Bandwidth limitations of ultra high frequency radio frequency identification tags

Transcription

1 Published in IET Microwaves, Antennas & Propagation Received on 0th November 01 Revised on 3rd May 013 Accepted on 19th May 013 ISSN Bandwidth limitations of ultra high frequency radio frequency identification tags Gerard Zamora 1, Ferran Paredes 1, Francisco Javier Herraiz-Martínez, Ferran Martín 1, Jordi Bonache 1 1 GEMMA/CIMITEC, Departament d Enginyeria Electrònica, Universitat Autònoma de Barcelona, Bellaterra (Barcelona), Spain Department of Signal Theory and Communications, Carlos III University in Madrid, Leganés (Madrid), 8911, Spain gerard.zamora@uab.cat Abstract: Fundamental bandwidth limitations on the design of ultra high frequency (UHF) radio frequency identification (RFID) tags are investigated. This study was conducted by considering the optimum equivalent-circuit network necessary for bandwidth broadening in single resonant UHF RFID tags with conjugate matching. This equivalent-circuit network is simply a parallel combination of an inductor and a resistor cascaded to the UHF RFID chip. Bandwidth optimisation of the resulting network is validated by means of the Bode criterion. According to this analysis, a broadband UHF RFID tag prototype able to operate worldwide is designed and fabricated. In order to reduce tag size and cost, the external matching network is omitted in the reported implementation. The measured read ranges are above 5 m within the whole UHF RFID frequency band ( MHz), with a peak value of 9 m at 870 MHz. 1 Introduction Radio frequency identification (RFID) is a rapidly developing technology that provides wireless identification and tracking capability [1]. Although the first paper on the basic principles of passive RFID technology was published in 1948 [], it took a long time before the technology advanced to the current level [3]. Nowadays, many applications, such as electronic toll collection, asset identification, retail item management, access control, animal tracking and vehicle security, among others, require RFID systems [4]. A passive back-scattered RFID system consists of a reader and a tag, which includes an antenna and an application specific integrated circuit (ASIC) chip. The reader transmits a modulated signal with periods of un-modulated carrier, which is received by the tag antenna. The chip converts the un-modulated signal to DC (for external feeding) and sends back its information by varying its complex RF input impedance. This impedance typically toggles between two different states, corresponding to conjugate matching and to some other impedance, effectively modulating the back-scattered signal [5]. The UHF RFID regulated bands vary in the different world regions, and include frequencies between 840 and 960 MHz. Thus, UHF RFID is operated at MHz in China, at MHz in Europe, at MHz in USA and at MHz in Japan. Appropriate impedance matching between the antenna and the chip is of substantial importance in RFID [5]. It directly influences RFID system performance, such as the read range of the tag, that is, the maximum distance at which a reader can read information from the tag. Since the impedance of RFID chips is complex and depends on the manufacturer, specific RFID tag antennas must be designed for the considered chips available on the market. RFID tags are typically designed to achieve conjugate matching between the chip and the antenna (this maximises the read range). Several techniques for achieving complex impedance matching by means of matching networks can be found in the literature [6 8]. In [8], matching at two of the UHF RFID regulated bands (Europe and USA) was considered, and it was demonstrated that optimum matching can be achieved by using dual-band tags, rather than broadband ones. Nevertheless these matching networks introduce losses, which may degrade the tag performance. However, apart from the losses introduced by the matching network, cost and size issues sometimes force to omit external matching networks and direct matching between the antenna and the chip is a due. The qualitative behaviour of antenna impedance, chip impedance and read range as a function of frequency for a typical UHF RFID tag is illustrated in Fig. 1 [5]. It can be seen that chip impedance is always capacitive and has a small resistance [9, 10]. Obviously, this complex impedance is frequency dependent, and adds difficulty in obtaining broadband matching with simple matching networks or without matching networks. For this reason, the design of worldwide UHF RFID tags without external matching networks represents an important challenge in RFID technology. In this paper, we investigate the fundamental bandwidth limitations on the design of single resonant UHF RFID 788 IET Microw. Antennas Propag., 013, Vol. 7, Iss. 10, pp & The Institution of Engineering and Technology 013

2 Fig. 1 Antenna impedance, chip impedance and read range as functions of frequency for a typical RFID tag Fig. extracted from [5] tags with complex conjugate matching at roughly the intermediate frequency of the UHF RFID worldwide band (915 MHz). According to this analysis, we also propose a broadband tag exhibiting conjugate matching at this frequency, without using an external matching network. The presented tag operates worldwide, with a significant read range ( >5 m) within the whole UHF RFID band ( MHz), and a peak value of 9 m. Physical bandwidth limitations and optimisation.1 Bode criterion The input impedance of a UHF RFID chip is mainly determined by the voltage multiplier stage of the integrated circuit transponder, which can be modelled by a parallel combination of a resistance R C (that accounts for the losses of the multiplier) and a capacitance C C (that includes all the capacitive effects) [9, 10], as it is shown in Fig.. According to Bode s limit [11, 1], the reflection coefficient achievable by any passive and lossless network placed between a purely resistive generator and an RC parallel load is limited by 1 0 ( ln 1 ) dv p (1) s R C C C Fig. Schematic of a typical UHF RFID tag in the most general case An impedance matching network is located in between the chip (modelled as a RC parallel load) and the tag antenna where s denotes the reflection coefficient, ω the angular frequency and R C and C C are the equivalent resistance and capacitance, respectively, of the chip. Since ln(1/ s ) can be expressed in terms of the return loss (in db) at the input of the matching network, the expression (1) yields 1 0 s db df 10 log e R C C C () Expression () forces the area given by the return loss curve to be less than, or equal, to a constant. Bandwidth optimisation implies assuming s to be a constant over the band of interest, and ṡ =1 (RL = 0 db) elsewhere. This is the response with maximum bandwidth, but it cannot be realised in practice because it would require an infinite number of elements in the matching network [13]. The necessity of multiple stages to approximate the optimum response makes this objective prohibitive because of the increment of cost and tag dimensions. In fact, UHF RFID tags are typically designed with a single resonant frequency in their frequency response. Therefore in this work, the main objective is bandwidth optimisation by considering a single resonant UHF RFID tag with complex conjugate matching at the intermediate frequency.. Choice of passive circuit network for conjugate matching and bandwidth optimisation The purpose of this subsection is to obtain the passive circuit network which provides complex conjugate matching at the intermediate frequency and the broader possible bandwidth when it is cascaded to a given UHF RFID chip. Let us consider the schematic depicted in Fig, where the load to be matched consists of a parallel combination of a capacitance C C and a resistance R C. An eventual impedance matching network is inserted between the antenna and the load, to consider the most general case. Maximum power transfer between the antenna and the chip is achieved for conjugate matching, that is, Z chip = ZN (or Y chip = YN), since this forces the power reflection coefficient, given by IET Microw. Antennas Propag., 013, Vol. 7, Iss. 10, pp & The Institution of Engineering and Technology 013

3 the authors of [14, 15] s = Z chip ZN = Y ( ) N Y +B chip G N G chip = Z chip +Z N Y N +Y chip ( ) +B G N +G chip to be zero. In (3), Y N = G N +jb N, Y chip = G chip +jb chip and B = B N + B chip. Since the chip impedance depends on the input power applied to the chip [5], for the evaluation of the maximum read range, Z chip must be considered at the threshold power. Obviously, conjugate matching cannot be fulfilled in a broad band. Therefore we choose the intermediate frequency of the UHF RFID frequency span, f 0, to force conjugate matching. This means that the conductances and susceptances must satisfy G N = G chip and B N = B chip, respectively, at f 0. In order to minimise the power reflection coefficient in the vicinity of f 0, B must be minimised and G N must be constant and equal to G chip. This result is inferred from the second-order Taylor expansion of the power reflection coefficient in the vicinity of f 0 ( ) +B ( ) G D sg N, B N G chip = 4G chip Then, the frequency derivative of the susceptance B db dv = db N dv + db chip dv must be as small as possible. Significant for our discussion, UHF RFID chips exhibit high-impedance phase-angle and small resistance within the UHF RFID frequency range [6, 9, 10]. Thus, it can be considered a low-loss one port network in which the energy loss per period is a small fraction of the stored energy. This can be expressed in terms of the quality factor of the chip equivalent-circuit model as Q 1 [16, 17], which is well satisfied by the typical values of the RFID integrated circuits available on the market today [18 0]. It follows that a low-loss network (with impedance Z N ) cascaded to the chip, such that Z N = Z chip at the operating frequency f 0, should be considered to minimise the power reflection coefficient around the tag resonance frequency. It was proven in [16] that the Foster s reactance theorem [1 3] is also valid for low-loss networks at frequencies far from the network resonance frequencies. From the Foster s theorem, both terms in the right-hand side of (5) are positive; therefore to minimise (5), it is necessary to reduce the slope of B N,atf 0 and in the vicinity, as much as possible (note that the second term is determined by the chip capacitance). In the case of a Foster network, it was demonstrated in [4] that the minimum frequency derivative of any negative susceptance corresponds to the value introduced by a simple inductor. From the previous analysis, it follows the relevant conclusion that the maximum bandwidth achievable by a typical (single resonant) RFID tag, exhibiting conjugate matching at the central frequency of the UHF band, is obtained by cascading a parallel combination of a resistor (which provides a pure conductance G N such that G N = G chip ) and an inductor, with susceptance B N = B chip at f 0, to the chip. This should be (ideally) the network describing (3) (4) (5) the antenna or the antenna plus the matching network (see Fig. ) for maximum bandwidth. It is important to point out that a series combination of an inductor and a resistor cascaded to the chip, providing conjugate matching at f 0, can be considered to achieve a bandwidth very close to the optimum. This is a direct consequence of the fact that a low-loss network is required for bandwidth optimisation. In order to demonstrate it, let us consider a low-loss network formed by a series combination of a resistor with resistance R s and an inductor with reactance χ s (Q 1, therefore x s R s ). The conductance G N and susceptance B N of such a network can be approximated by R s G N = R s + x s B N = x s R s + x s R s x s (6) 1 x s (7) Note that B N in (7) is approximated by the minimum achievable susceptance in a Foster network, obtained in the previous analysis. However, the conductance G N in (6) is approximated by a frequency dependent function, rather than a constant value which is required for achieving maximum bandwidth, as mentioned before. A first-order Taylor expansion of G N in the vicinity of f 0 ( G N = G chip 3 f ) (8) f 0 shows that the maximum variation of G N from the optimum value G chip, that is evaluating (8) at the UHF RFID frequency limits, leads to G N = 0.86G chip at the maximum frequency (960 MHz) and G N = 1.13G chip at the minimum frequency (840 MHz). From (4) it follows that a variation < in comparison with the optimum power reflection coefficient is expected at any frequency in the vicinity of f 0..3 Limitation of the 3 db bandwidth By considering the required network for maximum bandwidth obtained in the previous section, the optimum 3dB bandwidth (inferred by forcing (3) to be 1/) is found to be Df 3dB 1 pr C C C (9) and db/dω =C C at f 0. For a real network, deviations from the ideal network (a parallel LR load) providing the optimum bandwidth are expected, and the derivative of the susceptance B with frequency at f 0 can be expressed as db/dω =(k +1)C C, where k 1 is a factor that indicates how much the susceptance seen by the chip deviates from the purely inductive behaviour (k =1). By using the susceptance slope concept for real resonators [13, 5], expression (9) can be generalised as Df 3dB = = max Df 3dB (10) p( k + 1)R C C C ( k + 1) Expression (10) provides a useful formula for RFID designers that allows one to predict the degradation of the 3 db bandwidth, as compared with the maximum achievable 790 IET Microw. Antennas Propag., 013, Vol. 7, Iss. 10, pp & The Institution of Engineering and Technology 013

4 bandwidth (given by (9)), that is, the bandwidth that results by using the optimum network (a parallel LR network). Note that k can be easily determined from the antenna input impedance at f 0. Indeed, k is given by k = 1 db N (v) C C dv (11) v0 with ω 0 =πf 0. 3 Design of a broadband UHF RFID tag 3.1 Equivalent-circuit model for tag bandwidth optimisation The design of the complete system for tag bandwidth maximisation is carried out in this subsection. Since the design of a global band tag is pursued, the central frequency of the operating band was chosen to be the same as the intermediate frequency of the UHF RFID band ( f 0 = 915 MHz). Let us now consider a typical integrated circuit for the RFID tag (the NXP UCODE GXM chip). The impedance reported by the manufacturer of this integrated circuit at the intermediate frequency is Z IC ( f 0 )=16 j148 Ω. As indicated in Section, the chip can be modelled by a parallel combination of a resistance R C and a capacitance C C. These values were calculated from the input impedance of the chip transformed to its equivalent RC parallel circuit, giving R C = 1385 Ω and C C = 1.16 pf. Since the required resonance frequency is f 0 =(LC C ) 1/ /π, the inductance value corresponding to the susceptance B N was found to be L = 6 nh. The whole system is depicted in Fig. 3 and the simulated power reflection coefficient was obtained by using the Agilent ADS circuit simulator. Evaluation of () shows that the maximum area achievable between the return loss curve and the s = 1 axis for the chip considered in this work is 700 MHz db. For comparison, a numerical simulation, based on the trapezoidal rule, in order to obtain the area for the reflection coefficient of the circuit depicted in Fig. 3, was carried out by using the Matlab commercial software. As it was expected from the analysis of Section, the value of that area is exactly the same than the upper limit determined by the Bode criterion. Thus, it has been demonstrated that the circuit of Fig. 3 gives the upper limit for the operating bandwidth by achieving conjugate matching at the intermediate frequency. Obviously, the bandwidth evaluated when the power reflection coefficient equals to 3 db reaches the upper limit calculated from (9), which corresponds to 198 MHz. On the other hand, as indicated in Section., a series combination of an inductor and a resistor cascaded to the Fig. 3 Equivalent-circuit model of an UHF RFID tag for bandwidth maximisation and complex conjugate matching at the UHF RFID intermediate frequency (915 MHz) For the RFID chip considered in this work, the value of the circuit parameters are R C = 1385 Ω, C C = 1.16 pf and L =6nH Fig. 4 Power reflection coefficient obtained from (4) in the case of considering a parallel and a series RL circuit cascaded to the NXP UCODE GXM chip chip, providing conjugate matching at f 0, can be used to obtain roughly the optimum bandwidth. This can be observed in Fig. 4, where a plot of the power reflection coefficient in the vicinity of f 0 obtained from (4) is depicted, considering the cases of a parallel RL circuit (using G N = G chip ) and a series RL circuit [using G N given by (8)] cascaded to the chip. Note that the linear difference between both coefficients is in the order of RFID tag antenna design Considering the previous requirements to achieve optimum bandwidth and conjugate matching at the operating frequency, a broadband RFID tag is designed in this subsection. In order to avoid the use of an external matching network, the tag antenna should have a lumped-element equivalent-circuit model based on the parallel RL circuit of Fig. 3, at the frequency range of interest. As it is well known, to obtain an inductive behaviour by means of a planar semi-lumped element, physical dimensions must be much smaller than the wavelength, thus minimising distributed effects. According to Chu s work [6] there exists a tradeoff between size miniaturisation of antennas and maximum achievable gain. It means that reducing the electrical size of the tag antenna (in order to accurately describe it by a parallel RL circuit) involves unavoidable gain degradation, which directly influences the tag performance. Taking it into account, the following design procedure was used to obtain the layout of the proposed tag antenna. A narrow open rectangular loop with 1 mm strip width, connected to the chip at the aperture (located at the centre of one of the longer sides), was considered as a starting point. Such a loop, with an input impedance essentially inductive, cascaded to the chip provides an RLC shunt resonance. By varying the length (longer side) of the rectangular loop, the resonance frequency can be adjusted to the desired frequency, f 0. Finally, the required antenna resistance is achieved by tailoring the length and width of the loop, in the opposite region to chip location (without modifying the width of the arms connected to the chip). By this means, a tag antenna directly matched to the chip, based on a folded dipole, has been designed on a commercial low loss microwave substrate, the Rogers RO3010 substrate with dielectric constant ε r = 10. and thickness h = 1.7 mm. The proposed IET Microw. Antennas Propag., 013, Vol. 7, Iss. 10, pp & The Institution of Engineering and Technology 013

5 MHz in the lossless case and 14.5 MHz in the lossy case. These results were corroborated by obtaining the 3 db bandwidths from simulation, giving 16 and 144 MHz (73% of the upper limit) for the lossless and lossy case, respectively. The simulated gain reaches the value of 1 db at the operating frequency and the radiation pattern is similar to that of a conventional λ/ dipole. Fig. 5 Layout of the proposed RFID tag antenna for bandwidth optimisation and conjugate matching a Dimensions are l = 90 mm, w = 7 mm, g = 4 mm and d = 0.5 mm b Current distribution of the proposed tag antenna at the UHF RFID intermediate frequency (915 MHz) tag antenna, depicted in Fig. 5a, is an asymmetric, coplanar strip, folded dipole. The dimensions of the tag antenna are roughly 0.7λ 0.0λ, λ being the wavelength at 915 MHz. These small dimensions force the direction of the antenna current density to be different at both arms, as in the case of a conventional folded dipole working at the so-called transmission line mode [7 30]. Although a conventional folded dipole is usually designed for working at the supported mode (antenna mode), where the current density direction is the same at both arms and consequently the radiation efficiency is typically higher, an asymmetrical design of the folded dipole can improve substantially the radiation efficiency achieved at the transmission line mode. Hence, the asymmetrical folded dipole radiates as a result of its unbalanced condition. The power reflection coefficient of the presented RFID tag is depicted in Fig. 6a, where complex conjugate matching is achieved at the frequency of interest. Fig. 6b shows the input admittance of the tag antenna in comparison with the input admittance of the RL circuit of Fig. 3. Both results were obtained by using the Agilent Momentum commercial software. It can be seen that, at the UHF RFID frequency range, the input admittance of the tag antenna can be approximated by the RL circuit required for bandwidth optimisation and conjugate matching at f 0. The area for the reflection coefficient of Fig. 6a has been calculated by using the trapezoidal approach used in Section 3(a). The value of this area for the lossy and lossless case was found to be more than 80% of the upper limit determined by the Bode criterion. The 3 db bandwidth can be obtained through (10), where, inferred from (11), k = 1.44 for the lossless case and k = 1.78 for the lossy case, corresponding to 8 and 71%, respectively, of the upper limit determined by (9). Then, the resulting theoretical bandwidths are Read range The most important tag parameter indicative of its performance is the read range. Since the reader sensitivity is typically high in comparison with that of the tag, the read range is defined by the tag response threshold. The read range is also sensitive to tag orientation, the material where the tag is attached and environment [5]. The read range r can be calculated using the Friis free-space formula (valid only in the far-field region) as r = l EIRP G r t 4p P chip (1) where λ is the wavelength and EIRP is the equivalent isotropically radiated power, determined by local country regulations (EIRP = 3.3 W in Europe and EIRP = 4 W in USA). P chip is the minimum threshold power necessary to activate the RFID chip, G r is the gain of the receiving tag antenna, and t is the power transmission coefficient, which is related to the power reflection coefficient s by t =(1 s ). The tag gain and the power transmission coefficient are inferred from simulation, considering the chip impedance as the port impedance. It is important to emphasise that (1) gives the maximum distance in free space at which the reader can detect the backscattered signal from a given tag, that is, both reader antenna and tag antenna are oriented towards each other in the direction of maximum gain and their polarisations are matched. 4 Fabrication and measurement 4.1 Read range measurement The RFID setup available in our laboratory has an Agilent N518A vector signal generator, which creates RFID frames and plays the role of a reader with variable frequency and variable output power. Such a generator is connected to a TEM cell by means of a circulator. The tag Fig. 6 Power reflection coefficient for the a Lossless (dashed line) and lossy case (bold line) of the proposed RFID tag b Input admittance of the proposed tag antenna compared with the input admittance of the RL circuit of Fig IET Microw. Antennas Propag., 013, Vol. 7, Iss. 10, pp & The Institution of Engineering and Technology 013

6 under test is located inside the TEM cell and it is excited by the frame created by the generator. Then the tag sends a backscatter signal to an Agilent N900A signal analyser through the circulator. At each frequency, the minimum power at the input of the TEM cell required to communicate with the tag is recorded. Finally, an electric probe is placed into the TEM cell, in order to determine the root mean square of the incident electric field, E rms, corresponding to the minimum power at each frequency. This electric field is related to the power delivered to the chip according to P chip = SA ef t = E rms h l G r 4p t (13) where S is the incident power density, A ef is the effective area of the tag antenna and η is the wave impedance of free space. The measured read range can be inferred by introducing (13) into (1), resulting in the following expression 30EIRP r = (14) E rms 4. Implementation and experimental results First, it is important to highlight that the minimum power level necessary to activate the chip used in this work reported by the manufacturer is P chip = 15 dbm. However, a measure of the sensitivity of this chip was carried out in [31], resulting in a frequency dependent value which oscillates between 13 dbm at 800 MHz and 1.5 dbm at 1 GHz. By using this value and from electromagnetic and circuit co-simulation results, an evaluation of the theoretical read range was obtained and depicted in Fig. 7b. The proposed RFID tag was fabricated and the read range was measured (see Fig. 7) through the procedure explained in the previous subsection. Very good performance through the entire UHF RFID frequency band is achieved. The Fig. 7 Fabricated UHF-RFID tag a) photography b) simulated and measured read range of the proposed tag and other commercially available tags, considering EIRP=4 W in the whole frequency band proposed tag exhibits a significant read range, higher than 5 m within the whole band ( MHz) with roughly 9 m peak range. A 40 MHz frequency shift between the simulated and measured peak range can be observed in Fig. 7b. Obviously, the peak read range of a typical RFID tag occurs at the frequency where conjugate matching between the chip and the antenna is achieved (see Fig. 1). The main cause of shift is attributed to the power reflection coefficient s. This displacement can be attributed to a combination of different sources of error, such as variations in the imaginary part of the packaged chip impedance because of the flip-chip attachment process and to variations in ASIC assembly, which may cause changes in the chip impedance reported by the manufacturer [5]. Other reasons explaining this frequency shift are related to the accuracy of the electromagnetic simulators, and to tolerances in both geometrical and substrate parameters. Nevertheless, the measured read ranges of the proposed global band tag are very reasonable. A comparison of our fabricated tag with other commercially available tags only makes sense if the considered chips are identical. This is because the optimum bandwidth achievable by any tag depends on the chip impedance, and the activation power is fundamental in determining the read range. Hence, for comparison purposes, we include in Fig. 7b the measured read range of a wideband commercial tag (UPM Web) that uses identical chip to that of our proposed tag. Although the commercial tag has a higher area (50 mm 30 mm), our tag exhibits much higher read range in the whole UHF RFID band. From these results, it follows that the strategy for bandwidth optimisation is efficient, and the specific designed tag is competitive. 5 Conclusions In this paper, bandwidth optimisation in single resonant UHF RFID tags has been considered. It has been demonstrated that, for a certain tag chip modelled by a parallel RC circuit, the optimum network for bandwidth enhancement, providing conjugate matching, is simply a parallel combination of an inductor and a resistor cascaded to the chip. Bandwidth optimisation has been validated by means of the Bode criterion. It has been also shown that a folded dipole antenna behaves as a parallel combination of an inductor and a resistor, and by designing this antenna to provide conjugate matching at f 0, a bandwidth very close to the optimum value results. A formula that predicts the maximum achievable 3 db bandwidth (i.e. the one that results by considering the ideal RL network with conjugate matching) for a given UHF RFID chip has been derived. We have also inferred a generalised expression for the 3 db bandwidth when any real network, corresponding to the antenna and matching network (if it is considered), is cascaded to the UHF RFID chip. According to this analysis, a broadband RFID tag prototype able to operate within the different worldwide regulated UHF RFID bands has been designed and fabricated. The experimental results show a significant read range, over 5 m within the whole band ( MHz) and almost 9 m peak read range. The area of the power reflection coefficient is > 80% of the Bode area limit, and the corresponding 3 db bandwidth is 73% of the upper limit for complex conjugate matching between the antenna and the chip at f 0 = 915 MHz. Thus, with the reported IET Microw. Antennas Propag., 013, Vol. 7, Iss. 10, pp & The Institution of Engineering and Technology 013

7 example, it is demonstrated that it is possible to evaluate if the designed tag is close to the maximum achievable bandwidth or if it still has some margin for improvement. Moreover, the proposed tag has been designed without using an external matching network, for cost and size reduction. 6 Acknowledgments This work was supported by Spain MICIIN under projects CONSOLIDER CSD and METATRANSFER TEC , and by AGAUR (Generalitat de Catalunya) through the project 009SGR-41. The authors thank J. Selga for developing the numerical simulations used in this work. 7 References 1 Finkenzeller, K.: RFID handbook: radio-frequency identification fundamentals and applications (Wiley, 004, nd edn.) Stockman, H.: Communication by means of reflected power, Proc. IRE, October 1948, pp Landt, J.: The history of RFID, IEEE Potentials, 005, 4, (4), pp Brown, D.: RFID implementation (New York, McGraw-Hill, 007) 5 Seshagiri Rao, K.V., Nikitin, P.V., Lam, S.F.: Antenna design for UHF-RFID tags: a review and a practical application, IEEE Trans. Antennas Propag., 005, 53, pp Marrocco, G.: The art of UHF-RFID antenna design: impedance matching and size-reduction techniques, IEEE Antennas Propag. Mag., 008, 50, (1), pp Deavours, D.D.: Analysis and design of wideband passive UHF-RFID tags using a circuit model. IEEE Int. Conf. RFID, May 009, pp Paredes, F., Zamora, G., Bonache, J., Martin, F.: Dual-band impedance-matching networks based on split-ring resonators for applications in RF identification (RFID), IEEE Trans. Microw. Theory Tech., 010, 58, (4), pp Bergeret, E., Gaubert, J., Pannier, P., Gaultier, J.M.: Modeling an design of CMOS UHF voltage multiplier for RFID in a EEPROM compatible process, IEEE Trans. Circuits Syst., 007, 54, pp De Vita, G., Iannaccome, G.: Design criteria for the RF section of UHF and microwave passive RFID transponders, IEEE Trans. Microw. Theory Tech., 005, 53, (9), pp Bode, H.W.: Network analysis and feedback amplifier design (D. Van Nostrand Co., New York, NY, 1945), pp Fano, R.M.: Theoretical limitations on the broadband matching of arbitrary impedances, J. Frank1Inst., 1950, 49, pp and Matthaei, G.L., Young, L., Jones, E.M.T.: Microwave filters, impedance-matching networks and coupling structures (Artech House Books, Dedham, Mass, 1980) 14 Kurokawa, K.: Power waves and the scattering matrix, IEEE Trans. Microw. Theory Tech., 1965, MTT-13, (3), pp Nikitin, P.V., Seshagiri Rao, K.V., Lam, S.F., Pillai, V., Martinez, R., Heinrich, H.: Power reflection coefficient analysis for complex impedances in RFID tag design, IEEE Trans. Antennas Propag., 005, 53, pp Nedlin, G.: Energy in lossless and low-loss networks, and Foster s reactance theorem, IEEE Trans. Circuits Syst., 1989, 36, pp Boghosian, W.H.: Comments on Energy in lossless and low-loss networks, and Foster s reactance theorem, IEEE Trans. Circuits Syst., 1989, 36, (1), p Impinj RFID chips [Online]. Available at 19 Alien Technology RFID Ics [Online]. Available at alientechnology.com 0 NXP UCODE smart label ICs [Online]. Available at 1 Foster, R.A.: A reactance theorem, Bell Syst. Tech. J., 194, 3, pp Guillemin, E.A.: Communication networks (Wiley, New York, 1935), vol., Ch. 6 3 Balabanian, N., Bickart, T.: Linear network theory (Matrix, Chesterland, OH, 1981), Chs. 11 and 1 4 Sisó, G., Gil, M., Bonache, J., Martin, F.: On the dispersion characteristics of metamaterial transmission lines, J. Appl. Phys., 007, 10, 1 7, paper Hong, J.S., Lancaster, M.J.: Microstrip Filters for RF/microwave applications (John Wiley, New York, 001) 6 Chu, L.J.: Physical limitations of omni-directional antennas, J. Appl. Phys., 1948, 19, pp Weeks, W.L.: Antenna engineering (McGraw-Hill, New York, 1968), pp Jasik, H.: Antenna engineering handbook (McGraw-Hill, New York, 1961) 9 Thiele, G.A., Ekelman, E.P., Henderson, L.W.: On the accuracy of the transmission line model of the folded dipole, IEEE Trans. Antennas Propag., 1980, AP-8, pp Clark, A.R., Fourie, A.P.C.: An improvement to the transmission line model of folded dipole, IEE Proc. Microw. Antennas Propag, 1991, 138, pp Nikitin, P.V., Seshagiri Rao, K.V., Martinez, R., Lam, S.F.: Sensitivity and impedance measurements of UHF-RFID Chips, IEEE Trans. Microw. Theory Tech., 009, 57, (5), pp IET Microw. Antennas Propag., 013, Vol. 7, Iss. 10, pp & The Institution of Engineering and Technology 013