IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 61, NO. 1, JANUARY 2013 633
Extending Reading Range ofCommercial RFID Readers
Alírio J. Soares Boaventura, Student Member, IEEE, and Nuno Borges Carvalho, Senior Member, IEEE
Abstract—In this paper, multisine excitation signals are used toextend the reading range of commercial RF identification readers.To do so, a commercial reader is equipped with an external multi-sine front-end that implements previous mathematical proposals.A mathematical description is presented in order to show theability of multisine signals to communicate data, with minimalchanges in the downlink path, while no changes are required inthe conventional tag architecture. Moreover, and most important,if a proper multisine design is performed, a conventional readerreceiver is still able to demodulate and decode the backscatteredmultisine signal from the tag, without any hardware change. Thus,guidelines are presented for multisine design, including multisinenature, central tone positioning, tone separation, and bandwidthrequirements.In order to evaluate the reading range improvement, when com-
pared with the conventional single carrier approach with the sameaverage power, two experiments are conducted: in the first one,an oscilloscope is used to measure the tag response and to decidewhether the tag does or does not respond. In the second measure-ment scenario, the downlink path is implemented by the readercombined with the front-end and the uplink is implemented solelyby the reader. In this case, the decision on successful tag responseis taken when the reader reads the tag identification. The first mea-surement scenario has pointed out for a maximum reading rangeimprovement of near 43% for an eight-tone multisine signal with2-MHz tone separation. In the second scenario, a more realisticone, a reading range improvement of almost 25% has been ob-tained for a 8 1 tones multisine.
A CENTRAL issue in passive RF identification (RFID) sys-tems is the communication distance between reader and
tag. Unlike most of the radio communication systems, which aremainly limited by the satisfactory signal-to-noise ratio (SNR),in passive RFID, the reading range is primarily imposed bythe forward power link. Thus, the distance through which thereader is able to remotely power up the tag must be maximized
Manuscript received July 09, 2012; revised October 28, 2012; acceptedNovember 06, 2012. Date of publication December 11, 2012; date of cur-rent version January 17, 2013. This work was supported by the PortugueseFoundation for Science and Technology (FCT) under Doctoral ScholarshipSFRH/BD/80615/2011 and Project TACCS PTDC/EEA-TEL/099646/2008.This paper is an expanded paper from the IEEE MTT-S International Mi-crowave Symposium, Montreal, QC, Canada, June 17–22, 2012.The authors are with the Institute of Telecommunications and the Department
of Electronics, Telecommunications and Informatics, University of Aveiro,Aveiro 3819-193, Portugal (e-mail: [email protected]; [email protected]).Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.Digital Object Identifier 10.1109/TMTT.2012.2229288
without compromising the maximum regulated power. Equiva-lently, the transmitted power can be reduced while keeping thesame distance. Some approaches have been proposed in the lit-erature to improve the reading range of RFID readers. In [1], areconfigurable antenna architecture has been proposed, wherean antenna array with a steerable high gain beam is used toimprove the power transfer efficiency, and consequently, to in-crease the reading distance. In [2], the reading range is increasedby boosting the reader signal with an external auxiliary contin-uous wave (CW) transmitter.Recently, there has been a growing interest in waveform de-
sign for efficient wireless power transmission (WPT). For in-stance, the use of high peak-to-average power ratio (PAPR) sig-nals has been proposed in several works as a way for increasingthe efficiency of RF-dc converters and thus for increasing thesystem coverage range. One of the first application of this con-cept can be found in [3]. In this work, a high PAPR signal wasused to reduce the average power transmitted by an UHF RFIDreader, while keeping the same communication distance. Insteadof keeping the RF carrier always turned on during the entirebit period of a high-level ASK symbol, the RF carrier is inter-mittently switched on and off with a given duty cycle. There-fore, the transmitted average power is reduced according to theduty cycle. A power reduction of about three-quarters (75%)was achieved for a duty cycle of 40%. However, the main draw-back of this approach is related to the method itself. Switchingon and off the carrier with a square shape introduces new fre-quency components and degrades the spectral efficiency of thesystem.Following the same reasoning, ultra-wideband (UWB) sig-
nals were proposed for efficient low-power transmission [4],[5]. A train of short duration pulses with relatively high ampli-tude exhibits very low average power (and consequently, highPAPR). In [5], a conversion efficiency of 50% was achievedin a Schottky diode voltage doubler fed with a relatively lowpower level of 3 dBm. The obvious disadvantage of the UWBapproach is the increased spectral bandwidth and the need forwideband antennas and components. Furthermore, the UWBscheme cannot be directly applied to conventional RFID sys-tems, which are inherently narrowband.Multicarrier signals are well known from orthogonal fre-
quency division multiplexing (OFDM) systems because of theiroutstanding spectral efficiency and robustness against fadingand multipath effects. The benefits of these signals have beenalso explored for WPT and RFID systems. This was firstlydone in [6], where a multisine scheme was used to extend thereading range of UHF tags. In [7], a survey was presentedwhere several commercial UHF RFID tags were tested with
634 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 61, NO. 1, JANUARY 2013
multisine signals. In [8], the nonlinear behavior of RF-dc con-verters and the efficiency enhancement when using multisinesignals was investigated. Accordingly, a mathematical modeland description was provided to explain the nonlinear phenom-enon behind the improvement of RF-dc conversion efficiencyin Schottky diode detectors excited with multisine excitations.To do so, a Schottky diode detector was excited with a singlecarrier and several multisine signals. Multisine signals providedhigher RF-dc efficiency than single carrier. Basically, the highPAPR multisine signals are able to excite the nonlinearity ofdiode detectors (and charge pumps) in a more efficient way,forcing them to produce much more dc power [6], [8]. This canartificially improve the sensitivity of RFID tags and ultimatelyincrease the system communication range.In the experiments, conducted in [9], a multisine front-end
scheme was integrated with a commercial RFID reader in orderto extend its reading range. However, in that work, only thedownlink path was implemented and an oscilloscope was usedto measure the tag response. The present work is an extendedversion of [9]. In this work, both the downlink and the uplinkare fully implemented.Moreover, amultisine design is proposedto guarantee that the reader can successfully interrogate the tagand read its identification (ID) with minimal hardware changes:the multisine can be designed in such way that no additionalchanges are needed in the traditional receiver hardware archi-tecture to operate with multisine signals. This means that a com-mercial RFID reader will be able to demodulate and decode aproperly designed multisine signal reflected back from the tag.In this sense, only a few hardware changes will be required,namely, to generate the multisine signal. In order to evaluatethe reading range improvement, when compared with the con-ventional single carrier approach with the same average power,two experiments are conducted. In the first scenario, a commer-cial RFID reader is combined with the multisine front-end toimplement the downlink path. Concerning the uplink, an oscil-loscope is used to visualize the tag response in order to decidewhether the tag does or does not respond. In the secondmeasure-ment scenario, the downlink path is implemented as previouslydescribed and the uplink is implemented solely by the RFIDreader. In this case, the decision on successful tag response istaken when the reader successfully interrogate the tag and readits ID. The main novelties of this paper compared to previousworks can be summarized as follows.1) A mathematical model and description are provided fordownlink and uplink using multisine signals.
2) Themultisine design requirements are presented, includingtone separation, bandwidth, and central tone positioning.
3) The multisine scheme is integrated with a commercialRFID reader. According to 1) and 2), the reader is stillable to receive a backscattered multisine signal withoutthe deployment of new hardware in the receiver path.
II. RADIO LINK USING MULTISINE SIGNALS
The radio link of passive RFID systems exhibits very peculiarcharacteristics [10], [11]. Since a passive tag has no battery andit is entirely powered by the RF power radiated by the reader,
Fig. 1. Block diagram of: (a) RFID reader and external multisine front-endand (b) passive RFID tag. With proper multisine design, the tag response canbe directly forward to the reader receiver. The attenuator in the forward path isused to find the activation power level of the tag.
the downlink operation has a twofold goal: data communica-tion and power transfer. Moreover, since the passive tag has nolocal oscillator (LO), the uplink communication is implementedby power reflection/backscattering mechanism [11]. In this sec-tion, we analyze the data downlink and data uplink when usingmultisine excitations instead of a conventional single carrier.For the following mathematical description, consider Fig. 1(a)(basic configuration of the RFID reader andmultisine front-end)and Fig. 1(b) (block diagram of a typical passive tag).
A. Multisine Data Downlink
As proven before in [8], from the viewpoint of WPT, multi-sine signals are beneficial since they provide increased RF-dcefficiency in diode detectors. The same conclusion can be ap-plied for charge pump commonly used in passive RFID tags.This can potentially improve the communication range of RFIDsystems. However, in order to still guarantee data communica-tion the multisine must be properly designed. In this section, weanalyze the downlink data communication and we present somedesign rules to guarantee communication with multisine signals.In the following analysis, all link factors are ignored. Addition-ally, the tag’s diode envelope detector and charge pump [seeFig. 1(b)] will be modeled by an even-order polynomial series,as done previously in [8]. The externally generated baseband
BOAVENTURA AND BORGES CARVALHO: EXTENDING READING RANGE OF COMMERCIAL RFID READERS 635
multisine [see Fig. 1(a)] is a sum of subcarriers harmonicallyspaced by
(1)
where is the number of tones, is the frequency separationbetween tones and is the phase of each tone. In order to fa-cilitate the analysis, from now on, all signals will be representedby its Euler’s equivalent. Thus, the multisine modulated signalat the transmitter antenna [see Fig. 1(a)] is given by the mixtureof the baseband signal with an RF carrier at frequency ,followed by a mixture with a zero phase1 baseband multisine
(2)For instance, if a two-tone baseband multisine is used
, the bandpass modulated multisine (afterbeing shifted in frequency) will have spectral components at
,
(3)
The frequency spectrum of is depicted in Fig. 2. As canbe seen, the frequency separation between tones must behigher than the bandwidth of the data signal so that the in-formation can be successfully recovered at the tag side. Ideally,
must be much higher than . As an ex-ample, the EPC standard imposes a baseband data rateof 26.7 to 128 kb/s [14], which requires a bandwidth of53.4–256 kHz. In our experiments, the minimum tone separa-tion used is 500 kHz, which is higher than the maximum band-width allowed by the standard. The increased bandwidth is themain drawback of the multisine scheme. However, the max-imum bandwithd used in this work is compliant with indus-trial–scientific–medical (ISM) UHF regulations.In order to understand the multisine downlink opera-
tion, assume a unitary channel response (no attenuationor phase shift) from reader to tag. Additionally, considerthat the tag’s envelope detector behaves as a pure squarer,
(a second-order model is sufficient todescribe the basic envelope detection operation). Hence, thesquare of the signal (3) will provide dc/baseband componentscoming from the product of symmetric frequency compo-nents, ,
. Thus,
1Zero phase arrangement is proven to be the most efficient multisine arrange-ment [8].
Fig. 2. Spectrum of a modulated multisine signal. Tone separation in the mul-tisine must be higher than the bandwidth of the reader-to-tag baseband signal inorder to avoid spectrum overlapping. If overlapping occurs, the tag will not beable to recover the reader’s baseband information.
the baseband component at the tag’s detector output (afterlow-pass filtering) is given by
(4)
As can be seen by (4), the tag’s detector is able to recoverthe baseband information sent by the reader through the“multisine carrier” provided that the condition isverified. Precisely, if the baseband information is an ASK signal[14] with period , , simply composed by two levels( , ), then will also be between twolevels providing essentially the same information as . Therecovered information is proportional to the square of the base-band signal, thus, proportional to its power. More important, nochanges are needed in the architecture of a typical tag to de-modulate these new kind of signals. In the transmission path,changes are only required on the reader side, namely, to incor-porate the multisine signal. In this work, this is done by usingthe external front-end, as in Fig. 1(a); however, it could be ef-ficiently implemented by a software-defined radio (SDR) ap-proach simply by updating the RFID reader’s firmware.In Section II-B, it will be proven that if the multisine is prop-
erly designed, then no changes are needed in the conventionalarchitecture of the reader receiver. This design requirements im-poses a multisine with an odd number of tones, where the base-band multisine has a dc component and the central componentof the bandpass multisine coincides with the reader’s LO.
B. Multisine Data Uplink
As said before, the tag-to-reader data communication is madeby power reflection. This communication is performed by atwo-step operation: first, the reader illuminates the tag with anunmodulated carrier (a multisine in this case), and second, thetag modulates its antenna reflection coefficient according to theinformation to be sent to the reader. This information can be rep-resented by a time-varying reflection coefficient . Considernow that the tag is illuminated with an unmodulated three-tonemultisine signal where the amplitudes of the subcarriers are, re-spectively, A1, A0, and A2, and the phase relation between thesubcarriers is 0 . This multisine is designed in such way thatthe central tone with amplitude A0 is placed at the same fre-quency as the reader LO. Thus, unlike signal (3) that has anodd number of tones and has no subcarrier at the LO frequency,the bandpass version of this signal will have additional spectralcomponents at and , as can be seen in (5). Actually, this
636 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 61, NO. 1, JANUARY 2013
is the necessary condition to allow a conventional receiver to de-modulate the reflected signal from the tag without any hardwarechange. In order to better understand this operation mechanism,let us consider the backscatter modulator of the tag. In this case,the backscatter modulator will be operated by using a basebandsignal containing the tag information that will modulate themul-tisine subcarriers by acting in the reflecting coefficients createdby the backscattering transistor [see Fig. 1(b)]. Hence, the re-flected signal from the tag is given by
(5)
where , , , , and are the time-varyingreflection coefficients seen by each multisine frequency compo-nent. Assuming again no link losses, the signal reaches thereader receiver antenna and it is forward to the down-conversionmixer [see Fig. 1(a)] where it is mixed with the LO at frequency. The output of the down-conversion mixer is given by
(6)
The only components of falling on baseband frequen-cies result from the product of the components at and
. After passing the signal through a low-pass filter(present in traditional receiver architectures) with a cut-fre-quency of , a sample of the tag’s baseband information
—reflection coefficient of the central frequencycomponent—can be obtained
(7)
The result (7) shows that the received signal at the output ofthe low-pass filter is related to the amplitude of the LO andit is also related to the amplitude of central component of themultisine signal reflected from the tag . This secondcomponent contains the tag baseband information. It should alsobe stressed that no further changes are needed in the architectureof the receiver. Nevertheless, the use of a narrowband filter atthe input of the receiver, which accepts only the multisine cen-tral tone, might improve the system performance.
Fig. 3. (a) Even and (b) odd multisine schemes. Low-pass filter is representedby dashed line. In the first case (even multisine, the ) conventional reader isnot able to recover the tag baseband information, while in the second case (oddmultisine), the reader is allowed to access and process tag baseband information.Multisine is required to have an odd number of tones and the central tone mustcoincides with the reader’s LO.
The previous description is illustrated in Fig. 3: in thefirst case, a multisine with an even number of tones and nocomponent at the LO frequency is used. As can be seen inFig. 3(a), in this case, a conventional reader is not able torecover the tag baseband information. However, in the secondcase [see Fig. 3(b)], the tag is illuminated with a multisinewith an odd number of tones where the central tone coincideswith the reader’s LO. Thus, the tag will modulate all of themultisine subcarriers with its baseband information, and thus,a baseband sample will also be present at the central tone.This allows a conventional reader to demodulate and decodethe tag baseband information in the same manner it woulddo for a single carrier. In fact, the necessary condition thatenables conventional readers to demodulate/decode multisinesis the frequency positioning of the multisine central componentthat must coincides with the LO frequency. The additionalrequirement concerning the odd nature of the multisine (oddnumber of tones) is not a mandatory condition even though itis convenient to guarantee the symmetry of the signal spectrumand to simplify the multisine generation and hardware design.Moreover, the bandwidth of the low-pass filter of the readerreceiver has to be tight enough to reject all the lateral multi-sine components, leaving only the central one. While these
BOAVENTURA AND BORGES CARVALHO: EXTENDING READING RANGE OF COMMERCIAL RFID READERS 637
Fig. 4. Laboratory setup used to measure tag sensitivity gain.
considerations were not important in the previous downlinkanalysis (since the tag receiver uses envelope detection ratherthan coherent detection), they are crucial in the uplink case.
III. LABORATORY TEST BEDS AND MEASUREMENTS
In order to evaluate the reading range improvement, whencompared with the conventional single carrier approach with thesame average power, two experiments are conducted: in the firstone, an oscilloscope is used to measure the tag response and todecide whether the tag does or does not respond. In the secondmeasurement scenario, the downlink path is implemented bythe reader combined with the front-end and the uplink is imple-mented solely by the reader. In this case, the decision on suc-cessful tag response is taken when the reader reads the tag ID.
A. Measurement Setup 1
In this measurement scenario, a commercial RFID reader[12], [13], compliant with ISO/IEC 18000-6 and EPC GlobalClass-1 Greneration-2 protocols [14], will be equipped withan external front-end developed to incorporate the multisinewaveform in the original signal. Fig. 4 depicts the laboratorytest bed used in the measurements, which is similar to the oneused in [6]; however, using a commercial RFID reader.Reader’s transmitter output is mixed with a baseband mul-
tisine and then amplified and transmitted. Two power splittersare used, the first one to have a sample of the LO to be usedas an LO in the receiving path and a second one used to pro-vide a measurement of the radiated average power. The signalreflected from the tag is down-converted (mixed with ) so thatthe tag baseband response can be visualized in an oscilloscope.This will provide us with the information of whether the tag is ornot activated by identifying the tag’s response after each readercommand (Fig. 5). A power attenuator is used to control the ra-diated power. The receiving antenna and the tag (Class1 Gen2compliant) are placed at a fixed distance of the transmitterantenna. Fig. 6 depicts the experimental setup used.In order to evaluate the improvements of the multisine
scheme, a power gain figure of merit,(in decibels), is defined as the difference between the min-imum average power needed to activate the tag (and have aresponse) using a single carrier signal and the minimumaverage power required to activate the tag using a multisinesignal for the same distance . can be seen as
Fig. 5. Baseband signal visualized in the oscilloscope: RFID reader commandsfollowed by tag responses.
Fig. 6 RFID reader and multisine front-end setup. 1: RFID reader, 2: powersplitter, 3: variable attenuator, 4: mixer, 5: signal generator, 6: power amplifier,7: power splitter, 8: power meter, 9: cable to the transmitting antenna, 10: cablefrom the receiving antenna, 11: mixer, 12: oscilloscope, 13: spectrum analyzer.
a sensitivity improvement. To measure such gain, first thetag is illuminated with a single carrier and by varying theattenuation of the transmission path the minimum averagepower to activate the tag at a distance is found. Inthe second experiment, we repeat the previous procedure butnow using a multisine excitation. By using the measured gain
and the propagation law, we can estimate the expectedcommunication range gain (in meters). The minimum dcpower needed to activate the tag at a distance when usinga single carrier is , where
is the RF-dc conversion efficiency, is the transmittedpower, and are the transmitter and receiver antennagains, respectively, is the wavelength, and is the distancebetween the reader and tag antennas. On the other hand,when using a multisine with the same transmitted power ,a power gain and an extra reading range will be ob-tained. Thus, the Friis formula can be rewritten as follows:
. From the twoprevious Friis equations, an estimation for the reading rangegain can be obtained, as in [6] (in meters).
638 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 61, NO. 1, JANUARY 2013
TABLE IEXPECTED SENSITIVITY GAINS AND READING RANGE GAINS
OBTAINED IN THE FIRST MEASUREMENT SCENARIO
Table I shows the measurement results. Four different signalswere used in the measurements: a single carrier and three multi-sine signals with two, four, and eight tones. Each multisine wasalso tested with three different tone separations. The referencesignal (single carrier) required a minimum transmitted power
dBm for a reference fixed distance of m.For the fixed distance and based on the reference powerand on the minimum power for each multisine , we are ableto obtain the tag sensitivity gain . Finally, we can estimatethe expected reading range gain . The best result is obtainedfor the eight-tone multisine with an optimal tone separation of2 MHz pointing out for a communication distance improvementof 43%.
B. Measurement Setup 2
In this measurement setup, instead of evaluating the tag re-sponse using an oscilloscope, the tag response is directly for-warded to the reader receiver, as in Fig. 1(a), and the readerreads the tag ID. In this case, the downlink path is implementedby the reader combined with the front-end and the uplink is im-plemented solely by the reader. The decision on successful tagresponse is taken when the reader reads the tag ID. Contrary tothe complex receiver configuration (using a matched filter mul-tisine demodulator) that has been proposed in [7], we propose avery simple scheme that only implies a proper multisine designand does not require any additional hardware to be deployed.Nevertheless, to do so, the multisine requirements imposed inthe previous section must be fulfilled, namely, the multisine na-ture (odd number of tones), the central tone positioning (coin-ciding with the LO), and tone separation and bandwidth require-ments. The latter requirements should avoid spectrum overlap-ping and guarantee compliance with the internal low-pass filterof the reader.Again, the multisine tone spacing plays an important role in
this scenario since the tone spacing should allow the RF-dc fil-tering stage to power up the tag as previously discussed, butalso to avoid aliasing between the modulation bandwidth ofeach modulated subcarrier, as seen in Fig. 3. Moreover as saidbefore, the bandpass multisine signal must have a componentat the LO frequency, and consequently, the baseband multisinesignal must have a dc component. Since the baseband multi-sine is being provided by a signal generator (that employ a dcblock at the output), it has no dc component. For this reason,
Fig. 7. Transmitter configurations. (a) Setup used in the experiments, wherea dc component is added to the baseband multisine in order to create a cen-tral component in the bandpass multisine. (b) Alternative setup configuration,where a sample of the reader signal is directly transmitted to provide a centralcomponent in the multisine.
experimental setup [see Fig. 7(a)] employs a function generatorand a bias-tee to add a dc component in the baseband multi-sine. Hence, the bandpass multisine will have a frequency com-ponent at the LO frequency and it will fulfill the requirementsrelated to the odd number of tones and central frequency posi-tioning. The dc component is varied in order to level the ampli-tude of the central tone and to equal it to the other subcarriers.Fig. 7(b) presents an alternative configuration of the transmitter,which uses a sample of the reader signal directly forwarded tothe power amplifier (PA) in order to provide the multisine withcentral component. In this case, there is no need for a dc compo-nent in the baseband multisine. In this alternative setup, an at-tenuator can be used to level the central tone. In order tomeasurethe signal spectrum at the output of the PA [see Fig. 7(a)], theASK-modulated RFID signal was generated by a vector signalgenerator (VSG). This was done because the RFID reader uti-lized in the experiments employs a frequency hopping protocol[slow frequency hopping (SFH)] that makes the trigger difficult.Depicted in Fig. 8(a) is the frequency spectrum of an 8 1 tonesmultisine. Fig. 8(b) shows the expected shape of the normalizedtime domain waveform.In Table II are the measurement results obtained in the second
setup, where the RFID reader is actually reading the tag ID.The annotation 1 tones stands for a multisine with an oddnumber of tones with lateral subcarriers plus a central sub-carrier. In this case, the distance between reader and tag was
BOAVENTURA AND BORGES CARVALHO: EXTENDING READING RANGE OF COMMERCIAL RFID READERS 639
Fig. 8. Modulated 8 1 tones multisine signal. (a) Frequency spectrum: the866.6-MHz RF carrier is modulated by an arbitrary ASK signal at 100 kb/s andthe resultant signal is then mixed with the baseband multisine. In this example,the baseband multisine is a four-tone (centered at 5 MHz, with 2-MHz toneseparation) plus a dc component externally imposed by a function generator. (b)Normalized time-domain waveform, only five periods of the multisine envelopeare depicted.
TABLE IISENSITIVITY GAINS AND READING RANGE GAINS
m and the minimum power to activate the tag withthe single carrier signal was 16.5 dBm. Although the gains arelower than the first case setup (the reason for this is detailed inSection IV), an improvement of almost 25% has been achievedwith an 8 1 tones multisine with 2-MHz tone separation.
IV. DISCUSSION
Comparing the results of Tables I and II, it can be observedthat the second experiment (Table II), which is more realistic,provides less expressive results. However, it should be pointedout that in the second case, the objective is to guarantee that
Fig. 9. Tone separation effect illustrated in the time domain. (a) Higher averagedc voltage obtained with a lower envelope period (so higher tone separation) andb) lower average dc voltage for higher envelope period. The pulsed signal is theenvelope of the high PAPR signal after being rectified, maximum ripple valuedepends both on tone separation and on the output filter of the RF-dc converter.
the tag information is successfully read (demodulated and de-coded) by the reader, contrary to the previous case where weare only interested in observing the tag response in the oscillo-scope without any consideration on the quality of the response.For instance, in some situations the tag only returns a partialresponse and the reader is not able to properly decode the infor-mation (please refer to [15, Fig. 4]). This can somehow explainwhy the results in the second setup are less expressive than theresults in the previous one.In both tables, the gains are degraded with the decreasing of
the tone separation and it can also be observed that in some casesthe tag does not respond. This is due to the ripple effect on thetag’s RF-dc converter imposed by the low-frequency separationbetween tones in the multisine. This effect has been addressedin [6]. Fig. 9 illustrates how the tone separation imposes dif-ferent time-domain behavior at the RF-dc converter output. Ac-tually, as was seen in [8], the use of multisines with synchronousphases will impose the existence of high PAPR, and thus of highpeaks. These time-domain peaks will be temporally separatedby , and thus as the tone separation is reduced, thetime-domain peaks will separate from each other. This separa-tion will force the output dynamics of the charge pump, mainlythe filtering stage, to start decreasing its output voltage, and thusgo below the tag activation level. This can be observed in the lastcolumns of Tables I and II, where the tag does not respond fora very low tone separation. Thus, there exist a minimum toneseparation value that is strongly related with the filtering stageof the tag RF-dc converter. This minimum value can be opti-mized (lowered) by reducing the ripple effect through the useof a larger capacitance at the output of the RF-dc converter. Itshould be noticed that this ripple problem does not exist in theconventional approach (single carrier) because in that case thesignal envelope is constant. However, in the multisine scheme,the ripple is imposed by the tone separation (and by the envelopeof the multisine signal) rather than the RF signal. For instance, atone separation of 2 MHz imposes a peak time separation equalto 0.5 s [see Fig. 8(b)].As can be concluded from the experimental results, the gains
are directly influenced by the combination of the multisine pa-rameters such as number of tones, bandwidth, and tone spacing:in principle, as the number of tones increases, the PAPR alsoincreases which leads the efficiency of the charge pump to in-crease, and consequently, the coverage range increases (as seen
640 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 61, NO. 1, JANUARY 2013
by the tables). However, at a certain point, the increase of thenumber of tones would provide no extra gain. This is because,for the same tone spacing, as the number of tones increases, themultisine bandwidth also increases and this has a negative im-pact on the efficiency because of the limited input bandwidth ofthe tags. An optimal frequency separation value exists, whichis limited by two effects: first of all, for the same number oftones, the higher the frequency separation, the higher the band-width and some components will be attenuated due to the lim-ited bandwidth of the tag. On the other hand, the frequencyseparation imposes the separation between the signal peaks inthe time domain. Thus, the lower the frequency separation, thehigher the ripple effect, which degrades the performance, as pre-viously explained and as can be observed in Fig. 9.Finally, further improvements can be done in the system,
namely, the introduction of a bandpass filter at the PA outputto avoid out of band emissions. The use of a narrowband filterat the input of the low-noise amplifier can prevent the trans-mitter-to-receiver leakage of the multisine lateral components,as well as can improve the gains. This can be done because in theuplink the multisine lateral components are unimportant. Theshape of the time-domain waveform can also be optimized (asdone in [8]) in order to improve the gains. In order to efficientlyamplify the high PAPR signal, an improved PA/transmitter ar-chitecture must be employed. This will avoid signal clippingand will guarantee that the high PAPR signal reaches the tag. Itshould be noticed that in this work a linear amplifier with a highcompression point has been used for this purpose.
V. CONCLUSION
This paper has presented an RF front-end that can be appliedto commercial RFID readers in order to increase the reading dis-tance using multisine signals. In this work, both the forward link(reader-to-tag) and the reverse link (tag-to-reader) have beenimplemented. Guidelines are provided to guarantee that no addi-tional changes are needed in conventional receiver architectureof commercial RFID readers in order to receive and demodulatethe multisine signal backscattered from the tag.
REFERENCES[1] M. Abbak and I. Tekin, “RFID coverage extension using microstrip
[2] J.-S. Park, J.-W. Jung, S.-Y. Ahn, H.-H. Roh, H.-R. Oh, Y.-R. Seong,Y.-D. Lee, and K. Choi, “Extending the interrogation range of a pas-sive UHF RFID system with an external continuous wave transmitter,”IEEE Trans. Instrum. Meas., vol. 59, no. 8, pp. 2191–2197, Aug. 2010.
[3] H. Matsumoto and K. Takei, “An experimental study of passive UHFRFID systemwith longer communication range,” in Proc. Asia–PacificMicrow. Conf., 2007, pp. 1–4.
[4] C.-C. Lo, Y.-L. Yang, C.-L. Tsai, C.-S. Lee, and C.-L. Yang, “Novelwireless impulsive power transmission to enhace the convertion effi-ciency for low input power,” presented at the Microw. Workshop Se-ries on Innovative Wireless Power Transmission, 2011.
[5] Y.-L. Yang, C.-L. Yang, C.-L. Tsai, and C.-S. Lee, “Efficiency im-provement of the impulsive wireless power transmission,” presented atthe Microwave Workshop Series on Innovative Wireless Power Trans-mission, 2011.
[6] M. S. Trotter, J. D. Griffin, and G. D. Durgin, “Power-optimized wave-forms for improving the range and reliability of RFID systems,” inIEEE Int. RFID Conf., 2009, pp. 80–87.
[7] M. S. Trotter and G. D. Durgin, “Survey of range improvement of com-mercial RFID tags with power optimized waveforms,” in IEEE RFID,2010, pp. 195–202.
[8] A. J. S. Boaventura and N. B. Carvalho, “Maximizing DC power inenergy harvesting circuits using multisine excitation,” in IEEE MTT-SInt. Microw. Symp. Dig., Baltimore, MD, Jun. 2011.
[9] A. S. Boaventura and N. B. Carvalho, “Enhanced front-end to extendreading range of commercial RFID readers using efficient multisinesignals,” in IEEE MTT-S Int. Microw. Symp. Dig., Montréal, QC,Canada, Jun. 2012.
[10] Finkenzeller and Klaus, RFID Handbook, 2nd ed. New York: Wiley,2003.
[11] D. M. Dobkin, The RF in RFID: Passive UHF in Practice.Burlington, MA: Newnes, 2008.
[12] “Hardware Setup Guide ALR-8800,” Alien Technol., Morgan Hill,CA, 2007.
[13] “Reader Interface Guide,” Alien Technol., Morgan Hill, CA, 2007.[14] EPC Class-1 Generation-2 UHF RFID, Protocol for Communications
at 860 MHz–960 MHz, ver. 1.2.0, 2008.[15] X. Xu, L. Guy, J. Wang, and G. Xing, “Negotiate power and perfor-
mance in the reality of RFID systems,” in IEEE Int. Pervasive Comput.Commun. Conf., Mannheim, Germany, Mar. 2010, pp. 88–97.
Alírio J. Soares Boaventura (S’11) was born inSanto Antão, Cape Verde, in 1985. He received theMaster degree in electronics and telecommunicationengineering from the University of Aveiro, Aveiro,Portugal, in 2009, and is currently working towardthe Ph.D. degree at the University of Aveiro.From 2008 to 2010, he was with Acronym-IT,
a Portuguese manufacturing company devoted toRFID, WSN, and consumer electronics. In 2010, hejoined the Institute of Telecommunications, Aveiro,Portugal, as a Researcher. He has been a Reviewer
for the International Journal of Emerging and Selected Topics in Circuitsand Systems (JETCAS). His main research interests include passive RFIDand sensors, low-power wireless systems, WPT and energy harvesting, andcomputer-aided design (CAD)/modeling for RFID.Mr. Boaventura has served as a reviewer for the IEEE TRANSACTIONS ON
MICROWAVE THEORY AND TECHNIQUES. He was the recipient of a Merit Schol-arship for graduation from the Gulbenkian Foundation (2004–2009). In 2011, hewas a finalist of the Student Paper Competition of the IEEE Microwave Theoryand Techniques Society (IEEE MTT-S) International Microwave Symposium(IMS). He was the recipient of the 2011 URSI/ANACOM Prize, awarded bythe URSI-Portugal Section and the Portuguese National Authority of Commu-nications (ANACOM) for the best 2011 work in radio communications.
Nuno Borges Carvalho (S’97–M’00–SM’05) wasborn in Luanda, Angola, in 1972. He received theDiploma and Doctoral degrees in electronics andtelecommunications engineering from the Universityof Aveiro, Aveiro, Portugal, in 1995 and 2000,respectively.He is currently an Associate Professor with
“Agregação” and a Senior Research Scientist withthe Institute of Telecommunications, Universityof Aveiro. He coauthored Intermodulation in Mi-crowave and Wireless Circuits (Artech House,
2003). He has been a reviewer and author of over 100 papers in magazinesand conferences. He is the coinventor of four patents. His main research inter-ests include software-defined radio front-ends, wireless power transmission,nonlinear distortion analysis in microwave/wireless circuits and systems,and measurement of nonlinear phenomena. He has recently been involvedin the design of dedicated radios and systems for newly emerging wirelesstechnologies.Dr. Borges Carvalho is the chair of the IEEE MTT-11 Technical Committee.
He is the chair of the URSI-Portugal Metrology Group. He was the recipient ofthe 1995University of Aveiro and the Portuguese Engineering Association Prizefor the best 1995 student at the University of Aveiro, the 1998 Student PaperCompetition (Third Place) of the IEEE Microwave Theory and Techniques So-ciety (IEEEMTT-S) International Microwave Symposium (IMS), and the 2000IEE Measurement Prize.
