A Long-Distance RF-Powered Sensor Node with Adaptive ... - MDPI

sensors Article

A Long-Distance RF-Powered Sensor Node with Adaptive Power Management for IoT Applications Matteo Pizzotti 1,2, * ID , Luca Perilli 1 ID , Massimo del Prete 2 , Davide Fabbri 2 , Roberto Canegallo 3 , Michele Dini 1 , Diego Masotti 2 ID , Alessandra Costanzo 1,2 , Eleonora Franchi Scarselli 1,2 ID and Aldo Romani 1,2 ID 1

2

3

*

Advanced Research Center on Electronic Systems, University of Bologna, Via Toffano 2/2, Bologna 40126, Italy; [email protected] (L.P.); [email protected] (M.D.); [email protected] (A.C.); [email protected] (E.F.S.); [email protected] (A.R.) Department of Electrical, Electronic, and Information Engineering, University of Bologna, Via Risorgimento 2, Bologna 40126, Italy; [email protected] (M.d.P.); [email protected] (D.F.); [email protected] (D.M.) STMicroelectronics, Via Camillo Olivetti, Agrate Brianza 20864, Italy; [email protected] Correspondence: [email protected]; Tel.: +39-054-733-9537

Received: 30 June 2017; Accepted: 26 July 2017; Published: 28 July 2017

Abstract: We present a self-sustained battery-less multi-sensor platform with RF harvesting capability down to −17 dBm and implementing a standard DASH7 wireless communication interface. The node operates at distances up to 17 m from a 2 W UHF carrier. RF power transfer allows operation when common energy scavenging sources (e.g., sun, heat, etc.) are not available, while the DASH7 communication protocol makes it fully compatible with a standard IoT infrastructure. An optimized energy-harvesting module has been designed, including a rectifying antenna (rectenna) and an integrated nano-power DC/DC converter performing maximum-power-point-tracking (MPPT). A nonlinear/electromagnetic co-design procedure is adopted to design the rectenna, which is optimized to operate at ultra-low power levels. An ultra-low power microcontroller controls on-board sensors and wireless protocol, to adapt the power consumption to the available detected power by changing wake-up policies. As a result, adaptive behavior can be observed in the designed platform, to the extent that the transmission data rate is dynamically determined by RF power. Among the novel features of the system, we highlight the use of nano-power energy harvesting, the implementation of specific hardware/software wake-up policies, optimized algorithms for best sampling rate implementation, and adaptive behavior by the node based on the power received. Keywords: wireless sensor networks; RF power transfer; energy harvesting; nano-power DC/DC converter; rectifying antenna; ultra-low power sensor node; adaptive power management

1. Introduction Increasing interest in distributed sensor networks [1] and IoT applications has driven research into the scope of energy harvesting mechanisms towards a more precise field of application. The combination of smart nodes, able to interact with standard wireless communication infrastructures, and energy scavenging modules, which allow nodes to work in a standalone scenario, has proved crucial for the development of both technologies [2]. Since autonomous nodes are expected to operate in very dissimilar surroundings with different energy sources and power densities, the importance of finding efficient ways to exploit available energy is evident, as in many cases the power available from the environment is in the order of microwatts [3] or less. In this context, radio-frequency power harvesting [4,5] represents both a fascinating solution, due to the opportunity of selectively providing energy through dedicated RF Sensors 2017, 17, 1732; doi:10.3390/s17081732

www.mdpi.com/journal/sensors

Sensors 2017, 17, 1732

2 of 21

Sensors 2017, 17, 1732

2 of 21

energy showers augmented by smart power beaming techniques [6,7], and a tough challenge because of the limitations imposed by regulations causing very low voltage and power levels as the distance from the source increases. a matter of fact, the most[6,7], prohibitive obstacle to the because diffusion energy showers augmented by smart As power beaming techniques and a tough challenge RFlimitations harvesting nodes by is the lack of causing dedicated power converters able tolevels operate under these ofofthe imposed regulations very low voltage and power as the distance conditions. However, simply specific power converters or designing more efficient from the source increases. As a developing matter of fact, the most prohibitive obstacle to the diffusion of RF rectennas nodes is notissufficient: management must goable hand handunder not only obvious harvesting the lack ofpower dedicated power converters to in operate thesewith conditions. requirements in developing terms of ultra-low but also with the development of specific policies for However, simply specific power, power converters or designing more efficient rectennas is not adjustingpower the behavior of the node to the availability of energy. Active interactioninbetween sufficient: management mustaccording go hand in hand not only with obvious requirements terms power module andalso smart node integral to achieving thispolicies target. for Another important aspect that ofthe ultra-low power, but with theisdevelopment of specific adjusting the behavior of reinforces the need for such interaction is the considerable difference between the power profiles the node according to the availability of energy. Active interaction between the power module andof the harvesting sourcetoand the active the important former canaspect be considered constant or slowly smart node is integral achieving this node; target.while Another that reinforces the need for changeable, theis latter is characterized by short peakthe consumptions, by long source inactive such interaction the considerable difference between power profilesfollowed of the harvesting periods. Thisnode; mismatch calls for can the beintroduction of storage elements, e.g., capacitors and the active while the former considered constant or slowly changeable, the latter isor super-capacitors, which be dimensioned and considered as an integralcalls partforof characterized by short peakmust consumptions, followedcarefully by long inactive periods. This mismatch optimization policies. the introduction of storage elements, e.g., capacitors or super-capacitors, which must be dimensioned Thus, holistic approach can bestpart tackle this type of issue. On the one hand, circuit design aims carefully anda considered as an integral of optimization policies. at minimizing theapproach power consumption electronic devices on hand, the other optimized Thus, a holistic can best tackleofthis type of issue. Onand, the one circuithand, design aims at behavioral policies for active node modules need investigating with a view to exploiting the minimizing the power consumption of electronic devices and, on the other hand, optimized behavioral available at its modules best. Theneed solution presentedwith is designed an ultra-low powerenergy approach, policies forenergy active node investigating a view tofrom exploiting the available at introduction of a custom rectenna, a dedicated nano-power DC/DC with converter designed to itswith best.the The solution presented is designed from an ultra-low power approach, the introduction down to 250amV and 1 μW, and a multi-sensor node with a low-power profile and ofoperate a custom rectenna, dedicated nano-power DC/DC converter designed to operate down to adopting 250 mV a standard wireless communication protocol for IoT. In order to implement the necessary and 1 µW, and a multi-sensor node with a low-power profile and adopting a standard wireless interactions between power management subsection and the active sensor node, specific communication protocolthe for IoT. In order to implement the necessary interactions between the power circuitry hassubsection been designed, dynamically changing the behavior of the according to different management and the active sensor node, specific circuitry has system been designed, dynamically scenariosthe ofbehavior availableof environmental energy. to The peculiar achievement obtained through such active changing the system according different scenarios of available environmental energy. interaction is the possibility to modulate the transmission data-rate as a consequence of variations The peculiar achievement obtained through such active interaction is the possibility to modulatein thetransmission power harvested, so that higher amount power causes automatic increase theadata-rate the data-rate as a consequence ofofvariations in theanpower harvested, soin that higher and vice-versa, such aan way that the highest always obtained. Suchaan outcome amount of powerincauses automatic increasefeasible in the data-rate is and vice-versa, in such way that may have a considerable rebound on those applications where high transmission rates are not the highest feasible data-rate is always obtained. Such an outcome may have a considerable rebound strictly but canhigh helptransmission in building a rates moreare extensive information database and therefore on those indispensable applications where not strictly indispensable but can help in a more precise behavioral model for the monitored system. building a more extensive information database and therefore a more precise behavioral model for the In Figure 1 the overall system presented consists of two sub-systems: a harvesting module and a monitored system. sensing node.1 the Theoverall interaction presented obtained through specific control signals exchanged In Figure system presentedisconsists of two sub-systems: a harvesting module and between the power module and the microcontroller, designed to provide information about the a sensing node. The interaction presented is obtained through specific control signals exchanged current the state of the harvesting module, e.g., the voltage level and theinformation amount of power andcurrent energy between power module and the microcontroller, designed to provide about the extracted. Based on module, this, thee.g., microcontroller canand adapt its behavior in order to optimize the state of the harvesting the voltage level the amount of power and energy extracted. transmission rate. Based on this, the microcontroller can adapt its behavior in order to optimize the transmission rate.

Figure Figure1.1.General Generalscheme schemeofofthe theoverall overallsystem. system.

Twodifferent differentoperating operatingfrequencies frequenciesare areused usedfor forharvesting harvestingand andcommunication communicationchannels channels Two (868MHz MHzand and433 433MHz MHzrespectively). respectively).On Onone onehand, hand,harvesting harvestingmodule moduleisisoptimized optimizedfor fora aconstant constantRF RF (868 source, so communication signal cannot be effectively used and two separate antennas are required. source, so communication signal cannot be effectively used and two separate antennas are required. Onthethe other hand, two different for harvesting and communication allows On other hand, usingusing two different antennasantennas for harvesting and communication allows optimization

Sensors 2017, 17, 1732

3 of 21

of the harvesting antenna to best fit the DC/DC specifications, at the expense of a reasonable increase in area. Using a single antenna would be more complicated and would result in worse harvesting performance, mainly because of the losses introduced by the multiplexing circuitry. An overview on recent rectenna solutions for generic RF wireless energy transfer applications is available in [8], whereas a comprehensive perspective on RF harvesting techniques specifically devoted to wireless sensor nodes can be found in [3]. In this field of application most solutions concentrate the greatest effort in optimization of the harvesting module, while other studies are proposed to define advanced algorithms to find most efficient transmission rates in complex scenarios. Differently, this work aims at optimizing the full energy chain from the rectenna to power management. Enhanced rectenna designs are proposed in [9] with 40% (simulated) efficiency at −20 dBm input power at 868 MHz, in [10] with 20% (measured) efficiency at −20 dBm input power at 2450 MHz, and in [11], where both solar and RF sources are deployed by a compact structure. Concerning integrated RF-to-DC converters, an interesting solution is presented in [12] with a 65% efficiency at −20 dBm in UHF band and output voltage up to 1.6 V. Another notable solution is presented in [13], where 1.0 V output is obtained at 27 m distance. The nano-power design of an integrated 1 µW – to 5 mW power management circuit is shown in [14]. The DC/DC converter proposed achieves a peak conversion efficiency of 77% and a minimum start-up voltage of 223 mV. The trend for quiescent power consumption of commercial and academic PMIC implementation is shown in [15]. A complete solution with harvesting module and sensing node is proposed in [16], showing an operative range of 5 m with −8 dBm input power. Furthermore, in [6] a simple adaptive solution is adopted, with solely two possible transmission data-rates and a solar cell as harvesting source. Advanced adaptive solutions are studied in [17], for a general-purpose sensor node in a multi-antenna power transfer scenario. These solutions lie on the development of complex recursive algorithms based on the knowledge of the evolution of stored energy in time, which implies the use of additional power-consuming electronics incompatible with micropower scenarios. In fact, experimental results demonstrate just a 5-m operative range. A similar approach is used in [18], with an accurate analysis of harvest-and-use and harvest-store-use schemes. Again, the resulting optimization algorithm requires the development of dedicated energy meters interacting with the node MCU to estimate the correct transmission rate. Both solutions are based on an adaptive estimation of best data-rate accomplished by the MCU on the basis of complex monitoring of available energy. Although DASH7 protocol performs a communication range up to 5 km, it is also widely used for indoor applications and, in combination with RF harvesting, it can provide an optimum solution for enclosed sensors like smart meters placed inside walls or generic structures. Moreover, having two different interfaces for harvesting and communication allows placing the power source regardless the location of the receiving gateway, which can be even far farther away if enough energy is provided. The proposed solution is a fully autonomous RF energy-harvesting node, where all components (rectenna, DC/DC converter and sensing platform) are jointly optimized in order to obtain the best performances within distances up to 17 m. This work also proposes a micropower-compatible adaptive power management, based on both dedicated circuitry and microcontroller firmware. The introduction of smart voltage supervisors allows the harvesting module to adaptively configure the data transmission period of the node, and to accordingly change the sleep policy. The combination of a nano-power DC/DC converter, smart supervisors and a complex wireless sensor nodes allows operation with negligible input power and state-of-the-art operative distances up to 17 m. 2. Materials and Methods 2.1. The Harvesting Module Incident RF power in the standard 868 MHz band (scavenged or intentionally transmitted) is captured from the environment by a rectifying antenna, or rectenna, which converts it into DC power. Due to the unregulated nature of the output voltage and current, and to the strong dependence of the

Sensors 2017, 17, 1732 Sensors 2017, 17, 1732

4 of 21 4 of 21

output voltage on the loadvoltage current, power module is also required toisregulate, store, dependence of the output ona the loadmanagement current, a power management module also required and distribute the harvested power to the sensor node. to regulate, store, and distribute the harvested power to the sensor node. 2.1.1. Rectenna Rectenna 2.1.1. This section section describes describes the the rectenna rectenna unit unit we Figure 22 This we devised devised and and reports reports on on its its performance. performance. Figure shows a photo of the discrete-component prototype: it consists of a PIFA (Planar Inverted F shows a photo of the discrete-component prototype: it consists of a PIFA (Planar Inverted F Antenna)-like printed which is matched to thetoRFthe detector input port by means of ameans multistage Antenna)-like printedantenna antenna which is matched RF detector input port by of a distributed-element matching network. multistage distributed-element matching network.

Figure 2. Rectenna Rectenna prototype prototype with with highlighted different sections. Figure

The the space space constraints constraints The first first step step towards towards rectenna rectenna optimization optimization is is antenna antenna design. design. To To satisfy satisfy the of typical sensors for WSN applications [19,20], we selected a compact PIFA-like antenna designed of typical sensors for WSN applications [19,20], we selected a compact PIFA-like antenna designed to operate in the 868 MHz band (Figure 2). The topology is similar to that proposed for a completely to operate in the 868 MHz band (Figure 2). The topology is similar to that proposed for a completely different rather than joint maximization of different type type of ofapplication application[21] [21]ininwhich whichthe thetarget targetwas wassensitivity sensitivity rather than joint maximization output power and voltage. It consists of two branches in which the lengths have been optimized to of output power and voltage. It consists of two branches in which the lengths have been optimized tune the antenna to the required resonant frequency and to meet the specifications on radiation to tune the antenna to the required resonant frequency and to meet the specifications on radiation efficiency, radiation pattern pattern and and input input return return loss. loss. The efficiency, radiation The overall overall substrate substrate size size of of the the final final design design is is 2.2 The commercial substrate chosen was a Roger 4350B (εr = 3.55, tan δ = 0.0031, thickness 73 × 55 mm 73 × 55 mm . The commercial substrate chosen was a Roger 4350B (εr = 3.55, tan δ = 0.0031, thickness = =1.51.5 mm). Figure 3a shows the simulated and measured reflection coefficient of the PIFA mm). Figure 3a shows the simulated and measured input input reflection coefficient of the PIFA antenna. antenna. In addition, Figure 3b shows the simulated Eand H-plane at the operating frequency. As In addition, Figure 3b shows the simulated E- and H-plane at the operating frequency. As expected, expected, the antenna has an omnidirectional radiation pattern in the horizontal plane: the the antenna has an omnidirectional radiation pattern in the horizontal plane: the low-directivity low-directivity behavior is afor desirable feature for RF energywhere harvesting scenarios, the behavior is a desirable feature RF energy harvesting scenarios, the direction of thewhere incoming direction of the incoming power is typically unknown. The corresponding peak gain achieved is 1.4 power is typically unknown. The corresponding peak gain achieved is 1.4 dBi. All simulations were dBi. All simulations were carried out using a commercial full-wave simulator (CST Microwave Studio carried out using a commercial full-wave simulator (CST Microwave Studio 2016). 2016).The second step consists in designing the rectifying section, which involves choosing a rectifier The second step consists in designing the rectifying section, which involves a rectifier topology and a nonlinear design for the rectifier-antenna matching network. As choosing regards the former, topology and a nonlinear design for the rectifier-antenna matching network. As regards the former, single-stage full-wave voltage doubler topology represents the best option for ultra-low power single-stage doubler topology represents option for ultra-lowDickson power applications,full-wave as is well voltage documented in the literature [22,23]. the For best this reason, a full-wave applications, as is well documented in the literature [22,23]. For this reason, a full-wave Dickson rectifier based on low-threshold Skyworks SMS7630-079 Schottky diodes was selected. rectifier based on low-threshold Skyworks SMS7630-079 Schottky diodes was selected.

Sensors 2017, 17, 1732 Sensors 2017, 17, 1732

5 of 21 5 of 21

0

H-plane 868MHz

a)

b)

-20 Simulated

-30

E (V/m)

(dB)

-10

1

270

0

E-plane

0,5

θ°

0

90

Measured

-40 820

840

860

(MHz)

880

900

180

Figure Figure3.3. PIFA-like PIFA-like antenna antenna performance: performance: (a) (a) reflection reflection coefficient coefficient and and (b) (b) EEand andH-fieldradiation H-fieldradiation pattern at 868 MHz. pattern at 868 MHz.

Oncethe therectifier rectifiertopology topologyisisselected, selected,an anoptimum optimummatching matchingnetwork networkbetween betweenthe therectifier rectifierand and Once theantenna antennaneeds needsto tobe bedesigned designedby bynonlinear nonlinearcircuit circuittechniques, techniques,in inorder orderto toensure, ensure,throughout throughoutthe the the powerrange rangeavailable, available, that thatmaximum maximumRF RFpower powerenters entersthe therectifier rectifierinput. input. The Thetopology topology chosen chosen isis power thedistributed distributed matching matching network network shown shown in in Figure Figure4a: 4a: itit consists consists of oftwo twoimpedance impedance steps stepsand andtwo two the stubs, the the first first being beingshort-circuited short-circuited to toground groundwhereas whereasthe thesecond secondisisopen. open. Impedance Impedance steps steps are are stubs, employed to guarantee broadband matching between the RF detector (i.e., the rectifying circuit) and employed to guarantee broadband matching between the RF detector (i.e., the rectifying circuit) and theantenna. antenna.For Forthe the operating frequency MHz) chosen herefor and the targeted RFpower, input the operating frequency (868(868 MHz) chosen here and thefor targeted RF input power, from ranging to −10 dBm, a nonlinear is optimized accounting for the dispersive ranging −20from to −−20 10 dBm, a nonlinear regime isregime optimized accounting for the dispersive behavior behavior of sub-network the linear (antenna sub-network (antenna network), and matching network), represented by a of the linear and matching represented by a frequency-variable frequency-variable complex computed reflection coefficient computed by The full-wave simulation. complex reflection coefficient by full-wave simulation. nonlinear model ofThe the nonlinear diodes is model of with the their diodes is completed with istheir package model,optimization which is essential for rectenna accurate completed package model, which essential for accurate of the entire optimization of the entire rectenna at these operating frequencies. Optimization is based the at these operating frequencies. Optimization is based on the Harmonic Balance (HB) method andon aims Harmonic Balance (HB) method and aims at maximizing the RF-to-dcofconversion process, defined through at maximizing the RF-to-dc conversion process, through optimization rectenna efficiency, optimization of rectenna efficiency, defined as [24]: as [24]: V ·I ηRF−dc = RECT ∙ RECT (1) = PAV (1) where VRECT and IRECT are the dc components of the rectified voltage and current at the output port where VRECT and IRECT are the dc components of the rectified voltage and current at the output port when a certain load is applied, while PAV is the power available at the rectenna location, and represents when a certain load is applied, while PAV is the power available at the rectenna location, and the maximum power the antenna is able to deliver to the rectifying circuit. Commercial simulator ADS represents the maximum power the antenna is able to deliver to the rectifying circuit. Commercial was used for the integrated nonlinear/EM design optimization. Figure 4a shows the circuit layout, simulator ADS was used for the integrated nonlinear/EM design optimization. Figure 4a shows the with details on the microstrip line dimensions and on the optimum load resistor. Such a rectenna circuit layout, with details on the microstrip line dimensions and on the optimum load resistor. Such system has been fabricated and the corresponding performance is reported in Figure 4b, where the a rectenna system has been fabricated and the corresponding performance is reported in Figure 4b, simulated and measured behavior of the RF-to-dc conversion efficiency, computed as in (1), and the where the simulated and measured behavior of the RF-to-dc conversion efficiency, computed as in output dc voltage are plotted as a function of the input power (PAV ): the rectenna exhibits a conversion (1), and the output dc voltage are plotted as a function of the input power (PAV): the rectenna exhibits efficiency better than 25% throughout the entire range of low RF power, starting from −20 dBm. a conversion efficiency better than 25% throughout the entire range of low RF power, starting from In particular, at −18 dBm the efficiency is greater than 30%, corresponding to a dc power of about 6 µW, −20 dBm. In particular, at −18 dBm the efficiency is greater than 30%, corresponding to a dc power of which represents a practical limit for current state-of-the-art DC/DC converters [25]. At the same about 6 μW, which represents a practical limit for current state-of-the-art DC/DC converters [25]. At time, a dc output voltage greater than 200 mV has been experimentally verified for PAV = −20 dBm. the same time, a dc output voltage greater than 200 mV has been experimentally verified for All these values were obtained with respect to an optimum dc load of 22 kΩ. PAV = −20 dBm. All these values were obtained with respect to an optimum dc load of 22 kΩ.

Sensors2017, 2017,17, 17,1732 1732 Sensors

of21 21 66of

820μm 15mm

1060μm 9.8mm

650μm 23mm

IRECT

D1

ROPT

10pF

VRECT

820μm 15mm

10pF D2

(22kOhm)

850μm 15mm 600μm 17mm 890μm 9mm

Dickson Rectifier

Simulated

60

Measured

1.2

b)

50

ηRF-dc (%)

a)

Matching Network

1

40

0.8

30

0.6

20

0.4

VRECT (V)

RF Input

0.2

10 -20

-18

-16

-14

-12

-10

PAV (dBm)

Figure 4. 4. Rectifying Rectifying section and rectifier schematic topologies and Figure section performance: performance:(a) (a)matching matchingnetwork network and rectifier schematic topologies (b) RF-to-dc conversion efficiency and output dc voltage versus input poweravailable. and (b) RF-to-dc conversion efficiency and output dc voltage versus input poweravailable.

In Inorder order to to estimate estimate the the maximum maximumachievable achievabledistance distanceof ofaa sensor sensorequipped equippedwith withthe theproposed proposed 22 laboratory environment. rectenna from an RF source, the system was tested in a 15 × 15 m rectenna from an RF source, the system was tested in × 15 m laboratory environment. Experiments took place placepurposely purposelyoutside outside anechoic chamber in order to account for arealistic more Experiments took an an anechoic chamber in order to account for a more realistic scenario including limited multipath effects; nevertheless measured received power was scenario including limited multipath effects; nevertheless measured received power was tolerably tolerably close to the theoretical valuefrom derived Friis equation. close to the theoretical value derived Friisfrom equation. The The measurement measurement procedure procedure was was carried carried out out in in two two steps: steps: (i) (i) first, first, the the receiver-to-transmitter receiver-to-transmitter distance was set; (ii) then, measurement of the output dc voltage on the rectenna distance was set; (ii) then, measurement of the output dc voltage on the rectenna load load was was carried carried out. inin thethe measurements ranged from 4 to413 power was out.The Thedistances distancesinvolved involved measurements ranged from tom. 13The m. transmitted The transmitted power selected following the current regulations for for RFID applications [26]: 0.50.5 WWininthe was selected following the current regulations RFID applications [26]: the863–870 863–870MHz MHz frequency frequency range range and and 22 W W for for aa narrow narrow band band starting starting from from 869.4 869.4 MHz MHz to to 869.65 869.65 MHz. MHz. The Theoutput outputDC DC voltage (22 kΩ) was measured by a digital multimeter. This setup lends voltage on on the the optimum optimum load load R ROPT (22 kΩ) was measured by a digital multimeter. This setup lends OPT itself effectiveexperimental experimentalverification verification of the actual behavior the entire rectenna, the itself to effective of the actual behavior of theofentire rectenna, i.e., the i.e., antenna antenna as thesource powerofsource of the rectifier. create a reference comparison the acting asacting the power the rectifier. In order In to order create to a reference comparison with the with expected expected the link adopted was first by a Friis equation the actual simulatedsimulated results, theresults, link adopted was first modeled by modeled a Friis equation and the actual and rectified power rectified power (PRECTforwardly ) was straight forwardly computed byequation: the following equation: (PRECT ) was straight computed by the following 2  (2) λ = ∙ ∙ ∙ ∙ PRECT = PTX · GTX · GRX · 4π ∙ ·ηRF−dc (2) 4π· D where PTX is the power transmitted at the remote location (the power exciting the transmitting where PTXGis transmitted location (the power exciting the transmitting antenna), TXthe andpower GRX are the gains at ofthe theremote receiving and transmitting antennas, respectively,antenna), D is the G and G are the gains of the receiving and transmitting antennas, respectively, D is the link distance, TX distance, RX λ is the wavelength at the operating frequency. It is noteworthy that experimental and link λ is the wavelength thebe operating It is noteworthy that gain experimental theoretical analyses at may carried frequency. out considering the maximum direction,and fortheoretical both the analyses mayand be carried out considering thethe maximum gain direction, the transmitting and transmitting receiving antennas. As transmitting antenna, for weboth adopted a commercial receiving antennas. As the transmitting antenna, we adopted a commercial logperiodic antenna (PCB logperiodic antenna (PCB VA5JVB), with GTX = 6 dBi. As mentioned before, the receiving antenna VA5JVB), GTX = 6Figure dBi. As mentioned thedcreceiving antenna gain is G = 1.4different dBi. Figure RXtwo gain is GRXwith = 1.4 dBi. 5 reports the before, rectified power versus distance, for ERP5 reports the rectified dc power versus distance, for two different ERP levels. Considering a minimum levels. Considering a minimum required power of 6 μW for the dc-dc converter [27], it can be required power of rectenna 6 µW foristhe converter [27], it can be that experiment, the rectennaand is able concluded that the abledc-dc to operate at 6-m distance for concluded the 0.5-W-ERP up to operate at 6-m distance the 0.5-W-ERP experiment,that and up 12-m distance for the rectenna 2-W-ERP. to 12-m distance for the for 2-W-ERP. It is noteworthy thetomodels used during It is noteworthy that the usedlinear during rectenna and optimization for both the dispersive linear optimization for both themodels dispersive sub-circuit the nonlinear devices allowed one to sub-circuit and the nonlinear devices allowed onegood to obtain predicted performances that were very obtain predicted performances that were in very agreement with the experimental one in over a goodrange agreement with the RF experimental onebeover a wideinrange of transmitted RF power, as can be wide of transmitted power, as can observed the plots of Figure 5 where the measured observed in therectenna plots of Figure where the are measured and predicted rectenna dc power are and predicted output 5dc power compared over a number of output RF source-rectenna compared over a number of RF source-rectenna distances. distances.

Sensors 2017, 2017, 17, 17, 1732 1732 Sensors Sensors 2017, 17, 1732

of 21 21 77 of 7 of 21

Figure Measured and Figure 5. 5. Measured and simulated simulated rectified rectified power power for for different different link link distances. distances. Figure 5. Measured and simulated rectified power for different link distances.

2.1.2. Converter 2.1.2. DC/DC DC/DC Converter 2.1.2. DC/DC Converter The converter used used in the system features an ultra-low The DC/DC DC/DC converter ultra-low power power buck-boost buck-boost converter converter The DC/DC converter used in the system features an ultra-low power buck-boost converter designed in STMicroelectronics 0.32 µm CMOS microelectronic technology [14]. The overall designed in STMicroelectronics 0.32 μm CMOS microelectronic technology [14]. overall designed in STMicroelectronics 0.32 μm CMOS microelectronic technology [14]. The overall architecture architecture can be be divided divided into into two two main main blocks: blocks: a start-up circuit, which allows for IC IC bootstrap bootstrap architecture can be divided into two main blocks: a start-up circuit, which allows for IC bootstrap with with RF RF input input sources sources typically typically providing providinglow lowvoltages, voltages,and andthe themain mainDC/DC DC/DC converter which also also with RF input sources typically providing low voltages, and the main DC/DC converter which also provides voltage (FOCV) MPPT algorithm in order to adapt to the best power providesaafractional fractionalopen-circuit open-circuit voltage (FOCV) MPPT algorithm in order to adapt to the best provides a fractional open-circuit voltage (FOCV) MPPT algorithm in order to adapt to the best transfer condition. The ICThe dynamically decidesdecides whether to route to the to load to aorsmall power transfer condition. IC dynamically whether to power route power theorload to a power transfer condition. The IC dynamically decides whether to route power to the load or to a self-supply capacitor C : this achieves very fast activation times even in the presence of large buffer small self-supply capacitor conv Cconv: this achieves very fast activation times even in the presence of large small self-supply capacitor Cconv: this achieves very fast activation times even in the presence of large capacitors at the load output When charged,charged, all power routedistorouted the load. buffer capacitors at the load port. output port.CWhen Cconv is sufficiently allispower to conv is sufficiently buffer capacitors at the load output port. When Cconv is sufficiently charged, all power is routed to Should get C excessively discharged, all power routedishere to replenish it beforeitthe IC fails. the load.Cconv Should conv get excessively discharged, allispower routed here to replenish before the the load. Should Cconv get excessively discharged, all power is routed here to replenish it before the A of the circuit ICcircuit is reported in Figurein 6. Figure 6. ICblock fails. diagram A block diagram of the IC is reported IC fails. A block diagram of the circuit IC is reported in Figure 6.

Figure 6. DC/DC low-power converter block diagram. Figure6.6.DC/DC DC/DClow-power low-powerconverter converterblock blockdiagram. diagram. Figure

The start-up module consists of a 16-stage charge pump implemented with low-threshold The start-up module consists of a 16-stage charge pump implemented with low-threshold MOSFETs and driven by an internal oscillator. A minimum voltage of approximately 250 mV from MOSFETs and driven by an internal oscillator. A minimum voltage of approximately 250 mV from Vrect is required to keep it operating. During the start-up phase, when the input voltage (i.e., the Vrect is required to keep it operating. During the start-up phase, when the input voltage (i.e., the rectenna output voltage) approaches 250 mV, the charge pump circuit starts working, and the rectenna output voltage) approaches 250 mV, the charge pump circuit starts working, and the output voltage on the self-supply capacitor Cconv is boosted. However, internal devices are not fully output voltage on the self-supply capacitor Cconv is boosted. However, internal devices are not fully switched on until the output voltage gets to 600 mV, so that the output charging rate is initially switched on until the output voltage gets to 600 mV, so that the output charging rate is initially limited by the sub-threshold state of the system. As soon as the generated voltage reaches 0.6 V the limited by the sub-threshold state of the system. As soon as the generated voltage reaches 0.6 V the

Sensors 2017, 17, 1732

8 of 21

The start-up module consists of a 16-stage charge pump implemented with low-threshold MOSFETs and driven by an internal oscillator. A minimum voltage of approximately 250 mV from Vrect is required to keep it operating. During the start-up phase, when the input voltage (i.e., the rectenna output voltage) approaches 250 mV, the charge pump circuit starts working, and the output voltage on the self-supply capacitor Cconv is boosted. However, internal devices are not fully switched on until the output voltage gets to 600 mV, so that the output charging rate is initially limited by the sub-threshold Sensors 2017, 17, 1732 8 of 21 state of the system. As soon as the generated voltage reaches 0.6 V the start-up circuit becomes fully operational and the chargefully pump improvesand its the charging until the output exceedsrate 1.36until V, which start-up circuit becomes operational chargerate pump improves its charging the is the minimum operating voltage of the main DC/DC converter. At this point the charge pump output exceeds 1.36 V, which is the minimum operating voltage of the main DC/DC converter. At is disabled and power conversion occurs through the buck-boost DC/DC converter. Although the overall this point the charge pump is disabled and power conversion occurs through the buck-boost DC/DC efficiency of the start-upthe stage settles between 15%,stage its sole purpose is to5% initially bootstrap converter. Although overall efficiency of5% theand start-up settles between and 15%, its solethe main DC/DC so that its impact on operative efficiency be considered negligible. purpose is toconverter, initially bootstrap the main DC/DC converter, so thatcan its impact on operative efficiency Once the 1.36 Vnegligible. threshold is reached, an in-rush current of about 11 µA is absorbed from the can be considered Once the V threshold reached, current of about 11 is absorbed from the energy source by1.36 the module for aisshort timean to in-rush complete bootstrapping theμA converter functionalities. energy source by can thebemodule forwith a an short time to of complete bootstrapping the converter After this, the module sustained input power just 935 nW showing a quiescent current functionalities. After this, the module can be sustained with an input power of just 935 nW showing of 121 nA. The different functional modes are summarized with the corresponding supply voltages a quiescent of 121 nA. functional modes are summarized the in Table 1. It is current worth recalling that The thesedifferent values refer to the voltage on the self-supplywith capacitor corresponding supply voltages in Table 1.isItthe is worth recalling that these refer to which the voltage Vconv . Another aspect worth consideration high output resistance ofvalues the rectenna, causes on the self-supply capacitor V conv. Another aspect worth consideration is the high output resistance significant voltage drops on its output node as the current increases. of the rectenna, which causes significant voltage drops on its output node as the current increases. Table 1. DC/DC converter functional modes. Table 1. DC/DC converter functional modes.

Vrect Vconv mode

0 V–0.250 V

0.250 V–1.6 V 0 V–0.600 V

0.600 V–1.36 V

Switched

Charge pump

Charge pump

off

(limited efficiency)

(fully functional)

>1.36 V DC/DC converter fully functional  121 nA quiescent current  935 nW minimum input power 11 μA in-rush current from source needed to bootstrap the DC/DC converter

In order to extract the maximum power from the source, the IC adopts an FOCV MPPT In order to extract the maximum power from the source, the IC adopts an FOCV MPPT technique. technique. The input source is kept at 50% of the open-circuit voltage of the rectenna, which actually The input source is kept at 50% of the open-circuit voltage of the rectenna, which actually represents represents the maximum power transfer condition for the system. The open-circuit voltage is the maximum power transfer condition for the system. The open-circuit voltage is sampled for 2 µs sampled for 2 μs every 8 conversion cycles of the DC/DC converter. The buck-boost DC/DC every 8 conversion cycles of the DC/DC converter. The buck-boost DC/DC converter operates in converter operates in discontinuous current conduction mode, and is switched when the source discontinuous current mode,voltage. and is switched when the source crosses the voltage crosses the conduction reference MPPT The overall efficiency is voltage highly affected by reference source MPPT voltage. The overall efficiency is highly affected by source impedance and open-circuit voltage. impedance and open-circuit voltage. Thereafter, a full was performed with the with aim ofthe obtaining exhaustivean characterization Thereafter, a analysis full analysis was performed aim ofan obtaining exhaustive of the DC/DC converter in the specific running conditions of the system proposed. An equivalent characterization of the DC/DC converter in the specific running conditions of the system proposed. model of the rectenna in Section 2.1.1 in was extracted the staticfrom voltage-current An equivalent model described of the rectenna described Section 2.1.1 from was extracted the static characteristics andcharacteristics used for characterization. consists in a DC (VOC ) withsource a 22 kΩ voltage-current and used forThis characterization. Thisvoltage consistssource in a DC voltage (VOCresistance. ) with a 22The kΩ output series resistance. The of output voltage isVharv the V, DC/DC 2.3 V, which is series voltage Vharv the DC/DC set of at 2.3 whichisisset theatchosen maximum the chosen maximum operating voltage of the sensing node as described in Section 2.3. During operating voltage of the sensing node as described in Section 2.3. During operation, the variations variations Vharvhundred will be limited a few hundred mV theload worstwas case. A 10 MΩ to on operation, Vharv will the be limited to on a few mV astothe worst case. A 10as MΩ connected load was connected Vharv.the Results show that whenofthe voltage offrom the rectenna OC V, Vharv . Results show thattowhen open-circuit voltage theopen-circuit rectenna VOC ranges 0.32 V toV1.4 ranges from 0.32 V to 1.4 V, the related efficiency grows from 35% to 73%, with a plateau reached the related efficiency grows from 35% to 73%, with a plateau reached early on at 0.8 V. The whole early on at 0.8 is V.reported The whole is reported in Figure 7. characterization incharacterization Figure 7.

Sensors 2017, 17, 1732 Sensors 2017, 17, 1732

9 of 21 9 of 21

Sensors 2017, 17, 1732

9 of 21

Figure efficiency vs. open circuit Figure7.7.DC/DC DC/DC efficiency vs. open circuit input input voltage. voltage. Figure 7. DC/DC efficiency vs. open circuit input voltage.

WithWith respect to Figure Figure 4b,4b, VRECT will be kept by bythe theIC ICatat atapproximately approximately half of VOC, VOC, as the the With respect to 4b, VRECT by the IC approximately of as respect to Figure VRECTwill willbe be kept kept halfhalf of VOC, as the DC/DC converter works to providing an input input DC/DC converter works totomaintain maintain maximumpower powertransfer transfer providing an DC/DC converter works maintainaa state state of maximum by by providing an input impedance equal to the source impedance(22 (22 kΩ kΩ in thethe extracted model was was impedance equal to the the source impedance (22 kΩ in this thiscase). case).Moreover, Moreover, the extracted model was impedance equal to source impedance this case). Moreover, extracted model considered valid throughout the operative range of the system. Figure 8 shows the dedicated board considered valid valid throughout throughout the the operative operative range range of of the the system. system. Figure Figure 88 shows shows the the dedicated dedicated board board considered containing the DC/DC converteralong alongwith with its its subsidiary subsidiary components. containing the DC/DC DC/DC converter components. containing the

Figure moduleboard. board. Figure8.8.Harvesting Harvesting module

2.2.Active The Active Node 2.2. The Node

Figure 8. Harvesting module board.

The sensing and communication node, whose internal architecture is shown in Figure 9, The sensing and communication node, whose internal architecture is shown in Figure 9, includes 2.2. The Activea Node includes temperature and relative humidity sensor, a low-power microcontroller and a sub-GHz a temperature and relative humidity sensor, a low-power microcontroller and a sub-GHz radio radio device for data communication. STMicroelectronics HTS221 is an ultra-compact sensor based The and communication node, whose HTS221 internalisarchitecture is shown in based Figureon 9, device forsensing data communication. STMicroelectronics an ultra-compact sensor on a planar capacitance technology that integrates humidity and temperature sensing with a mixed a temperature and relative humidity sensor, a low-power microcontroller and a sub-GHz aincludes planar capacitance technology that integrates humidity and temperature sensing with a mixed signal ASIC to provide data measured through standard digital serial interfaces. The ultra-low radiopower device data communication. STMicroelectronics HTS221 isinterfaces. anCortex-M3 ultra-compact sensorpower based signal ASICSTMicroelectronics tofor provide data measured standard digital serial The ultra-low STM32L1 through microcontroller, based on the ARM core, interfaces on a with planar technology thatand integrates andCortex-M3 temperature sensing withwith a mixed STMicroelectronics microcontroller, basedhumidity oncommunication the ARM interfaces the thecapacitance sensorSTM32L1 for data acquisition manages with thecore, STMicroelectronics signal ASIC toacquisition provide data measured through standard digital serial interfaces. The ultra-low sensor for data and manages with the STMicroelectronics SPIRIT1 module, SPIRIT1 module, a low-power sub-GHzcommunication RF transceiver.

STMicroelectronics STM32L1 microcontroller, based on the ARM Cortex-M3 core, interfaces apower low-power sub-GHz RF transceiver. with the sensor for data acquisition and manages communication with the STMicroelectronics SPIRIT1 module, a low-power sub-GHz RF transceiver.

Sensors 2017, 17, 1732 Sensors 2017, 17, 1732

10 of 21 10 of 21

Figure Figure9.9.Block Blockdiagram diagramofofthe thesensing sensingnode. node.

Wireless communication is based on DASH7 [28], an open source Wireless Sensor and Actuator Wireless communication is based on DASH7 [28], an open source Wireless Sensor and Actuator Network (WSAN) protocol; the microcontroller implements OpenTag, a DASH7 protocol stack and Network (WSAN) protocol; the microcontroller implements OpenTag, a DASH7 protocol stack and a a minimal Real-Time Operating System (RTOS) designed to be light and compact and targeted to minimal Real-Time Operating System (RTOS) designed to be light and compact and targeted to run on run on resource-constrained microcontrollers. resource-constrained microcontrollers. The DASH7 network architecture has a star structure where all the nodes, which are typically The DASH7 network architecture has a star structure where all the nodes, which are typically low-power devices able to transmit and receive data, communicate only with a gateway that is never low-power devices able to transmit and receive data, communicate only with a gateway that is never offline and connects the DASH7 network to other networks and to the web. offline and connects the DASH7 network to other networks and to the web. DASH7 supports two communication models: pull and push [29]. The pull model consists of a DASH7 supports two communication models: pull and push [29]. The pull model consists of a request-response mechanism initiated by the gateway; it uses an advertising protocol for rapid request-response mechanism initiated by the gateway; it uses an advertising protocol for rapid ad-hoc ad-hoc node synchronization before sending an addressed request to a node and waiting for the node synchronization before sending an addressed request to a node and waiting for the response [29]. response [29]. The data transfer to the gateway initiated by the nodes is based on the push model The data transfer to the gateway initiated by the nodes is based on the push model (e.g., beaconing); (e.g., beaconing); this approach is implemented as an automated message or beacon that is sent at this approach is implemented as an automated message or beacon that is sent at specific time intervals. specific time intervals. In this work, the beaconing approach is used to send the temperature and humidity data from the In this work, the beaconing approach is used to send the temperature and humidity data from sensor node to the gateway; this method is the least power hungry and the node can send the message the sensor node to the gateway; this method is the least power hungry and the node can send the as soon as the harvesting module provides enough power to power-up the node or wake up it from a message as soon as the harvesting module provides enough power to power-up the node or wake deep sleep state. up it from a deep sleep state. The node can be supplied with a voltage ranging from 1.8 V to 3.6 V, and is programmed to The node can be supplied with a voltage ranging from 1.8 V to 3.6 V, and is programmed to perform two different stop policies: off-mode and standby-mode. The policy is selected by the harvesting perform two different stop policies: off-mode and standby-mode. The policy is selected by the module, as will be explained in the following section. harvesting module, as will be explained in the following section. As can be seen in Figures 10 and 11, the two modes behave in the same way at power on, when, As can be seen in Figures 10 and 11, the two modes behave in the same way at power on, when, after the start-up phase, the node acquires data from the sensor and transmits them to the gateway. after the start-up phase, the node acquires data from the sensor and transmits them to the gateway. Afterwards, if off-mode is selected, the node generates a signal for the harvesting module to inform it Afterwards, if off-mode is selected, the node generates a signal for the harvesting module to inform it that the power supply can be switched-off. When the power supply is once more provided, the node that the power supply can be switched-off. When the power supply is once more provided, the node will perform a new start-up phase, data acquisition and message transmission. This phase is associated will perform a new start-up phase, data acquisition and message transmission. This phase is with a significant energy overhead. Figure 10 shows a schematic of the current consumption of the associated with a significant energy overhead. Figure 10 shows a schematic of the current node and its behavior in off-mode indicating the average current in each of the three phases (start-up, consumption of the node and its behavior in off-mode indicating the average current in each of the data acquisition and transmission); in all each activation phase lasts 680 ms and the average current three phases (start-up, data acquisition and transmission); in all each activation phase lasts 680 ms consumption over this time is 4.7 mA at 1.9 V. and the average current consumption over this time is 4.7 mA at 1.9 V. On the contrary, if the standby-mode is selected, after the first start-up phase and message On the contrary, if the standby-mode is selected, after the first start-up phase and message transmission, the node informs the harvesting module that transmission is completed and that it transmission, the node informs the harvesting module that transmission is completed and that it is is entering a deep sleep state in which its current consumption is 3 µA. The node then waits for an entering a deep sleep state in which its current consumption is 3 μA. The node then waits for an external interrupt, generated by the harvesting module, to wake up and perform a new data acquisition external interrupt, generated by the harvesting module, to wake up and perform a new data and message transmission. In standby-mode, therefore, the power supply is never switched-off, and the acquisition and message transmission. In standby-mode, therefore, the power supply is never overhead consists in constant power consumption. The schematic of the node’s current consumption switched-off, and the overhead consists in constant power consumption. The schematic of the node’s and its behavior in standby-mode is shown in Figure 11. In this configuration, with the exception current consumption and its behavior in standby-mode is shown in Figure 11. In this configuration, of the first power-on in which the current consumption is the same as each activation in off-mode, with the exception of the first power-on in which the current consumption is the same as each the start-up phase does not have to be repeated for each activation phase because the power supply activation in off-mode, the start-up phase does not have to be repeated for each activation phase because the power supply is never switched-off. As can be seen in Figure 11, in standby-mode each

Sensors 2017, 17, 1732

11 of 21

Sensors 2017, 17, 1732

11 of 21

is never switched-off. As can be seen in Figure 11, in standby-mode each activation phase following Sensors 2017, 17, 1732 11 of 21 theactivation first one phase lasts 38 ms, andthe thefirst average current during thisconsumption time is 3.81during mA atthis 1.9 V. following one lasts 38 ms,consumption and the average current In activation standby-mode, thefollowing charge for each data message transmission is this much phase the first one lasts 38 andacquisition the averageand current consumption during time is 3.81 mA at 1.9 V.consumption In standby-mode, thems, charge consumption for each data acquisition and time is in 3.81 mA at but 1.9 isV. In standby-mode, the charge consumption for each data acquisition less than off-mode, a much current ofthan 3 µAinhas to be guaranteed the alive in theand deep message transmission less off-mode, but a currenttoofkeep 3 μA hasnode to be guaranteed to message than in off-mode, but a current of 3 μA has to be guaranteed to keep the transmission node alive in is themuch deep less sleep state. sleep state. keep the node alive in the deep sleep state.

Figure 10.10. Schematic, not inoff-mode; off-mode;the theaverage average charge Figure Schematic, nottotoscale, scale,ofofthe thenode nodecurrent current consumption consumption in charge consumption in in each activation phase isisQnode ==I ∗current 4.7 mA∗∗680 ms 680 ms 3.196 mC. Figure 10. Schematic, not to scale, of the consumption in== off-mode; the average charge consumption each activation phase ∗T = = 4.7 mA 3.196 mC. consumption in each activation phase is

= ∗

= 4.7 mA ∗ 680 ms = 3.196 mC.

The selection betweenoff-mode off-modeand andstandby-mode standby-mode has power budget. The selection between has aa significant significantimpact impactononthe the power budget. The selection betweenwith off-mode and standby-mode has a significant impact on the power budget. The off-mode is associated a constant energy overhead, consisting in the energy consumed forfor The off-mode is associated with a constant energy overhead, consisting in the energy consumed The off-mode associated with a constant energy overhead, consisting the energy consumed for starting the is system. By contrast, the standby-mode is associated with ainconstant power overhead. starting the system. By contrast, the standby-mode is associated with a constant power overhead. Hence, starting the system. the standby-mode is associated with a constant power overhead. Hence, which of the By twocontrast, modes costs less will depend on the frequency of activation. For frequent which of the two modes costs less will depend on the frequency of activation. For frequent activation, Hence, which the two modes costs less will the frequency activation.ofFor frequent activation, theofstand-by mode consumes less depend energy.on Furthermore, theoffrequency activations the stand-by mode consumes less energy. Furthermore, the frequency of activations strictly depends activation, the stand-by mode consumes strictly depends on the harvested power. less energy. Furthermore, the frequency of activations onstrictly the harvested power. depends on thethe harvested power. Figure 12 shows sensor node connected to the harvesting module describer in Section 2.1. Figure 12 shows the sensor node harvestingmodule moduledescriber describerininSection Section 2.1. Figure 12 shows the sensor nodeconnected connected to to the the harvesting 2.1.

Figure 11. Schematic, not to scale, of the node current consumption in standby-mode; the average Figure Schematic, not to to scale, of the node current consumption standby-mode; charge consumption each activation following the consumption powerinon phase is the = average ∗ =the Figure 11.11. Schematic,in not scale, of phase the node current in standby-mode; charge consumption in each activation following power onthe phase is on = phase ∗ = is 3.81 mA ∗ 38 ms ~ 145 μC. average charge consumption in each phase activation phasethefollowing power 3.81 mA ∗ 38 ms ~ 145 μC. Q = I ∗ T = 3.81 mA ∗ 38 ms ∼ 145 µC.

Sensors 2017, 17, 1732

12 of 21

Sensors 2017, 17, 1732

12 of 21

Sensors 2017, 17, 1732

12 of 21

Figure 12. Sensing node and harvesting module boards.

Figure 12. Sensing node and harvesting module boards. Figure 12. Sensing node and harvesting module boards.

2.3. Policies for Power Optimization 2.3. Policies for Power Optimization 2.3. Policies Power Optimization In this for section, the proposed adaptive feature is described in detail, as well as the architectural choices andsection, dimensioning of control circuits shown in Figure 13. The interface the In this section, thethe proposed adaptive feature isdescribed described detail, as well as between the architectural In this proposed adaptive feature is inin detail, as well as the architectural harvesting module and theof active node circuits iscircuits implemented four13. control signals connected to the the choices dimensioning ofcontrol control shownthrough in interface between choices and and dimensioning shown in Figure Figure 13.TheThe interface between microcontroller (Policy, Reset_V DDnode , node Startisand Reset_Start), besides the ground and connected positive harvesting module andthe the active implemented through four control signals to the harvesting module and active is implemented through four control signalssupply connected (V DD). microcontroller (Policy, Reset_V DD, Start and Reset_Start), besides the ground and positive supply to the microcontroller (Policy, Reset_VDD , Start and Reset_Start), besides the ground and positive (VDD).The Policy signal selects one of the two stop policies of the active node (off-mode and supply (VDD ). standby-mode). The Policy signal selects one of the two stop policies of the active node (off-mode and The Policy signal selects one of the two stop policies of the active node (off-mode and standby-mode). standby-mode).

Figure 13. Schematic of the harvesting module and its interface with the active node. Figure 13. Schematic of the harvesting module and its interface with the active node. Figure 13. Schematic of the harvesting module and its interface with the active node. The first factor to be analyzed is the buffer capacitance Cbuff, which plays the role of energy storage by the to offset requestCof nodeplays during active phase. Theneeded first factor tosystem be analyzed is the total bufferpower capacitance buffthe , which theitsrole of energy The first factor besystem analyzed is the buffer capacitance Cbuff which plays the rolephase. of energy Such requests are summarized inoffset Table 2, which integrates the current absorbed byits the node when storage needed byto the to the total power request of the, node during active storage needed byare the system to in offset total request of theabsorbed node during active switching into active mode from each ofthe thewhich otherpower two modes. Such requests summarized Table 2, integrates the current by the its node whenphase. mode from each of2,the otherintegrates two modes.the current absorbed by the node when Suchswitching requests into are active summarized in Table which 2. Charge requests of node during states according to mode. switching into activeTable mode from each of the other twoactive modes. Table 2. Charge requests of node during active states according to mode.

Mode Charge (μC) Operating Voltage (V) Table 2. Charge requestsCharge of 3196 node during active 2.3–1.9 states according Mode (μC) Operating Voltage (V) to mode. OFF STANDBY 145 ~2 OFF 3196 2.3–1.9 Mode Charge (µC) Operating STANDBY 145 ~2 Voltage (V) OFF 3196 2.3–1.9 STANDBY 145 ~2

Sensors 2017, 17, 1732

13 of 21

In order to have a voltage drop of about 0.4 V, the resulting Cbuff must be at least Cbu f f =

3196 µC Charge(OFF ) = = 7990 µF ∼ = 8 mF ∆V 0.4 V

(3)

where the voltage drop was chosen as a trade-off between system power consumption and wake-up time. In this instance, a lower bound of 1.9 V was chosen by considering a 100 mV margin to the minimum 1.8 V supply voltage required by the node to operate, while the upper bound was chosen by considering that a lower voltage implies lower power consumption, though at the same time it requires a larger buffer capacitance which causes a longer start-up time when the system has to be booted for the first time. Hence a maximum operating voltage of 2.3 V was chosen and, as a result, an 8 mF Cbuff must be considered. Three control circuits are necessary to generate the correct control signals for the micro-controller (Figure 13). The first one is the Control circuit for OFF mode. This block has the task of monitoring the voltage across Cbuff capacitance voltage and consequently to provide power supply to the sensing node only when voltage is in the acceptable range (1.9 V–2.3 V). In this way, the sensing module is only switched on when VDD exceeds 2.3 V, i.e., when the buffer capacitance has sufficient energy to sustain at least one complete data transmission. Likewise, power supply is taken off as soon as VDD drops below 1.9 V, as the sensing module cannot work at lower voltages. During off-mode, a reset signal (Reset_VDD ) must be issued by the microcontroller at the end of each transmission stage, so that power supply is simultaneously turned off. This behavior ensures that the next transmission will only be held when the buffer capacitance is fully recharged, i.e., when it reaches 2.3 V again. A NCP303 low-power voltage supervisor was used to implement the block, along with two AS11P2 analog switches and two resistors in order to obtain the desired hysteresis, respectively 2.4 MΩ and 500 kΩ. The total quiescent current of the block is 500 nA, mostly due to the NCP303 supervisor. The second one is the Control circuit for STANDBY mode. This block has the task of monitoring the capacitance voltage and consequently providing the start signal only when voltage is above 2.0 V. During standby-mode, the sensing node is always powered, and subsequent transmissions are regulated by the start signal, so that it must only be issued when the buffer capacitance is storing enough energy to sustain transmission. Actually, the expected voltage drop due to a single transmission in standby-node is a mere 18 mV as explained in following equation ∆V (STANDBY ) =

Charge(STANDBY ) 145 µC = = 18 mV Cbu f f 7990 µF

(4)

so a threshold voltage of 1.9 V + ∆V = 1.918 V could be enough to sustain standby-mode, but a safer margin of 2.0 V − 1.9 V = 100 mV was chosen for this condition. A reset signal (Reset_Start) from the microcontroller is also requested at the end of each transmission in order to force the start signal to a low level, even if the related capacitance voltage drop has not been sufficient to trigger the voltage supervisor. This happens because the microcontroller is sensitive to a positive start signal edge, so that a low-to-high transaction is always required to awaken the sensing node. An NCP303 low-power voltage supervisor was used to implement the block, along with one AS11P2 analog switch and a pull-up resistor of 2.4 MΩ. The total quiescent current of the block is 400 nA, mostly due to the NCP303 supervisor. The ability to dynamically select whether to operate in off-mode or standby-mode is a crucial aspect of the solution presented, as it enables the system to fully exploit the available environmental energy under any circumstance without human intervention. In both cases the minimum transmission period is obtained, as the microcontroller is either supplied (off-mode) or awoken (standby-mode) as soon as the buffer capacitance has recovered the energy lost, but depending on the variable amount of extracted energy, one solution can be less advantageous than the other, if not altogether unfeasible.

Sensors 2017, 17, 1732

14 of 21

Transmission Period (TP) is defined as the time interval between two consecutive transmissions and can be calculated as: Qneeded (5) TP = ( Iextracted − Iqharv − Iqnode ) Sensors 2017, 17, 1732 14 of 21 where Qneeded is the charge lost during each transmission, Iextracted is the current provided by the where converter, Qneeded is theIqharv charge lostquiescent during each transmission, Iextracted is the currentthat provided by the DC/DC by DC/DC is the current of the harvesting module is directly supplied converter, I qharv is the quiescent current of the harvesting module that is directly supplied by VHARV VHARV while Iqnode is the quiescent current of the rest of the system supplied by VDD . These values while Iqnode is the the two quiescent current of the rest the system supplied differ between operative modes, and areofsummarized in Table 3.by VDD. These values differ between the two operative modes, and are summarized in Table 3. Sensors 2017, 17, 1732

14 of 21

Table 3. Quiescent currents of harvesting module and sensor node during recharge phase. Table 3. Quiescent currents of harvesting module and sensor node during recharge phase.

where Qneeded is the charge lost during each transmission, Iextracted is the current provided by the DC/DC converter, Iqharv is the quiescentIqharv current directly supplied by VHARV (µA) of the harvesting module that Iqnodeis(µA) while Iqnode is the quiescent current of the rest of the system supplied by VDD.for These values differ Qneeded (µC) Control circuit Control circuit Control circuit Sensor node OFF-mode STANDBY-mode between the two operative modes, and are summarized in Table 3. Policy detection OFF

0.5

switched-off 0.5

Iqharv(μA) STANDBY

OFF

switched-off

switched-off

Table 3.3196 Quiescent currents of harvesting module and sensor node during recharge phase. switched-off

Qneeded(μC) 145

3196

Control 0.5 circuit OFF-mode 0.5

0.4 Control circuit STANDBY-mode switched-off

4.5

Iqnode(μA) 0.6 Control circuit for 4 Policy detection

3 Sensor node

switched-off

switched-off

switched-off

The most significant difference is that in off-mode the charge needed is significantly higher than The most significant difference is that in off-mode the charge0.5needed is significantly higher than in in standby-mode, but the total quiescent current is only 500 nA, since all the other modules are 0.6 modules are switched 3 standby-mode, but the total quiescent current is only 5000.4 nA, since all the other off 0.5 switched off during recharge periods. As a consequence, in off-mode nearly all the current from the 4 145 STANDBY during recharge periods. As a consequence, in off-mode nearly all the current from the power converter power converter can be used to recharge the buffer capacitance. 4.5 can be used to recharge the buffer capacitance. Thus, this last mode proves more suitable when the extracted current is relatively low, whereas The most significant difference is that in off-mode the charge needed is significantly higher than the standby-mode becomes more cost effective at higher extracted currents, when its lower value of in standby-mode, but the total quiescent current is only 500 nA, since all the other modules are activation charge becomes predominant. A detailed graph of transmission data rates versus extracted switched off during recharge periods. As a consequence, in off-mode nearly all the current from the currents is shown in Figure 14, analytically obtained through Equation (5). power converter can be used to recharge the buffer capacitance.

Figure 14. Transmission period vs. Extracted currents in different operating modes.

Thus, this last mode proves more suitable when the extracted current is relatively low, whereas the standby-mode becomes more cost effective at higher extracted currents, when its lower value of activation charge becomes predominant. A detailed graph of transmission data rates versus extracted currents is shown in Figure 14, analytically obtained through Equation (5). Figure Transmissionperiod periodvs. vs.Extracted Extracted currents currents in Figure 14.14. Transmission in different differentoperating operatingmodes. modes. The available current is defined as the effective current that can be used by the active part of the system or, in words,proves the whole current by Iqharv,iswhich is always drawn Thus, thisother last mode moreextracted suitable when thesubtracted extracted current relatively low, whereas Thethe available current is defined the effective current that cangraphs, be usedasby the active part of the from DC/DC converter (I available = as I extracted − I qharv). As shown in the the available current the standby-mode becomes more cost effective at higher extracted currents, when its lower value of system or, in other words, the whole extracted current subtracted by Iqharv , which isdata always drawn from rises above 4.190 µA, standby-mode becomes the best choice in terms of minimum rates, and the activation charge becomes predominant. A detailed graph of transmission data rates versus related threshold value is about 13 min. extracted currents is shown in Figure 14, analytically obtained through Equation (5). With this assumption, Control for policy detection task of The available current isthe defined ascircuit the effective current thathas canthe be used bycorrectly the activeidentifying part of the whether the current available is either above or below the threshold value ofis4.190 µA,drawn and system or, in other words, the whole extracted current subtracted by Iqharv, which always communicating this to the microcontroller through the policy signal. Since the correct relationship from the DC/DC converter (Iavailable = Iextracted − Iqharv). As shown in the graphs, as the available current between the4.190 power received by thebecomes rectennathe andbest thechoice powerinextracted the DC/DC been fully rises above μA, standby-mode terms ofby minimum datahas rates, and the

Sensors 2017, 17, 1732

15 of 21

the DC/DC converter (Iavailable = Iextracted − Iqhar v ). As shown in the graphs, as the available current rises above 4.190 µA, standby-mode becomes the best choice in terms of minimum data rates, and the related threshold value is about 13 min. With this assumption, the Control circuit for policy detection has the task of correctly identifying whether the current available is either above or below the threshold value of 4.190 µA, and communicating this to the microcontroller through the policy signal. Since the correct relationship between the power received by the rectenna and the power extracted by the DC/DC has been fully investigated (see details in Results section), the least power-consuming solution is to simply monitor the average voltage at the rectenna output. This value is solely related to the extracted current through the efficiencies of the interposed stages, and experimental results show that a value ≥585 mV for VRECT is required to obtain an available current of 4.190 µA. Since the low-power voltage supervisor 2V NCP303 has a threshold value of 2 V, a dedicated low-power voltage amplifier with a 0.585 V = 3.42 gain was added, composed of an LPV521 op-amp and two resistors of 20 MΩ and 4 MΩ. Moreover, a simple RC filter was inserted with τ = 100 ms, in order to reject the periodic fluctuations caused by the FOCV algorithm of the DC/DC converter, which actually disconnect the rectenna for about 4 µs every 8 extraction cycles. The overall quiescent current of the block is 600 nA, as already reported in Table 3. 3. Results The fully autonomous DASH7 IoT sensor node with RF harvesting capability was characterized experimentally. Moreover, we tested the capabilities of implementing the adaptive behavior. The available energy is mainly affected by two factors: the transmission power and the distance between transmission and receiving antennas. For the former, two different transmitter powers were selected to be compliant with the RFID standard [26], respectively 0.5 W ERP and 2.0 W ERP. Then, a full analysis with respect to values of distance was performed. Starting from experimental results for the rectenna (Figure 4) and DC/DC (Figure 7), a precise analytic model of the harvesting module was extracted, and this provides the power extracted by the DC/DC converted as a function of RF source power and node distance. The related results are presented in Section 3.1. Furthermore, model reliability was proved by two experiments accomplished with a regulated RF source placed at a defined distance from the harvesting module (logperiodic directive antenna “PCB VA5JVB”, with GTX = 6 dBi). Combining the harvesting analytic model described in Section 3.1 with Equation (5) and the measured currents reported in Table 3, yielded a detailed estimate of startup times and transmission periods as reported in Section 3.2. The foregoing results are summarized in Section 3.3. Section 3.4 presents the experimental results of the sensor node, while Section 3.5 describes the tests performed on voltage supervisors. 3.1. Harvesting Module Results First of all, a specific graph reporting the available power versus node distance is presented in Figure 15. The available power is the total power extracted from the DC/DC converter. It also accounts for losses on the capacitor and on instrumentation. The graph shows two different curves for 0.5 W ERP and 2.0 W ERP, which intersect at threshold values. The value of 1.373 µW is the threshold value at which the system presented can be considered self-sustained, as it can successfully supply the minimum actual load of the system, which is the control circuit for off-mode described in Figure 13. The far from inconsiderable result obtained is that with a transmitted power of just 0.5 W ERP the system presented can start at 8.4 m, while with 2.0 W ERP the distance can rise up to 16.8 m. Minimum distance of 8.4 m with 0.5 W ERP transmitted power was verified through a dedicated experiment.

Sensors 2017, 17, 1732 Sensors2017, 2017,17, 17,1732 1732 Sensors

16 of 21 16of of21 21 16

Figure 15. Power extracted vs. Node distance. Figure15. 15.Power Powerextracted extractedvs. vs.Node Nodedistance. distance. Figure

3.2. Startup/Transmission Startup/Transmission Period Results Startup/TransmissionPeriod PeriodResults Results 3.2. When from a completely discharged situation, an initial time is required When the thesystem systemstarts starts from completely discharged situation, an startup initial startup startup time isis When the system starts from aa completely discharged situation, an initial time to charge the 8-mF buffer capacitance whose value depends on the harvested current. required to charge the 8-mF buffer capacitance whose value depends on the harvested current. required to charge the 8-mF buffer capacitance whose value depends on the harvested current. Once booted, booted, the the node node isis automatically automatically configured operate in the best best operating operating mode, mode, configured to to operate in the Once providing the transmission period. A detailed graph of startup and transmission times providing shortest transmission period. A detailed graph of startup and transmission times providing the shortest transmission period. A detailed graph of startup and transmission times isis is reported in Figure 16. reported in Figure 16. reported in Figure 16.

Figure 16. Transmission data rate and policy relationship with node distance. Figure 16.Transmission Transmission datarate rateand andpolicy policyrelationship relationshipwith withnode nodedistance. distance. Figure 16. data

resultsshow show that the minimum transmission period about 2121 corresponding toan an Analyticresults showthat thatthe theminimum minimumtransmission transmission period is about s, corresponding to Analytic period isisabout 21 s,s,corresponding to extracted power of 26.5 μW obtained with an antenna open circuit voltage of 1.650 V. Similarly, the an extracted power of 26.5 obtained an antenna circuit voltage of 1.650 V. Similarly, extracted power of 26.5 μW µW obtained withwith an antenna openopen circuit voltage of 1.650 V. Similarly, the operatingmode modeisisswitched switchedfrom fromoff-mode off-modeto tostandby-mode standby-modeaccording accordingto toFigure Figure13 13when whenthe theextracted extracted operating powerisis11.011 11.011μW μW(open (opencircuit circuitvoltage voltage1.170 1.170V), V),with withaaperiod periodof of12′43″. 12′43″. power

Sensors 2017, 17, 1732

17 of 21

the operating mode is switched from off-mode to standby-mode according to Figure 13 when the extracted power is 11.011 µW (open circuit voltage 1.170 V), with a period of 120 43”. Sensors 2017, 17, 1732 17 of 21 3.3. Power/Transmission Power/Transmission Period 3.3. Period Combined Combined Results Results Figure 17 17 shows and Figure shows the the relationship relationship between between node node distance, distance, power power extracted extracted from from DC/DC DC/DC and obtainable transmission periods. Starting from distance (bottom horizontal axis), the single blue line obtainable transmission periods. Starting from distance (bottom horizontal axis), the single blue line represents extracted is 0.5 W ERP, the triple pinkpink line isline extracted powerpower when represents extractedpower powerwhen whenthe thesource source is 0.5 W ERP, the triple is extracted the source is 2.0 W ERP. Once the available power is known, the obtainable period can be found on when the source is 2.0 W ERP. Once the available power is known, the obtainable period can the be dotted on green whose values reported the top horizontal Moreover,axis. threeMoreover, operative found the line, dotted green line, are whose valuesalong are reported along the axis. top horizontal regions of extracted power are identifiable: a turn off regionawhere the node where cannotthe be switched on, three operative regions of extracted power are identifiable: turn off region node cannot an off-mode operating region and a standby-mode one. be switched on, an off-mode operating region and a standby-mode one.

Figure 17. Combined Figure 17. Combined graph graph showing showing the the relationship relationship between between distance, distance, extracted extracted power power and and transmission transmission periods. periods.

Table 4 recaps the threshold values shown in Figure 17. Table 4 recaps the threshold values shown in Figure 17. Table 4. Minimum distances in the three operating conditions. Table 4. Minimum distances in the three operating conditions. Distance (m) Antenna Voc Transmission Extracted Power (μW) Distance (m) (V) Period Extracted 0.5 W ERP 2.0 W ERP Antenna Voc Transmission Power Period→ ∞ 0.5 W ERP 8.4 2.0 W ERP 16.8 Switch-on 0.490 (V) 1.373(µW) Policy Threshold 1.170 0.490 11.011 4.6 9.1 Switch-on 1.373 → ∞0:12:43 8.4 16.8 Policy Threshold 1.650 1.170 11.011 0:12:43 4.6 9.1 Min. Trans Period 26.531 0:00:21 3.5 7.0 Min. Trans Period

1.650

26.531

0:00:21

3.5

7.0

3.4. Sensor Node Results 3.4. Sensor Node Results The effective operation of the DASH7 node was tested, in terms of power consumption, The effective operation of the DASH7 node was18a tested, terms of power consumption, transmission capability and adaptive data rate. Figure showsinwaveforms of signals related to transmission capability rate. Figure 18aVshows signalsofrelated off-mode operation. Supplyand VDD adaptive decreasesdata as expected from 2.3 to 1.9 Vwaveforms over a timeofinterval about to off-mode operation. Supply V decreases as expected from 2.3 V to 1.9 V over a time interval of about 680 ms, which correspond toDD a single transmission. With reference to Figure 10, this time accounts 680 the ms,microcontroller which correspond to a single transmission. With reference to Figure this time accounts for for start-up phase, the data acquisition and finally data10, transmission. The reset the microcontroller phase, theend dataofacquisition and finally data transmission. resetcontrol signal signal (Reset_VDD) isstart-up activated at the each transmission. Similarly, in Figure The 18b the (Reset_Vfor ) is activated at end ofVoltage each transmission. Similarly, in expected, Figure 18bwhile the control for signals arethe shown. drop is about 20 mV as in thissignals mode the DDstandby-mode standby-mode are shown. Voltage dropofisReset_V about 20 as end expected, while in this mode the Reset_Start Reset_Start signal is activated instead DDmV at the of transmission. signal is activated instead of Reset_VDD at the end of transmission.

Sensors 2017, 17, 1732

18 of 21

Sensors 2017, 17, 1732

18 of 21

Sensors 2017, 17, 1732

18 of 21

(a)

(b)

Figure 18. Voltage drop (a) and standby-mode (b). (a) and control signals of micro-controller in off-mode(b) Figure 18. Voltage drop and control signals of micro-controller in off-mode (a) and standby-mode (b). Figure 18. Voltage drop and control signals of micro-controller in off-mode (a) and standby-mode (b).

Transmitted data are then received by a dedicated gateway, which displays data directly onto a Transmitted data thenthe received bydata a dedicated gateway, which displays data directly onto a PC terminal. Figure 19are shows received bytes with different sensor values. Transmitted data thenthe received bydata a dedicated gateway, which displays data directly onto a PC terminal. Figure 19are shows received bytes with different sensor values. PC terminal. Figure 19 shows the received data bytes with different sensor values.

Figure 19. Data received format. Figure 19. Data received format.

FigureResults 19. Data received format. 3.5. Voltage Supervisor and Policy Selection

3.5. Voltage and Policy Selection Finally,Supervisor power optimization circuits Results for policy selection were tested. In Figure 20 control signals 3.5. Voltage Supervisor and Policy Selection Results behavior can be observed with respect to DC/DC output voltage VHARV. Considering signals reported Finally, power optimization circuits for policy selection were tested. In Figure 20 control signals Finally, circuitsthe for activation policy selection were of tested. In Figure 20 control signals in Figure 13,power VHARVoptimization is initially below threshold 2.3 V, so all other signals are behavior can be observed with respect to DC/DC output voltage VHARV. Considering signals reported behavior can observed with respect DC/DC output voltage VHARV . Considering signals reported exceeds 2.3toV, the node can be enabled so V DD (=VHARV) is provided, and disabled. As be soon as VHARV in Figure 13, VHARV is initially below the activation threshold of 2.3 V, so all other signals are in Figure 13, VHARVare is supplied initially below the V activation threshold V, so otherissignals other supervisors through DD too. Start signalof is 2.3 high, as all VHARV above are 2.0 disabled. V, while disabled. As soon as VHARV exceeds 2.3 V, the node can be enabled so VDD (=VHARV) is provided, and As soon VHARV V, above the node canVbe so(see VDDTable (=VHARV ) isRF provided, other policy is as also high exceeds as VOC is 2.3 kept 1.170 in enabled this stage 4). As source isand switched other supervisors are supplied through VDD too. Start signal is high, as VHARV is above 2.0 V, while supervisors arevoltage supplied through VDD Start signal is high, VHARV is above 2.0low. V, while off, the buffer VHARV begins to too. decrease, and policy is as immediately forced Thenpolicy start policy is also high as VOC is kept above 1.170 V in this stage (see Table 4). As RF source is switched is also goes highlow as Vwhen kept 1.170 2.0 V in stage (see Table 4). Asvoltage RF source switched off, signal HARVabove falls below V this as expected, while supply VDD is disconnected OC is V off, the buffer voltage VHARV begins to decrease, and policy is immediately forced low. Then start the buffer voltage VHARV1.9 begins when VHARV goes below V. to decrease, and policy is immediately forced low. Then start signal signal goes low when VHARV falls below 2.0 V as expected, while supply voltage VDD is disconnected goes low when V falls below 2.0 V as expected, while supply voltage VDD is disconnected when when VHARV goes HARV below 1.9 V. VHARV goes below 1.9 V.

Sensors 2017, 17, 1732

19 of 21

Sensors 2017, 17, 1732

19 of 21

Figure Figure20. 20.Power Poweroptimization optimizationcircuits circuitsand andpolicy policydetection detectionwaveforms. waveforms.

4.4.Conclusions Conclusions In this work workwe we proposed an autonomous battery-less node RF harvesting In this proposed an autonomous battery-less sensorsensor node with RFwith harvesting capability capability in the 868 MHz UHF band. Temperature and humidity data are acquired and transmitted in the 868 MHz UHF band. Temperature and humidity data are acquired and transmitted using a using a standard protocol, which the allows the to be compatible IoT infrastructure. standard DASH7DASH7 protocol, which allows node tonode be compatible with IoTwith infrastructure. An optimized harvesting module, combining a dedicated rectifying antenna An optimized harvesting module, combining a dedicated rectifying antennaand andan anultra-low ultra-low power DC/DC converter with MPPT, is used to obtain considerable results at long distances. power DC/DC converter with MPPT, is used to obtain considerable results at long distances.Besides Besides that, that, innovative innovative solutions solutions for for adaptive adaptive behavior behavior are are introduced. introduced. In Inparticular, particular, by by monitoring monitoring the the rectifying antenna open-circuit voltage, a dedicated low-power management circuitry can select two rectifying antenna open-circuit voltage, a dedicated low-power management circuitry can select two different operatingmodes modes of sensing the sensing Furthermore, in both cases, thetransmission minimum different operating of the node. node. Furthermore, in both cases, the minimum transmission period is achieved accordingly to the available RF power through the power period is achieved accordingly to the available RF power through the power management module. management module. The node achieves operation with standard wireless protocol in micropower The node achieves operation with standard wireless protocol in micropower condition. condition. As a result, the node is able to operate with an antenna open-circuit voltage of solely 520 mV, As a result,tothe node ablepower. to operate with an antenna open-circuit voltage of solely 520can mV, corresponding 12.3 µW is input Consequently, with a RF source of 2.0 W ERP, the node be corresponding to 12.3 μW input power. Consequently, with a RF source of 2.0 W ERP, the node can turned on up to 16.8 m distance. be turned on up to 16.8 m distance. Acknowledgments: This work was performed within the activities of the joint research laboratory of University of Bologna and STMicroelectronics at the Advanced Center on Electronic Systems of the of university. Acknowledgments: This work was performed withinResearch the activities of the joint research laboratory University of Bologna and STMicroelectronics at the Advanced Research Center on Electronic Systems of the university. Author Contributions: Matteo Pizzotti and Davide Fabbri designed, implemented and tested the electronic section of the “harvesting and power management module”, and wrote the related sections of the paper. Luca Perilli Author Contributions: Matteo Pizzotti and Davide Fabbri designed, implemented and tested the electronic designed, implemented and tested the “active sensing node”, and wrote the related sections of the paper. section of the and power management module”, and wrote the of the paper. of Luca Massimo del “harvesting Prete designed, implemented and tested the “rectenna”, andrelated wrotesections the related sections the Perilli implemented the the “active sensing node”, and wrote the related sectionsAldo of theRomani, paper. paper.designed, Michele Dini designed and tested provided “DC/DC converter”. Eleonora Franchi Scarselli, Roberto Canegallo, Alessandra Costanzo and supervised thewrote development of the system Massimo del Prete designed, implemented andDiego testedMasotti the “rectenna”, and the related sections of and the of individual and RF blocks, and the the “DC/DC writing ofconverter”. the paper. Eleonora Matteo Pizzotti the common parts, paper. Micheleelectronic Dini designed and provided Franchiwrote Scarselli, Aldo Romani, merged and coordinated the whole paper. Roberto Canegallo, Alessandra Costanzo and Diego Masotti supervised the development of the system and of Conflicts of Interest: and The RF authors declare individual electronic blocks, and no theconflict writingofofinterest. the paper. Matteo Pizzotti wrote the common parts, merged and coordinated the whole paper.

References

Conflicts of Interest: The authors declare no conflict of interest. 1.

Yick, J.; Mukherjee, B.; Ghosal, D. Wireless sensor network survey. Comput. Netw. 2008, 52, 2292–2330.

References [CrossRef] 2. 1. 2.

Sudevalayam, S.; Kulkarni, P. Energy harvesting sensor nodes: Survey and implications. IEEE Commun. Yick, J.; Mukherjee, B.; Ghosal, D. Wireless sensor network survey. Comput. Netw. 2008, 52, 2292–2330. Surv. Tutor. 2011, 13, 443–461. [CrossRef] Sudevalayam, S.; Kulkarni, P. Energy harvesting sensor nodes: Survey and implications. IEEE Commun. Surv. Tutor. 2011, 13, 443–461.

Sensors 2017, 17, 1732

3.

4. 5. 6. 7.

8.

9. 10.

11. 12. 13. 14. 15. 16.

17. 18. 19. 20. 21.

22. 23.

24.

20 of 21

Del Prete, M.; Masotti, D.; Costanzo, A.; Magno, M.; Benini, L. A 2.4 GHz–868 MHz dual-band wake-up radio for wireless sensor network and IoT. In Proceedings of the IEEE 11th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob), Abu Dhabi, UAE, 19–21 October 2015; pp. 322–328. Vullers, R.J.M.; van Schaijk, R.; Doms, I.; van Hoof, C.; Mertens, R. Micropower energy harvesting. Solid-State Electron. 2009, 53, 684–693. [CrossRef] Lu, X.; Wang, P.; Niyato, D.; Kim, D.I.; Han, Z. Wireless networks with RF energy harvesting: A contemporary survey. IEEE Commun. Surv. Tutor. 2015, 17, 757–789. [CrossRef] Gollakota, S.; Reynolds, M.S.; Smith, J.R.; Wetherall, D.J. The emergence of RF-powered computing. Computer 2014, 47, 32–39. [CrossRef] Oh, S.; Wentzloff, D.D. A −32 dBm sensitivity RF power harvester in 130 nm CMOS. In Proceedings of the 2012 IEEE Radio Frequency Integrated Circuits Symposium, Montreal, QC, Canada, 17–19 June 2012; pp. 483–486. Carvalho, N.; Georgiadis, A.; Costanzo, A.; Stevens, N.; Kracek, J.; Pessoa, L.; Roselli, L.; Dualibe, F.; Schreurs, D.; Mutlu, S.; et al. Europe and the future for WPT: European contributions to wireless power transfer technology. IEEE Microw. Mag. 2017, 18, 56–87. Chaour, I.; Fakhfakh, A.; Kanoun, O. Enhanced passive RF-DC converter circuit efficiency for low RF energy harvesting. Sensors 2017, 17, 546. [CrossRef] [PubMed] Eid, A.; Costantine, J.; Tawk, Y.; Ramadan, A.H.; Abdallah, M.; ElHajj, R.; Awad, R.; Kasbah, I.B. An efficient RF energy harvesting system. In Proceedings of the 2017 11th European Conference on Antennas and Propagation (EUCAP), Paris, France, 19–24 March 2017; pp. 896–899. Collado, A.; Georgiadis, A. Conformal hybrid solar and electromagnetic (EM) energy harvesting rectenna. IEEE Trans. Circuits Syst. I Regul. Pap. 2013, 60, 2225–2234. [CrossRef] Ouda, M.H.; Khalil, W.; Salama, K.N. Wide-range adaptive RF-to-DC power converter for UHF RFIDs. IEEE Microw. Wirel. Compon. Lett. 2013, 26, 634–636. [CrossRef] Stoopman, M.; Keyrouz, S.; Visser, H.J.; Serdijn, K.P.W.A. Co-design of a CMOS rectifier and small loop antenna for highly sensitive RF energy harvesters. IEEE J. Solid-State Circuits 2014, 49, 622–634. [CrossRef] Dini, M.; Romani, A.; Filippi, M.; Tartagni, M. A nano-current power management IC for low voltage energy harvesting. IEEE Trans. Power Electron. 2015, 31, 4292–4304. [CrossRef] Romani, A.; Tartagni, M.; Sangiorgi, E. Doing a lot with a little: Micropower conversion and management for ambient-powered electronics. Computer 2017, 50, 41–49. [CrossRef] Syed, Y.; Hegde, B.G.; Prabhakar, T.V.; Manjunath, M.; Vinoy, K.J. RF energy harvesting chip powered sensor node. In Proceedings of the 2016 IEEE International Conference on Electronics, Circuits and Systems, Monte Carlo, Monaco, 11–14 December 2016; pp. 748–751. Choi, K.W.; Ginting, L.; Rosyady, P.A.; Aziz, A.A.; Kim, D.I. Wireless-powered sensor networks: How to realize. IEEE Trans. Wirel. Commun. 2017, 16, 221–234. [CrossRef] Li, T.; Fan, P.Y.; Chen, Z.C.; Letaief, K.B. Optimum transmission policies for energy harvesting sensor networks powered by a mobile control center. IEEE Trans. Wirel. Commun. 2016, 15, 6132–6145. [CrossRef] Kamalinejad, P.; Mahapatra, C.; Sheng, Z.; Mirabbasi, S.; Leung, V.C.M.; Guan, Y.L. Wireless energy harvesting for the Internet of things. IEEE Commun. Mag. 2015, 53, 102–108. [CrossRef] Mason, A.; Al-Shamma’a, A.I.; Shaw, A. An antenna for wireless sensor applications. In Proceedings of the 2009 Loughborough Antennas Propagation Conference, Leicestershire, UK, 2 February 2009. D’Elia, A.; Perilli, L.; Viola, F.; Roffia, L.; Antoniazzi, F.; Canegallo, R.; Cinotti, T.S. A self-powered WSAN for energy efficient heat distribution. In Proceedings of the 2016 IEEE Sensors Applications Symposium (SAS), Catania, Italy, 2016; pp. 1–6. Moon, J.-I.; Sim, D.-U.; Park, S.-O. Compact PIFA for 2.4/5 GHz dual ISM-band applications. Electron. Lett. 2004, 40, 844–846. [CrossRef] De Carli, L.G.; Juppa, Y.; Cardoso, A.J.; Galup-Montoro, C.; Schneider, M.C. Maximizing the power conversion efficiency of ultra-low-voltage CMOS multi-stage rectifiers. IEEE Trans. Circuits Syst. I Regul. Pap. 2015, 62, 967–975. [CrossRef] Essel, J.; Brenk, D.; Heidrich, J.; Weigel, R. A highly efficient UHF RFID frontend approach. In Proceedings of the 2009 IEEE MTT-S International Microwave Workshop on Wireless Sensing, Local Positioning, and RFID, Cavtat, Croatia, 24–25 September 2009.

Sensors 2017, 17, 1732

25. 26.

27. 28. 29.

21 of 21

Analog Devices. Ultralow Power Energy Harvester PMU with MPPT and Charge Management; ADP5091; Analog Devices: Norwood, MA, USA, 2016. Report ITU-R SM.2255-0(10/2012). Technical Characteristics, Standards, and Frequency Bands of Operation for Radio-Frequency Identification (RFID) and Potential Harmonization Opportunities; International Telecommunication Union: Geneva, Switzerland, 2015. Masotti, D.; Costanzo, A.; del Prete, M.; Rizzoli, V. Genetic-based design of a tetra-band high-efficiency radio-frequency energy harvesting system. IET Microw. Antennas Propag. 2013, 7, 1254–1263. [CrossRef] DASH7 Alliance Protocol 1.0: Low-Power, Mid-Range Sensor and Actuator Communication. Available online: http://www.dash7-alliance.org/ (accessed on 1 April 2017). Ergeerts, G.; Nikodem, M.; Subotic, D.; Surmacz, T.; Wojciechowski, B.; de Meulenaere, P.; Maarten, W. DASH7 alliance protocol in monitoring applications. In Proceedings of the 2015 10th International Conference on P2P, Parallel, Grid, Cloud and Internet Computing (3PGCIC), Krakow, Poland, 4–6 November 2015; pp. 623–628. © 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).