan. 2019. Bluetooth low energy: An emerging low-power wireless technology,”
[5] Terms and Definitions for Interconnecting and Packaging Electronic Sensors, vol. 12, no. 9, pp. 11734–11753, 2012.
Circuits, Revision F, document IPC-T-50, 1996. [33] S. W. Song, Y. S. Lee, F. Imdad, M. T. Niaz, and H. S. Kim, “Efficient
[6] C. Brunetti and R. W. Curtis, “Printed-circuit techniques,” Proc. IRE, advertiser discovery in Bluetooth low energy devices,” Energies, vol. 12,
vol. 36, no. 1, pp. 121–161, Jan. 1948. no. 9, p. 1707, May 2019.
[7] Z. Zhang, X. Zhang, Z. Xin, M. Deng, Y. Wen, and Y. Song, “Synthesis [34] S. K. Gharghan, R. Nordin, and M. Ismail, “Energy efficiency of ultra-
of monodisperse silver nanoparticles for ink-jet printed flexible electron- low-power bicycle wireless sensor networks based on a combination of
ics,” Nanotechnology, vol. 22, no. 42, Oct. 2011, Art. no. 425601. power reduction techniques,” J. Sensors, vol. 2016, pp. 1–21, Jul. 2016.
[8] A. Russo, B. Y. Ahn, J. J. Adams, E. B. Duoss, J. T. Bernhard, and [35] L. G. Manzano et al., “An IoT LoRaWAN network for environmental
J. A. Lewis, “Pen-on-paper flexible electronics,” Adv. Mater., vol. 23, radiation monitoring,” IEEE Trans. Instrum. Meas., vol. 70, pp. 1–12,
no. 30, pp. 3426–3430, 2011. 2021.
[9] Q.-H. Lu and F. Zheng, “Polyimides for electronic applications,” in
Advanced Polyimide Materials. Amsterdam, The Netherlands: Elsevier,
2018, pp. 195–255. Dung (Daniel) Nguyen received the B.Sc. degree
[10] R. V. Martinez-Catala and J. Barrett, “A modular wireless sensor in computer engineering from Vietnam National
platform with fully integrated battery,” IEEE Trans. Compon. Packag. University, Hanoi, Vietnam, in 2012, and the M.Sc.
Technol., vol. 32, no. 3, pp. 617–626, Sep. 2009. degree in computer science and engineering from
[11] K. Myny, “The development of flexible integrated circuits based on thin- the Faculty of Information Technology and Electri-
film transistors,” Nature Electron., vol. 1, p. 30, Jan. 2018. cal Engineering, University of Oulu, Oulu, Finland,
[12] A. Nathan et al., “Flexible electronics: The next ubiquitous plat- in 2019.
form,” Proc. IEEE, vol. 100, no. 5, pp. 1486–1517, May 2012, doi: He is currently working as the Research Scien-
10.1109/JPROC.2012.2190168. tist of the Flexible Electronics Integration Research
[13] T. M. Kraft, L. Leppanen, T. Kololuoma, S. Lahokallio, L. Frisk, and Team, VTT Technical Research Centre of Finland,
M. Mantysalo, “High density R2R screen printed silver interconnections Oulu. His research focus is on smart embedded
for hybrid system integration,” in Proc. 6th Electron. Syst.-Integr. software and electronics design.
Technol. Conf. (ESTC), Sep. 2016, pp. 1–6.
[14] G. Tong, Z. Jia, and J. Chang, “Flexible hybrid electronics: Review and Colm Mc Caffrey (Member, IEEE) received the
challenges,” in Proc. IEEE Int. Symp. Circuits Syst. (ISCAS), May 2018, B.Eng. degree (Hons.) in electronic engineering and
pp. 1–5. the M.Eng.Sc. degree in microelectronics from Uni-
[15] W. Gao, H. Ota, D. Kiriya, K. Takei, and A. Javey, “Flexible electronics versity College Cork, Cork, U.K., in 2006 and 2008,
toward wearable sensing,” Acc. Chem. Res., vol. 52, no. 3, pp. 523–533, respectively.
Mar. 2019. He has worked as an Intern Application Engi-
[16] Y. Liu, M. Pharr, and G. A. Salvatore, “Lab-on-skin: A review of flexible
neer with Analog Devices, Wilmington, MA,
and stretchable electronics for wearable health monitoring,” ACS Nano,
USA, from 2005 to 2006; a Graduate Researcher
vol. 11, no. 10, pp. 9614–9635, 2017.
and an Instrumentation Engineer with the Tyndall
[17] Y. Zhao and J. Guo, “Development of flexible Li-ion batteries for flexible
National Institute, Cork, from 2006 to 2010; the
electronics,” InfoMat, vol. 2, no. 5, pp. 866–878, Sep. 2020.
[18] L. Yin et al., “High performance printed AgO-Zn rechargeable battery Research Scientist of VTT Technical Research Cen-
for flexible electronics,” Joule, vol. 5, no. 1, pp. 228–248, Jan. 2021. tre of Finland, Espoo, Finland, from 2012 to 2019; and the Senior Scientist
[19] J. D. MacKenzie and C. Ho, “Perspectives on energy storage for flexible of VTT Technical Research Centre of Finland from 2019 to 2021, where this
electronic systems,” Proc. IEEE, vol. 103, no. 4, pp. 535–553, Apr. 2015. work was carried out. Since 2021, he has been working as a Principal Engineer
[20] M. Losada, A. Cortés, A. Irizar, J. Cejudo, and A. Pérez, “A flexible and a Wearable Prototype Architect with Huawei Technologies Finland Oy,
fog computing design for low-power consumption and low latency Helsinki, Finland.
applications,” Electronics, vol. 10, no. 1, p. 57, Dec. 2020.
[21] O. Kanoun et al., “Energy-aware system design for autonomous wireless Olli Silvén (Senior Member, IEEE) is currently
sensor nodes: A comprehensive review,” Sensors, vol. 21, no. 2, p. 548, a Professor of signal processing engineering with
Jan. 2021. the University of Oulu, Oulu, Finland. His current
[22] E. Sardini and M. Serpelloni, “Self-powered wireless sensor for air tem- research interests range from environment wireless
perature and velocity measurements with energy harvesting capability,” self-powered vision sensors to embedded machine
IEEE Trans. Instrum. Meas., vol. 60, no. 5, pp. 1838–1844, May 2011. learning in diverse applications. His past contribu-
[23] M. D. Prieto, D. Z. Millan, W. Wang, A. M. Ortiz, J. A. O. Redondo, and tions include solutions for industrial visual inspec-
L. R. Martinez, “Self-powered wireless sensor applied to gear diagnosis tion systems to video codecs.
based on acoustic emission,” IEEE Trans. Instrum. Meas., vol. 65, no. 1,
pp. 15–24, Jan. 2016.
[24] B.-Y. Ooi and S. Shirmohammadi, “The potential of IoT for instrumen-
tation and measurement,” IEEE Instrum. Meas. Mag., vol. 23, no. 3,
pp. 21–26, May 2020. Martin Kögler received the Dr.Ing. degree from
[25] M. A. Nasab, S. Shamshirband, A. Chronopoulos, A. Mosavi, and the Technical University of Berlin, Berlin, Germany,
N. Nabipour, “Energy-efficient method for wireless sensor networks in 2018.
low-power radio operation in Internet of Things,” Electronics, vol. 9, Since then, he has been working as the Senior Sci-
no. 2, p. 320, Feb. 2020. entist of VTT Technical Research Centre of Finland,
[26] F. Hutu, A. Khoumeri, G. Villemaud, and J.-M. Gorce, “A new wake-up Oulu, Finland, in the field of sensing and integration.
radio architecture for wireless sensor networks,” EURASIP J. Wireless Apart from customer contract research, his focus
Commun. Netw., vol. 2014, no. 1, pp. 1–10, Dec. 2014. area and research interests are applied sensor tech-
[27] F. Moreno-Cruz, V. Toral-López, A. Escobar-Molero, V. U. Ruíz, nologies, electronics integration, and data analytics.
A. Rivadeneyra, and D. P. Morales, “TreNch: Ultra-low power wire- His work at VTT includes strong collaboration with
less communication protocol for IoT and energy harvesting,” Sensors, various customers and research institutes across the
vol. 20, no. 21, p. 6156, Oct. 2020. globe.