1
Quadrature Amplitude Modulation Backscatter
for Passive Wireless Sensors
Ricardo Correia∗ , Nuno Borges Carvalho∗
∗
Instituto de Telecomunicações, Departamento de Electrónica, Telecomunicações
e Informática, Universidade de Aveiro
Abstract
Backscatter RFID sensors are beneficial for sensing applications where small, low-maintenance and
no-battery sensor’s nodes are needed. In this work two different approaches for battery free sensors are
presented. The first approach is based on ASK backscatter modulation and a prototype is presented and
measured. With this circuit it is shown that is possible to have a complete passive WSN by combining
backscatter modulation with WPT. In order to increase the data rate of the sensors, a new approach
based on QAM modulation was implemented. A backscatter modulator based in a wilkinson power
divider was proposed which contains a delay in one of the design branch and two transistors acting as
switches to generate M-QAM backscatter modulation. It is foreseen that the modulator is the basis for
a truly passive WSN.
Index Terms
Wireless sensor network, backscatter communication, wireless power transfer, quadrature amplitude
modulation, amplitude shift keying, internet of things.
I. I NTRODUCTION
Radio Frequency Identification (RFID) technology has been used for logistical purposes as
a replacement for bar codes so as to improve reliability and speed of item identification, as
well as enabling additional features such as enhanced inventory management. The promise of
RFID and its eventual widespread successful deployment within a commercial supply chain is
predicated on low cost, passive tags that support reliable, fast reading of tag IDs by readers. Along
with this price reduction, advances in semiconductor technology have allowed the possibility of
passive RFID enabled sensors to be integrated in wireless sensor networks (WSNs) [1]. However,
2
Power
data
RFID Tag
Antenna
RFID Reader
Modulated
backscatter
Fig. 1. Traditional RFID system with load modulation
ultra-cheap tags and a reliable, efficient system are very difficult to combine, due to extreme
limitations on tag capabilities. This includes very limited memory and on-tag data processing, and
(most relevant for our purposes), seriously constrained power availability. Nevertheless, RFID
tag designs are being continually improved on various fronts - via improved circuitry and energy
management approaches, among others. The major advantage of passive RFID-enabled sensors
over traditional sensors is their low cost and the maintenance free. These advantages are only
possible due to the use of backscattered radio waves.
In most commonly deployed RFID systems, the backscatter link uses a simple modulation
scheme, such as amplitude shift keying (ASK) or phase shift keying (PSK) backscatter generated
by a two-state modulation of the impedance presented to a transponder’s antenna (Fig. 1). Since
tag to reader distance is generally unknown, most readers employ homodyne I/Q demodulation.
In the case the reader rotate the received constellation until it falls on the real axis before
demodulation as if the signal were ASK regardless of whether the tag’s modulator is configured
to generate ASK or PSK modulation. The reflected portion of the carrier signal incident on
the tag antenna is impressed with an envelope (altering amplitude, phase, or both) carrying
the binary backscatter data. Backscatter radio technology has grown is the last decade because
the applications that do not require battery have increased. Using backscatter communication
allows for extremely low power dissipation for remote devices, which enables new areas of
sensor development. The backscatter tags, by receiving the local oscillator over the air, do not
need their own radio frequency (RF) oscillator or phase-locked loop (PLL). By removing these
3
circuits, one reduces the power consumption and the cost of a tag chip [2]. Nevertheless, the
backscattered signal received by the readers is weak, thus limiting communication range. In
conventional backscatter communication, the tags must harvest enough power from the reader to
turn on and modulate data, and the readers must receive strong backscattered signal to operate.
A recent work [3] has shown that modulated backscatter can be extended to include higher
order modulation schemes, such as 4-QAM. While ASK and PSK transmit 1 bit of data per
symbol period, 4-QAM based can transmit 2 bits per symbol period, thus increasing the data
rate and leading to reduced on-chip power consumption and extended read range. The work
presented in [3], [4] refers to a 4-QAM backscatter in semipassive systems, by using a coin cell
battery as a power source for the modulator and a microcontroller that needs 3 V of supply. This
way, the authors proved the QPSK modulator and battery powered system, by using an approach
with a four lumped impedances connected to a RF switch that is controlled by a microcontroller.
The same authors developed a 16-QAM modulator for UHF backscatter communication with a
consumption of 1.49 mW at a rate of 96 Mbps only in the modulator (not the overall system
with data generation logic feeding the modulator) [5]. This modulator was implemented with 5
switches with lumped terminations as a 16-to-1 Mux to modulate the load between 16 different
states.
Some recent work [6], [7] show the possibility of using one frequency to continuously power
the wireless sensor and other to transfer data by backscatter means with ASK modulation. Fig. 2
presents a potential solution presented in [6], [7] for this implementation. An energy harvester
and power management circuitry are responsible for collecting sufficient energy to power the
tag and any additional sensor. As previously mentioned, this energy can come from a variety of
sources but is typically reader-delivered RF power. Over the last years, RF energy harvesting,
which is the capability of converting RF signals into DC energy, has gained a lot of interest [8]
and resulted in some commercial solutions. Thus, there is a strong motivation to enable a WSN
with wireless power transmission (WPT) in order to afford all the operational costs.
As illustrated in Fig. 3, without the need to replace energy depleted sensors in conventional
WSNs, a passive WPT WSN can achieve a continuous operation with a large number of
sensors powered by fixed energy transmitters used for both wireless charging and data collection.
Nevertheless, due to the more ample power supply from WPT, RFID devices can now expect
much longer operating lifetimes, and afford to transmit actively at a much larger data rate and
from a longer distance than conventional backscatter based RFID communications. Therefore, we
4
Passive Sensor
Backscatter modulator
Control
S
External
Sensors
G
D
Microcontroller
uC
Matching Network 1
Matching Network 2
Rectifier
Energy Storage
DC out
Fig. 2. Block diagram for a passive sensor.
Passive WPT WSN
Conventional battery powered WSN
Active sensor
Data
transmission
Battery depleted
sensor
Information
access point
Wireless powered
sensor
Data
transmission
Fixed wireless
power transmitter
Fig. 3. Example applications of a conventional battery powered WSN and a passive WPT WSN system. The red lines represent
information, and the green represent energy.
envision that a WSN with WPT will be an important building block of many popular commercial
and industrial systems in the future, including the upcoming Internet of Things/Everything
(IoT/IoE) systems consisting of millions of sensing/RFID devices as well as large-scale WSNs.
In this article we first provide an overview of state-of- the-art in WPT and backscatter
modulation in order to face the above technical challenges. Section II highlights the definition
and the importance of backscatter. Section III shows the first approach for the passive sensor
5
using ASK modulation with measured results. Section IV describes the approach used for the
16-QAM RF circuit design and the final solution for the QAM backscatter implementation.
Section V presents the simulated and experimental results as well as the achievable data rates
obtained for the implemented backscatter modulator. Finally, we conclude the work in Section
VI.
II. BACKSCATTER C IRCUITS R EVISITED
The idea of communication by means of reflected power was revisited in development of
an animal tracking system that involved into a robust design for automated tracking of railway
trains [9]. The use of backscatter as a low-power communication means is possible due to the
capability of supplying the remote device by the continuous wave (CW) signal emitted from the
reader.
Backscatter systems were first used in the late 1990s-early 2000s with the advent of singlechip UHF RFID tags. The first designs were battery-free devices that used RF-to-DC converters
to supply system DC power and employed modulated backscatter to transfer ID information
[10]. Some recent work, shows the possibility to transfer data and continuously deliver power to
the sensor by using two different frequencies, one to supply the sensor and the other to transfer
data by backscatter means [7].
To understand the principle of backscatter, a schematic is shown in Fig. 4 where a current
flowing on a transmitting antenna leads to a voltage induced on a receiving antenna. In the
figure there are two different cases to analyse. In the first case, the smallest possible load (short
circuit), is shown that connecting the antenna to a load leads to an induced current on the
receiving antenna. The radiated wave can make its way back to the transmitting antenna, induce
a voltage, and therefore, produce a signal that can be detected: a backscattered signal. In the
second case, the largest possible load (open circuit), is shown that little or no induced current
will result, that leads to negligible backscattered wave. Therefore, the signal on the transmitting
antenna depends on the load connected to the receiving antenna [11].
A backscatter RFID system relies on the wireless communication between an RFID reader
and a backscatter tag [12]. A schematic of the backscatter communication system with emphasis
on the RFID tag is depicted in Figure 1. The wireless communication link between the RFID
reader and the backscatter tag can be divided into a forward link and a backward link. In the
forward link, the RFID reader transmits RF power and data to the tag. In the backward link,
6
transmit antenna
current
backscattered
sinal
radiated wave
receive antenna
current
backscattered wave
receive antenna
current
short circuit load
no backscattered
sinal
no backscattered wave
open circuit load
Fig. 4. Physics of backscatter signaling. Figure taken from [11].
whenever a reader command requires a tag’s response, the tag starts its data transfer using a
modulated backscatter signal, i.e., the signal from the reader is reflected by the tag depending
on the transmitted data. Typically, some tags achieve their data transmission by changing the
properties of the tags themselves, while others switch a load resistor in and out of the antenna
circuit and the reflected signal switches between two states, representing a logical ‘0‘ and ‘1‘.
Over short ranges, the amount of power reaching the tag from the reader is sufficient to allow
operation of small low current circuits within the tag. This can be used to drive an electronic
switch, e.g. a FET that can switch an antenna load resistor in and out of circuit. This will
effectively modulate the returned signal and allow data to be passed back to the reader.
The limitation of such a system is that the read range is reduced and the data rate is low.
In this work, an high order backscatter modulator is implemented in order to overcome these
limitations.
III. PASSIVE S ENSOR WITH ASK MODULATION BACKSCATTER CAPABILITIES
Fig. 2 presents the block diagram of the proposed system. This system is composed by two
main blocks, which are the backscatter modulator (switch transistor) and the RF harvesting circuit
with a power management unit. The main objective of the RF harvesting block is to generate
7
ASK Backscatter
Rectifier
Matching Network
Matching Network
Energy Storage
+
Microcontroller
+
D
G
S
Backscatter
Rectifier
(a)
(b)
Fig. 5. a) Configuration of the proposed system. b) Photograph of the proposed system.
()*+,-./01234.
5)2)-3+,!"#)+
<2*0=)2+.>3?)
92+)223.$
!"#)*
+566!,78
1.6 GHz
2.3 GHz
*5*9!,78
+-32FA0++)=
F01234
92+)223.H
$%
!"#
678
!"#$%&'''%
&'(
,-#.)/012304
")!)*+)=.>3?)
B,C)-.9AE40")B,C)!,AD02)$%
/01234.9234:;)-
D3*GF*3++)-)=
F01234
!"#
>3?)@,-A.
5)2)-3+,-
Fig. 6. Block diagram of the measurement setup
dc output power to supply the microcontroller to modulate the information acquired from an
external sensor, by means of switching ON and OFF the transistor (ASK modulation).
Our proposed system was based on the circuit presented in [6] and it is illustrated in Fig. 5.
The circuit includes a switch transistor to modulate the impedance of the antenna and causes a
change in the reflection coefficient. It also includes a five-stage Dickson multiplier to maximize
the dc output voltage collected and RF Schottky diodes (SKYWORKS, SMS7630-006LF) were
8
2.30
Fig. 7. Backscattered signal obtained in signal analyzer
350
300
Distance (cm)
250
200
150
100
50
0
0
5
10
15
20
Power Transmitted (dBm)
25
30
35
Fig. 8. Power transmitted as function of distance.
employed. The designed matching network has an important role in this circuit, since it needs
to be designed for the backscatter load modulation in one frequency and for the continuous
energy beam for the WPT in other frequency. The frequency selected for the communication
was 2.3 GHz and for the WPT was 1.6 GHz.
Fig. 6 presents the block diagram of the measurement setup used to acquire the reflection wave
from our proposed circuit. We used a vector signal generator (ROHDE&SCHWARZ, SMW200A)
to generate 1.6 GHz and 2.3 GHz. At the output of the vector signal generator we used a power
combiner (Mini-Circuits, ZFRSC-183-S+) to combine 1.6 GHz and 2.3 GHz to the input of the
power amplifier. A coupler (Marki, CBR16-0012) was used to measure the reflected backscattered
9
signal, using a signal analyzer (ROHDE&SCHWARZ, FSQ). A waveform generator (Agilent,
3325OA) was used to switch the transistor’s gate voltage to provide the backscatter modulation.
Fig. 8 represents the distance at which is possible to supply the microcontroller and the
sensors for different values of transmitted power. The reflected wave from the backscattered
wave is illustrated in Fig. 7, which is the sensor subcarrier. The measures were made for the
frequency of 1.6 GHz (WPT) and 2.3 GHz (Backscatter).
IV. QAM BACKSCATTER M ODULATOR
In most RFID systems and passive sensors, the reader to tag communication is an ASK or
phase shift keying (PSK) that modulates either the amplitude, or both the amplitude and phase,
of the reader’s transmitted RF carrier.
The proposal presented in this manuscript follows the same approach as in [6] and Fig. 2,
combining WPT with a higher order backscatter modulation format, by designing an RF circuit
that is able to be replicated to include more modulation levels. In this work a novel strategy
will be proposed in order to increase the modulation order to 16-QAM, by using the transistor
model to create different backscatter impedances.
Backscatter radios are based on the reflected waveform that should switch between a fully
matched and short circuit solution built on a switch that is directly connected to the receiving/transmitting antenna of the sensor. For a higher level modulation format, the number of
different backscattered waves should change among several states, thus the switch combinations
should clearly create a specific impedance for that symbol combination.
Fig. 9 presents the design for the 16-QAM solution. This model employs a wilkinson power
divider and each branch is terminated with a line and an ideal impedance. The lines present a
45◦ phase shift respect to each other, so as to allow that the reflected wave from each branch
has 90◦ phase difference from the other. Equation 1 shows the reflected wave as a sum of both
reflected waves from each branch. Using (2) and (3) is possible to determine the reflection
coefficient of the model for different impedances.
π
b2
a1 Γ1 ej 2
a2 Γ2
b1
+ √
b= √ +√ = √
2
2
2
2
Γ1 =
R1 − Z0
;
R1 + Z0
Γ2 =
R2 − Z0
R2 + Z0
(1)
(2)
10
a1
Γ
Z0√2 90º
Z0 45º
a
R1
b1
100 Ohm
Z0 0º
b
a2
Z0√2 90º
Γ
Z0 0º
b2
R2
Fig. 9. Model of the QAM solution
Γ1 √12 j π Γ2 √12
b
Γ1
Γ2
Γ= = √ e 2 + √ ⇒Γ=
+j
a
2
2
2
2
(3)
Considering two different values for each resistance (0 or ∞) we obtain four different combinations with four different reflection coefficients:
R1 = 0 and R2 = 0 ⇒ Γ = − 12 − j 21 ;
R1 = 0 and R2 = ∞ ⇒ Γ =
1
2
− j 21 ;
R1 = ∞ and R2 = 0 ⇒ Γ = − 12 + j 21 ;
R1 = ∞ and R2 = ∞ ⇒ Γ =
1
2
+ j 12 .
Fig. 10 presents the previous reflection coefficients in the smith chart.
A similar reasoning can be used to create any type of multi-level higher order modulation. In
order to do that, arbitrary real impedances could be used, instead of the 0 and ∞ impedance. Thus
different impedances are implemented using an active load based on a transistor, one example is
presented in Fig. 11, where different impedances were used in the scheme presented in Fig. 9,
the impedances were 0 Ω, 10 Ω, 100 Ω and 300Ω.
The last approach can be extended, where the load presents different impedance values,
consider for instance a, b, c and d real impedances. The impedances are then matched with
the overall circuit using a two branch approach and a microstrip line scheme, were each branch
imposes a specific delay. By combining these two degrees of freedom, 16 impedance levels can
be created. Equation 4 was deduced, where R represents the four different impedances in one
branch and Q represents the four impedances in the other brach. It was based on Fig. 12 and
presents sixteen possibilities of impedances from a φ+a θ to d φ+d θ, using all combinations.
11
+j1.0
+j2.0
+j0.5
+j5.0
5.0
2.0
1.0
0.5
0.0
0.2
+j0.2
∞
-j5.0
-j0.2
-j2.0
-j0.5
-j1.0
Fig. 10. Simulation of the model for 4-QAM solution
+j1.0
+j2.0
+j0.5
+j5.0
5.0
2.0
1.0
0.5
0.0
0.2
+j0.2
∞
-j5.0
-j0.2
-j2.0
-j0.5
-j1.0
Fig. 11. Simulation of the model for 16-QAM solution
X =R φ+Q θ
(4)
12
R
Φ
Wilkinson
Power
Divider
a
b
c
d
Q
X
θ
a
b
c
d
Fig. 12. Block diagram for different impedance terminations
Wilkinson Power Divider
Matching Network 1
Mosfet 1
Matching Network 2
Mosfet 2
Fig. 13. Photograph of the QAM backscatter circuit
Fig. 13 illustrates the diferences of line length in each branch, which are related to 45 ◦ phase
shift. By using this approach, the circuit in Fig. 13 was designed and each transistor is switched
for different voltage levels to achieve different reflection coefficients.
V. R ESULTS
Fig. 14 presents the photograph of measurement setup used to acquire the reflection coefficient
from our proposed circuit, illustrated in Fig. 13. We used a power supply and a Performance
Network Analyzer (PNA) (E8361C, from Agilent Technologies) that was calibrated at 0 dBm
for the frequencies from 2 GHz to 3 GHz. The results of the simulations and measurements
are presented in Fig. 15 and in Table I are presented the different reflection coefficients for
different voltages at the transistor. Table I shows also the simulated and experimental voltage
levels used to obtain the matching between the reflection coefficients. The switches were changed
according to the simulated results, with different voltage levels at each gate of transistor. Table I
shows different voltage levels for simulated and measured results, due to the models used for
the transistor.
13
Power Supply
DUT
Vector
Network
Analyzer
Fig. 14. Photograph of measurement setup
+j1.0
+j2.0
+j0.5
✴
Measured S-parameters
Simulated S-parameters
+j5.0
5.0
2.0
1.0
0.5
0.0
0.2
+j0.2
∞
-j5.0
-j0.2
-j2.0
-j0.5
-j1.0
Fig. 15. Simulated and measured S11 for different voltage levels at the gate of each transistor for 2.45 GHz at 0 dBm
14
TABLE I
FOR DIFFERENT LEVELS AT THE GATE OF TRANSISTORS
Measured
(V)
(V)
0.32
0.21
0.36
Simulated
M1
M2
(V)
(V)
0.36 + j*0.24
0.15
0.44
0.35 + j*0.28
0.35
0.09 + j*0.06
0.32
0.40
0.09 + j*0.06
0.36
0.21
0.09 + j*0.21
0.20
0.39
0.09 + j*0.21
0.31
0.37
0.32 + j*0.05
0.33
0.45
0.31 + j*0.04
0.43
0.27
-0.36 + j*0.07
0.42
0.45
-0.34 + j*0.09
0.39
0.30
-0.14 + j*0.08
0.42
0.45
-0.14 + j*0.08
0.39
0.15
-0.14 + j*0.25
0.42
0.45
-0.15 + j*0.25
0.45
0.13
-0.39 + j*0.27
0.42
0.45
-0.39 + j*0.27
0.46
0.38
-0.38 - j*0.15
0.39
0.04
-0.30 - j*0.15
0.44
0.48
-0.35 - j*0.34
0.42
0.05
-0.30 - j*0.32
0.37
0.48
-0.18 - j*0.36
0.43
0.14
-0.30 - j*0.15
0.39
0.42
-0.17 - j*0.17
0.39
0.17
-0.18 - j*0.16
0.33
0.42
0.08 - j*0.17
0.39
0.35
0.07 - j*0.16
0.21
0.41
0.37 - j*0.15
0.39
0.35
0.37 - j*0.18
0.00
0.45
0.35 - j*0.32
0.39
0.35
0.34 - j*0.32
0.30
0.45
0.12 - j*0.31
0.39
0.35
0.11 - j*0.29
S11
+j1.0
-j5.0
-j2.0
-j1.0
(a)
-j0.2
∞
-j5.0
-j2.0
-j0.5
-j1.0
(b)
+j0.2
0.0
+j5.0
0.5
5.0
2.0
1.0
0.5
0.0
+j5.0
0.2
5.0
2.0
1.0
0.5
0.2
-j0.2
∞
+j0.2
+j2.0
+j0.5
+j2.0
+j0.5
+j5.0
-j0.5
+j1.0
0.2
+j2.0
+j0.2
0.0
S11
+j1.0
+j0.5
M2)
5.0
M2
AND
2.0
M1
(M1
1.0
S IMULATED AND MEASURED S11
-j0.2
∞
-j5.0
-j2.0
-j0.5
-j1.0
(c)
Fig. 16. Simulated S11 for different voltage levels (from 0 V to 0.6 V with a step of 0.01 V) at the gate of each transistor for
2.45 GHz (a) Pin = -10 dBm. (b) Pin = 0 dBm. (c) Pin = 5 dBm.
In order to give more precise and accurate results in real scenario applications we present
Fig. 16, which can be seen that for higher input power levels the constellation will be more
15
difficult to implement. For lower input powers it is possible to obtain 16-QAM as in Fig. 16b.
Fig. 16 presents the simulated results optimized for 0 dBm. The results could be improved for
higher input power levels if the circuit was optimized for that levels.
A. Demodulation and potential data rates
Incident Wave
OUT
Vector Signal
Generator
IN
OUT
2.450GHz
IN
Coupler
GATE 1
DUT
CPL
Reflected Wave
GATE 2
OUT 1
Waveform
Generator
OUT 2
RF IN
Signal Analyzer
Fig. 17. Block diagram of measurement setup for demodulation and achievable data rates.
The maximum data rate of the communication on the modulator link was studied and evaluated
by using the measurement setup presented in Fig. 17. This measurement setup was used to
view the received constellation at multiple transmission rates. A vector signal generator (ROHDE&SCHWARZ SMJ 100A) to generate the transmitter signal at 2.45 GHz was used. An
arbitrary waveform generator (TEKTRONIX AWG5012C) was used to generate the different
voltage levels at the gate of each transistor. The voltage levels presented in Table I were applied
into the gate of each transistor and 16 different impedances were generated and analyzed in the
signal & spectrum analyzer. The received constellations are shown in Fig. 18 for bit rates of
8 Mb/s, 24 Mb/s and 80 Mb/s. The constellation points lie in the appropriate quadrants, and
each point is clearly visible. In order to perform better results, the system can be calibrated by
changing the gate voltage level.
16
(a)
(b)
(c)
Fig. 18. Received 16-QAM constellation for a center frequency of 2.45 GHz for different data rates. (a) 8 Mb/s. (b) 24 Mb/s.
(c) 80 Mb/s.
17
Fig. 19 shows the final implementation of the system, combining the QAM backscatter modulator with WPT. This solution presents a potential performance improvement when compared
with conventional battery-powered wireless sensors network, because it eliminates the need
of battery replacement or recharging. Using WPT reduces the operational cost and increase
the communication performance. Since the wireless sensors need to harvest enough energy
before sending data, with this solution it is possible to overcome these limitations by achieving
continuous power delivery and increase the communication performance. The solution uses two
different frequencies, one for WPT and the other for the QAM backscatter modulation and
therefore it will have an important role in many popular commercial and industrial systems in
the future of IoT systems.
Microcontroller
uC
V1
V2
Matching Network
Rectifier
Energy Storage
DC out
Control
D
Wilkinson
power
divider
S
G
V1
D
S
G
V2
16-QAM Backscatter modulator
Fig. 19. Final block diagram of WPT with 16-QAM backscatter modulator
VI. C ONCLUSIONS
Tags employing vector backscatter modulations guarantee higher data throughput while running at a lower on-chip clock frequency, reducing on-chip power consumption and extending
read range. Tags that employ an M-ary modulation can achieve log2 M higher data throughput
at essentially the same dc power consumption as an ASK or PSK tag.
With the solution proposed it was shown that the combination of WPT and backscatter can
provide a continuous power flow to the wireless sensor. This way the sensors can be continuously
powered during the operation mode.
18
The solution now presented combined with a WPT scheme can actually be used to increase
bit rate in fully passive WSN and be one of the enablers of the IoT paradigm. It was described
a modulation technique that enables a high-bandwidth wireless communication while requiring
very low power demands. This solution can be combined with WPT for ultra low-power wireless
applications requiring high bandwidth communications such as remote camera sensors or wireless
audio.
ACKNOWLEDGMENT
This work is funded by National Funds through FCT - Fundação para a Ciência e a Tecnologia
under the project PEst-OE/EEI/LA0008/2013; UID/EEA/50008/2013.
R EFERENCES
[1] J. R. Smith, A. P. Sample, P. S. Powledge, S. Roy, and A. Mamishev, “A wirelessly powered platform for sensing and
computation,” in 8th International Conference on Ubiquitous Computing (Ubicomp), ser. Lecture Notes in Computer
Science, vol. 4206, 2006, pp. 495–506. [Online]. Available: http://link.springer.com/chapter/10.1007/11853565 29
[2] D. Kuester and Z. Popovic, “How Good Is Your Tag?: RFID Backscatter Metrics and Measurements,” IEEE Microwave
Magazine, vol. 14, no. 5, pp. 47–55, jul 2013.
[3] S. J. Thomas, E. Wheeler, J. Teizer, and M. S. Reynolds, “Quadrature amplitude modulated backscatter in passive and
semipassive UHF RFID systems,” IEEE Transactions on Microwave Theory and Techniques, vol. 60, no. 4, pp. 1175–1182,
2012.
[4] S. Thomas and M. S. Reynolds, “QAM backscatter for passive UHF RFID tags,” in 2010 IEEE International Conference
on RFID (IEEE RFID 2010).
IEEE, apr 2010, pp. 210–214.
[5] S. J. Thomas and M. S. Reynolds, “A 96 Mbit/sec, 15.5 pJ/bit 16-QAM modulator for UHF backscatter communication,”
in 2012 IEEE International Conference on RFID (RFID).
IEEE, apr 2012, pp. 185–190.
[6] R. Correia, N. B. D. Carvalho, G. Fukuda, A. Miyaji, and S. Kawasaki, “Backscatter Wireless Sensor Network with WPT
Capabilities,” in International Microwave Symposium, 2015, pp. 1–4.
[7] R. Correia, N. B. Carvalho, and S. Kawasaki, “Continuously Power Delivering for Passive Backscatter Wireless Sensor
Networks,” IEEE Transactions on Microwave Theory and Techniques, pp. 1–9, 2016.
[8] H. J. Visser and R. J. M. Vullers, “RF Energy Harvesting and Transport for Wireless Sensor Network Applications:
Principles and Requirements,” Proceedings of the IEEE, vol. 101, no. 6, pp. 1410–1423, jun 2013.
[9] A. Koelle, S. Depp, and R. Freyman, “Short-range radio-telemetry for electronic identification, using modulated RF
backscatter,” Proceedings of the IEEE, vol. 63, no. 8, pp. 1260–1261, 1975.
[10] U. Karthaus and M. Fischer, “Fully integrated passive uhf rfid transponder ic with 16.7-µ minimum rf input
power,” IEEE Journal of Solid-State Circuits, vol. 38, no. 10, pp. 1602–1608, oct 2003. [Online]. Available:
http://ieeexplore.ieee.org/document/1233745/
[11] D. M. Dobkin, The RF in RFID: Passive UHF RFID in Practice.
Newnes, sep 2007. [Online]. Available:
http://dl.acm.org/citation.cfm?id=1543456
[12] J.-P. Curty, M. Declercq, C. Dehollain, and N. Joehl, Design and Optimization of Passive UHF RFID Systems. Springer
Publishing Company, Incorporated, oct 2010. [Online]. Available: http://dl.acm.org/citation.cfm?id=1942137