`
`Where Is the Tag?
`
`Robert Miesen, Randolf Ebelt, Fabian Kirsch, Tobias Schäfer,
`Gang Li, Haowei Wang, and Martin Vossiek
`
`R adio-frequency identification (RFID)
`
`has penetrated logistics, manufactur-
`ing, production, ticketing, access con-
`trol, baggage tagging and various other
`areas of our daily life [1]. RFID technol-
`ogy has experienced tremendous growth and develop-
`ment since its humble beginnings back in the 1940s [2].
`These remarkable technical advances have resulted in
`enhanced performance and novel application areas,
`which in turn stimulate new needs and spawn exciting
`new research initiatives. A current hot research topic
`in the RFID fi eld is RFID localization [3], [4].
`
`In this article we present a broad overview of RFID
`localization developments and trends. In this con-
`text, we use RFID to denote a system that is primar-
`ily intended for the identification of an object tagged
`with a transponder. The basic functionality of an RFID
`system is a bilateral interaction between a reader and
`a single transponder. The reader incorporates most of
`the power-consuming signal generation and signal pro-
`cessing capability. Reader and transponder comprise a
`single antenna or an antenna array. Transponder com-
`plexity is limited because its price and power consump-
`tion are critical for most applications. Up to now passive
`
`Robert Miesen (miesen@iei.tu-clausthal.de), Randolf Ebelt, Gang Li, Haowei Wang, and Martin Vossiek are with the Institute for Microwave
`Technology, University of Erlangen-Nuremberg, D-91058 Erlangen, Germany. Fabian Kirsch and Tobias Schäfer are with the Institute of
`Electrical Information Technology, Clausthal University of Technology, D-38678 Clausthal-Zellerfeld, Germany.
`
` Digital Object Identifier 10.1109/MMM.2011.942730
` Date of publication: 15 November 2011
`
`December 2011 Supplement
`
` 1527-3342/11/$26.00©2011 IEEE
`
`
`
`S49
`
`RFC - Exhibit 1018
`
`
`
`and chipless RFID transponders meet this requirement
`best [1], [5]. However, many custom applications use
`semi-passive and active transponders, too [6], [7].
`We have chosen the above definition of the RFID
`localization system to differentiate the scope of this
`article from other publications dealing with wire-
`less local positioning systems (Figure 1). In a wireless
`local positioning system, localization and tracking is
`the primary intention, with localization relying on a
`network of several spatially distributed transceivers or
`transponders [8]. Multilateration and multiangulation
`are typical localization techniques used in wireless
`local positioning systems [9]–[13]. While many basic
`wireless local positioning techniques are very similar
`to those used for RFID localization, system and com-
`ponent layouts and target applications are different.
`RFID localization relative to a reader can be broken
`down into two different tasks: ranging and direction
`finding. A system with both of these capabilities can
`provide two-dimensional (2-D) or three-dimensional
`(3-D) localization and an RFID system with only one of
`these capabilities can still provide useful functionalities.
`The ranging and direction finding functions come
`to us from radar technology [14] and we will subse-
`quently see that many RFID localization solutions are
`based on radar principles.
`
`Usage and Application of RFID Localization
`
`Distance Bounding for Secure RFID
`Authentication and Access Systems
`Standard RFID systems are vulnerable to relay attacks
`[15]. During a relay attack an attacker uses two trans-
`ceiver relay stations to relay the information exchanged
`between a reader and an RFID tag during a crypto-
`graphic challenge-response protocol communication.
`One relay transceiver powers up the RFID tag and
`enables contact with the reader. The other relay trans-
`ceiver communicates with the reader. As the relay can
`bridge long distances the unauthorized authentication
`is difficult to detect by the holder of the RFID device.
`As presented in [15] and [16] it is impossible to impede
`
`relay attacks effectively by countermeasures based
`solely on cryptographic protocols that operate at higher
`layers of the RFID protocol stacks. Hancke stated:
`“The only effective defense are distance-bounding or
`secure-positioning protocols that are tightly integrated
`into the physical layer of the communication protocol,
`so as to obtain high-resolution timing information
`about the arrival of individual data bits.”
`Distance bounding and the relay attack in passive
`keyless entry systems and car starting systems attract
`a lot of attention [17], [18]. Hands-free computer sys-
`tems, terminal access systems or automatic door open-
`ers for buildings are just some of the applications
`posing similar challenges.
`
`Automatic Range-Dependent Functionalities
`If objects or persons equipped with RFID tags are
`detected in designated zones, automatic responses like
`opening doors in garages, turning on lights or signaling
`alarms can be triggered. Some of these ideas have been
`applied in practice and are commercially available. More
`examples are automated parking lot access and payment
`[19] or a system that recognizes the presence of infants
`in a car and deactivates air bags or issues an alarm signal
`if children have been left in a car in a dangerous situa-
`tion—for example, if the temperature in the car reaches
`dangerous levels [20]. Another application is booting
`and logging on/off computer terminals [21]. Localiza-
`tion of long-range RFID tags could improve ease of oper-
`ation by improving reading range and security as users
`can leave their IDs in their pockets.
`Load/unload detection for industrial trucks and
`the correct identification of loaded items in warehouse
`management systems is one of the key functions in
`automatic stock localization. Because of the long read-
`ing range of UHF RFID tags it has become a challenge
`to identify only the loaded item and the precise time of
`loading/unloading in order to establish the location of
`the item for the inventory control system.
`Locatable long-range RFID tags can potentially
`improve and broaden the scope of these functionalities
`by more precisely determining the “trigger zone.”
`
`Tag
`
`i o n ,
`C o m m u n i c a t
`M e a s u r e m e n t
`
`RFID
`Reader
`
`(a)
`
`T
`
`T
`
`T
`
`n ,
`n t
`
`a ti o
`e m e
`
`n i c
`r
`u
`
`C
`
`o m m u
`s
`a
`M e
`
`T
`
`T
`
`(b)
`
`Figure 1. The main differences between (a) RFID-like localization systems and
`(b) wireless local positioning systems are lower transponder complexities and the absence
`of a transponder network.
`
`Software-Defi ned
`Boundaries of the
`Detection Volume
`If RFID is applied in envi-
`ronments where multipath
`fading and shadowing are to
`be expected, it is necessary
`to provide a considerable
`margin in the link budget to
`ensure transponders in the
`vicinity of the reader are cor-
`rectly identified. Path loss
`and thus, reader detection
`range, can vary drastically
`
`S50
`
`
`
`December 2011 Supplement
`
`
`
`with time or location. If the path loss is low a tagged
`object may be identified that is not in the vicinity of
`the reader and is, in fact, not part of the process. In this
`case false conclusions may be drawn.
`Standard RFID systems feature only coarse locating
`capabilities with a granularity given by the dimension
`of a cell formed by a reader. The cell is irregularly
`shaped and its bounds may vary with time. To assure
`a spatial and situational context it is common practice
`to place readers with short reading ranges at key loca-
`tions, such as at gates and in selected storage areas.
`However, with increasing reading range this approach
`becomes more and more impractical.
`Only if the RFID system features a ranging capabil-
`ity is it possible to define a maximum reading range per
`software independently of a link budget margin. With
`RFID localization, detection areas and volumes can be
`defined and the benefits from an improved RFID read-
`ing range can be exploited as depicted in Figure 2.
`
`Generating Spatial Object Maps for Precise
`Real-Time Inventory and Augmented/Virtual
`Reality Applications
`If tags can be localized in 2-D or even 3-D, a completely
`new and exciting application area opens up. A mobile
`reader capable of localizing RFID tags can build an
`object map of its surroundings. If the reader is moved
`it can add areas to its map and its application range
`can be extended beyond the reader’s actual reading
`range. This is similar to the simultaneous localization
`and mapping (SLAM) approach [22]. Reference tags
`can provide a framework for reliable reader localiza-
`tion. A position-aware reader can generate an inven-
`tory of all tags and instantly put them in the spatial
`context. When sufficient spatial information is avail-
`able, operators can be guided to items of interest using
`augmented-reality enabled devices that create a com-
`mon spatial context of operator and tagged items and
`provide guidance in an intuitive and efficient way.
`Furthermore, the data from false positive identi-
`fications which heretofore had to be suppressed can
`now be used. In its spatial context a false positive is
`additional data which is easy to distinguish from the
`intended data. Consequently, the additional data can
`be used for inventory verification and correction or
`for security purposes. Correct location of hazardous
`goods can be verified and blocked pathways reported.
`Large areas inside production facilities can be scanned
`by antennas mounted on cranes or vehicles to gather
`additional information. Hospital operating rooms can
`be scanned to monitor the presence and correct loca-
`tion of vital equipment.
`
`Finding and Retrieving Tagged
`Objects or Persons
`The fact that an item is present might be sufficient
`information for a stock inventory system; it is most
`
`RFID
`Reader
`
`Transponder
`with Sensor
`
`Valid
`Volume
`
`Maximum Reading Range
`
`Figure 2. Automatic range-dependent pairing of remotely
`readable medical sensors in the vicinity of a reader can
`prevent incorrect assignments. Ranging as well as direction
`estimation is needed in this example to check the valid
`volume.
`
`certainly inadequate for a production planning sys-
`tem. The steel billets in Figure 3, for example, might
`all be known to the inventory control system but with
`standard RFID systems the most accurate positional
`information available is the cradle number where a
`steel billet has been stored.
`Therefore a production planning system has to
`assume the worst-case time needed to retrieve a bil-
`let for scheduling purposes and the billet cannot be
`retrieved automatically. By attaching low-cost tags
`to items that can be localized by the reader, opera-
`tor search times can be reduced drastically. Finding
`tagged objects—especially small objects in densely
`tagged environments such as shelves—can be speeded
`up and object retrieval checked automatically [23].
`Tagging metal objects as shown in the example
`above is requested as steel producers strongly want to
`benefit from RFID technology. Proximity to metal sur-
`faces detunes RFID tag antennas, shifts the resonance
`frequency and lowers its amplitude. Consequently,
`
`Figure 3. A cradle of steel billets implies only vague
`positional information. By individually tracking the
`billets’ locations search times can be cut and optimized
`storing strategies can be implemented. Metal objects pose
`additional challenges to UHF RFID systems.
`
`December 2011 Supplement
`
`
`
`S51
`
`
`
`Chipless TDR
`(Time-Domain
`Reflectometry)
`
`Backscatter
`
`Bidirectional
`Transceiver
`
`Unidirectional
`Transmitter
`
`Round-Trip Time
`of Flight
`
`— Broadband CW
` Radar (Frequency
` Modulated/Phase
` Coded) or Pulse
` Radar Principles
`— Short-Range Devices
` and Ultra Wideband
` Transceivers with
` Nearly Delayless
` Response or
` Sophisticated Clock
` Synchronization
`
`Received Signal
`Strength
`
`— RSS to Distance
` Mapping
`— Reference
` Transponders
`— Direction
` Scanning
`— Amplitude
` Monopulse
` Antenna
`
`Phase
`Evaluation
`
`— Low-Frequency
` Round-Trip Phase
`— Multifrequency
` Round-Trip Phase
`— Phase Difference
` of Arrival; Phase
` Monopulse Antenna
`— Antenna Array
` Direction Finding
`— Synthetic Aperture
` Radar/Holography
`
`Figure 4. Classification of RFID ranging techniques. The boxes on the top represent the transponder types used for RFID. As
`the lines illustrate, almost every transponder type can be used with all three measurement principles.
`
`the reading range of detuned tags is reduced. Special
`on-metal tags are commercially available [24]. While
`some older on-metal tags solemnly rely on a gap
`between metal surface and tag antenna in the order of
`10–20 mm, recently developed tags are tuned to per-
`form very close to metal surfaces [25].
`RFID is used for maintenance purposes mainly to
`reliably identify system components—for example, for
`maintenance work on airport fire security equipment
`like fire shutters, fire doors and smoke detectors [26].
`It is also used to find these components in complex
`environments. If numerous tagged components are
`packed densely together or stowed behind hatches in
`airplanes, for instance, localization will help mainte-
`nance personal work more efficiently.
`Passive radar reflectors and active beacons are used
`by avalanche rescue services [27]. Radar reflectors
`require custom equipment, and beacons available on
`the market are difficult to use. Improved localization
`technologies can potentially reduce search and rescue
`times and save lives.
`
`RFID Localization Techniques
`
`Classifi cation of RFID
`Localization Techniques
`RFID ranging techniques may be categorized accord-
`ing to three criteria. The first and second are the type
`of transponder and the fundamental principle behind
`tag localization. The transponder types are chipless
`time-domain reflectometry (TDR), e.g., surface acoustic
`
`wave (SAW) transponders; backscatter transponders,
`e.g., UHF RFID tags; bidirectional transceivers; and
`unidirectional transmitters. Measurement principles
`are round-trip time of flight (RTOF), in which the time
`a signal needs to travel between interrogator and tran-
`sponder is retrieved; received signal strength (RSS),
`utilizing the relationship between signal strength loss
`and distance; and phase evaluation, which allows the
`distance to be determined as a fraction of the signal
`wavelength, whereby the total number of signal peri-
`ods is unknown. The third criterion is the actual tech-
`nique applied based on one of the three principles. Not
`all of these methods are suited for use with all types
`of transponders; this is shown with colored lines in
`Figure 4.
`The most important limitations influencing the
`localization performance of the various techniques
`are summarized in “Constraints on Performance and
`Localization Accuracy.”
`
`Round-Trip Time of Flight Principles
`For an RTOF measurement the reader transmits a
`signal to the RFID transponder at time t1TX. The tran-
`sponder retransmits a response to the reader after a
`predefined processing period t2p that must be known
`by both units. The time of flight (TOF) to the tag and
`back to the reader are denoted as t12 and t21. The basic
`principle and timing are illustrated in Figure 5.
`Given the transmit time t1TX and the measured
`receive time t1RX, the reader can calculate the distance
`to the transponder as
`
`S52
`
`
`
`December 2011 Supplement
`
`
`
`Constraints on Performance and Localization Accuracy
`
`radar to separate two closely spaced echoes. It is
`directly linked with the radar signal bandwidth B via:
`drad < c0
`2B
`
`,
`
`(S1)
`
`
`
`where c0 is the free space RF signal phase velocity.
`An investigation focusing on the impact of the
`available bandwidth in a real world RFID scenario is
`presented in [28].
`The plots in Figure S3 show that the ranging
`uncertainty is on the order of some decimeters if the
`radar bandwidth is 80 MHz. Achievable accuracies
`may be considerably worse in severe multipath
`scenarios—especially if the line of sight is damped or
`blocked. Typically, the ranging uncertainty is inversely
`proportional to the radar bandwidth B as predicted by
`the rule of thumb (S1).
`A well-designed system operating in an
`environment with low-multipath distortion, however,
`delivers much better ranging precision. An expression
`for the lower bound of precision sr in ideal
`situations—where multipath and all other distortions,
`except noise, is negligible—can be derived from the
`Cramér-Rao Lower Bound (CRLB) [80].
`
`1B, Es, N0
`
`2 $
`
`
`
` s r
`
`c0
`
`2BÅ 1
`
`p2Es/N0
`
`a1 1 1
`
`Es/N0
`
`b ,
`
`(S2)
`
`where Es and N0 denote the signal and noise power.
`Provided that the bandwidth is fi xed by
`
`Tag A Measurement No. 1
`Tag A Measurement No. 2
`Tag B Measurement No. 1
`Tag B Measurement No. 2
`
`−55
`
`−60
`
`−65
`
`−70
`
`−75
`
`RSS (dBM)
`
`1
`
`2
`
`4
`3
`Distance (m)
`
`5
`
`6
`
`Figure S2. Multipath fading can cause large
`differences in field strength. The plot shows received
`signal strength versus distance in a direct line of sight
`situation. Tag B is placed 0.2 m below tag A.
`
`There is a multitude of constraints on RFID
`localization systems—constraints that may prohibit
`reasonable usage because of inadequate accuracy
`levels, complex transponder hardware or additional
`expensive infrastructure.
`One major challenge for all localization techniques
`is multipath propagation. When the impulse response
`of a real channel is recorded, an echo profile like the
`one shown in Figure S1 is frequently encountered.
`Strong multipath reflections from the ceiling, the
`floor and from metal objects coupled with a blocked
`line of sight between reader and transponder often
`occur in industrial environments, such as warehouse
`gates. Therefore, time of flight (TOF) measurements
`can fail to deliver accurate distance estimation if,
`for example, the LOS echo is masked by noise or a
`strong multipath echo. The same applies to received
`signal strength-based techniques. Reflections can
`lead to interference which increases or decreases
`signal strength regardless of the distance, while a
`blocked line of sight reduces signal strength and leads
`to greater distance estimations. This can even happen
`with a clear, unobstructed line of sight in an open
`space when ground reflections severely impact the
`measured signal strength (see Figure S2). Note that
`two measurements with the same tag are very similar
`as long as the environment does not change. Many of
`the systems mentioned above utilizing reference tags
`are based on this fact.
`The measured phase is altered by superposition of
`different signal paths as well. Furthermore, the phase
`of the transponder signal depends on the modulation
`properties of the transponder which again depend on
`carrier frequency and transponder power level [75],
`[79]. Compensation and calibration techniques are
`available.
`Multipath perturbation effects can be
`counteracted to some extent by using a higher
`bandwidth. The minimum width of an echo and thus
`the range resolution drad quantifies the ability of a
`
`Weak LOS Echo
`
`10
`
`30
`20
`Distance (m)
`
`40
`
`50
`
`1
`0.8
`
`0.6
`
`0.4
`
`0.2
`
`0
`
`0
`
`Normalized Amplitude
`
`Figure S1. Echo profile with weak line of sight and
`extensive multipath components [28].
`
`December 2011 Supplement
`
`
`
`S53
`
`
`
`Movable
`Palette
`
`rx1
`rx2
`
`NLOS
`
`tx1
`
`tx1
`
`RFID Gate Top View
`
`(b)
`
`Low SNR,
`Blocked LOS
`
`tx1/rx1
`tx1/rx2
`tx2/rx1
`tx2/rx2
`
`10.0
`6.0
`4.0
`
`2.0
`
`1.0
`0.6
`0.4
`
`0.2
`
`Ranging Uncertainty (m)
`
`10
`
`20
`
`40 60 100
`Bandwidth (MHz)
`
`200 300 500
`
`(a)
`
`Figure S3. RFID-localization uncertainty is inversely proportional to the signal bandwidth. The propagation paths
`of the signals from interrogator tx1 are blocked by a movable palette, while tx2 has a line of sight to rx1 and rx2. Even
`under non-line of sight conditions the ranging uncertainty drops with increasing bandwidth [28].
`
`legal spectrum regulations, expression (S2) shows
`that the signal-to-noise ratio is the limiting factor in
`ideal operating conditions. In real-world multipath
`scenarios, model-based/super-resolution techniques
`or antenna diversity multiple input multiple output
`
`(MIMO) may improve the localization performance
`beyond the limits set by the signal bandwidth.
`Robustness may also improve with SAR techniques
`[73] that greatly suppress multipath disturbances but
`also considerably increase the signal processing load.
`
`
`
`d 5
`
`t21 1 t12
`2
`
`# c 5
`
`1t1RX 2 t1TX
`
`2 2 t2p
`
`2
`
`# c.
`
`(1)
`
`The quality of the TOF measurement eTOF is primar-
`ily determined by the signal bandwidth and signal-to-
`noise ratio (SNR) [28] and the quality of the receiver. To
`meet these demands, broadband correlating receivers
`are the receivers of choice for RTOF systems.
`A high-quality clock is not typically integrated in
`a low-cost transponder. Even a standard crystal oscil-
`lator may not always suffice. Given a standard oscilla-
`tor with 50 ppm frequency stability, a 1-ms processing
`time would cause a ranging error of up to 15 m. There
`are two options to limit the impact of the clock skew.
`One is to run a synchronization protocol that enables
`the clock skew to be estimated and its effect to be miti-
`gated. The other option is to use a transponder with-
`out processing time, viz. the modulated backscatter
`transponder or the switched injection-locked oscilla-
`tor (SILO) transponder [29]. Modulated backscatter
`transponders can be realized as passive transponders
`whereas active SILO transponders have a notably
`increased operating range.
`
`Chipless Time Domain
`Reflectometry Transponder
`A class of RFID system with inherent distance mea-
`surement capability is TDR-based chipless RFID [30].
`TDR-based chipless RFID transponders are usually
`interrogated by a reader similar to a common pulse
`or frequency modulated continuous wave (FMCW)
`radar. The radar transmits a signal and receives a train
`of delayed echoes. The echoes are delayed by a specific
`
`In an RTOF measurement, two different challenges
`need to be faced. First the processing delay needs to
`be known exactly because slight deviations will lead to
`significant distance estimation errors. Thus the reader
`can only determine an estimate (denoted as t1p) of
`the real transponder processing time t2p. The second
`task is to determine the transmit and receive times as
`accurately as possible. Based on the uncertainty eTOF of
`the TOF measurement and considering the clock skew
`eclock, the ranging error eRTOF of the RTOF measurement
`can be determined as
`# eClock
`
` eRTOF 5 0 eTOF
`
`0 1 0 t2p
`
`0 with eclock 5 1 2 t2p/t1p. (2)
`
`TOF
`There
`
`τ12
`
`Processing
`Time
`τ1p
`
`τ2p
`
`TOF
`Back
`
`τ21
`
`Interrogator
`
`Transponder
`
`t1TX
`
`t2RX
`
`t2TX
`
`t1RX
`Time
`
`Figure 5. Round-trip time of flight measurement
`between an RFID reader (interrogator) and an RFID tag
`(transponder). The interrogator can only estimate the
`processing time t2p.
`
`S54
`
`
`
`December 2011 Supplement
`
`
`
`Chipless TDR Tag
`Tag Antenna
`
`e
`
`u ls
`
`d P
`
`e
`
`s
`
`n
`
`o
`
`p
`
`s
`
`e
`
`g R
`
`a
`
`T
`
`Reflector
`
`m itt e
`
`s
`
`n
`
`T r a
`
`Reader
`
`Figure 6. Time-domain reflectometry-based SAW RFID
`system [31].
`
`However, the maximum delays achievable thus far are
`much lower than for SAW transponders. Short delay
`periods limit the available code space and the abil-
`ity to separate and suppress direct echoes from the
`environment. A considerable amount of research is
`still needed to bring these new transponder technolo-
`gies to maturity. That said, they are very promising
`candidates for next generation locatable RFID tran-
`sponders. If ultrawideband (UWB) signals and UWB
`radar systems are used to read TDR transponders, the
`task of precisely localizing such a transponder—even
`if the multipath conditions are extreme—should be
`straightforward.
`
`Backscatter Transponder
`Chipless TDR transponders separate the desired sig-
`nal from environmental reflections by a time delay.
`Another popular method of separating the response
`signal is the use of modulated reflection [37], [38].
`Backscatter transponders modulate their response by
`varying their reflection coefficient.
`In typical indoor scenarios, as shown in Figure 8,
`unmodulated multipath reflections may be received
`with higher amplitude than the desired line-of-sight
`transponder reflection. The modulated multipath
`reflections still contribute to the received signal, but
`are typically weaker than the line-of-sight signal.
`
`Environmental
`Echoes Tag Reflector Echoes
`
`Start Bit
`
`Stop Bit
`
`−30
`−40
`−50
`−60
`−70
`−80
`
`RX Echo Level (dBm)
`
`0
`
`1
`
`2
`RTOF (μs)
`
`3
`
`4
`
`Figure 7. Time-domain response of a pulse-coded 2.45 GHz
`SAW RFID transponder.
`
`delay structure on the transponder. Pulse position,
`pulse amplitude and pulse phase modulation concepts
`form the modulation principle applied to encode the
`transponder ID. The delay times of all echoes gener-
`ated on the transponder have a constant offset that is
`given by the RTOF of the interrogating signal between
`the radar and transponder antenna. A start bit that is
`located at a defined position on every transponder, or
`several bits in known positions, or even a correlation
`of the complete transponder response can be used to
`measure the RTOF between the radar and transponder
`antenna.
`The best developed TDR-based chipless RFIDs are
`based on SAW devices [31]. A schematic of a TDR-
`based chipless SAW RFID system is shown in Figure 6.
`The pulse received by the transponder is fed into an
`interdigital transducer which forms the SAW. Marks on
`the transponder partially reflect the wave back into the
`transducer and are retransmitted to the reader. Figure 7
`shows an example of a transponder modulated signal
`as received by the reader.
`To ensure that the transponder reflector echoes are
`not disturbed by environmental echoes, they need to
`be separated well from the environmental echoes by a
`proper delay. In addition, a high SNR and a wide band-
`width are required to accurately measure the position
`of the reflector echoes. Temperature variations often
`need to be considered because the RTOF of the tran-
`sponder reflector echoes changes significantly with
`temperature [32], [33].
`In [32] a localization accuracy of approximately
`20 cm was achieved with a SAW RFID system. This sys-
`tem runs at 2.45 GHz and uses a 40 MHz bandwidth.
`In another system a rotating antenna in combination
`with the inherent RTOF measurement capabilities of
`the SAW RFID system was deployed for 2-D localiza-
`tion of a SAW transponder [33]. Sub-decimeter local-
`ization accuracy within a maximum detection range
`of 10 m was demonstrated in this paper. These results
`were achieved with an antenna having a vertical beam
`width of 17°.
`The maximum frequency of SAW devices is limited
`by technological issues and the available bandwidth
`is governed by legislation. Typical SAW RFID systems
`have a bandwidth of less than 80 MHz. Given this
`small bandwidth, good ranging accuracies can only be
`achieved in a channel with very low multipath distor-
`tion [28]. To improve the multipath resistance, diver-
`sity techniques or other multiantenna concepts may be
`applied, but precise localization remains challenging
`in real world indoor environments.
`Recently, innovative designs for TDR chipless
`RFID transponders have been proposed based on
`transmission delay lines and artificial left-handed
`or right-handed transmission lines [34]–[36]. These
`novel approaches promise greatly increased band-
`width and lower insertion loss compared to SAW.
`
`December 2011 Supplement
`
`
`
`S55
`
`
`
`backscatter transponder remains in (4). The vector
`DSBB,mod is rotated around the origin of the IQ plane
`depending on the RTOF tBS and the used carrier fre-
`quency vc. At typical carrier frequencies on the order
`of 900 MHz this ranging information has a small
`unambiguous range of 33 cm, which does not give
`usable distance information for most applications. To
`overcome this limitation the measurement can be car-
`ried out with different frequencies.
`Using only two different carrier frequencies gives
`reliable distance measurements as long as the line-of-
`sight signal is significantly stronger than the modu-
`lated multipath signals. To mitigate multipath effects
`a multitude of carrier frequencies can be used to mea-
`sure the phase information—the so-called frequency
`stepped continuous wave (FSCW) approach [40].
`Instead of using discrete stepped frequencies, the
`FMCW approach uses a sweep signal that linearly
`changes frequency over time. With the carrier fre-
`quency vc being time dependent, the baseband signal
`now takes the form of
`# ejvc
`
`# ejvc
`
`aSR,i
`
`N i
`
`51
`
`1t2 5 aBS
`
` SBB
`
`1t2tBS # s1t2t2 1 a
`
`The linearly changing carrier frequency vc
`results in each reflection generating an individual
`complex oscillation in the baseband signal. These so-
`called beat signals have frequencies typically below
`1 MHz. To separate the unwanted static reflections
`from the transponder’s reflection, a periodic modula-
`tion with period 1/ f mod is chosen for FMCW instead of
`using payload data. This modulation moves the beat
`signals of the transponder response to frequencies at
`tBS
`# dwc
`2p
`dt
`
`
`
`f r
`beat 5
`
`6 fmod.
`
`(6)
`
`By choosing fmod high enough, the static reflections
`can be fully separated from the RFID transponder’s
`beat signal [41]. There is no need to use a quadrature
`
`1t2tSR,i. (5)
`1t2
`
`RFID
`Transponder
`
`d
`
`u l a t e
`d
`n m o
`a t h
`M u lti p
`
`U
`
`RFID
`Reader
`
`M odulated
`M ultipath
`
`Figure 8. Signal paths present in communication with
`modulated backscatter transponders. The unmodulated
`multipath is the reader’s signal reflected by walls or objects.
`The modulated multipath is the transponder’s signal
`reflected by walls or objects. The line-of-sight signal is
`shown in the middle [28].
`
`The basic setup of a backscatter RFID system is
`shown in Figure 9. For the sake of simplicity, modu-
`lated multipath reflections are ignored in order to
`obtain the basic phase relationship. After quadrature
`demodulation (IQ) and low-pass filtering the complex
`baseband signal is comprised of unmodulated com-
`ponents from the static reflectors and the modulated
`response from the RFID transponder. Calculating the
`difference of the base band signal SBB with
`
`# ejvctSR,i,
`
`(3)
`
`aSR,i
`
`N i
`
`51
`
`1t2 5 aBS
`
`# ejvctBS # s1t2t2 1 a
`
`
`
`SBB
`
`and the two different modulation states s0 and s1, we
`obtain
`
`# ejvctBS # 1 s12s0
`
`2,
`
`
`
`(4)
`
`
`
`DSBB,mod 5 aBS
`
`with aBS, tBS and aSR, tSR, denoting the amplitude and
`RTOF values of the backscatter and static reflectors,
`respectively. All static reflections cancel out of equa-
`tion (3) and only the distance-dependent term of the
`
`LO
`
`ϕ = 90°
`
`S Tx(t )
`
`σB(t )
`
`SR
`
`Q
`
`I
`
`SRx(t ) = αSR STx(t − τ SR)
`+ αBS STx(t − τ BS)σB(t − τ BS)
`
`
`S BB(τ, σ)
`
`Figure 9. Block diagram of interrogator and backscatter transponder [39].
`
`S56
`
`
`
`December 2011 Supplement
`
`
`
`mixer in the FMCW scheme. As seen in Figure 10, the
`peaks corresponding to the transponder reside at
`tBS
`# dwc
`2p
`dt
`
`fbeat 5 0 f r
`
`beat
`
`0 5 fmod 6
`
`
`
`.
`
`(7)
`
`signal in the transponder. The recovered code is then
`shifted to another frequency channel by a modulator
`and retransmitted to the reader. By correlating the
`transmitted code sequence and the received demod-
`ulated signal, the reader can determine the signal
`round-trip phase or time delay and thus, the distance
`to the transponder. The basic principle is illustrated in
`Figure 11.
`Many variations of such FDD RTOF ranging sys-
`tems have been investigated in the past. Some typical
`solutions can be found in [43], [44].
`Despite the advantage that no synchronization is
`required, FDD approaches are not always practical
`for broadband systems due to their inefficient use of
`the spectral mask assigned by legal regulations and
`challenges posed by the required band selection/
`duplexing filters. TDD approaches often provide more
`practical solutions for the efficient use of an available
`spectral band.
`TDD protocols usually require synchronization of
`the clocks of the reader and transponder, unless the
`response time in the transponder is kept very small. In
`[45] a UWB system with round-trip ranging protocol
`lasting only 20 ms is presented. With a 2 ppm time base
`this allows for an accuracy of around 1 cm without any
`additional synchronization efforts.
`Typical signals used for RTOF ranging are direct-
`sequence spread spectrum pseudo-random codes—
`such as Gold or Kasami codes—that have good
`autocorrelation and cross-correlation properties. An
`immediate response strategy is impractical, if lon-
`ger correlation code sequences are deployed or if a
`
`Transmit
`Signal
`
`Modulated
`Response
`
`RFID
`Transponder
`
`Backscatter
`Modulator
`
`This enables easy ranging by evaluating the frequency
`difference between the peaks surrounding the modu-
`lation frequency [37].
`Recently, efforts have been made to incorporate this
`distance