CRC-16-BASED COLLISION RESOLUTION IN EPCGLOBAL

CLASS1 GENERATION2 RFID SYSTEMS

Jun-Bong Eom and Tae-Jin Lee

School of Information and Communication Engineerin, Sungkyunkwan University, Suwon 440-746, Korea

Keywords: Anti-collision algorithm, CRC-16, EPCglobal Class1 Generation2 protocol, RFID, RN16.

Abstract: The 900 MHz UHF passive RFID systems have drawn attraction because they have long access distance

and rapid identification speed. One of the most representative international standards is EPCglobal Class1

Generation2 (EPCglobal Gen2). RN16 which is used in the EPCglobal Gen2 standard assists a reader to

detect tag collision rapidly, but it may be removed because it does not have information of products. In this

paper, we propose an algorithm to replace RN16 with CRC-16 to reduce the identification time. We show

that CRC-16 has similar characteristics of RN16 and our proposed scheme reduces the number of bits

required for one tag identification.

1 INTRODUCTION

Radio Frequency Identification (RFID) is a

technology by which a reader recognizes the

information of objects to which tags are attached for

wireless communications with the reader. RFID

systems have become prevalent in supply chain

management, in manufacturing process, in the

industries to require the identification of products,

and so on, because they are able to transmit much

data rapidly with the use of wireless channel. Indeed

they are increasingly being used for automated

identification systems and come into the spotlight

for the realization of the ubiquitous society (R. want,

2006).

In RFID systems, a basic process for tag

identification is that if a reader queries tags for

information, the tags receiving the query signal

transmit their IDs to the reader. When only one tag

is in the reading range of the reader, tag

identification is very simple and easy. When,

however, many tags are in the reading range, tag

collisions may occur due to the simultaneous

transmission of their information. Thus an efficient

RFID tag anti-collision algorithm is necessarily

required in RFID systems to identify a lot of objects

in real time. One of the most popular algorithms

used for tag anti-collision is Framed-slotted ALOHA

(FSA).

In FSA-based tag anti-collision algorithms, to reduce

the probability of tag collision, a frame is divided

into several timeslots. And each tag selects

randomly one timeslot for the transmission of the

ID. When a tag transmits its information in a

timeslot, a reader can read the tag successfully.

When, however, a timeslot experiences collision by

more than one tag, the reader can not identify their

information. Unread tags must retransmit their

information in the next frame. This is inefficient due

to waste of collision timeslots. Also if there are

many idle timeslots in a frame, it is not efficient too.

The throughput of FSA may become low as the

number of tags increases. Thus it is necessary to

change the next frame size appropriately for high

efficiency after estimating the number of tags in the

reading range as in Dynamic FSA (DFSA) (S. Lee, S.

Joo and C. Lee, 2005). EPCglobal Class1

Generation2 (EPCglobal Gen2) (EPCTM Radio-

frequency Identification Protocols Class-1

Generation-2 UHF RFID Protocol For

Communications at 860MHz-960MHz Version1.0.9,

2005) is a representative international standard using

DFSA.

Unlike general DFSA having fixed timeslots for

success, idle, and collision, DFSA of EPCglobal

Gen2 adopts different timeslots. To reduce the waste

of time when a collision occurs, a relatively short

Random Number with 16 bits (RN16) is transmitted

before Electronic Product Code (EPC). Depending

on the success or failure of RN16 reading, a reader

decides whether it reads in the EPC of a tag or not.

The RN16 does not contain any information of a

Eom J. and Lee T. (2008).

CRC-16-BASED COLLISION RESOLUTION IN EPCGLOBAL CLASS1 GENERATION2 RFID SYSTEMS.

In Proceedings of the International Conference on Wireless Information Networks and Systems, pages 61-65

DOI: 10.5220/0002021700610065

 SciTePress

Figure 1: Tag identification mechanism using RN16 and having different time duration for success, idle, and collision.

product, and may consume extra time. Thus it would

be useful to reduce it if possible.

In this paper, we propose to use CRC-16 instead

of RN16 in EPCglobal Gen2 protocol, resulting in

reduced identification time. The outline of this paper

is organized as follows. In Section II, we summarize

the EPCglobal Class1 Generation2 standard. Our

proposed scheme is presented in Section III.

Performance evaluation and comparison via

simulations are presented in Section Ⅳ, and finally

Section Ⅴ draws conclusion.

2 EPCGLOBAL C1 GEN2

In EPCglobal Gen2, RN16 is used by a tag when it

sends the information to a reader. If long Protocol

Control (PC), EPC, and CRC-16 are transmitted

directly without transmitting short RN16 first, when

a collision occurs, it may waste time since another

PC, EPC and CRC-16 needs to be transmitted.

Figure 1 shows the mechanism using RN16 with

different time duration in success, idle, and collision.

In Figure 1, each tag randomly selects a timeslot,

and it transmits the 16 bit RN16 in the timeslot. If

the reader does not detect any such signal, it tries to

read tags in the next timeslot after waiting for few

time slots. If the reader identifies RN16 successfully,

it attempts to read the tag information by sending an

ACK command to the tag. When, however, RN16 is

not identified correctly, or there is no reply to ACK,

the reader comes to a conclusion that a collision

occurs.

In EPCglobal Gen2, communication between a

reader and tags is conducted through the steps of

Figure 2. A reader uses a Select command to select a

particular slot in which it identifies a tag. To identify

tags, the reader transmits some commands to the

tags in the Inventory round. The Access state of a

reader means that the reader reads from or writes to

individual tags after the tags are uniquely identified

Do not add any text to the headers (do not set

running heads) and footers, not even page numbers,

because text will be added electronically.

Figure 2: Reader/Tag operation and the state of tags.

The tags receiving the Select command prepare

for operation and stay at the Ready state until it

receives a Query command. The tags which receive

the Query command generate their slot counters

(0~2

-1), i.e., RN16, depending on the Q value

(0~15) during the Query command by their Random

Number Generator (RNG). When the value of the

slot counter of a tag is decremented to 0, the tag

goes to the Reply state and backscatters its RN16 to

the reader. If the reader does not detect a collision of

RN16, it issues an ACK command including the

RN16 of the tag. After the ACK is received, the tag

changes its state to the Acknowledged state. If the

received ACK is valid, the tag backscatters its PC,

EPC and CRC-16, and the reader conducts the CRC

(Reader)

WINSYS 2008 - International Conference on Wireless Information Networks and Systems

Figure 3: The proposed tag identification mechanism using CRC-16 and having different time duration for success, idle,

and collision.

check. Without any error, the reader identifies the

information of the tag successfully. Until the value

of the slot counter becomes 0, tags stay in the

Arbitrate state, holding state. And, the tags in the

Arbitrate state do not participate in the current

Inventory round until their slot counters become 0.

The tags shall decrement their slot counters by 1,

every time they receive a QueryRep command. If a

round to identify tags is over, but there are still

unread tags, the reader broadcasts a QueryAdj

command to the tags.

3 PROPOSED IDENTIFICATION

SCHEME

In this paper, we propose to use CRC-16 which has

the similar role of RN16 in the EPCglobal Gen2

protocol. In EPCglobal Gen2, when tags transmit

their information to a reader, RN16 shall be

generated before tags backscatter the information.

RN16 generated by RNG shall meet the following

randomness criteria (J. Choi, D. Lee, H. Lee, 2007).

▪ Probability of a single RN16:

The probability of an RN16 shall be bounded by

0.8/2

< P[RN16 = j] < 1.25/2

, where j is a

number created by RNG.

▪ Probability of simultaneously identical

sequences:

For tag population up to 10,000 tags, the

probability that any two or more tags

simultaneously generate the same sequence of

RN16s shall be less than 0.1%.

▪ Probability of predicting an RN16:

An RN16 shall not be predictable with the

probability greater than 0.025%.

We first investigate the property of CRC-16

whether it resembles that of RN16, and present the

proposed identification mechanism

3.1 A Uniformly Random

Characteristics of CRC-16 as in

RN16

All burst errors of length less than or equal to the

degree of the polynomial can be detected by CRC (B.

A. Forouzan, 2003). In CRC-P, assuming random

information of P bits and applying CRC-P

mechanism to the information, the CRCs from two

different random information should be different. If

the CRCs of two different information bit sequences

are the same, a burst error can not be detected by the

CRC check when one information bit sequence

becomes the same with another by channel error. It

is against the fact that all burst errors of length less

than or equal to the degree of P can be detected. For

example, let us consider two arbitrary information

part of PC+EPC, 1111000011110000 and

1100110011001100, and assume that they have the

same CRC R made by the CRC generator. When a

sender transmits the 1111000011110000 and R to a

receiver, and the altered information,

1100110011001100 and R by channel error, is

received, the receiver can not identify the error.

Thus each information with sequential P bits should

have only one corresponding and unique CRC-P.

(Reader)

CRC-16-BASED COLLISION RESOLUTION IN EPCGLOBAL CLASS1 GENERATION2 RFID SYSTEMS

In EPCglobal Gen2, a tag uses CRC-16 to check

the integrity of PC and EPC, backscattered from the

tag to a reader. And, as we examined before, all of

the CRC-16s are uniquely different for every

sequential 16 bits. It means that the probability of

any CRC-16 is 1/2

and it has uniformly random

characteristics as in RN16. Thus, we can know that

probability of a single CRC-16 is bounded by 0.8/2

< P[CRC-16=j] < 1.25/2

, where j is a CRC-16

number, like that of a single RN16. It also meets the

second characteristic of RN16, because the number

of tags that all 16 bits of CRC-16 are the same is

found by multiplying 10,000 by the probability of a

single CRC-16 and it is less than 0.1% of 10,000

tags. When a reader identifies tags, the tags might be

randomly selected and the probability of predicting a

CRC-16 might be 1/2

. And the probability that a

CRC-16 is predictable is not greater than 0.025%.

3.2 Proposed Identification Mechanism

using CRC-16

Figure 3 shows a simple proposed mechanism using

CRC-16 instead of RN16. In the proposed scheme,

slot counters of tags are also determined and

decremented by Query, QueryRep or QueryAdjust

commands. When the value of the slot counter of a

tag is 0, the tag backscatters the CRC-16 to the

reader. Note that the CRC-16 is generated from the

PC+EPC. After the tag backscatters its CRC-16, if

the reader receives CRC-16 of the tag correctly

without collision, the reader requests PC and EPC

information of the tag by transmitting an ACK

command to the tag. The reader receiving PC and

EPC conducts the CRC check with the already

backscattered and received CRC-16. In the case that

a collision of CRC-16 occurs or an invalid ACK

command is transmitted to the tag, the tag does not

reply to the reader. And, if there is no reply during

the predetermined time, the reader tries to read the

tag with the next smallest slot counter.

4 PERFORMANCE

EVALUATION

We evaluate the performance of the mechanism

using proposed CRC-16 and the existing mechanism

with RN16. We set the length of tag ID and the

commands of a reader as follows (see Table 1).

In the EPCglobal Gen2 standard, the Q selection

algorithm that tags choose Q values from Query

command is used for tag anti-collision. The Q

selection algorithm based on Framed Slotted

ALOHA (FSA) adjusts the number of timeslots in

the next round as the number of collision and idle

timeslots and tags select different random counter

values again from the reader reissuing Query or

QueryAdjust command. However, the Q selection

algorithm is not specified in the standard (G.

Khandelwal, K. Lee, A. Yener, and S. Serbetli,

2007). So we apply Dynamic Framed Slotted

ALOHA (DFSA) instead of the Q selection

algorithm to solve the problem of tag collision. And

we assume that the reader can knows the number of

tags accurately.

Table 1: The length of the tag ID and the reader

commands.

Command No. of bits Command No. of bits

RN16 16 bits Select 44 bits

PC 16 bits QueryAdj 9 bits

EPC 64 bits QueryRep 4 bits

Query 22 bits ACK 18 bits

NAK 8 bits RN16 16 bits

Figure 4 depicts that CRC-16 is uniformly

random. When an arbitrary CRC-16 is given, the x

axis denotes the simulation number, and y axis

represents how many repetitions are needed until the

CRC-16 of PC and EPC of a randomly selected tag

and the given CRC-16 becomes the same. If CRC-16

is uniformly random, the same CRC-16 might be

repeated every 2

=65536 on average. The

difference between

the expected mean, 65536, and

the simulation mean, 65869, is very close within

about 0.5%. Thus, we can verify the uniformly

random characteristics of CRC-16.

Figure 4: The uniformly random characteristics of CRC-

16.

Figure 5 shows the number of used bits per one

tag identification with the varying number of tags

under no channel error. When the number of tags is

WINSYS 2008 - International Conference on Wireless Information Networks and Systems

256, the proposed scheme needs 149.7 bits for one

tag identification and the scheme using RN16

requires 165.48 bits for one tag identification. The

difference between the number of bits required in

the proposed scheme and that in the existing scheme

is about 16 bits because it does not use as many bits

as RN16. Thus the performance of the proposed

scheme is improved up to 9.53%.

Figure 5: The number of used bits for one tag

identification with the varying number of tags, PER=0.

Figure 6: The number of used bits for one tag

identification with the varying number of tags, PER=0.3.

There may be lots of channel errors in RFID

systems because of low power backscattering of

tags. Figure 6 demonstrates how much improvement

in our proposed scheme is achieved under error-

prone environment. The figure shows the number of

bits per one tag identification with the varying

number of tags. Under channel error environment,

when the packet error rate (PER) is 0.3 and the

number of tags is 256, the proposed scheme uses

186.09 bits for one tag identification, while the

scheme using RN16 uses 212.62 bits for one tag

identification. This result also depicts that the

proposed scheme consumes less time than the

scheme using RN16, about 16 bits because 16 bits of

RN16 are saved as in the case of no channel error.

Thus, we can see that our proposed scheme is shown

to improve the performance about 12.5%.

5 CONCLUSIONS

We have proposed a simple and efficient tag

identification mechanism using CRC-16 instead of

RN16 in EPCglobal Gen2. In this paper, we have

verified that CRC-16 is uniformly random like

RN16, and thus it can provide the similar role of

RN16. According to the simulation results under

both error-free and error-prone environment, the

proposed algorithm using CRC-16 instead of RN16

requires less time than that using RN16 as in

EPCglobal Gen2

ACKNOWLEDGEMENTS

This work was supported by Korea Research

Foundation grant funded by the Korean Government

(MOEHRD) (KRF-2006-311-D00683), by a grant

under the Brain Korea 21 Initiative of the Korean

Ministry of Education and Human Resources, and

by the Korea Science and Engineering Foundation

(KOSEF) grant funded by the Korea government

(MOST) (No. R01-2006-000-10402-0).

REFERENCES

R. Want, “An Introduction to RFID Technology,” IEEE

Pervasive Computing, vol. 5, no. 1, pp. 25-33, Jan.

2006.

“EPCTM Radio-frequency Identification Protocols Class-1

Generation-2 UHF RFID Protocol For

Communications at 860MHz-960MHz Version1.0.9,”

EPCglobal, Jan. 2005.

S. Lee, S. Joo and C. Lee, “An Enhanced Dynamic

Framed Slotted ALOHA Algorithm for RFID Tag

Identification,” in Proc. of MobiQuitous, pp. 166-172,

Jul. 2005.

G. Khandelwal, K. Lee, A. Yener, and S. Serbetli, “ASAP:

A MAC Protocol for Dense and Time-Constrained

RFID Systems,” EURASIP Journal on Wireless

Communications and Networking, vol. 2007, Article

ID 18730, 13 pages, 2007. doi:10.1155/2007/18730,

2007.

B. A. Forouzan, Data Communications and Networking,

Third Edition, pp. 249-253, 2003.

J. Choi, D. Lee, H. Lee, “Query Tree-Based Reservation

for Efficient RFID Tag Anti-Collision,” IEEE

Communications Letters, Vol.11, No.1, pp. 85~87,

Jan. 2007.

CRC-16-BASED COLLISION RESOLUTION IN EPCGLOBAL CLASS1 GENERATION2 RFID SYSTEMS