RFID Reader Antennas for Tag Detection in

Self-Confined Volumes at UHF

1 12 1
Carla R. Medeiros , Jorge R. Costa , , and Carlos A. Fernandes
1
1nstituto Telecomunica90es, Instituto Superior Tecnico
Av. Rovisco Pais 1, 1049-001 Lisboa, Portugal
Tel: +351218418480; Fax: +351218418472; E-mail: {carla.medeiros, jorge.costa, carlosJernandes}@lx.it.pt

2
ISCTE-IUL Instituto Universitario de Lisboa
Av. das For9as Armadas, 1649-026 Lisboa, Portugal

Abstract

Confinement of the detection region is a critical issue for some important RFID applications, where the coarse location of the
object is required along with its identification. In the UHF band, it is a challenge to confine antenna radiation to reasonably
?
sharp interrogation volumes, < lOA , without resorting to physical barriers. This paper presents a novel approach for RFID
reader radiating structures that self-confine tag detection to a desired volume, avoiding undesired readings outside the
interrogation volume. The proposed solution simultaneously covers the three world-assigned RFID frequency subbands at
UHF. To demonstrate the effectiveness of the solution, the proposed configuration was integrated into bookshelves, a
conveyor belt system, and a proximity point reader. Together with an appropriate control application, these form the building
blocks of a smart store with automatic inventoryLn9 and billing capability. The radiating structure concept was explored by full­
wave simulations, and it was validated with both near-field measurements and with tag-reading scores. Short demonstration
videos are available online.

Keywords: RFID; UHF antennas; microstrip antennas; RFID antennas; near-field UHF RFID; smart RFID reading surfaces

1. Introduction
adio-frequency identi fication (RFID) is a rapidly expanding
Rtechnology that enables radio detection and recognition of
obj ects associated with an un ivocal identi fication code carried by
an electronic chip attached to an antenna: the RFI D tag. These tags
send back their univocal code whenever they are interrogated by an
RFI D reader within range. The basic concept has been in use since
the 1 980s, initial ly for automatic tol l systems . However, the many
advantages of thi s technology and the rapid dropping of associated
manufacturing costs extended its applicabi lity to other mass appli­
cations, such as pet or object identification, inventorying, asset
management, contact-less access control , anti-counterfeiting and
anti-theft systems , health care, etc. [I, 2]. Recently, RFID technol­
ogy has been extended to simultaneously al low some degree of tag
tracking and location [3 -5].
RFIO tags can be passive, retrieving the necessary energy
from the interrogating wave, or can include a battery (active tags),
to enhance reading range. Passive tags are the cheapest to manu­
facture, do not require maintenance, and are more compact and
lighter. World-assigned RFID frequency bands range from high
frequency (HF) up to microwaves [I, 2]. The al located band at
ultra-high frequency (UHF) ranges from 860 MHz to 960 MHz,
divided into three subbands corresponding to the three world
regions. In general, UHF passive RFID tags [6] present the lowest
unit cost - of the order of cents - which makes U H F RFID the pre­
ferred choice for mass appl ications. Its tag cost is expected to
decrease within a few years from $0.20 to $0.05 USO [7], making
it affordable to attach to very-low-val ue products. It has the poten­
tial to be used (for instance) on all items from retai l stores and
supermarkets (item-level tagging, or IL T), to allow automatic real­
time inventorying, tracking of misplaced items, and unattended
billing at cashiers. Thi s would totall y replace the present manned
optical-barcode reading for every instance of store items and asset
contro l . Unlike optical reading, RFI D technology enables simulta­
neous multiple-tag reading, and does not require tag visibility.
The change of paradigm for item tracking in stores or ware­
houses requires the possibility of confining RFI D tag reading to
predefined volumes, such as those associated with individual
shelves, cashier conveyor belts, proximity point readers, etc . , with­
out unwanted readings from immediate neighboring volumes.
However, it is a chal lenge at UHF to l imit detection to smal l sharp
regions of the order of < lOA? without resorting to intrusive and
costly metal plates, absorbing shields, or gates [8]. Unlike HF
RFIO, coupling between the reader and tag antennas at UHF is
made through a propagating electromagnetic wave, which inher­
ently enhances the tag-detection range.
Thi s paper presents a very simple, effective, and l ow-cost
RFI D solution for smart reading applications at UHF. Thi s allows
retrieving tag information from objects placed on shelves, prox­
imity point readers, or sliding over conveyor belts. The proposed
configuration overcomes the field-confinement problem at UHF,
intrinsically limiting tag detection to the boundaries of the region

IEEE Antennas and Propagation Magazine, Vol. 53, No.2, April 2011 ISSN 10 45-9243120 111$25©20 11 IEEE 39
of interest, insuring near-range univocal detection over the entire
surface without the need for physical barriers. The structure is very RF
RFID
simple with mild fabrication tolerances, appropriate for extremely Switch
Reader
low-cost mass production. Furthermore, the solution can be easily
attached to pre-existing store furniture, reducing the cost and time
RFID Application
required for a technol ogy change. To validate the proposed solu­
tion, prototypes of bookshelves, an isolated point reader, and a
cashier conveyor belt were characterized, fabricated, and tested in
the UHF band, using commercially avai lable RFID reader equip­
ment and tags. The authors first proposed in [9] a preliminary
RFI D bookshelf solution, and in [ 1 0], the smart conveyor-belt con­
figuration. A patent is pending [II], and a demonstration video can Point Reader
Conveyor Belt
be seen online [ 12]. Shelves or Tables

Thi s paper i s organized as fol lows. Section 2 states the vol­
ume confinement problem for UHF RFID, which is the motivation
of the present paper. The subsequent section is dedicated to the
characterization of the radiating structure proposed for smart-shelf,
proximity-point-reader, and conveyor-belt solutions for UHF
RFID. Section 4 presents the fabricated prototypes and the corre­
sponding experimental reading scores using commercial RFI D tags
and readers, demonstrating the effectiveness of the proposed solu­
tion. Conclusions are drawn in Section 5 .
2. Near-Rang e RFID Reading
Figure 1. An RFID-based smart reading system configuration
for a retail shop.
[ 1 3 , 1 4]. The main difference with UHF is that energy coupling
between the reader's antenna and the tag at HF is essentially made
through the magnetic field: coils or loops are typically used for the
HF reader and tag [ 1 3 , 1 4]. The detection region is inherently very
short, requiring in some configurations the repetition of the reading
antenna element along the shelf length to enable its full coverage
[ 1 5]. Thi s becomes intrusive, adds complexity to the feeding cir­
cuit, and increases cost. M oreover, HF passive tags are larger than
UHF tags, have very low data rates, and are more expensive.

Near-range UHF RFID antennas have been reported for tag

Coarse location can be accompl i shed by creating sub-areas
detection in shallow regions with footprints of the order of
with sharply confined reading range, without overlapping
A x A For instance, an inverted-L with a parasitic patch element
neighboring volumes. Tag location can then be identified through
.

was proposed in [16], but its use for smart shelves again required
association with the interrogated volume. The main challenges in
multiple antennas to cover the whole length of the shelf, repeating
near-range UHF RFI D are as fol lows :
the same problem identi fied above. An alternative solution, based
on multiple meandered open-circuited microstrip lines with a slot­
I. First, to ensure that the antenna's reading capability is
ted ground plane was proposed in [8]. The structure was moder­
confined to the volume above the reading surface, that
ately complex, and no coverage results were presented for larger
its coverage is uniform without blank spots to ensure a
surfaces, such as those from typical shelves. As for conveyor-belt
1 00% reading score, and that it does not extend to unde­
applications, existing UHF RFID solutions employ patch-type
sired adjacent volumes. Self-confinement must be sharp
antennas [ 1 7], either as single elements combined with shielding
enough so that the reader can discriminate (for instance)
barriers, or as multiple elements arranged in a tunnel-l ike structure
adjacent shelves or adjacent cashier conveyor belts In
to force confinement of the radiated field [ 1 8]. Either configuration
supermarkets.
is very intrusive and inappropriate for a store front-end.

2. The solution must be compatible with the existing com­

The solution presented in thi s paper incorporates a single
mercial passive UHF tags, which are much cheaper than
leaking microstrip line laid along the structure's length, with an
active tags . It is desirable that the antenna's operating
extended ground plane. M icrostrip lines are common ly used at
band be wide enough to cover the three RFI D world
microwave frequencies as transmission l ines, due to their ease of
UHF bands.
fabrication and integration with other devices. However, depend­
ing on substrate parameters and strip shape, the leaking fields may
3. The reader's antenna must b e simple enough for very­
be enhanced. Although undesired for usual microwave circuits
low-cost mass production, and easy integration into dif­
[ 1 9], this leaking characteri stic can be explored for the present
ferent shapes and sizes of store structures (shelves,
RFID application to produce controlled close-range tag detection at
tables, conveyors), inclusive of existing structures.
UHF, without readings from items located outside the desired vol­
ume. One end of the microstrip line i s connected to the reader, and
Figure I shows the generic configuration of an RFID system
the other end is connected to a matched load. It will be shown that
that explores the proposed UHF RFID smart-reading concept. A
this very simple solution ful fi l l s all the requirements stated above.
single reader can be used with multiple smart devices (shelves,
tables, conveyors, point readers), using a standard electronically
actuated RF switch to sequentially read each RFI D smart device in
the store. In this way, the RF power output from the single reader
is enough to read an arbitrary number of smart devices in the store.
3. Radiating Structure Characterization
A dedicated application is run on the computer to manage the
reading sequence, data logging, and data processing. Of course, Thi s section describes in detail the basic radiating structure
thi s configuration can be replicated several times in a large store. applied to a shelf, and its modified meandered versions applied to a
cashier conveyor belt and to an isolated point reader. These are
Some solutions limited to RFI D smart tables and shelves can three of the main types of equipment required for an RFI D-based
be found in the l iterature, but most of these operate in the HF band retail shop. Full-wave simulations and model optimizations were

40 IEEE Antennas and Propagation Magazine, Vol. 53, No.2, April 2011
perfonned using the transient solver from CST Microwave Stu­
dio ™ [20], based on the Finite Integration Method. The concept
was validated through RF measurements, and using commercial
RFI D readers and tags to assess the reading score.
versely to the vertical component. At the sheWs surface (Fig­
ure 4b), the y = 0 plane, the intensity of the Ey and Ez compo-
nents was reasonably constant within the sheWs area, and.
decreased rapidly outside the shelf boundaries. Experiments have
shown that the minimum field level for detection of a common
commercial tag - which i s described in the next section - i s of the

3.1 RFID Smart Shelf

load,,
,
An example of a smart-shelf structure [9] is shown in Fig­
ure 2. It uses a straight microstrip l ine running across the sheWs
length, along the x direction. The microstrip l ine is probe fed at one
z '. .
end, and it is tenninated on the other end with a matched load.
substrate
It i s well known that the main electric-field lines in a micro­ ground plane _
_
--
---
--
�
100 err!
strip flow from the metal strip to the ground plane through the sub­
strate layer, whi le some radiation leaks into the air, as shown in
Figure 2. The R FID smart shelf configuration.
Figure 3. A low-pennittivity Styrofoam (Gr "" 1 .05 ) substrate layer
is used to enhance these leaking fields. The electric-field orienta­
tion is mainly transverse to the microstrip line. Although this
places a restriction on tag orientation for reliable detection, it is
j ust perfect for a bookshelf application, where books tend to be
placed according to a preferred orientation. As wi l l be seen ahead,
this i s not a limitation of the concept, since some meandering of
the strip path is enough to enable reading arbitrary tag orientations.
Figure 3. The electric-field behavior of the microstrip line.
The prototype overall dimensions were typical for book­
shelves : 30 cm x 1 00 cm. The substrate thickness was I cm, and dB Vim
the transmission-line width was 5 cm, leading to a 50 n line char­ 30
acteristic impedance. The line was fed by a 50 n feeding coaxial 5
cable (EZ- 1 4 1 -AL), and it was tenninated by a 5 0 n resistor. The 20
15
strip's width, along with the fact that the ground plane was
10
extended up to the edges of the shelf, contributed to obtaining an 5
adequate field di stribution above the shelf, without significant 0
-20 o 20
spil lover to the volume below. Vertical decay of the fields also Z[cmj
ensured isolation with respect to an upper shelf. Depending on the Ez dB Vim
shelf size, other combinations of strip width and substrate thick­ 30 30
25
ness may be more favorable, eventual ly leading to different values 20
E 20
of the characteristic impedance. The relationship between strip .£. 10 15

•••• 10
>-
width and substrate thickness is determined by the known equation
o
of the microstrip characteristic impedance [2 1 ]. The higher the 5
-20 0
substrate thickness, the higher is the vertical detection. The width o 20 -20 o 20
Z[cmj Z[cmj
of the detection range within the bookshelf is mainly controlled by
the width of the ground plane. In fact, it is possible to ensure tag
Figure 4a. The simulated near-field components (Ex. Ey• and
detection up to a bookshelf width of about seven to eight times the
width of the microstrip line, as long as the ground plane covers the Ez) and the total near-field (E,) amplitude distribution at
entire bottom face of the bookshelf. A word of caution is note­ 866 MHz over an empty shelf in the yz plane.
worthy if the strip is chosen too wide, because the null of the
transverse E-field component along its axis may compromise tag
detection over this line. E
y dB Vim

Although not required, in the presented example an addi­ E

tional layer of Styrofoam substrate was added on top of the trans­ �
N
mission line to increase mechanical stability, and to protect the
o
radiating structure from contact with the books.
X [cm)
E
To evaluate the shelf perfonnance and analyze the electric­ z dB Vim

field behavior, near-field calculations were perfonned for the UHF �-20 �·20
band (at 866 MHz). Considering the coordinate frame from Fig­
1 0 1 0 �
ure 2, shelf boundaries corresponded to z = ±1 5 cm, x ±5 0 cm ,= N 20 N 20
and the shelf surface corresponded t o y = O . The electric-field o 50 -50 0 50
amplitude distribution in the yz and xy planes for I W of RF source X [cm) X [cm)
power is shown in Figure 4. The results at the x = 0 plane, shown
in Figure 4a, demonstrated that the leaking fields were strong Figure 4b. The simulated near-field components (Ex. Ey' and
above the shelf, reasonably confined to the near-zone, and that the Ez) and the total near-field (E,) amplitude distribution at
dominant electric-field components were Ey and Ez• The inten-
866 MHz over an empty shelf in the xz plane (the shelf's sur­
sity of the latter component was low at the center of the shelf, con- face).

IEEE Antennas and Propagation Magazine, Vo l . 53, No.2, April 2011 41

order of 20 d8V/m. So, Figure 4 anticipated that tag detection
would be well confined to the top surface of the shelf, preventing
detection outside the interrogated shelf without the need of EM
isolation barriers. Experimental results shown below confirmed
this.

It i s interesting to analyze how the presence of books affects

the previous field distribution . S imulations were repeated with 2 1
books placed o n the shelf. The books were modeled a s b locks of
Er = 2.3 dielectric spaced 0.5 cm apart, with 4 cm width, 23 cm
height, and 1 8 cm depth. The results are presented in Figure 5 .
Thi s showed that the field intensity was enhanced everywhere in
the shelf, including into the books, as if they drain power out of the
transmission line. Even the Ex component was greatly enhanced.
Nonetheless, the results demonstrated that the total electric field,
Er, remained at a low level outside the shelf, compared to field
levels over the shel f.

The simulated results from Figures 4 and 5 suggested that

circularly polarized or dual-polarized tags are the preferred choice
for detection of arbitrarily oriented obj ects above the shelf. H ow­
ever, full detection is also insured with linearly polarized tags
placed paral lel to the yz plane; other orientations may reduce the
tag-reading score, in this case. This will be analyzed below. Figure 6a. The finished fabricated prototype and application
screen of the smart shelf.

dB Vim
30
25
20
15
10
5
o
Z[cmj Z[cmj
Ez Et dB Vim

25
20
15
10
5
o

Z[cmj

Figure Sa. The simulated near-field components (Ex, Ey' and

Ez) and the total near-field (Er) amplitude distribution over
Figure 6b. An opened edge of the smart shelf, showing the
the shelf at 866 MHz when loaded with 21 books, in the yz detail of the line-feeding point.
plane.

dB Vim

·50 o 50 ·50 0
X [cm) X [cm)
Ez Et dB Vim

�·20 20
,[ ,[
�·

0 0
N 20 N 20
·50 o 50 0
X [cm) X [cm)
Figure 5b. The simulated near-field components (Ex' Ey' and
Ez) and the total near-field (Er) amplitude distribution over
the shelf at 866 MHz when loaded with 21 books, in the xz
plane (the shelf's surface). Figure 6c. A detail of the line termination of the smart shelf.

42 IEEE Antennas and Propagation Magazine, Vol. 53, No, 2, April 2011
 Figure 6a shows two manufactured shelf prototypes and the
Or-------�--�--r>--�_,
monitoring appl i cation screen. Each shelf comprised a wooden
-5 rectangular frame for mechanical support and two sheaths of ply­
wood, which were assembled to form a hollow structure that
housed the microstrip l ine, and the Styrofoam substrate and super­
strate. Detail s of the microstrip feeding point and of the termina­
a) tion can be seen in Figures 6b and 6c. The reader was located out­
side the shelf according to the scheme of Figure I, but a smaller
-5 commercially available reader module [2 1 ] could be accommo­
CD 1-----
-- - dated inside the shelf along with the proposed antenna, to obtain a
� -1 O ---
-- ""'
--::----+- --- __ -
__ __
� ___ ---------
___
•

standalone smart-shelf solution.

The measured and simulated input return losses of the

embedded straight transmission line are presented in Figure 7, both
for the empty shelf and when it was fully loaded with 2 1 books.
The agreement between simulation and measurements was quite
Figure 7. The measured and simulated reflection coefficient satisfactory. It was noted that when the 2 1 books were present, the
amplitudes: (a) empty shelf; (b) shelf fully loaded with 21 line ' s input return loss improved s lightly, consistent with the pre­
books. vious interpretation that more power was drained out from the line
in this case. It was noted that the operating band simultaneously
covered the three world-assigned RFI D subbands at UHF, bounded
50 by the two vertical dashed lines.
_E simulated
45 y
40 - Ez simulated
Near-field measurements of the empty shelf were performed
35 • E measured
y using an Agil ent PNA E836 1 A network analyzer with a sman­
E 30 E measured
electric-field probe. The near field was scanned across two cut
:> 25
•
z

!g 20 planes: a z-axis scan (plane x = 0 , y = 0 ), and a vertical scan

15 (plane x = 0 and z = 3 ). The measured results for Ey' Ez, and the
total electric field, shown in Figures 8 and 9, demonstrated good
agreement with simulations. Results for the Ex component are not
0
shown because its amplitude was very low. Consistent with the
z[cmJ
simulation results in Figure 4, the transverse scan from Figure 8
50 showed faster field-intensity decay outside the sheWs boundaries
-Et simulated
45 (beyond the ground plane, marked with vertical dashed l ines). This
" E measured
40 t showed the relevance of extending the ground plane towards the
35 shelf boundary. Figure 9 shows the total electric-field magnitude
E 30 measured across the closed ABCD contour, drawn 1 0 cm apart
:> 25 from the shelf boundary (at y = 0 ). The results demonstrated that
!g 20 the total electric field remained at a low level compared to the field
15 l evel over the shelf and, in fact, mostly below the referred
20 dBV/m threshold. Figure 1 0 shows the vertical decay of the
" " .. " fields, confirming the predicted extension of the vertical coverage.
-�o -20 -10 0 10 20 30
z[cmJ No measurements of this type are shown for the shelf loaded
with books, but the demonstrated -good agreement between previ­
Figure 8. The measured and simulated near-field components ous simulations and measurements allowed trusting the results pre­
and total near-field amplitudes at the (x = 0 , y = 0 ) plane of sented in Figure 5. Nevertheless, experimental tag-detection scores
the empty shelf. are thoroughly analyzed below, both for this case and for the
empty shelf.

50 r---�----�'-�-'----����====9
45
0 y - Et simulated

E ��.
30
��� ::�: �= . �=-;;;;�
I
3.2 RFID Smart Conveyor Bel t

The overall width and length of the studied conveyor-belt

:>
CD 25 A structure was 40 cm x 1 20 cm, typi cal dimensions for a cashier
�- 20 application. The same type of radiating structure was proposed for
..,.
w 15 integration with the conveyor belt, as shown in Figure I I. In thi s
10 case, t h e transmission l ine was sl ightly meandered t o enhance the
5 leakage of the previous less-dominant field components, and to
0 extend the tag-detection height for al l tag orientations in a con­
0 50 100 250
trolled way [ 10]. As before, a 1 cm-thick Styrofoam substrate was
used. The width of the microstrip line was 5 cm, leading to a 5 0 n
Figure 9. The measured and simulated total electric near field characteristic impedance. The meandered line was probe fed at one
around the shelf at a distance of 10 cm away from the shelf's end, and terminated at the other end with a matched load (a 50 n
boundary. resistor). The microstrip line and ground plane (30 cm x 100 cm)

IEEE Antennas and Propagation Magazine, Vol. 53, No.2, April 2011 43
were cut from thin metal sheet. This was smaller than the size of
the conveyor belt so as to fit inside it, mounted directly below the
belt, as shown in Figure lib. an

To evaluate the performance of the proposed solution, near­

field calculations were performed at the UHF band. Referring to "
Ground plane
the coordinate frame from Figure II, the embedded structure ( ;)
boundaries were z = ±1 5 cm, x = ±50 cm. The belt surface corre­
sponded to y = 0 cm. Microwave Studio [20] was again used to
Figure l1a. The R FID smart conveyor belt configuration: the
calculate the volume distribution of the electric-field components meandered microstrip line.
above an empty conveyor belt for I W of power fed to the micro­
strip line. The results are presented in Figure 1 2. It was clear that
the presence of the meanders significantly enhanced all the near­
field components (Figure 1 2a), while stil l remaining confined to
the structure's boundary. However, the field distribution at the
conveyor's surface presented isolated low field spots near the
edges of each meander of the microstrip l ine. This was not a set­
back, because the conveyor belt's continuous movement and the
consequent travel of the obj ects over the microstrip line was j ust
appropriate to insure that the tag was exposed to different field
values and orientations originated in different parts of the mean­
dered line. Thus, tags traveling over the device at least for one
period of the meander could be detected for any arbitrary orienta­
Figure llb. The R FID smart conveyor belt configuration:
tion, provided they were exposed to the appropriate field level
assembly.
during the travel .

Given the above, Figure l2b shows the maximum E-field

value for each (y, z) tag position that could be found when scan­
ning the x coordinate in the interval [-50 cm, 50 cm]. The results
showed that Ez was stil l the dominant leaking electric-field -40

component, but it now extended well above the belt's surface. 'E-20
.£ 0
However, the Ex and E field components were also enhanced,
y N 20
although they decreased with height faster than did Ez. All field 40
-50 0 50
components were reasonably confined to the sheWs near-zone, X [cm)
preventing detection of adj acent conveyor belts without the need E
z
for isolation barriers. -40
'E-20
.£ 0
50 N 20
40
45 -E simulated
y -50 0 50
40 -E simulated x [cm) X [cm)
z
35 • E
y
measured

E 30 • E measured Figure 12a. The simulated empty conveyor belt near-field

:> 25
z
components (Ex. Ey• and Ez) and the total near-field (E/)
!g 20 amplitude at 866 MHz in the xz plane.
15 . .
10 E Ey
x dB VIm
5
0 30
0 2 4 6 8 10 16 18 20 25
y[cm] E E 20
15
.2- .2-
50 >- >-
10
45 5
40 o
35
E 30 dB VIm
:> 25 30
25
!g 20 E 20
15 . . .2- 15
>-
10 10
5
5
o
9) 2 4 6 8 10 12 14 16 18 20
y[cm]
Figure 12b. The simulated empty conveyor belt cumulative
Figure 10. The measured and simulated near-field components maximum of the E field in the yz plane for a-50 cm < x < 50 cm
at the planes x = O. x = 3 . scan.

44 IEEE Antennas and Propagation Magazine, Vol. 53, No.2, April 2011
 were very similar to those found for the conveyor belt, reasonably
confined to the reader' s near-zone. The yz plane simulations (not
shown) anticipated tag detection up to 1 5 cm above the structure.
Figure 17 shows the fabricated prototype, wherein the meandered
microstrip line was encased in a shallow acrylic box.

4. RFID UHF Tag-Reading Score

To evaluate the effectiveness of the proposed structures,

experimental tag-reading-score tests were performed with the setup
shown in Figure 1 8a. The ALI EN ALR-8800 reader [22] (Fig­
ure 18b) was used to feed I W to all the proposed smart devices
Figure 13. The fabricated conveyor-belt prototype with through a routing switch, as indicated in Figure I . ALI EN native
embedded meandered microstrip line (located beneath the belt software was used to determine the tag-reading scores. Commer­
in the region bounded by the dashed line). cial UHF tags were selected for the test: A L I EN ALN-9540
"Squiggle" GEN 2 tags [22] (Figure 1 8c). The size of this tag was
9.7 cm x 1. 1 cm, the polarization was linear, and it operated from

o
-2
beasured
=::m ulated
860 MHz up to 920 MHz.

-4 We next present reading scores for each of the developed

-6 devices. Videos with experimental demonstrations of these devices
� -8r can be found in [ 1 2].
:::: -10 ---------- ....
......
....

Vl;::: -12---­
- - -_ -
-.... _--.....

-14
- -.... _----_ ...............

-16
-18
-tP.S2 0.84 0.86 0.88 0:9 0.92 0.94 0.96 0.98
Freq. (GHz)
load
Figure 14. The measured and simulated input reflection coeffi­
cient amplitudes of the conveyor-belt radiating structure.
feed
The fabricated prototype is shown in Figure 13. A thin
acrylic layer was glued on the top face of the microstrip assembly
to favor smooth sliding of the belt over it. Neither the acrylic nor
the belt layers affected the antenna 's performance.
ground
.
The input return loss of the conveyor-belt prototype was plane ,
measured, and the results agreed well with the meandered-line
< :>
simulations (Figure 14). It was seen that the device could operate 27.5 em
simultaneously with any of the three world-assigned RFID sub­
bands at UHF (bounded by the two vertical dashed lines). Tag­ Figure 15. The RFID isolated point-reader configuration.
reading scores for this structure are presented below.

Ey dBV/m
30
3.3 RFID Smart Point Reader ·20 25
20
E
K 0 .£ 15
Large retail shops and stores usually provide individual opti­ N N 10
cal barcode readers to enable users to self-check the prices of 20 5
0
items. It is possible to extend this concept for RFID-tagged items. ·20 0 20 ·20 0 20
X [cm] X [cm]
For this application, the RFID reader must be able to read tags with
E E/ dBV/m
any orientation, while confining the detection to avoid unwanted z
30
reading of neighboring tags. Desirably, it should allow simultane­
·20 25
ously reading more than one item held by the user in a single scan . 20
E
A point reader can be easily obtained by appropriately reducing the K 0 .£ 15
total length of the meandered microstrip line and the width of the N N 10
20 5
ground plane, as shown in Figure 1 5 . A demonstration example
0
was designed and optimized using a I cm-thick Styrofoam sub­ ·20 0 20
x [cm]
strate and a 5 cm-wide microstrip line to obtain a 50 n character­
istic impedance. The overall dimensions were 27.5 cm x 22 cm.
Figure 16. The simulated empty point reader's xz-plane near­
The near-field distribution at the isolated point-reader's sur­ field components (Ex' Ey' and Ez) and total near-field (E1)
face is shown in Figure 16. The amplitudes of all field components amplitude at 866 MHz.

IEEE Antennas and Propagation Magazine, Vol. 53, No.2, April 2011 45
 Figure 18a. The testing environment.

Figure 17. The fabricated point reader prototype (black box)

with embedded meandered microstrip line.

4.1 RFID Smart Shelf

Three different score tests were performed for the smart-shelf

structure [9]. In one test, the reading score was evaluated for a
standalone tag that was positioned successively at different uni­
formly di stributed locations in the volume above the shelf. The
second test was performed in a similar way, but with the tag
attached to a book. In the third test, multiple books with attached
tags were ranged across the shel f, to evaluate the simultaneous
reading score.
Figure 18b. The ALIEN ALR-8800 reader.

Figure 1 9 shows experimental reading scores for the first

type of tests, with the tags oriented along the z-axis above the
shelf. A large number of tag interrogations was performed auto­
matically for each tag position, and the corresponding percentage
of detections was recorded. F i l l ed circles indicate the positions
where the tag exhibited 100% detection, while the cross marker, x ,
indicates the positions without detection. Each dot corresponds to
the center of the tag. Figure 1 9 showed full tag detection over the
shelf with good lateral isolation. The height above the sheWs sur­
face where the tag could be detected was controlled by adj usting
the RF signal's attenuation on the reader. With a full 1 W of feed
power, a 100% detection score was obtained up to 7 em above the
sheWs surface, without undesired detection of tags placed below
the interrogated shelf or above the adj acent upper shelf. The 7 em
maximum detection height is sufficient for the bookshelf applica­
tion . A lthough the Ez component' s intensity was very low at the
center of the microstrip l ine, thi s effect was very local ized at z = 0 ,
and did not deter rel i able tag detection . I t i s mentioned for refer­
ence that tags were detected where simulations predicted field lev­
els above 20 d8V/m. Figure 18c. The ALIEN ALN-9S40 "squiggle" tag.

46 IEEE Antennas and Propagation Magazine, Vol. 53, No.2, April 2011
 Tag
orientati,!" pOSItIOns where the tag reading was intermittent Experimental
results (not shown) also confirmed detection for x-axis tag orienta­
tion (Ex component in Figure 5b) above the microstrip line, where
the other electric-field components were also significant

Thus, books must be preferably placed in the upright position

on the shelves, with the respective tags attached to the bottom of
the cover along the z axis_ Nonetheless, good flexibility is allowed
(x) Undetected for tag orientation and for its height in the book, without compro­
'Iem) mising reading score. This flexibility is attractive for practical use
of the proposed solution. Similar tests were performed for multiple
Figure 19. The measured detection score of a standalone tag books, either arranged in clusters or fully loading the shelf. The
oriented along the z axis which scanned the volume above the previously presented conclusions for the single book at different
shelf without books. positions on the shelf held equal ly for these cases_

4.2 RFID Smart Conveyor Belt

�
, f � f
'
� -r 1 -·"'-i
....... -_x__
�
3O'lJ
20
A slightly different approach was adopted for the validation
--- _ .

10 of the smart conveyor belt's performance [ 1 0]. As previously

--.;;;,:r.t;r�;:f�:7Sf-1-��f1J -1�
-

(.) Detected
-20 -10 0
(x) Undetected 110 70 -30
-20 ��
IV
explained, fully reliable detection within a reasonable height above
the belt requires that the tagged items slide over the device. Several
tags were thus uniformly distributed in one transverse plane above
x [cmJ
the belt The cumulative reading score at UHF was then registered
Figure 20a. The measured detection score of a tag placed inside for each tag (Yn' Zn) as it slid along x using the conveyor's move­
a book at different positions in the volume above the shelf, with ment. Several complete x sweeps along the belt's length were per­
the tag oriented along the z axis. formed, in order to verify repeatability of the results.

Results obtained in this way are represented in Figure 2 1 .

Filled circles correspond to 1 00% tag detection; unfi lled circles
correspond to tags detected only for some of the sweeps: intermit­
tent readings. The cross marker, " x ", indicates failed tag detection
for all the x sweeps. The color and size of the fi lled circles gives an
additional measure of the detection redundancy (larger circles cor­
respond to a higher number of times that the same tag was detected
by the reader in a single x sweep)_ The results demonstrated the
-10 0 10 20 30 effectiveness of tag detection, and the good detection confinement
(.) Detected (x) Undetected

x [cm] to a smal l volume above the conveyor-belt structure, as predicted

(0\ Intenllinent
Figure 20b. The measured detection score of a tag placed inside
a book at different positions in the volume above the shelf, with
the tag oriented along the y axis.
In the second type of reading-score tests, the tag was attached
to the inner side of the back cover of a book, near its bottom part
The book was placed upright at different positions above the shelf,
with the tag parallel to the z axis. This was the preferred orienta­
tion to take advantage of the predominant and well-confined Ez
field component Results are shown in Figure 20a. Each dot corre­
sponds to the position of the center of the book. As before, fi l led
circles indicate the positions where the tag exhibited 1 00% detec­
tion, while the cross marker, x, indicates the positions without
detection_
Book detection was reasonably contained within the limits of
the shelf for z-axis tag orientation. However, experiments showed
that a near- I OO% detection score was still possible when the tag
inside the book was rotated in the yz plane. Even for the extreme y­
axis tag orientation inside the book, detection was always possible
over most of the shelf's surface. This was true except in a narrow
strip located about 7 cm apart from the center of the shelf (marked
with unfilled circles in Figure 20b), where the Ey amplitude was
low and tag reading was unrel iable. Unfilled circles thus indicate
by simulations. Again, tags were detected where simulations pre­
dicted field levels above 20 dBV/m. The maximum detection
height with 1 00% score for all tag orientations and positions above
the belt's surface was 1 5 cm, although the detection height reached
30 cm at certain spots for x- and y-oriented tags, or 1 00 cm for z­
oriented tags. Fine control of the maximum detection height can be
achieved through the reader' s output power contro l .
Tests were repeated with multiple tagged objects, placed iso­
lated, in clusters, stacked, etc_ Simi lar results were obtained, with a
1 00% reading score insured for items placed up to 1 5 cm above the
conveyor belt for al l orientations, not deterring detections at higher
item positions for certain tag orientations. Lateral confinement of
the detection was perfectly compatible with normal distances
between paral lel cashiers in retail shops and supermarkets_ This
already effective solution can be further redesigned to improve the
detection range and to enhance the less-intense field components,
if required for other applications.
4.3 RFID Smart Point Reader
Tag-reading tests were also performed with the isolated point
reader. Since the tagged items are manually scanned in front of the
device by the user, the tag reading score study as a function of the
tag's orientation was not relevant The user intuitively adj usts the

IEEE Antennas and Propagation Magazine, Vol. 53, No.2, April 2011 47
 the books and tags in the usual preferred orientation. However,
x-orientation y-orientation
100 XXXXXXXXXXXXXXXXXXX 100 XXXXXXXXXXXXXXXXXXX other orientations were also all owed, which makes the proposed
solution adequate for practical application to retail stores. For the
90 90
conveyor-belt system, a 1 00% reading score was also achieved for
� XXXXXXXXXXXXXXXXXXX � XXXXX�XXXXXXXXXXXXX tagged items placed with arbitrary orientations up to 1 5 cm above
the belt's surface. Lateral confinement of the fields was very
70 70
effective for all tag orientations, which is appropriate for close par­
� XXXXX�XXXXXXX�XXXXX 00 xxxxx�xxxxxxoxxx.xx allel cashier conveyor operation in scenarios such as supermarkets
E without cross-reading. Experimental tests of the point reader (not
£ 50 xxxx.xx.xxxxxixxxxx 50 xxxx.�xxxxxx.ixo.xx
. .
>- shown here, but availabl e in [ 1 2]) also confirmed the predicted
� oxxx.�xoxxxx.xxxxxx � .xxx.�x.xxxx.i •••xx performance, allowing simultaneous reading of more than one
scanned obj ect.

� xxxx.xo ••••x.xxxxxx � xxxx.� •••••x.� •••xx The straight and meandered microstrip-Iine structures pre­
xxxx.� •••••••xxxxxx .xxo •••••••x •••••xx
sented can be redesigned, without frequency restriction, to adjust
10 xxxx.� •••••••xxxxxx W ox •••� ••••••• � •••xx
•xoo •••••••••oo.oxx for different device sizes, materials, and coverage requirements .
$0 -� 0 � 50 $O---- ����0��2�0---;'50 The transmission-line circuit can be embedded on the structures, as
Z[cm) Z [cm) demonstrated in this paper, or can be just assembled in a self-con­
tained layered structure, to be laid on existing store furniture, to
reduce the cost of the technology change. The very interesting
z-orientation
100 .x ••••••••••••••••x point to stress is the simplicity of the radiating structure and its
mild fabrication tolerances, requiring no sophisticated fabrication
90
environment. For instance, the microstrip line can be implemented
� .x ••••••••••••••••x Detected: with metallized adhesive tape. The concept is thus adequate for
extremely low-cost mass production at non-specialized facilities.
70
11 ID 15 counts
In fact, a pilot experience has started in the sequence of the present
•
00 ••••••••••••••••••x 61D 10 counts work, where the radiating structure is being fabricated and inte­
E • grated directly into a store furniture factory. A patent was filed
£ 50 ••••••••••••••••••x
lID 5 counts [ II].
>-
•
� ••••••••••••••••••x
Intermittent: Even though the presented work is focused on shelves, con­
� ••••••••••••••••••x 0 veyor belts, and point readers, the concept can been readily
� .o ••••••••••••••••x Undetected: extended to other applications where a tight detection-volume con­
.x •••••••••••••••xx X finement at UHF is required. It can replace some functions that are
10 .x ••••••••••••••oxx presently performed with manual optical-barcode readers, enabling
ox ••••••••••••• o ••x
$O�---��--�O---��--�50 more-efficient complete RFI D-based solutions in a retail shop
Z[cm) environment.

Figure 2 1. The experimental reading scores of tags positioned

at each represented point after x-axis sliding over the belt. The
filled circles represent full detection. The filling color repre­
sents the detection redundancy.
items' orientation for best detection_ The maximum detection
range achieved experimental ly for up to three simultaneously
scanned tagged books was about 1 5-20 cm. This value insures
good isolation with respect to neighboring smart furniture.
5. Conclusions
6. Acknowledgments
The authors would like to thank Creativesystems for discus­
sions about the motivation and system specifications, and for pro­
viding the commercial RFI D tags and reader equipment. The
authors also thank V. Fred and C. Brito for prototype fabrication,
A. Almeida for prototype measurements, and E. B. Lima for
designing the "RFID Store Manager" software for system control ,
data logging, and information processing. The work was partly
funded by Funda9ao para a Ciencia e Tecnologia, through proj ect
RFID-Local PTDC/EEA-TEllI023 90/2008.

This paper has presented new low-cost U H F RFI D radiating

structures that can be used with existing commercial readers to
enable reliable tag detection in selective self-confined interrogation
volumes. The proposed approach uses the leaking fields from
shaped microstrip transmission lines in a controlled way. It was
demonstrated that good isolation was achieved with respect to non­
desired adjacent regions without the need for physical barriers.
Prototypes of shelves, a cashier conveyor belt, and a point
reader were fabricated and tested together, emulating a retail shop
environment. Test results using a commercial reader and commer­
cial tags demonstrated a very good match between the experimen­
tal reading score and the simulated near-field distribution. For the
smart-shelf prototypes, a 1 00% reading score was achieved with
7. References
I. V. Chawla and Dong Sam Ha, "An Overview of Passive RFlD,"
IEEE Communication Magazine, 4, September 2007, pp. 1 1 -/7.
2. K. Finkenzeller, RFID Handbook, New York, Wiley & Son,
Inc., 2000.
3. J. Hightower, R. Want, and G. Borriello, "SpotON : An Indoor
3D Location Sensing Technology Based on RF Signal Strength,"
University of Washington, Department of Computer Science and
Engineering, WA, Seattle, Technical Report, February, 2000.

48 IEEE Antennas and Propagation Magazine, Vol. 53, No.2, April 2011
4. L. M . Ni, Y. Liu, Y. C . Lau, and A. P. Patil , "LANDMARC : 1 8 . Z.-M. Liu and R. R. H i l l egass, "A 3 Patch Near Field Antenna
Indoor Location Sensing Using Active RFID," 1 st IEEE Interna­ tor Conveyor Bottom Read in RFID Sortation Application," I E E E
tional Conference on Pervasive Computing and Communications International Symposium o n Antennas and Propagation Digest,
(PerCom '03) Digest, Texas, U SA, March 2003 , pp. 407-4 1 5 . July 2006, pp. 1 043 - 1 046.

5 . P . Bahl and V . N . Padmanabhan, "RADAR: An Inbuilding RF­ 1 9. D. M. Pozar, Microwave Engineering, Second Edition, New
Based User Location and Tracking System," 1 9th Annual Joint York, John Wiley & Sons, Inc . , 1 99 8 .
Conference of the I EEE Computer and Communications Societies
(IEEE INFOCOM 2000) Digest, 2, Tel Aviv, I srael, March, 2000, 20. CST Computer Simulation Technology, http://www.cst.com/.
pp. 775-784.
2 1 . S . Ramo, 1. Whinnery, and T. Van Duzer, Fields and Waves in
6. K. Rao, P. Nikitin, and S . Lam, "Antenna Design for UHF RFID Communication Electronics, Third Edition, New York, John Wiley
Tags: A Review and a Practical Application," IEEE Transactions & Sons, Inc., 1 994, p. 4 1 2 .
on Antennas and Propagation, AP-53, 1 2 , December, 2005 , pp.
3 870-3 876. 22. Skyetek Embedded RFID Readers, avai lable at
http ://www. skyetek.com/. December, 2009.
7. M . Lieshout, L. Grossi, G . Spinelli, S . Helmus, L. Kool, L.
Pennings, R. Stap, T. Veugen, B. der Waaij , and C . Borean, "RFI D 23. ALIEN Technology Web site,
Technologies: Emerging I ssues, Challenges and Policy Options," http://www . alientechnology.com/. December, 2009.
Institute for Prospective Technological Studies, European Com­
mission, Tech . Report, EUR 22770 EN, 2007.

8. W. Choi, "Near-Field Antenna for RFID Smart Shelf in UHF," Introducing the Feature Article Auth ors
I EEE International Symposium on Antennas and Propagation, June
2009.

9. C . R. Medeiros, 1 . R. Costa, and C. A. Fernandes, "RFI D Smart

Shelf with Confined Detection Volume at UHF," IEEE A ntennas
and Wireless Propagation Letters, 7, October, 2008, pp. 773-776.

1 0. C . R. Medeiros, J. R . Costa, and C . A . Fernandes, "UHF RFI D

Smart Conveyor B e l t with Confined Detection Range," I EEE
International Symposium on Antennas and Propagation, June
2009.

I I. C . R. Medeiros, 1 . R. Costa, and C . A. Fernandes; PT 1 04 1 2 1 ,
June 2008 and PCT 200900003 5 (pending).
1 2 . "UHF-RF I D Self-Confined Volume Detection," available at
http ://www.iLptiauto_temp_webyageyreview. asp?id=666,
December, 2009.
1 3 . X . Qing and Z. N . Chen, "Proximity Effects of Metallic Envi­
ronments on High Frequency RFID Reader Antenna: Study and
App l ications," IEEE Transactions on Antennas and Propagation,
AP-55, 1 1 , November, 2007, pp. 3 1 05-3 1 1 1 .
Carla R. Medeiros was born in Ponta Delgada, Ayores,
Portugal , in 1 982. She received the Licenciado and MSc degrees in
Electrical and Computer Engineering from the Instituto Superior
Tecnico (1ST), Technical University of Lisbon, Lisbon, Portugal,
in 2006 and 2007, respectively. Since 2006, she has been a
researcher at the Instituto de Telecomunicayoes (IT), focusing her
work on antennas for wireless communications. She collaborates in
national research proj ects. Her current research interests are in the
areas of reconfigurable, RFI D , and UWB antennas.

1 4. A. Cai, X. M. Qing, and Z. N. Chen, "H igh Frequency RFI D

Smart Table Antenna," Microwave and Optical Technology Let­
ters, 49, 9, September, 2007, pp. 2074-2076.

1 5 . X . Qing and Z. N. Chen, "Characteristics of a Metal-Backed

Loop Antenna and its Application to a High-Frequency RFI D
Smart Shelf," IEEE Antennas and Propagation Magazine, 5 1 , 2 ,
April 2009, p p . 26-3 8 .

1 6. C . Cho, J. Ryoo, I . Park, and H . Choo, "Design of a n RFI D
Reader Antenna for Near-Field Communications Using Opposite­
directed Currents," Asia Pacific M icrowave Conference Digest,
Hong Kong and Macau, China, December, 2008.
1 7 . 1 . M . Lee, N . S . Kim and C . S . Pyo, "A Circular Polarized
Metallic Patch Antenna for RFID Reader," Asia-Pacific Confer­
ence on Communications Digest, Australia, October, 200 5 , pp.
1 1 6- 1 1 8 .
Jorge R. Costa was born in Lisbon, Portugal , in 1 974. He
received the Licenciado and PhD degrees in Electrical and Com­
puter Engineering from the Instituto Superior Tecni co (1ST),
Technical University of Lisbon, Lisbon, Portugal, in 1 997 and
2002, respectively.
He is currently a Researcher at the Instituto de
Telecomunicayoes, Lisbon, Portugal . He is also an Assistant Pro­
fessor at the Departamento de Ciencias e Tecnologias da Informa-

IEEE Antennas and Propagation Magazine, Vol. 53, No.2, April 2011 49
9ao, Instituto Superior de Ciencias do Trabalho e da Empresa. H i s
present research interests include lenses, reconfigurable antennas,
M E M S switches, and U W B , and RFI D antennas. He i s the coau­
thor of four patent appl ications, and more than 5 0 contributions to
peer-reviewed j ournals and international conference proceedings.
More than ten of these papers have appeared in IEEE j ournals.
Changes of Address
or Delivery Problems
Information regarding subscriptions and addresses is man­
aged by I EE E headquarters. It i s not maintained, nor can it be
changed, by any member of the Magazine staff. I f you are a mem­
ber of the I EEE, your subscription i s sent to the address in your
I EEE member record. Your address can be confirmed or updated
by visiting the Web page dealing with delivery of I E E E publica­
tions:

http ://www . ieee. org/aboutlhelp/subscriptions. html

Carlos A. Fernandes received the Licenciado, MSc, and
PhD degrees in Electrical and Computer Engineering from Insti­
tuto Superior Tecnico (1 ST), Technical University of Lisbon,
Lisbon, Portugal, in 1 980, 1 98 5 , and 1 990, respectively. He j oined
the 1ST in 1 980, where he i s presently a full Professor in the
Department of Electrical and Computer Engineering in the areas of
microwaves, radiowave propagation, and antennas. He is a senior
researcher at the Instituto de Telecomunica90es and coordinator of
its Wireless Communications scientific area. He has been the
leader of antenna activities in National and European Proj ects,
such as RAC E 2067 - M B S (Mobile Broadband System), ACTS
AC23 0 - SAMBA ( System for Advanced Mobil e Broadband
Applications), and ESA/ESTEC - l LASH (Integrated Lens
Antenna Shaping). He has coauthored a book, a book chapter, and
more than 1 00 technical papers in international j ournals and con­
ference proceedings, in the areas of antennas and radiowave
propagation model ing. His current research interests include
dielectric antennas for m i l l imeter-wave applications, antennas and
propagation modeling for personal-communication systems, RFi D
antennas, artificial dielectrics, and metamaterials. EID
This page also has information about publ ication delivery, and a
l ink to an online form that can be used to inquire about missing or
delayed publications:
http ://www. ieee.org/about/help/publicationjnquiry.html
You can also update your address information by contacting
IEEE headquarters: M ember Address Records, I EEE
Headquarters, 445 Hoes Lane, Piscataway NJ 0 8 8 5 5 - 1 3 3 1 USA;
Tel : + 1 (908) 98 1 -0060 or + 1 (800) 678-43 3 3 ; Fax: + 1 (908) 98 1 -
9667; E-mai l : address.change@ieee.org. I f you are an institutional
or other non-member subscriber, contact IEEE Customer Service
at the above address, telephone, and fax numbers; E-mai l :
customer. service@ieee.org.
Please do not send requests related to the above items to any
member of the Magazine Staff.

Getting the Magazine

by Air Freig ht
Air freight delivery of the IEEE Antennas and Propagation
Magazine is avai lable to subscribers in I EEE Regions 8- 1 0. The
cost was recently $44 . 00 per year. Thi s can be added to a
member ' s subscription (and the current cost verified) via the Web
at http ://www . ieee.org/web/aboutus/help/member support.html .
_

50 IEEE Antennas and Propagation Magazine, Vol. 53, No.2, April 2011