Progress In Electromagnetics Research B, Vol.

45, 19–35, 2012

DESIGN CONSIDERATIONS FOR RADIO FREQUENCY

ENERGY HARVESTING DEVICES

D. Pavone1, 3 , A. Buonanno2 , M. D’Urso2, * , and F. D. Corte3

1 Consorzion per le Tecnologie Optoelettroniche delll’InP c/o SELEX

Sistemi Integrati S.p.A., Engineering Directorate, Via Circumval-
lazione Esterna di Napoli, zona ASI, Loc. Pontericci, Giugliano, Napoli
I-80014, Italy
2 SELEX Sistemi Integrati S.p.A., Engineering Directorate, Via
Circumvallazione Esterna di Napoli, zona ASI, Loc. Pontericci,
Giugliano, Napoli I-80014, Italy
3 DIMET, Mediterranea University of Reggio Calabria, Via Graziella-
Feo di Vito, RC 89100, Italy

Abstract—Radio Frequency Energy harvesting is a research topic of

increasing interest, related to sustainability, which could become a
promising alternative to existing energy resources. The paper will
show all the activities addressed to design a wideband system to
recover wideband energy from electromagnetic sources present in the
environment. The main idea is to develop battery-free wireless sensors
able to capture the available energy into the mentioned bandwidth.
The final goal is to realize self-powered Wireless Networks based on
Ultra Lower Power — ULP sensors minimizing the need of dedicated
batteries. This last feature is particularly attractive in different kind
of applications, ranging from military to civil cases. A first system
prototype is shown and discussed. Conclusion follows.

1. INTRODUCTION AND MOTIVATIONS

In recent years there has been an increasing interest in the development

of Wireless Sensor Networks (WSNs). These networks can be uses in
different scenarios as intelligent office spaces, medical monitoring and
military applications. A WSN is characterized by low bit-rates, volume
Received 29 June 2012, Accepted 15 October 2012, Scheduled 19 October 2012
* Corresponding author: Michele D’Urso (mdurso@selex-si.com).
20 Buonanno et al.

constraints for the nodes and hundreds microwatts power consumption.

The use of primary batteries limits the sensor lifetime. The low power
requirement of the network sensor node and the need of eliminating
primary batteries pushes toward the possibility of powering the nodes
by scavenging ambient power from the surroundings [1, 2]. This can
potentially extend the lifetime of a node to infinite duration of time
and reduce the maintenance costs associated with battery-operated
devices.
The applications for these WSN normally require an operating
distance of 3 to 100 meters and they usually have a backup battery
in case the power provided by the RF radiation is insufficient.
Other applications for RF powered devices include access control,
equipment monitoring and even personal identification. In all of these
applications, there must be a power conversion circuit that can extract
enough DC power from the incident electromagnetic waves for the
passive device to operate. In a far field RF energy harvesting system,
RF energy must be extracted from the air at very low power density
since the propagation energy drops off rapidly as distance from the
source is increased. In free space, the electric field and power density
drop off at the rate of, where is the distance from the radiating source.
The available power to the receiver decreases by 6 dB for every doubling
of distance from the transmitter.
In the following tables, are indicated the main electromagnetic
sources and the relative power density available in the input RF energy
harvesting architecture (before the conversion to DC) for outdoor-
indoor applications.
The paper wants to define the design guidelines of such systems
to optimize the overall performances. Note the amount of energy to
harvest can be limited in the problem at hand: note that the Italian
law DPCM of 8 July 2003, for example, fixes the limit values of the
electromagnetic radiations to 1 W/m2 between 3 MHz and 3 GHz [3],
thus reducing the energy available in the environment. Of course,
the use of wideband architectures can overcome such limitations, thus
maximizing the harvested energy.
For these reasons wideband system architecture based on

Table 1. Density power available for the 50 kW AM radio station.

50 kW AM RADIO STATION
DISTANCE DENSITY POWER AVAILABLE
5 km 159 µW/m2
10 km 40 µW/m2
Progress In Electromagnetics Research B, Vol. 45, 2012 21

Table 2. Density power values available for the 100 W GSM base
station.
100 W GSM BASE STATION
DISTANCE DENSITY POWER AVAILABLE
100 m 800 µW/m2
500 m 32 µW/m2
1000 m 8 µW/m2

Table 3. Density power available Table 4. Density power available

for the 0.5 W mobile phone. for the 1 W Wi-Fi router.
0.5 W MOBILE PHONE 1 W Wi-Fi ROUTER
DENSITY DENSITY
DISTANCE POWER DISTANCE POWER
AVAILABLE AVAILABLE
1m 40 mW/m2 1m 80 mW/m2
5m 1.6 mW/m2 5m 3.2 mW/m2
10 m 0.4 mW/m2 10 m 0.8 mW/m2

Figure 1. Energy harvesting architecture.

radiofrequency energy harvesting, able to recover energy from available

RF electromagnetic sources to power wireless sensor nodes is presented.
The paper is organized as follows. In Section 2 the overall
problem is introduced. Section 3 and Section 4 deal with the design,
simulation, realization and characterization of a first prototype working
at 900 MHz. In Section 4, the multiband solution is proposed and
realized. Section 5 concerns possible application to batteries-free sensor
networks.

2. OVERALL SYSTEM ARCHITECTURE

The basic structure of an energy harvesting system is shown in

Fig. 1 [4]. The front-end is the receiving antenna. It converts the
input microwave signal into AC voltages and currents. The matching
22 Buonanno et al.

network between the receiver antenna and rectifier circuit is necessary

to further reduce the transmission loss (high matching efficiency α
with 0 ≤ α ≤ 1) and to increase the input voltage of the rectifier
circuit. The output voltage of the matching network is AC type, but
the electronic loads require DC voltage for their operation. Therefore,
the AC voltage of the output has to be processed by a suitable power
converter to produce the required DC voltage for the load. The device
assigned to this aim is the rectifying circuit.

3. SINGLE FREQUENCY ENERGY HARVESTING

ARCHITECTURE

To obtain an effective energy harvesting system, the rectenna must

exhibit high RF-DC conversion efficiency η defined as:
PDC V2 /RL
η= × 100[%] = DCout × 100[%] (1)
PRF PRF
where the PDC is the output DC power harvested, PRF the input RF
power and RL is the load to supply (see Fig. 1). The diode is the main
part of the rectifying circuit; it changes the AC microwave signal into
DC signal. A diode with a lower built-in voltage would realize a higher
rectifying efficiency. Due to its nonlinear characteristics, the diode
will create harmonic signals, which can notably decrease the rectifying
efficiency of the system [5]. Moreover, this spurious harmonics could
to modify the typical working of the load. An effective design of the
above mentioned RF energy harvesting system cannot neglect these
issues. In the following, the design guidelines for the rectifying circuit
and the matching network are addressed.

3.1. Rectifying Circuit

Different ways to rectify AC signals have been introduced in [6]. In [6],
the voltage doubler rectifier (VDR) has been exploited. The latter
exhibits two main interesting features matching the addressed RF
energy harvesting application. The first one concerns the capability
to enhance the low RF input voltage; the second one is related to the
ability of working without external bias supply. The VDR circuit has
shown in Fig. 2. It consists of a peak rectifier formed by D1 and C2
and a voltage clamp formed by D2 and C1. The VDR output is:
VOU T = (2VRFamp − Vth1 − Vth2 ) (2)
where Vth is the threshold voltage of the diode and VRFamp is the RF
input voltage. The individual stages of VDR circuit can be arranged in
Progress In Electromagnetics Research B, Vol. 45, 2012 23

Figure 2. N -stage voltage multiplier as the cascade of the N -VDR.

cascade as reported in Fig. 2 (N-stage voltage multiplier) to increase

the output voltage of the rectifier [7].
µ ¶
(N − 1)ILOAD
VOU T = 2N VRFamp − 2N Vth − (3)
f0 ∗ C
where ILOAD is the load current, C are the blocking capacitor and f0
is the working frequency.
When the rectifier stages are cascaded, each rectifier stage acts
as a passive voltage level shifter in addition to the voltage shift in
voltage clamp and peak rectification. To better explain such a point,
let suppose that a sinusoidal voltage, say VRF , with a frequency f0
and amplitude VRFamp , is applied as input of the voltage multiplier.
To assure a small ripple in the output voltage, capacitances have to
be dimensioned so that their time constant is much smaller than the
period of the input signal, that is, 1/(2πCRL ) ≪ f0 , C being the
blocking capacitors and RL the equivalent load resistance. The average
input power, PIN required to obtain a given output voltage and power
can be evaluated by summing up the average power, PD , dissipated in
each diode and the power, PL , transferred to the load. By increasing
the number of stages N (or equivalently the number of diodes) the
value of average dissipated power PD increases. Taking into account
substrate losses, the average input power is given by,
µ ¶ µ ¶
VOUT VRFamp
PIN = 2N ID,SAT B1 exp −
VT 2N VT
N 2
+ VOUT RSUB (ω0 CSUB )2 (4)
2
where VT , is the thermal voltage and B1 is the modified Bessel function
of order one and RSUB and CSUB are the substrate resistance and
24 Buonanno et al.

capacitance, respectively [8]. Solving (4) by numerical iteration, for

a fixed output voltage and power consumption, it is possible to note
that higher is the number of stages smaller is the amplitude of the
input voltage required to obtain a given DC output voltage and power
consumption. However, optimal number of stages should be found with
a compromise between high DC output voltage and low power loss due
to power consumption of diodes and the loss of the substrate. The
tests have shown that the optimum number N of stages is between 1
and 2. To reduce the loss in the diode, the diode should have large
saturation current ID,SAT , low junction capacitance CJ which result in
low threshold voltage Vth , small series resistance Rs and small junction
resistance RJ [9]. The AVAGO HSMS family diodes represent a good
market solution for these applications.

3.2. Power Matching

The equivalent input impedance of the voltage multiplier can be
represented, as a zero-order approximation, by the parallel of a
resistance and capacitance [8]. In the high frequency analysis of the
voltage multiplier, the input capacitance is the parallel of all the
diodes capacitances while the equivalent input resistance, Rin eq , is
the resistance calculated from power consumption as
VRFamp
Rin eq = (5)
2PIN
In order to obtain a good matching between the RF source and
voltage multiplier, assigned the DC load in terms of voltage VOUT
and power POUT , the first step is to calculate the voltage VRFamp
(RF input voltage to the voltage multiplier) in order to obtain the
VOUT and the POUT . Performing a sweep of the RF input power,
it is possible calculate the input impedance in terms of equivalent
input resistance and capacitance. Fixed the voltage VOUT and power
POUT , in order to compensate the equivalent input capacitance of the
voltage multiplier, a proper inductance L′ , in parallel with the input
of the voltage multiplier, can be introduced. In this way, the input
impedance of voltage multiplier can be schematized as a resistive load.
Again, performing a sweep of the RF input power and fixed the output
power POUT , it is possible to calculate the input resistance, Rin eq (5)
and consequently the voltage VRFamp . In this way, by an impedance
transformer LC (classic λ/4 transformer with L referred to ground) it
is possible calculate the input voltage VIN RF of it that correspond at
the output voltage antenna.
To realize the λ/4 transformer, a discrete LC matching network
circuit has been used, thus achieving a voltage gain (between VIN RF
Progress In Electromagnetics Research B, Vol. 45, 2012 25

and VRFamp ) of a Q-factor [5] defined as

sµ ¶
Rin eq
Q= −1 (6)
RAN T
where RAN T is the antenna resistance.
Input capacitance due to the diodes is variable with the voltage
across diodes, L′ is chosen in order to resonate with its average value
at the operating frequency f0 . This leads to the condition,
Q
∆C ≪ (7)
2πf0 Rin eq
where ∆C is the maximum variation of the input capacitance of the
voltage multiplier with respect to its mean value.
Note the matching circuit is a key element of the harvesting
system. Indeed, the conversion efficiency of the system η [see (1)]
is closely related to the impedance matching between the antenna and
the rectifier circuit that is a nonlinear function of both power input
and frequency.
In particular, the matching network LC is a resonator circuit
(non-linear) working at the designed frequency according to the diodes
rectifier. Of course, the conversion efficiency η is high at the designed
frequency and extremely low elsewhere.
With regards to the input power, as the Rin eq and Vout of the
VDR is a non linear function with the power input (see (4)), the Q-
factor of the matching network changes and accordingly the conversion
efficiency η. In order to obtain good conversion efficiency η, it is
useful to design the matching network around a specific power input
(considering the losses of the circuit, it must be higher than output
power POUT ) because it is possible supposed that the circuit works
in conditions of linearity about it. Small shifts around this power
(more less ± 5 dBm) involves a good parameter η, conversely it drops
quickly. This explains why the characteristic efficiency curve presents
a bell-shaped.

4. SINGLE FREQUENCY PROTOTYPE

In the following, according to the guidelines, an interesting and

very compact rectenna working at 900 MHz is designed, realized and
characterized. The RF/DC converter is a 2-stage voltage rectifier.
It has been designed and simulated by using the Advanced Design
System (ADS) software, which properly uses the Harmonic-Balance
(HB) method. For the circuit realization, it has been used the
commercial zero biased Schottky diode HSMS2850, 1 nF blocking
26 Buonanno et al.

Figure 3. EM/circuit co-simulation setup at 900 MHz.

Figure 4. Co-simulation results: output voltage and efficiency

rectification of the 900 MHz rectenna varying the RF input power.

capacitors and 100 nF output capacitor (Fig. 3). A 60 kΩ load,

approximating an ultra-low power sensor, has been considered. The
matching circuit has been designed according to guidelines in Section 3.
To obtain very compact size a concentrated parameter implementation
of the matching circuit has been adopted. In order to obtain very
accurate numerical results, it has been co-simulated the rectifier with
momentum ADS including diode models and capacitor and inductor
packaging (Fig. 3).
Performing a sweep of the RF input power (between −20 dBm and
20 dBm), it is possible to plot the DC output voltage and the efficiency
rectification RF-DC of the circuit simulated with real component
(accounting of the substrate loss) (Fig. 4).
As it can be seen, the designed system exhibits a very high
efficiency, due to the effective guidelines previously derived. Notably,
the realized system is characterized by an improved efficiency with
respect to the ones reported in [10, 11].
Progress In Electromagnetics Research B, Vol. 45, 2012 27

Figure 5. Single frequency Figure 6. Measurements: output

prototype rectenna. voltage and efficiency rectification
of the 900 MHz rectenna varying
the RF input power.

4.1. Experiments Results

As simple prototype is shown in Fig. 5. The layout size of the
harvesting system is 2.1 cm × 1 cm. It has been fabricated using 0.5-
oz copper (0.017 mm thick), Rogers Duriod 5880 board material at
SELEX Sistemi Integrati S.p.A. facilities available at SELEX SIstemi
Integrati Giugliano site.
The overall performances have been evaluated by measuring the
DC output voltage and the overall efficiency by increasing the input
power of an HP Signal Generator from −20 dBm to 20 dBm at
900 MHz. The results are shown in Fig. 6.
The measured efficiency values, reported in Fig. 6 are slightly lower
than the simulated ones (see Fig. 6), according to the fact that lumped
parasitic components and the tolerances (±10%) of the capacitors
and the inductor have not been considered. As highlighted in the
previous section, a variation of these values involves a variation of the
Q-parameter and then the variation of the conversion efficiency η. In
particular, the critical component is the inductor L1 that must also
to delete the capacitance input impedance of the VDR. In order to
overcome this drawback, a solution could be the proper realization of
the distributed matching circuit rather than discrete components.

5. MULTIBAND ENERGY HARVESTING

ARCHITECTURE

This Section concerns the case of wideband energy harvesting systems.

The problem comes out from the circumstance that, often, a single
rectenna is not sufficient to supply the common device operations: the
28 Buonanno et al.

problem of wideband energy harvesting systems arises, and possible

solutions could be represented by a multi-antenna system [12]. In this
case, each antenna (operating at different frequencies) could have its
own rectifier to harvest separately DC power [13]. This configuration
is suitable when multi frequency sources are available. The drawback
of this configuration [13] is the need to use many antennas, thus
increasing the overall costs, weight and size of the harvesting system.
To overcome this drawback, a possible solution is to use only one Ultra
Wide Band — UWB (or Multiband) antenna. In this case, the problem
is to design the matching networks. To our knowledge, there is not a
solution to this problem due to the non-linearity of the diodes with
respect to both frequency and RF power input. In this paper, a novel
architectural solution (Fig. 7) able to avoid the problem is proposed.
In particular, knowing the N frequencies of RF sources, N matching
networks working at different frequencies fi are used. In this way, only
one UWB antenna (or multi-band) connected with N rectifying circuits
and relative matching network (Section 2) is considered. Note that the
matching circuits have a double function. They allow to achieve the
desired impedance matching and to filter the proper frequency content
on each rectifier channel. In this way, the matching networks operating
at different frequency band, consume a very low power when the RF
input signal doesn’t contain the correspondent frequency even if they
are connected in parallel (see Fig. 7). In fact, the input impedance of
the matching network (designed at frequency fi ) at the frequency fj
is very high (it is approximated to an open circuit). In this particular
case, the lost power is approximately 3% of the total input power. The
outputs from each rectifying circuit are connecting together by DC
Combining Circuit in order to sum the harvested power.
To obtain the mentioned system it is very important to design an
efficient DC combining circuit. An interesting solution is represented
by the modified version of the single-stage full-wave Greinacher
rectifier [11] showed in Fig. 8.

Figure 7. Multi-band energy harvesting architecture.

Progress In Electromagnetics Research B, Vol. 45, 2012 29

Figure 8. Schematic of the rectifier with DC combining circuit.

The modified Greinacher rectifier operates as follows. If Vrec (t)

is the induced voltage at the antenna port, then C1 and D1 shift the
voltage Vrec (t) up at node B. Subsequently, D2 and C2 rectify the
voltage at node B (both RF and DC components) to appear across
the DC load. Similarly, C3 and D3 shift the voltage Vrec (t) down at
node C. In turn, D4 and C4 rectify the voltage at node C to appear
across the DC load. Upon reaching equilibrium, the rectifier circuit
delivers a constant output current and voltage to the DC load.
In this architecture, the outputs from each rectifying circuit are
properly connecting in series. This architecture allows obtaining a good
rectification and DC-combining efficiency but has the disadvantage to
have a differential output and a limited number of input signal (N = 2).
In this paper, a new solution removing these limitations has been
introduced. The outputs of the N rectifier circuits have been connected
in series, but the DC load is connected between the N th rectifying
circuit (referred to ground) and the first. In this way, has been
extended to N the number of the RF input signals (see Fig. 7). In
more detail, for every rectifier circuits (RF channel), the series of the
blocking capacitors is cascaded with the previous and next RF channel
while the i-th matching network is already referred to ground.
This configuration (not present in literature) allows connecting
more rectifier circuits, working at different frequency, in order to
harvest more energy than a single frequency circuit.
Note that, in the proposed configuration, due to the non linearity
of the diode, the N output resistance of each rectifier circuit change
with the input RF signals. This means that it is not possible to identify
the optimum load in terms of harvested DC power. The choice of the
sub-optimal load value should account of the presence and power value
of RF sources.
The total harvested DC power of the proposed UWB (or multi
30 Buonanno et al.

Table 5. Comparison between a single and multi-band energy

harvesting architecture.
MULTIBAND ENERGY HARVESTING SYSTEM
ADVANTAGES DRAWBACKS
MORE POWER RECOVERABLE SAME CONVERSION
BY A MULTI-BAND ANTENNA EFFICIENCY
DC COMBINING
WORK FLEXIBILITY
EFFICIENCY
LOW (NO ULTRA LOW)
POWER LOAD

band) energy harvesting architecture is:

N
X
PDC = e Gi P iRF η (8)
i
where η refers to the RF to DC conversion efficiency, e denotes the
efficiency of the DC combining circuit and Gi refers to antenna’s gain
corresponding at i-th frequency. As it can be seen from (8), by properly
designing the rectifying and the DC combining circuits, is possible to
sensibly increase the DC power harvested through the adoption of a
larger number of RF sources and the antenna gain. In this way, the
energy harvested from the multi-band energy harvesting architecture
(Fig. 7) can be enough to power different kind of ULP load.
In order to manage the harvested energy, it is also mandatory to
exploit a power manager device. It allows to supply sensors requiring
short time high load current (low duty cycle). The power manager
is also able to charge an energy storage (e.g., supercapacitor) that
provides energy to the load when the input power may be lost.
Finally, in Table 5, the comparison between a single and multi-
band energy harvesting architecture is showed.

5.1. Simulation
In the following, an interesting and very compact rectenna working
at 900 MHz and 2.4 GHz is designed, realized and characterized. The
RF/DC converters are a 2-stage voltage rectifier. For the simulations,
it has been used the commercial zero biased Schottky diode HSMS2852
and HSMS2862 models respectively for the RF/DC converter at
900 MHz and 2.4 GHz (Fig. 9). A 60 kΩ load, approximating an ultra-
low power sensor, has been considered. The matching circuit has been
designed according to guidelines in Section 3. To obtain very compact
Progress In Electromagnetics Research B, Vol. 45, 2012 31

size a concentrated parameter implementation of the matching circuit

has been adopted.
As before, performing a sweep of the two RF inputs power between
−20 dBm and 20 dBm (the power is the same for each RF source), it is
possible to plot the DC output voltage and the efficiency rectification
RF-DC of the circuit simulated with real component (Fig. 10).
As it can be seen, the designed system exhibits a very high
efficiency, due to the effective guidelines previously derived.
Let analyze the efficiency rectification of the proposed architecture

Figure 9. EM/circuit co-simulation setup at 900 MHz and 2.4 GHz.

Figure 10. Co-simulation re- Figure 11. Co-simulation

sults: outputvoltage and efficiency results: output voltage and
rectification of the 900 MHz– efficiency rectification of the
2.4 GHz rectenna varying the RF 900 MHz–2.4 GHz rectenna
input power. varying the RF input power 1
(900 MHz) while, the RF input
power 2 (2.4 GHz) is −30 dBm.
32 Buonanno et al.

in some special cases varying the power of one of the two RF sources.
In particular, the Fig. 11, shows the case when the input power at
2.4 GHz is very low (−30 dBm) while the input power at 900 MHz vary
between −20 dBm and 20 dBm. Reversed case is showed in Fig. 12.
The simulations show a very good efficiency conversion for both
the shown cases.

5.2. Experiments Results

In order to demonstrate the validity of the broadband architecture,
it has been realize the layout of the circuit shown in Fig. 9. The
antenna considered has been the directional four band panel antenna
suitable for simultaneous GSM-DCS-UMTS-Wireless application in
the 900/1800/2200/2500 MHz bands. The RF sources (at 900 MHz
and 2.4 GHz) used are mentioned respectively in Tables 3 and 4. The
distance between the two sources and the antenna was about 1 m.
Before to show the achieved result, in order to evaluate the
soundness of the developed guide lines, the numerical simulation of
the considered scene is considered.
In particular, compatible with the values of Tables 3 and 4, the
RF input power1 (900 MHz) is 4.5 dBm (2.8 mW) and the RF input
power 2 (2.4 GHz) is 1.7 dBm (1.5 mW).
In this case is obtained the rectification efficiency of about 75%
with total DC output power of about 3 mW (4.7 dBm) and the DC
output voltage of about 13 V (see Fig. 13).

Figure 12. Co-simulation Figure 13. Co-simulation

results: output voltage and results: output voltage and
efficiency rectification of the efficiency rectification of the
900 MHz–2.4 GHz rectenna 900 MHz–2.4 GHz rectenna when
varying the RF input power 2 the RF input power1 (900 MHz)
(2.4 GHz) while, the RF input is 4.5 dBm and the RF input
power 1 (900 MHz) is −30 dBm. power 2 (2.4 GHz) is 1.7 dBm.
Progress In Electromagnetics Research B, Vol. 45, 2012 33

The calculation of the power input has been made, considering

the antenna’s gain and the distance between RF sources and receiving
antenna.
The described experimental setup is shown in Fig. 14. In this
experiment an output voltage of 12.2 V on a load of 60 kohm has been
measured. More in detail, the output power (2.48 mW) is slightly lower
than that simulated (3 mW) due to the numerical model approximation
and the not exact knowledge of the input parameters.
This means that the designed system works correctly and in line
with the required performances.

6. APPLICATION TO BATTERIES-FREE SENSOR

NETWORKS

The developed architectures have the aim to make any sensor batteries-
free. In particular, in this section, the eZ430-RF2500 sensor has been
powered with the single frequency prototype rectenna shows in Fig. 5
without the use of batteries. This last feature is particularly attractive
in different kind of applications, ranging from military to civil cases.
The eZ430-RF2500 is a complete USB-based MSP430 wireless
development tool providing a 2.4-GHz wireless transceiver. It is
preloaded with a reprogrammable wireless temperature sensor. The
considered network consists of an Access Point that measures its own
temperature and also wirelessly receives temperature measurements
from End Devices (ED). End Devices measure their temperature
once per second and then enter a low-power mode to reduce battery
usage. The average current calculation must also take into account

Figure 14. Multiband energy Figure 15. The overall harvest-

harvesting experimental. ing system including LTC3108,
eZ430-RF2500 and supercapaci-
tor.
34 Buonanno et al.

the sleep current of the ED. The average current consumption for the
considered application is 36.80 µA (for one second of the transmission
period). Typically, the eZ430-RF2500 sensor is supplied by two 1.5
volts alkaline batteries. The average power consummation of the sensor
node is roughly 100 µW (−10 dBm) corresponding to our measured
value using −4 dBm input RF power at 900 MHz (see Fig. 6). This
proves that having a low power RF source (less than a typical cell
phone output radiated power) is possible to self-power the considered
ULP temperature sensor node. The overall realized harvesting system
including the LTC3108 power management, a supercapacitor and the
sensor node as shown in Fig. 15.
Of course, we expect that the use of the multi-band architecture
can improve the performances.

7. CONCLUSIONS

This paper presents the design criteria for the multi-band RF energy
harvesting system. In particular the design tradeoffs of each block of
the proposed multi-band architecture have been derived and discussed.
Two systems enable to harvest RF power at 900 MHz GSM band
(single frequency) and at 900 MHz GSM–2.4 GHz Wi-Fi bands (multi-
frequency) have been designed and realized.
The prototype a single frequency (900 MHz) has been used to
supply eZ430-RF2500 wireless sensor node. In particular, at −4 dBm
input power, the system recovers 160 µW with 3 V DC output (32%
RF-DC rectification efficiency). It has been shown that the generated
DC power is sufficient to power sensor nodes (eZ430-RF2500). Two
aspects have to be underline. Firstly, the obtained results show the
proposed harvesting system is able to recover the available RF energy
by using opportunity sources and to supply ULP wireless sensor nodes.
Second, the proposed system could represent, on a large scale, a
suitable solution to environmental sustainability: it allows reducing
the use of batteries, sources of environmental pollution.

REFERENCES

1. Warneke, B. A., “An autonomous 16 mm3 solar-powered node

for distributed wireless sensor networks,” Proc. of IEEE Sensors,
Vol. 2, 1510–1515, Jun. 12–14, 2002.
2. Meninger, S., J. O. Mur-Miranda, R. Amirtharajah, A. Chan-
drakasan, and J. H. Lang, “Vibration-to-electric energy conver-
sion,” IEEE Trans. on VLSI Systems, Vol. 9, No. 1, 64–76,
Feb. 2001.
Progress In Electromagnetics Research B, Vol. 45, 2012 35

3. Decree of President of the Council of Ministers, Jul. 8, 2003.

4. Zhu, N., K. Chang, M. Tuo, P. Jin, H. Xin, and R. W. Ziolkowski,
“Design of a high-efficiency rectenna for 1.575 GHz wireless low
power transmission,” Department of Electrical and Computer
Engineering, University of Arizona, Tucson, Arizona, USA, 2011.
5. Akkermans, J. A. G., M. C. van Beurden, G. J. N. Doodeman, and
H. J. Visser, “Analytical models for low-power rectenna design,”
IEEE Antennas and Wireless Propagation Letters, Vol. 4, 2005.
6. Razavi, B., RF Microelectronics, Prentice Hall PTR, 1997.
7. Karthaus, U. and M. Fischer, “Fully integrated passive UHF RFID
transponder IC with 16.7-µm minimum RF input power,” IEEE
Journal of Solid-State Circuits, Vol. 38, 1602–1608, Oct. 2003.
8. De Vita, A. and G. lannaccone, “Design criteria for the RF section
of long range passive RFID systems,” Proc. Norchip Conference,
107–110, Oslo, Norway, 2004.
9. Buted, R. R., “Zero bias detector diodes for the RF/ID market,”
Hewlett-Packard Journal, Dec. 2005.
10. Yan, H., M. Popadic, J. C. Macias, L. C. N. de Vreede,
A. Akhnoukh, and L. K. Nanver, “Design of an RF power
Harvester in a silicon-on-glass technology,” Proc. of 19th Annual
Workshop on Circuits, Systems and Signal Processing, 287–290,
Ultrcht, 2008.
11. Wilas, J., K. Jirasereeamornkul, and P. Kumhom, “Power
harvester design for semi-passive UHF RFID tag using a tunable
impedance transformation,” Proc. of the IEEE Conference ISCIT,
1441–1445, Sep. 2009,
12. Olgun, U., C. C. Chen, and J. L. Volakis, “Investigation of
rectenna array configurations for enhanced RF power harvesting,”
IEEE Antennas and Wireless Propagation Letters, Vol. 10, 2011.
13. Hagerty, J. A., F. B. Helmbrecht, W. H. McCalpin, R. Zane, and
Z. B. Popovic, “Recycling ambient microwave energy with broad-
band rectenna arrays,” IEEE Trans. on Microwave Theory and
Techniques, Vol. 52, No. 3, Mar. 2004.