-
Fast splitting based tag identification algorithm for anticollision in UHF RFID System
Article (Accepted Version)
http://sro.sussex.ac.uk
Jian, Su, Sheng, Zhengguo, Xie, Liangbo, Li, Gang and Liu, Alex
X (2018) Fast splitting based tag identification algorithm for
anti-collision in UHF RFID System. IEEE Transactions on
Communications. ISSN 0090-6778
This version is available from Sussex Research Online:
http://sro.sussex.ac.uk/id/eprint/80427/
This document is made available in accordance with publisher
policies and may differ from the published version or from the
version of record. If you wish to cite this item you are advised to
consult the publisher’s version. Please see the URL above for
details on accessing the published version.
Copyright and reuse: Sussex Research Online is a digital
repository of the research output of the University.
Copyright and all moral rights to the version of the paper
presented here belong to the individual author(s) and/or other
copyright owners. To the extent reasonable and practicable, the
material made available in SRO has been checked for eligibility
before being made available.
Copies of full text items generally can be reproduced, displayed
or performed and given to third parties in any format or medium for
personal research or study, educational, or not-for-profit purposes
without prior permission or charge, provided that the authors,
title and full bibliographic details are credited, a hyperlink
and/or URL is given for the original metadata page and the content
is not changed in any way.
http://sro.sussex.ac.uk/
-
IEEE TRANSACTIONS ON COMMUNICATIONS 1
Fast Splitting Based Tag Identification AlgorithmFor
Anti-collision in UHF RFID System
Jian Su, Zhengguo Sheng, Liangbo Xie, and Gang Li
Abstract—Efficient and effective objects identification
usingRadio Frequency Identification (RFID) is always a challengein
large scale industrial and commercial applications. Amongexisting
solutions, the tree based splitting scheme has attractedincreasing
attention because of its high extendibility and feasibil-ity.
However, conventional tree splitting algorithms can only solvetag
collision with counter value equals to zero and usually resultin
performance degradation when the number of tags is large.To
overcome such drawbacks, we propose a novel tree-basedmethod called
Fast Splitting Algorithm based on ConsecutiveSlot Status detection
(FSA-CSS), which includes a fast splitting(FS) mechanism and a
shrink mechanism. Specifically, the FSmechanism is used to reduce
collisions by increasing commandswhen the number of consecutive
collision is above a threshold.Whereas the shrink mechanism is used
to reduce extra idleslots introduced by FS. Simulation results
supplemented byprototyping tests show that the proposed FSA-CSS
achieves asystem throughput of 0.41, outperforming the existing UHF
RFIDsolutions.
Index Terms—RFID, UHF, anti-collision, FS mechanism, sys-tem
throughput, time efficiency.
I. INTRODUCTION
RFID technology has greatly revolutionized tag based
appli-cations in retail industry such as warehouse management
andinventory control [1-2]. A typical RFID system is composedof a
reader and a large number of tags attached to trackedobjects. Each
object can be identified based on query responsefrom the attached
RFID tag. However, in large scale RFIDapplications, simultaneous
query responses from multiple tagscan cause significant tag
collisions. Since passive tags areunable to perceive or identify
such collisions, the developmentof anti-collision algorithms is of
great importance for fast tagidentification especially in
high-density ultra high frequency(UHF) RFID environment.
Existing tag anti-collision algorithms can be categorizedinto
dynamic framed slotted Aloha (DFSA)[3-5], query tree(QT)[6-8] and
binary splitting (BS) based [9-10] algorithms.Among them, QT and BS
algorithms are operated by recur-sively dividing responding tags
into smaller subsets until eachsubset has at most one tag. The
distinction between these twoapproaches is that, in QT solutions,
tags are separated by their
J. Su is with Nanjing University of Information Science and
Technology,Jiangsu 210044, China (e-mail: [email protected]).
Z. Sheng (corresponding author) is with the Department of
Engineer-ing and Design, University of Sussex, Brighton BN1 9RH, UK
(e-mail:[email protected]).
L. Xie is with Chongqing University of Posts and
Telecommunications,Chongqing 400065, China (e-mail:
[email protected]).
G. Li is with University of Electronic Science and Technology of
China,Chengdu 611731, China (e-mail: [email protected]).
Digital Object Identifier xxxx
IDs. Whereas in BS approaches, tags are divided by binaryrandom
numbers generated in the splitting process. Strictlyspeaking, QT
methods are derived from the bit trackingtechnology [11] which can
detect the position of collided bitby the reader. However, with an
increasing number of tags, theefficiency of such methods is
deteriorated because of the widedeviation of backscatter link
frequency among tags [12-14].
As a contrary, DFSA and BS algorithms are more preferablefor UHF
RFID systems. DFSA algorithms usually employa frame structure which
contains a certain number of timeintervals (called time slots) per
frame, and each tag responsesto the reader by randomly choosing a
time slot using itsID. During the identification process, the size
of frame isdynamically updated according to the number of unread
tags.When the frame size is equal to the backlog (number of
unreadtags), the maximum system throughput can be achieved.Recent
works in DFSA include improved linearized combi-natorial model
(ILCM) [15] based anti-collision algorithm,an efficient
anti-collision algorithm with early adjustment offrame length
(EAAEA) [16], and access probability adjustmentbased fine-grained
Q-algorithm (APAFQ) [17], etc. However,those algorithms have failed
to prevent collisions completely,because of the tag starvation
problem in which a specifictag may not be identified for a long
time [9]. Furthermore,the performance of DFSA algorithms highly
depends on theinitial frame size. When the number of tags is much
largerthan the size of frame, most of DFSA solutions are unable
toadjust frame size properly in order to cope with backlog,
thuslead to performance degradation [3, 15]. That is to say,
DFSAsolutions are cardinality sensitive and shows inconsistent
per-formance with a wide range of backlog. As a contrary,
BSalgorithms are insensitive to tag backlog particularly when
thenumber of tags is increased, the system throughput is
almostconverged to a constant value. Although the BS approach
cantackle tag starvation, it has a relatively long
identificationlatency due to the splitting process starting from a
singleset with all tags. Specially, the system throughput of
BSalgorithms is about 0.348 [10] when the number of tags islarger
than 100. Most of recent anti-collision algorithms withhigh
performance are based on the integration of Aloha andQT (or BS)
algorithms [18-19]. These methods usually needto estimate
cardinality of tag population. Many efforts havealso been made to
improve estimation accuracy [3-4][15][20].However, constant
estimation of number of tages with highaccuracy requires high
computation overhead and thus leadsto serious challenges in
implementation [16].
In this paper, we focus on the UHF RFID anti-collisionalgorithm
and propose a fast splitting algorithm based on
-
IEEE TRANSACTIONS ON COMMUNICATIONS 2
consecutive slot status detection (FSA-CSS) to improve
theidentification and stability performance of BS algorithm.
Theproposed solution is based on the pure BS algorithm whichis
rarely investigated given concerns of marginal
performanceimprovement by the existing literature. Different to the
existingsolutions in which the average performance is only
inves-tigated, i.e., incoming and outgoing tags are allowed to
beidentified multiple times [21-22], we focus on a
practice-drivenand challenging scenario in which a tag can only be
identifiedonce and propose to improve individual performance
foridentifying a batch of tags. It shows from the results that
FSA-CSS can still maintain high performance in large-scale
RFIDsystems without estimation of backlog. The contributions ofthis
paper can be summarized as follows.
1) We propose an enhanced anti-collision algorithm namelyFSA-CSS
for passive UHF RFID systems. In order to ac-celerate the splitting
process, the reader allows the tags withcounter value above zero to
be split into subsets in advance.Meanwhile, the reader can avoid
over-splitting by using theshrink mechanism to reduce the extra
idle slots.
2) We provide the theoretical analysis and carry out
simula-tions to evaluate the performance of FSA-CSS with a
massivenumber of tags. The results have been compared with
variouslegacy anti-collision algorithms. The optimal parameters
andperformance boundaries have been derived.
3) We implement FSA-CSS in a practical UHF RFIDsystem, which
includes a passive RFID reader and 20 tags.The identification time
of the proposed FSA-CSS is reduced by32.5% compared to the standard
BS used in ISO/IEC 18000-6B.
The rest of this paper is organized as follows. SectionII
reviews and analyzes the mainstream tag identificationstrategies
for UHF RFID systems. Section III discusses thenovel anti-collision
algorithm FSA-CSS and analyzes its per-formance. Section IV
illustrates the simulation results. Theexperimental results are
presented in Section V. Finally, thepaper is concluded in Section
VI.
II. RELATED WORKSA. Binary splitting (BS) algorithm
The principle of BS algorithm is to continuously dividecollided
tags into smaller subsets by using random binarynumbers. Each tag
has an integer counter Tc which is usedto record its position in
the splitting process and a randombinary number generator TR.
Different values of Tc lead tagsinto different subsets. Only tags
with Tc = 0 respond tothe reader immediately, while tags with
Tc>0 should waitin the pipeline. At the initial stage of
identification process,all tags in reader’s vicinity should respond
simultaneouslyand thus are formed as one set. Depending on
receivedresponses, the reader will send a feedback, e.g.,
ID-idle,ID-collision or ID-success, to all tags for furtheractions.
When the feedback is ID-success or ID-idle,all tags act Tc = Tc −
1. The tags already identified by thereader will be silent during
the next identification process. Ifthe feedback is ID-collision,
the collided tags will bedivided into two subsets. Tags with Tc = 0
add TR to its Tc,while tags with Tc>0 increase its Tc by 1.
The reader also uses a counter Rc to terminate the
identi-fication process. Rc denotes the number of unidentified
tagsand its initial value is 0. When a collision occurs, the
readerperforms Rc = Rc + 1, given the number of identifiabletags is
increased. Otherwise, Rc is decreased by 1. WhenRc < 0, the
identification process is ceased. Compared toAloha-based
algorithms, the BS algorithm is insensitive to tagcardinality
particularly when the number of tags is increased.The system
throughput is almost converged to a constant value.The work in [10]
reveals that the system throughput of BS canbe maintained at 0.348
when the number of tags is above 100.However, the BS algorithm has
a relatively long identificationlatency because the splitting
process is started from a singleset with all tags. Moreover, it
always use tag ID to performcollision arbitration, hence there is
significant space to furtherimprove time efficiency.
B. DFSA protocol with backlog estimation
In DFSA protocols, time is divided into a series of
discreteintervals called time slots. Several time slots are
packagedinto a frame [3][15]. Each tag can randomly select a slot
torespond to the reader in each identification round. At the endof
each frame, the reader counts the number of idle, success,and
collision slots. If collision slots exist, the reader willestimate
the number of unidentified tags and adjust the framelength
accordingly. The identification process continues untilno collision
occurs during a frame, i.e., all tags are
successfullyidentified.
It is noted that the reader needs to accurately estimate
thebacklog to achieve the best performance. To improve estima-tion
accuracy, most previous methods [3-5] are implementedwith high
complexity. However, typical RFID readers arecomputation
constrained due to the limited processing powerprovided by
single-chip microprocessor. As a result, estimationmethods with
high computation overhead are characterized asenergy inefficient.
Recently, many state-of-the-art algorithms[15-16] have been
proposed to achieve energy efficiency.However, the system
throughput of these solutions is stillbelow 0.36. In addition, it
is worth mentioning that most ofexisting DFSA studies are
simulation based [3-5][15-17], thusthe practical performance cannot
be verified.
C. Hybrid protocol combining DFSA with BS
So far it has been shown that the BS performs better whenthe
number of tags is relatively small [10]. This result hasinspired
further studies to design hybrid anti-collision algo-rithms
combining DFSA and BS. The authors in [23] proposean adaptive
binary tree slotted Aloha (ABTSA) algorithm. InABTSA, tags are
randomly assigned to slots in a frame. Ifmultiple tags are collided
in a slot, collision will be solvedby the BS algorithm while other
tags in the following slotswill be in waiting state. The benefit of
using ABTSA isthat the adjustment of frame size is simplified
because thereader has the knowledge of the status in every slot
andthus can obtain an appropriate frame size accordingly. Sincethe
ABTSA combines both DFSA and BS algorithms, it canachieve an
average system throughput of 0.40, which is higher
-
IEEE TRANSACTIONS ON COMMUNICATIONS 3
than that of DFSA or BS algorithm alone. However, it iswith a
complex structure, hence much more difficult to beimplemented.
In summary, aforementioned anti-collision algorithms doprovide
solutions to solve the tag collision problem, how-ever, with
sacrifices in identification efficiency, complexityand stability,
etc. The characteristics of existing anti-collisionalgorithms are
summarized in Tab. I. In the following, we willintroduce the
proposed solution based on the BS algorithmto improve the
identification efficiency of RFID system andreduce the
implementation cost.
(a,b,c,d)
(a,c) (b,d)
(0) (1)
c a
(0) (1) (2) (3)
(a,b,c,d)
(a,c) (b,d)
c a d b
a d b
(a,d)
(0) (1)
(0) (1) (2) (2) (3)
(2)
(0) (1) (2) (3) (4)
(a,b,c,d)
(a,c) (b,d)
(0) (1)
(a,c) d b
c a d b
(0) (3) (4)
(0) (1) (2) (8) (9)
… …
(10) (12) (13)
(a) FS with M=2, N=2 (M=N)
(b) FS with M=2, N=3 (MN)
Collided Successful Idle
Solved by the commands from the peer node(It will not occupy a
dedicated time slot during identification)
(*) Tc
d b
Depth 0
Depth 1
Depth 2
Depth 0
Depth 1
Depth 2
Depth 3
Depth 0
Depth 1
Depth 2
Depth 3
Depth is defined as the number of braches from that node to the
root (top) node
Fig. 1. The FS mechanism with different M and N
III. THE PROPOSED FSA-CSS ALGORITHM
A. Fast splitting mechanism
In the fast splitting (FS) mechanism, the counter value Tcof all
tags are initialized to zero. When a collision occurs, thecollided
tags (i.e., tags with Tc = 0) perform Tc = Tc + TR,while the tags
with Tc>0 perform Tc =M ·Tc+TR. That is,
Tc =
{Tc + TR, if Tc = 0M · Tc + TR, if Tc > 0 .
(1)
where TR ∈ {0, 1, · · ·, N − 1} denotes a random binarynumber
generator at an arbitrary time, N determines themaximum subsets
needed. That is to say, the collided tags canbe potentially
allocated into N subsets. Tc denotes the valueof counter at an
arbitrary time. An integer M>0 is a splittingcoefficient, which
denotes the splitting level of collided tags.The larger the M , the
more available subset slots can beprovided during a single
splitting process.
In order to illustrate the FS mechanism, we describe asimple
case with five tags using M = N = 2. Assume thetags a, b, c, d, e
collide in the slot 0, given Tc = 0 is for alltags, the reader will
randomly divide them into two subsets,e.g., S1 = {a, d, e} with
random number 0 and S2 = {b, c}with random number 1. The FS
mechanism will trigger thereader to split S2 into {b} and {c} in
slot 1. According to (1),no collision will happen between b and c.
Meanwhile, tags a,d, and e will collide again in slot 1 and the FS
will continueto assist the splitting until all collisions are
solved. It is notedthat we consider the case where a tag can be
identified onlyonce, whereas tags are identified repeatedly can be
referred to[9][20-21].
By generalizing cases with N>2, we can derive the follow-ing
result for the splitting coefficient M .
Result 1. The performance of FS mechanism depends on thechoice
of M and leads to the following result. M > N, yield extra idle
slotsM = N, no extra idle and collision slots
M < N, yield extra collision slots(2)
Proof: See the Appendix A.According to the Result 1, M = N is
the best choice for FS
mechanism. Examples of four tags using FS mechanism
withdifferent M and N are illustrated in Fig. 1. The depth of anode
is defined as the path length (number of solid branches)from that
node to the root (top) node. As can be seen in Fig.1 (a), all tags
collided in slot 0 are further divided into twosubsets, e.g., S1 =
{a, c}, S2 = {b, d}. Since tags in S1 arecontinually collided in
slot 1, they act Tc = Tc + TR, whilethe tags in S2 act Tc = 2 · Tc
+ TR. After reading in slot1, all tags are divided into four
subsets. Each subset containsonly one tag. In total, the FSA-CCS
consumes six slots toidentify four tags by using FS mechanism with
M = N = 2.Fig. 1 (b) shows the example of MN . It thus indicates
that
-
IEEE TRANSACTIONS ON COMMUNICATIONS 4
TABLE ITHE CHARACTERISTICS OF DIFFERENT ANTI-COLLISION
ALGORITHMS
Method Category implementation
tagstarvationcardinalitysensitive
systemthroughput
identificationlantency complexityDFSA BS Hybrid
MAP [3]√
medium√ √
medium medium highMFML-DFSA [4]
√medium
√ √medium medium high
FuzzyQ [5]√
medium√ √
low medium mediumILCM [15]
√easy
√ √low low low
ds-DFSA [13]√
difficult√ √
high low lowBS [10]
√easy × × medium high low
ABS [9]√
easy × × medium high lowABTSA [23]
√difficult
√ √high high high
a large M may cause too many idle slots. In our
proposedalgorithm, we define such a result with a large number of
idleslots as over-splitting.
B. The proposed FSA-CSS algorithm
According to the analysis above, the FS mechanism canaccelerate
the splitting speed and reduce collision during theidentification
process. However, the question is how to applythe FS mechanism and
avoid over-splitting, given the numberof tags is unknown at the
beginning? To tackle the over-splitting problem, we propose the
fast splitting algorithmwith consecutive slot status detection
(FSA-CSS). The mainidea is that the reader implements the FS
mechanism onlywhen it detects a consecutive collision. The
consecutive col-lision indicates a large number of tags coexist in
the sameidentification process. In contrast, when the reader
detects aconsecutive idle, it performs the shrink mechanism by
sendinga Decrease command to decrease counter value of tags.
Theshrink mechanism is defined as follows.
Tc = trunk (Tc/2) . (3)
where trunk() is a truncation function. The flowchart of
theproposed FSA-CSS algorithm is illustrated in Fig. 2.
Twothresholds µ and σ are denoted as the upper tolerance limitof
the number of consecutive collision slots and idle
slots,respectively, and two counters TCCN and TCIN are used atthe
reader to count the number of consecutive collision slotsand idle
slots. If TCCN ≥ µ, the reader should perform theFS mechanism. If
TCIN ≥ σ, the reader should perform theshrink mechanism due to the
occurrence of over-splitting.
As can be observed in Fig. 2, the reader sends the
cor-responding feedback commands according to the slot type.Similar
to BS, the reader also has a counter Rc to terminatethe
identification process when Rc < 0. By receiving
feedbackcommands from the reader, each tag acts as follows.•
Splitting 0 command: Tags with Tc = 0 act Tc = TR,
while tags with Tc>0 act Tc = Tc+1. It is similar to
theID-collision feedback of the BS algorithm. Aftersending this
command, the reader acts Rc=Rc+1.
• Splitting 1 command: Tags with Tc = 1 act Tc = TR.Tags with
Tc>0 will be split into two groups. Thepotential advantage is to
reduce the collisions.
• Increase command: All tags act Tc = 2·Tc+TR. Besidethe current
collided tags, tags with Tc>0 will be dividedafter receiving
this command. This is essentially different
to the BS algorithm. After sending this command, thereader acts
Rc=2· Rc+1.
• Decrease command: All tags act Tc = trunk(Tc/2).This command
is allowed to alleviate the over-splitting.After sending this
command, the reader acts Rc =round(Rc/2).
• QueryRep command: All tags act Tc = Tc − 1. This issimilar to
the ID-idle and ID-success of the BSalgorithm. After sending this
command, the reader actsRc=Rc-1.
It is worth noting that the optimal M cannot be directly
de-rived by the algorithm. In this paper, we choose an
appropriatevalue M by using experimental method. Intuitively, a
larger Mcan provide more splitting subsets and hence reduce
collisions.However, it also introduces idle slots. In our proposed
solution,we choose M = N = 2 to best achieve the simplicity
andbalance between the performance and complexity of usingrandom
number generator by each tag. Additional analysisand discussion can
be found from Section IV and Fig. 3which compares the number of
slots consumed by FSA-CSSby adopting different M .
Fig. 4 illustrates an example of seven tags by implementingthe
proposed FSA-CSS algorithm. Assume µ = 3 and σ = 3,the reader
performs the FS mechanism in the slot 3 due toTCCN = 3. Then the
reader performs the BS in each tagsubset. After the tag c is
identified, three consecutive idleslots occur, the reader performs
the shrink mechanism in slot11 to decrease the counter value of
unidentified tags in orderto alleviate over-splitting. Thereafter,
the tags g, e and f areidentified in slot 13, slot 15 and slot 16,
respectively. Thedetailed communication procedure is also
illustrated in Tab.II. In essence, compared to the conventional BS
algorithm,the proposed FSA-CSS algorithm can reduce collisions
butintroduce extra idle slots. However, the FSA-CSS can use
theIncrease command to alleviate the negative impact from theidle
slots.
C. Upper bound performance of FSA-CSS
In this section, we analyze the system throughput of FSA-CSS.
The system throughput Tsys is defined as follows
Tsys =m
Nm. (4)
where m is the number of tags waiting to be identified in
thereader vicinity, Nm is the required total slots to identify them
tags. The system throughput is actually equivalent to the
-
IEEE TRANSACTIONS ON COMMUNICATIONS 5
Initialize identification process
Tags with Tc=0 respond to the reader
Response status TCCN
-
IEEE TRANSACTIONS ON COMMUNICATIONS 6
100 250 400 550 700 850 10000
500
1000
1500
2000
2500
3000
3500(a) Comparison of total slots
Number of Tags
Num
ber o
f tot
al sl
ots
M=2M=4M=6M=8M=10
100 250 400 550 700 850 10000
200
400
600
800
1000
1200
1400
1600
1800
2000(b) Comparison of idle slots
Number of Tags
Num
ber o
f idl
e sl
ots
M=2M=4M=6M=8M=10
100 250 400 550 700 850 10000
100
200
300
400
500
600
700
800
900(c) Comparison of collision slots
Number of Tags
Num
ber o
f col
lisio
n slo
ts
M=2M=4M=6M=8M=10
Fig. 3. Comparison of the number of slots under different M
assume that the reader does not perform the FS after the
L-thdepth, which is a reasonable assumption since the tags willbe
divided into 2L subsets at L-th depth of the tree and eachsubset
only contains a few tags, which reduces the chances ofconsecutive
collisions. Thus, the total slots to identify m tagsexpended by
FSA-CSS can be expressed as
NFSA−CSSm = L+ 2L · [P (0|m, L) + P (1|m, L)]
+∑
r>1 2L · P (r|m, L) ·NBSr .
(9)
where NBSr denotes the number of slots used for BS toidentify r
tags. L represents the number of collided slotsand involves L − µ
times FS operations. When the FS isfinished, the collided tags can
be divided into 2L groups.2L · P (0|m, L) and 2L · P (1|m, L)
represent the numberof idle nodes and success nodes,
respectively.
Result 2. The upper bound system throughput of FSA-CSScan be
approximated as following when the number of tags mtends to
infinity
limm→∞
TFSA−CSSsys ≈ 0.4521 . (10)
Proof: See the Appendix B.
D. Discussion of µ and σ
It is noted that Result 2 reveals the upper performanceof
FSA-CSS. However, the priori knowledge of number oftags is usually
unknown to the reader and the frame sizeis typically not equal to
the number of tags. In such cases,the FS and shrink mechanisms are
necessary for the FSA-CSS to reduce collisions and avoid
over-splitting during theentire identification process.
Specifically, µ is the critical valueto enable the FS mechanism, σ
is the key value to performthe shrink mechanism. The system
throughput depends on thesetting of these two parameters.
We use the numerical method to search the optimal com-bination
of µ and σ. Experiments are performed under thefollowing cases: the
number of tags are set as 200, 400, 600,and 800, respectively. µ
and σ vary from 1 to 10 in stepsof 1. As can be observed from Fig.
5, the optimal systemthroughput relies on the combination of µ and
σ. In Fig. 5(a), the maximum system throughput of FSA-CSS is
0.4049which can be obtained when µ = 4 and σ = 9. Similarly inFig.
5 (b), the system throughput of FSA-CSS peaks at 0.4128when µ = 4
and σ = 9. Fig. 5 (c) shows that the pair of µ = 4and σ = 7 can
achieve the maximum system throughput of0.4178. In Fig. 5(d), µ = 4
and σ = 7 can also achieve thebest performance at 0.4188 when n =
800. In general, noconstant parameter setting of µ and σ can
maintain the bestperformance as the number of tags varies in a
large scale. Theoptimal values of µ and σ rely on exhaustive search
usingcomputer simulations.
The system throughput of FSA-CSS under specific pairs ofµ and σ
when the number of tags varies from 100 to 1000 areillustrated in
Fig. 6. The average throughput of six curves fromhighest to lowest
are 0.4128, 0.4123, 0.4122, 0.4102, 0.4086,and 0.4083,
respectively. The corresponding pairs of µ and σare (4, 7), (4, 8),
(4, 6), (4, 9), (3, 4), and (3, 5), respectively.We can found that
no constant parameter setting can alwaysmaintain the best
performance. When the number of tagsvaries in a large-scale, the
expectation of system throughputof FSA-CSS is maximum at (µ = 4, σ
= 7). In practicalimplementation of fast RFID systems, it might be
too costlyto search the optimal combination of these two parameters
andkeep real-time updating by introducing extra overhead
strategy.Therefore, a default parameter setting is preferable
during thewhole identification process with recommendation of (µ =
4,σ = 7).
-
IEEE TRANSACTIONS ON COMMUNICATIONS 7
(a,b,c,d,e,f,g)
(a,b,c,d)
(e,f,g)
(a,b,c,d)
(e,f,g)
(a,b) (c,d) g (e,f)
b a d g fc
(0) (1)
(0) (2)
(0) (1) (4) (5)
(0) (1) (2) (9) (11)(3)
(0)
g (e,f)
(10)
(2)
e
f
(4)~(6)
Shrink mechanism (Tc=trunk(Tc/2))
FS mechanism (Tc=2·Tc+TR)
(1)
Collision Success Idle
Solved by the commands from the peer node(It will not occupy a
dedicated time slot during identification)
(*) Tc
e
(0) (1)
(Start from slot3 to slot4)
…
(Start from slot11 to slot 12)
The probe order (slot index) is from top to bottom, if a left
node is visited, the FSA-CSS moves horizontally right
Fig. 4. An identification example by using FSA-CSS algorithm
IV. SIMULATION RESULTS
In this section, we compare FSA-CSS with the
existingstate-of-the-art including ABS [9], APAFQ [17],
ds-DFSA[13], EAAEA [16], ILCM [15], and ABTSA [23] in termsof
system throughput and time efficiency. Simulations witha reader and
a various number of tags have been evaluatedusing MATLAB, where the
tags are uniformly distributed inthe reader vicinity in order to
receive reader command directly.The evaluations are mainly focused
on the MAC layer, whereasphysical layer effects such as radio
propagation effects arenot considered in the proposed model. The
similar assumptionhas been widely applied in the literatures [4,
10, 13-16]. Thesimulation results are average over 1000
iterations.
Fig. 7 illustrates the simulation results of system
throughputunder different initial frame size. Since ABS and
FSA-CSSare not Aloha-based algorithms, their performances are
notaffected by varying initial frame size. The FSA-CSS onlyshows
minor improvement when the number of tags is below200. The reason
is that the frequent use of FS mechanismintroduces too many idle
slots when the size of tag is small,and hence decreases the system
throughput. As the numberof tags is increased, the FSA-CSS shows
its advantage overother algorithms. From Fig. 7, the FSA-CSS
outperforms allother algorithms and achieves an average system
throughputof 0.4128, where the average throughput of ABS,
APAFQ,ABTSA, ds-DFSA, EAAEA, and ILCM are 0.3448, 0.3573,0.4083,
0.4079, 0.3361, and 0.3252, respectively. It is noted
that although ABTSA adopts a complex hybrid structure toenhance
its performance, our proposed FSA-CSS still outper-forms ABTSA by
1.10%. Similarly, the FSA-CSS is superiorto the ds-DFSA and APAFQ
with lower implementation cost.
In Fig. 8, we compare FSA-CSS (µ = 4, σ = 7) withABS in terms of
total number of slots, collision slots and idleslots. The number of
unread tags is progressively increasedfrom 100 to 1000. Compared to
ABS, FSA-CSS has fewercollision slots but more idle slots. The
simulation resultsindicate that significant collision slots are
reduced by the FSmechanism, and hence the total slots consumed by
the FSA-CSS. Although the shrink mechanism is expected to
overcomethe over-splitting, the extra idle slots are unavoidable by
theFS mechanism. As a result, the practical performance of FSA-CSS
is slightly lower than its theoretical results.
Consider the disparity between slot durations such as
theduration of non-idle slot is always longer than that of idle
slot[22], the system throughput metric is ineffective to
evaluatethe performance of identification in terms of
identificationtime. Therefore, we use time efficiency in the
simulations.Assume the time durations of success, collision and
idle slotsare denoted as Ts, Tc and Te, the time efficiency is
defined as[16]
Teffi =S · Ts
S · Ts + E · Te + C · Tc. (11)
where E, S, and C are the statistics of success, idle
andcollision during the identification process measured by the
-
IEEE TRANSACTIONS ON COMMUNICATIONS 8
0 24 6
8 10
0246
810
0.250.3
0.350.4
0.45
(a) n=200
Syste
m th
roug
hput
0 24 6
8 10
0246
810
0.250.3
0.350.4
0.45
(b) n=400
Syste
m th
roug
hput
0 24 6
8 10
0246
810
0.250.3
0.350.4
0.45
(c) n=600
Syste
m th
roug
hput
0 24 6
8 10
0246
810
0.250.3
0.350.4
0.45
(d) n=800
Syste
m th
roug
hput
0.26
0.28
0.3
0.32
0.34
0.36
0.38
0.4
(μ=4, σ=9) (μ=4, σ=8)
(μ=4, σ=7) (μ=4, σ=7)
Fig. 5. Comparison of system throughput under different number
of tags
100 200 300 400 500 600 700 800 900 10000.39
0.395
0.4
0.405
0.41
0.415
0.42
0.425
0.43
Number of tags
Syst
em th
roug
hput
CCN=4,CIN=6CCN=4,CIN=7CCN=4,CIN=8CCN=4,CIN=9CCN=3,CIN=4CCN=3,CIN=5
Fig. 6. The system throughput of FSA-CSS under specific pairs of
µ and σ
reader.Fig. 9 shows the time efficiency of various methods
under
different ratios between the duration of success, collision
andidle slots. As can be seen, all algorithms present
fluctuatingperformances under different Ts : Tc : Te ratios. When
Tc =Te in Fig. 9 (a) and (b), all the algorithms present a
similarbehavior, and the performance ranking from high to low
isFSA-CSS, ds-DFSA, ABTSA, APAFQ, ABS, EAAEA, andILCM. Fig. 9 (c)
and (d) show differently with the ranking ofABS dropping down to
the last one and the ranking of APAFQmoving up to the third when Ts
: Tc : Te = 3 : 3 : 1. As theratio between Tc and Te is increased,
ABS shows a significantperformance degradation due to the large
number of collisionslots. Compared to the reference methods, the
proposed FSA-
CSS can always achieve the best time efficiency under
variousratios by reducing collision and idle slots.
V. EXPERIMENTAL RESULTS WITH A PRACTICAL RFIDTESTBED
TABLE IIILINK PARAMETERS USED FOR RF COMMUNICATIONS
Parameter ValueFrequency 922.875 MHz
Backscatter link frequency 64 KHzModulation DSB-ASKDeviation 20
HHz
Channel width 250 KHzRead-tag data coding TPP
Tag-reader data coding Miller-8Reader-tag preamble 137.5
µsTag-reader preamble 2000 µs
T1 154 µsT2 130 µsT3 436 µs
To further evaluate the performance of FSA-CSS algorithmin a
practical UHF RFID system, we conduct experimentsusing a testbed in
an indoor environment. Experiments includean active RFID reader and
20 passive tags. The reader isequipped with ARM Cortex A9
processor, which is a 32-bit reduced instruction set (RISC)
processor with a maximumoperating frequency of 1 GHz and an
off-chip memory 512Mto ensure high speed and stable operation of
programs. Theexternal interface of the reader includes UART, JTAG,
ETHand USB, which greatly facilitates the software and
hardwaredebugging. The UART interface is used for
communicationbetween the host computer and the reader. The JTAG
interfacefacilitates the developer to debug the reader hardware.
ETH
-
IEEE TRANSACTIONS ON COMMUNICATIONS 9
100 250 400 550 700 850 10000.3
0.32
0.34
0.36
0.38
0.4
0.42
0.44
Number of Tags(a) Q=4
Syste
m T
hrou
ghpu
t
100 250 400 550 700 850 10000.3
0.32
0.34
0.36
0.38
0.4
0.42
0.44
Number of Tags(b) Q=5
Syste
m T
hrou
ghpu
t
100 250 400 550 700 850 10000.3
0.32
0.34
0.36
0.38
0.4
0.42
0.44
Number of Tags(c) Q=6
Syste
m T
hrou
ghpu
t
100 250 400 550 700 850 10000.3
0.32
0.34
0.36
0.38
0.4
0.42
0.44
Number of Tags(d) Q=7
Syste
m T
hrou
ghpu
t
ABS APAFQ ABTSA ds-DFSA EAAEA ILCM FSA-CSS
Fig. 7. System throughput comparisons of various algorithms
under different initial frame size
100 250 400 550 700 850 10000
500
1000
1500
2000
2500
3000
Number of Tags
Num
ber o
f slo
ts
total slots (ABS)idle slots (ABS)collision slots (ABS)total
slots (FSA-CCS)idle slots (FSA-CCS)collision slots (FSA-CCS)
Fig. 8. Simulation results of the number of total slots,
collision slots and idleslots between FSA-CSS and ABS
and USB provide various ways to upload tag data by thereader.
Compared to the BS commercial tag, the custom taghas added some
control logic codes in its state machine, whichis used to support
the reader commands of FSA-CSS.
Tab. III lists the link parameters configured for radio
fre-quency communication between the reader and tags. In orderto
comply with ISO/IEC 18000-6B, the carrier frequency isset to
922.875MHz and Miller coding is used.
The experiments are carried out by placing 20 tags inthe antenna
interrogation zone of RFID reader with a fixedtransmitting power.
We evaluate and compare the performanceof standard BS used in
ISO/IEC 18000-6B and the proposedFSA-CSS in total identification
time (defined as the total time
required to identify all tags) and identification rate
(definedas the number of tags can be identified per second).
Theexperimental environment is captured in Fig. 10. We vary
thenumber of tags from 2 to 20 and repeat the experiment in
50trials to get average performance. In parallel, we also
performsimulations using the same parameters.
Fig. 11 compares the experimental results of FSA-CSS withthat of
ABS which is used in ISO/IEC 18000-6B. In theexperiments, the
proposed FSA-CSS reduces the identificationtime by 32.5% and
improves the average identification rate by50% compared with ABS.
The simulation results are closedto the experimental results. Both
simulation and experimentresults indicate that the proposed FSA-CSS
outperforms theABS constantly in the practical RFID system.
VI. CONCLUSIONS
We have proposed a novel design of BS-based
anti-collisionalgorithm namely FSA-CSS to improve the low
identificationefficiency of traditional BS-based algorithms. Unlike
the ex-isting methods, the FSA-CSS allows tags which are in
thewaiting state to participate in the splitting process via
fastsplitting mechanism, thus the number of collision slots canbe
reduced. Meanwhile, the shrink mechanism has been intro-duced to
avoid the over-splitting problem. The proposed FSA-CSS has been
shown to improve the identification efficiencywithout estimation of
number of tags and extra hardware cost.The simulation results have
shown that FSA-CSS outperformsABS by 23.3% in system throughput. We
have also prototypeda RFID system and evaluated its performance in
a real-worldRFID environment. The experimental results have been
shownthat the proposed FSA-CSS can reduce the total
identificationtime by 32.5% for identifying 20 tags. Both
simulations and
-
IEEE TRANSACTIONS ON COMMUNICATIONS 10
100 250 400 550 700 850 10000.64
0.66
0.68
0.7
0.72
0.74
0.76
Number of Tags(a) Ts:Tc:Te=4:1:1
Tim
e ef
fici
ency
100 250 400 550 700 850 10000.56
0.58
0.6
0.62
0.64
0.66
0.68
0.7
Number of Tags(b) Ts:Tc:Te=3:1:1
Tim
e ef
fici
ency
100 250 400 550 700 850 10000.38
0.4
0.42
0.44
0.46
0.48
0.5
0.52
Number of Tags(c) Ts:Tc:Te=3:3:1
Tim
e ef
fici
ency
100 250 400 550 700 850 10000.54
0.56
0.58
0.6
0.62
0.64
0.66
Number of Tags(d) Ts:Tc:Te=4:2:1
Tim
e ef
fici
ency
ABS APAFQ ABTSA ds-DFSA EAAEA ILCM FSA-CSS
Fig. 9. Simulation results of time efficiency for various
algorithms under varying Ts : Tc : Te
Power Board Baseband BoardRF Board
Antenna
custom tags (compatible with ISO/IEC 18000-6B and
our proposed solution)
tags
Antenna2 meters
Fig. 10. The hardware setup used in the experiments
experiments have indicated that the proposed FSA-CSS is
asuitable candidate for the resource-constrained RFID systems.
APPENDIX APROOF OF RESULT 1
Assuming the i-th and j-th (i
-
IEEE TRANSACTIONS ON COMMUNICATIONS 11
2 4 6 8 10 12 14 16 18 200
200
400
600
800
1000(a) comparison of total identification time
Number of tags
Tota
l ide
ntifi
catio
n tim
e (m
s)
FSA-CSS (simulation)FSA-CSS (experiment)ISO/IEC 18000-6B
(simulation)ISO/IEC 18000-6B (experiment)
2 4 6 8 10 12 14 16 18 2015
20
25
30
35
40(b) comparison of identification rate
Number of tags
Iden
tific
atio
n ra
te (t
ags/
s)
FSA-CSS (simulation)FSA-CSS (experiment)ISO/IEC 18000-6B
(simulation)ISO/IEC 18000-6B (experiment)
Fig. 11. Comparison of the experimental results
Since T (i, k)R , T(j, k)R ∈ {0, 1, · · · , N − 1}, we have
− (N − 1) ≤ T (j, k)R − T(i, k)R ≤ (N − 1) , (17)
Without loss of generality, we assume 0N , according to (20), we
have
2 ≤ T (j, k+1)c − T (i, k+1)c ≤ N(j + 1− i)− 1 . (23)
which indicates that the minimum gap between counter valuein
adjacent subsets is above 1. That means the extra idle slotis
generated when M>N compared to (22). Therefore, theResult 1 can
be yielded.
APPENDIX BPROOF OF RESULT 2
As the prior knowledge of number of tags is known to thereader,
it is straightforward that the FS should be repeateduntil all tags
have been divided into m subsets and each subsetcontains few tags
in equal probability. Then the conventionalBS can be adopted to
identify these tags in each subsets. Themaximum depth Lmax of the
binary tree using FSA-CSS toidentify m tags can be derived as
Lmax = blogm2 c , (24)
where b·c denotes the operation of round down to the
nearestinteger. Substituting Eq. (24) into Eq. (9) leads to
NFSA−CSSn = Lmax + 2Lmax · (1− 1
2Lmax)m
+m · (1− 12Lmax
)m−1
+m∑r=2
2Lmax · P (r|m, L) ·NBSr ,(25)
When 2Lmax = m, FSA-CSS algorithm consumes the leastnumber of
slots to identify m tags. When 2Lmax = m andm→∞, the Eq. (25) can
be further expressed as
NFSA−CSSn∣∣2Lmax=m, m→∞
= Lmax + 2Lmax · (1− 1
2Lmax)m +m · (1− 1
2Lmax)m−1
+m∑r=2
2Lmax ·m!r!(m−r)!
(1
2Lmax
)r · (1− 12Lmax
)m−r ·NBSr≈ Lmax + 2×2
Lmax
e +m∑r=2
2Lmax (e−1)e ·N
BSr
≈ 2.212m.(26)
According to Eqs. (4) and (26), the Result 2 can be yielded.
REFERENCES[1] J. Wang, H. Hassanieh, D. Katabi, P. Indyk,
“Efficient and reliable low-
power backscatter networks,” Proc. 2012 ACM SIGCOMM, pp.
61-72,Jul. 2012.
-
IEEE TRANSACTIONS ON COMMUNICATIONS 12
[2] X. Liu, B. Xiao, S. Zhang, K. Bu, and A. Chan, “Step: A
time-efficienttag searching protocol in large RFID systems,” IEEE
Trans. Comput., vol.64, no. 11, pp. 3265-3277, 2015.
[3] W.-T. Chen, “An accurate tag estimate method for improving
the perfor-mance of an RFID anticollision algorithm based on
dynamic frame lengthALOHA,” IEEE Trans. Autom. Sci. Eng., vol. 6,
no. 1, pp. 9-15, 2009.
[4] J. Vales-Alonso, V. Bueno-Delgado, E. Egea-Lopez, et al.,
“Multiframemaximum-likelihood tag estimation for RFID anticollision
protocols,”IEEE Trans. Ind. Informat., vol. 7, no. 3, pp. 487-496,
2011.
[5] L. Arjona, H. Landaluce, A. Perallos, and E. Onieva, “Fast
fuzzyanticollision protocol for the RFID standard EPC Gen-2,”
Electron. Lett.,vol. 52, no. 8, pp. 663-665, 2016.
[6] J. Shin, B. Jeon, and D. Yang, “Multiple RFID tags
identification with M-ary query tree scheme,” IEEE Commun. Lett.,
vol. 17, no. 3, pp. 604-607,2013.
[7] X. Jia, Q. Feng, and L. Yu, “Stability analysis of an
efficient anti-collisionprotocol for RFID tag identification,” IEEE
Trans. Commun., vol. 68, no.6, pp. 2285-2293, 2012.
[8] H. Landaluce, A. Perallos, E. Onieva, L. Arjona, and L.
Bengtsson,“An energy and identification time decreasing procedure
for memeorylessRFID tag anticollision protocols,” IEEE Trans.
Wireless Commun., vol.15, no. 6, pp. 4234-4247, 2016.
[9] J. Myung, W. Lee, and J. Srivastava, “Adaptive binary
splitting forefficient RFID tag anti-collision,” IEEE Commun.
Lett., vol. 10, no. 3,pp. 144-146, 2006.
[10] Y. Cui and Y. Zhao, “Performance evaluation of a
multi-branch treealgorithm in RFID,” IEEE Trans. Commun., vol. 58,
no. 5, pp. 1356-1364, 2010.
[11] Y. H. Chen, S. J. Horng. R. S. Run, et al., “A novel
anti-collisionalgorithm in RFID systems for identifying passive
tags,” IEEE Trans.Ind. Inform., vol. 6, no. 1, pp. 105-121,
2010.
[12] C. Angerer, R. Langwieser, and M. Rupp, “RFID reader
receives forphysical layer collision recovery,” IEEE Trans.
Commun., vol. 58, no.12, pp. 3526-3537, 2010.
[13] J. Su, Z. Sheng, D. Hong, and G. Wen, “An effective
breaking policyfor dynamic framed slotted aloha in RFID,” IEEE
Commun. Lett., vol.20, no. 4, pp. 692-695, 2016.
[14] Y. Lai, L. Hsiao, and B. Lin, “Optimal slot assignment for
binarytracking tree protocol in RFID tag identification,” IEEE/ACM
Trans.Netw., vol. 23, no. 1, pp. 255-268, 2015.
[15] P. Solic, J. Radic, and N. Rozic, “Energy efficient tag
estimation methodfor Aloha-based RFID systems,” IEEE Sensors J.,
vol. 14, no. 10, pp.3637-3647, 2014.
[16] W.-T. Chen, “Optimal frame length analysis and an efficient
anticollisionalgorithm with early adjustment of frame length for
RFID systems,” IEEETrans. Veh. Technol., vol. 65, no. 5, pp.
3342-3348, 2016.
[17] J. Su, X. Zhao, D. Hong, Z. Luo, and H. Chen, “Q-value
fine-grainedadjustment based RFID anti-collision algorithm,” IEICE
Trans. Commun.,vol. E99-B, no. 7, pp. 1593-1598, 2016.
[18] C. Qian, Y. Liu, R. H. Ngan, and L. M. Ni, “ASAP: Scalable
collisionarbitration for large RFID systems,” IEEE Trans. Parallel
Distrib. Syst.,vol. 24, no. 7, pp. 1277-1288, Jul. 2013.
[19] L. Zhang, J. Zhang, and X. Tang, “Assigned tree slotted
Aloha RFIDtag anti-collision protocols,” IEEE Trans. Wireless
Commun., vol. 12, no.6, pp. 5493-5505, 2013.
[20] I. Bratuz, A. Vodopivec, and A. Trost, “Resolving collision
in EPCglobalClass-1 Gen-2 system by utilizing the preamble,” IEEE
Trans. WirelessCommun., vol. 13, no. 10, pp. 5330-5339, 2014.
[21] Y. Lai and C. Lin, “”Two blocking algorithms on adaptive
binarysplitting: single and pair resolutions for RFID tag
identification,” IEEETrans. Netw., vol. 17, no. 3, pp. 962-975,
2009.
[22] Y. M. Hu, I. C. Chang, and J. S. Li, “Hybrid blocking
algorithm foridentification of overlapping staying tags between
multiple neighboringreaders in RFID systems,” IEEE Sensors J., vol.
15, no. 7, pp. 4076-4085,2015.
[23] H. Wu, Y. Zeng, J. Feng, and Y. Gu, “Binary tree slotted
Aloha forpassive RFID tag anticollision,” IEEE Trans. Parallel
Distrib. Syst., vol.12, no. 6, pp. 19-31, 2013.