An Efficient Adaptive Anticollision Algorithm Based on 4-Ary Pruning ...

Hindawi Publishing Corporation International Journal of Distributed Sensor Networks Volume 2013, Article ID 848746, 7 pages http://dx.doi.org/10.1155/2013/848746

Research Article An Efficient Adaptive Anticollision Algorithm Based on 4-Ary Pruning Query Tree Wei Zhang,1,2 Yajun Guo,2 Xueming Tang,1 Guohua Cui,1 Longkai Wu,3 and Ying Mei1 1

School of Computer Science and Technology, Huazhong University of Science and Technology, Wuhan 430074, China School of Computer, Central China Normal University, Wuhan 430079, China 3 National Institute of Education, Nanyang Technological University, 1 Nanyang Walk, Singapore 637616 2

Correspondence should be addressed to Wei Zhang; [email protected] Received 7 September 2013; Revised 19 November 2013; Accepted 25 November 2013 Academic Editor: Luis Javier Garc´ıa Villalba Copyright © 2013 Wei Zhang et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. In radio frequency identification system (RFID), the efficiency in which the reader identifies multiple tags is closely related to the methods to solve the collision of multiple tags. At present, a reasonable solution is the introduction of 4-ary query tree (or n-ary query tree) to reduce the collision time slots and additional query is used to decrease idle timeslots. The advantage of a 4-ary tree anti-collision algorithm is that it is able to reduce collision timeslots, but it also increases the idle timeslots. To reduce these excessive idle timeslots the 4-ary tree anticollision algorithm brings, an anti-collision algorithm based on adaptive 4-ary pruning query tree (A4PQT) is proposed in this paper. On the basis of the information of collision bits, some idle timeslots can be eliminated through pruning the 4-ary tree. Both theoretical analysis and simulation results support that A4PQT algorithm can significantly reduce recognition time and improve throughput of the RFID system.

1. Introduction In the Radio frequency identification (RFID) system, if multiple readers and multiple tags, or a single reader and multiple tags, transmit simultaneously in the readers’ working area, there are three types of collision as readers and tags operate on the same wireless channel. First, it is the frequency interference among readers which is due to the overlapping of two or more readers in the working area. The signals transmitted by these readers are causing mutual interference; thus the readers cannot read properly the tag data within the region. Second, when a tag appears in two or more readers’ working area, it will not know which readers to communicate with. Third, when many tags appear in the working area of one reader, these tags respond simultaneously to the queries of the reader, which will result in collision. Generally, the first two types of collision are called readerto-reader collision, and the third is known as reader-to-tag collision. When collision occurs, the reader cannot read the tag’s data. How to reduce the number of collisions? How to reduce the messages that tag transmits repeatedly and

reduce the communication overhead of the RFID system? Therefore, an efficient anticollision algorithm for identifying multitag is of great importance for the wireless RFID system. RFID system for solving multitag collision problem is currently divided into ALOHA-based anticollision algorithms and tree-based anticollision algorithms [1, 2]. ALOHA-based anticollision algorithms such as Slotted ALOHA, framed slotted ALOHA, dynamic framed slotted ALOHA, and enhanced dynamic framed slotted ALOHA are on the basis of a probabilistic method, in which each tag selects randomly a time slot in response to the query of reader in the event of a collision. If there are a large number of tags, the collision will happen when some tags repeatedly back off after the collision, so there will be a long time for the tag to be identified. Tree-based anticollision algorithms, such as binary search algorithms [2, 3], query tree algorithms [4–10], and tree-splitting algorithms [11] are based on deterministic methods. In the tree-based anticollision algorithm, the reader iteratively queries a subset of tags which match a given property until all tags are identified.

2 In the binary tree anticollision algorithm, the reader sends a query to tags. Upon receiving the query from the reader, tags compare their own ID numbers to the query number. Tags respond only when their ID numbers are less than or equal to the query ID number. When there are multiple tags simultaneously to respond, it will cause a collision. Then the reader will change the query number and set the highest collision bit to 0. Other bits after highest collision are set to 1. The query will be repeated until it can identify each tag. Some studies have improved identification efficiency of the binary tree anticollision algorithm by reducing the number of times to repeat searching [2, 3]. Wu et al. [3] proposed a binary tree algorithm based on ALOHA. The algorithm uses a dynamic, adaptive, and split method to adjust the frame length and brings frame length closer to the number of tags, so that the efficiency of the algorithm is very close to the optimal values. Query tree (QT) algorithm uses the query prefix of reader to split tags into two groups. The reader transmits a query string called prefix and the tags compare the string with its ID to see whether any of its ID contains the prefix. Only those tags whose IDs have a prefix matching to the string respond to the query. If the match succeeds, the tag transmits its ID to the reader. Collision occurs when multiple tags have the same prefix. In this case, the reader forms two new query prefixes by appending the old prefix with a binary 0 or 1. The reader then repeats the query with a different new prefix, and these tags are divided into two different groups until the number of tags in a group is one. However, in the basic query tree algorithm, the reader gradually inquires bit by bit. If the reader sends a query and no tag’s ID matches the query string, it will result in an idle time slot and increase the traffic and query time. Many researchers have improved the query tree algorithm. In the 4-ary query tree anticollision algorithm [6], the query prefix updates by increasing two binary bit each time and the query by the reader no longer constitutes a binary tree but splits into a 4-ary tree. The 4-ary query tree algorithm can reduce the collision time slots but increase the idle time slots. In some n-ary tree anticollision algorithms [1, 7], in order to reduce idle time slots once collision occurs, the reader will add an additional query to determine the status of the collision bit and avoid to query the nonexistent query branching. So it will avoid many idle time slots. This way can improve efficiency but adds an extra query time slot. Wu et al. [3] also introduce a bit collision detection mechanism based on basic query tree anticollision algorithm to reduce the idle time slots, and it is only suitable for mobile reader to identify tags. Tree-splitting anticollision [11] algorithm uses a random number generator to split tags into a plurality of subsets. Each tag has a random number generator and a counter and responds to the query of reader when the counter value is 0. The start value of the counter of each tag is 0. When the collision occurs, the reader broadcasts collision information. These tags just responding to the reader will generate a random number 0 or 1 and add to the counter. Those tags that has remained silent will set their counter value plus 1. Treesplitting anticollision algorithms are similar to the slotted ALOHA algorithm. These algorithms are random and their recognition efficiency is low. The new enhanced anticollision

International Journal of Distributed Sensor Networks algorithm [12] split the tags into 𝑚 + 1 groups, from 𝐺0 to 𝐺𝑚 , where 𝑚 is the tag ID. A tag belongs to 𝐺𝑘 only if the number of bit 1 in its ID is equal to k. When a collision occurs, the reader divides these collided tags into two subgroups, namely, subgroup 0 and subgroup 1. Each group or subgroup uses two counters, namely, 𝐶1 and 𝐶0 , which represent the number of 0 and 1 in the remaining tag IDs. When the 𝐶0 or 𝐶1 is equal to 1 in a group, the reader can identify the tags in the groups. In order to reduce idle time slots the 4-ary tree brings, this paper proposes an adaptive 4-ary pruning query tree (A4PQT) anticollision algorithm, which adaptively prune the idle time slot branches without any additional query, so it greatly improves the identification efficiency of RFID system. The rest of the paper is organized as follows. Section 2 introduces the basic idea to design A4PQT. Section 3 presents A4PQT. Section 4 describes the performance of the proposed algorithm. Section 5 analyzes the experimental results of A4PQT. Section 6 compares A4PQT with query tree algorithm (QT), collision tree algorithm (CT) [4], and the improved 4-ary query tree algorithm (I4QTA) [6]. Section 7 draws some conclusions.

2. Optimal Split Tree for Anticollision Algorithm In order to improve the recognition efficiency, many anticollision recognition algorithms have been proposed based on the binary tree, 4-ary tree, and n-ary tree. There are three types of time slots in the identification process of treebased anticollision algorithms: readable time slots (one tag response), collision time slots (multiple tags response), and idle time slots (no tag response). If a tree-based anticollision algorithm in identifying the same number of tags requires minimum time slots, then the algorithm is optimal. Suppose there are five tags, which are 0100, 1001, 0010, 1011 and 0110, the tree-split process of anticollision algorithm to identify multiple tags by using a binary tree and 4-ary tree is shown in Figure 1. As can be seen from Figure 1, a data frame of the treebased anticollision algorithm can be expressed by a tree. A data frame is constituted by a number of time slots and each time slot corresponds to a branch of the tree. For an anticollision algorithm, the more total time slots are to identify a certain number of tags, the lower identification efficiency the algorithm will bring. In Figure 1(a), there are five tags to be identified. The algorithm that uses a binary tree-splitting method requires five collision time slots, one idle time slot and eleven total time slots. While in Figure 1(b), by using a 4-ary tree method to identify the same five tags, the algorithm requires three collision time slots, five idle time slots and a total of thirteen time slots. As can be seen, the binary tree-splitting method to identify the same number of tags will result in a relatively large collision time slot and a relatively small idle time slot the anti-collision algorithm can reduce the collision time slots by using the 4-ary tree splitting method, but it also increases the idle time slots.

International Journal of Distributed Sensor Networks

5

3

5

1

0

00 2

3 00 2

1

1

2

1

11

10 2

2 011

010

11

10

01

01

100 1

1

0100

101 1

0101

1

0110 1

Collision timeslots

Collision timeslots

Readable timeslots

Readable timeslots

Idle

Idle

(a) Binary query collision tree

0111 1000 1001 1010

1011

1

1

(b) 4-ary query collision tree

Figure 1: The process of binary tree and 4-ary tree anticollision algorithm for identifying tags.

The total time slots of a n-ary tree anticollision algorithm and [5]

(1) If the mode is the highest collided bit +0, the two branches 01 and 11 can be cut. (2) If the mode is the highest collided bit +1, the two branches 00 and 10 can be cut.

𝑡 (𝑚) ∞

𝑚

= 1 + 𝐵 ∑ 𝐵𝐿 [1 − (1 − 𝐵−𝐿 ) − 𝑚𝐵−𝐿 (1 − 𝐵−𝐿 )

𝑚−1

],

(1)

𝐿=0

where 𝐵 is the number of branches of the tree, 𝑚 is the number of tags, and 𝐿 is the current level. When 𝐵 value is different, the total time slots of an anticollision algorithm are different. It has been proved that when the 𝐵 value is 3, the total time slots are the minimum; that is to say the identification process is most efficient, but it is not possible to construct a 3-ary tree. The advantage an anticollision algorithm has by using 𝑛-ary tree is that it is able to reduce collision time slots, but it increases the idle time slots. The basic idea of this paper is to take advantage of some information to cut some branches of the 4-ary tree (idle time slots). This corresponds to the way the 3-ary tree splits, which can guarantee minimum total time slots and greatly improve the anticollision recognition efficiency of 4-ary tree algorithm.

3. 4-Ary Pruning Query Tree Anticollision Algorithm A4PQT uses tag IDs to split tags into four groups. The reader transmits a query prefix and the tags in the work area of the reader respond to the query. Manchester code is used to detect collision bits. If a collision occurs, the reader updates query prefix. From the beginning of the highest collision bit, a collision node is split by a 4-ary tree and A4PQT will cut some idle time slot nodes according to the characteristics of collision bits. Assuming 𝑋 denotes the collision bit, the pruning principles are as follows.

(3) If the mode is the highest collided bit +𝑋, do no pruning. In the first case, according to the quadtree splitting method, the node 01 and node 11 are apparent idle time slot nodes. Similarly, in the second case, node 00 and node 10 are idle time slot nodes, and the two nodes can all be pruned. In the third case, since next the highest collided bit is also a collision bit, it is not possible to determine whether there is an idle time slot node and we cannot cut any branch. The A4PQT algorithm consists of rounds of queries and responses. In each round, the reader transmits query prefix. Tags receive the query prefix and check whether their own IDs contain the same prefix. If the tag’s ID contains the query prefix, the tag sends its ID except the part which is the same as the received prefix. If there is a collision, the reader updates query prefix based on the principle of pruning. For the reader, a stack is used as a prefix pool to hold prefixes. Firstly, a null string 𝜀 is pushed onto the stack. The reader starts a query with a null string 𝜀; if this causes a collision, the reader pushes new queries onto the stack and pops prefix to a new query until the stack is empty. The A4PQT algorithm is described as follows. (1) Initialize algorithm. An empty string 𝜀 is pushed onto the stack. (2) Determine whether the stack is empty. If it is empty, then go to step (7). (3) The reader pops a new query prefix from the stack, and sends to tags. Tags compare the query prefix with their IDs and check whether their IDs contain the prefix. If the matching is successful, tags respond to the reader query and send their IDs which do not

4

International Journal of Distributed Sensor Networks include the prefix. If the matching is not successful, then tags respond to nothing. (4) Reader receives response from tags. If only one tag responds, turn to step (6). If no tag responds, turn to step (7). If multiple tags respond simultaneously, then a collision occurs. The reader updates query prefix based on principle of pruning and pushes new prefixes onto the stack. Suppose 𝑆 is the set of query prefixes; the response of tag is 𝑏1 𝑏2 , . . . , 𝑏𝑢−1 𝑏𝑢 𝑏𝑢+1 , . . . , 𝑟, If the reader detects the highest collision bit which is 𝑏𝑢 , then new query prefixes produce the following principles (4.1) If the 𝑏𝑢+1 bit is 0, then cut two branches of 4-ary tree; only use 𝑆 𝑏1 𝑏2 , . . . , 𝑏𝑢−1 00 and 𝑆 𝑏1 𝑏2 , . . . , 𝑏𝑢−1 10 as two new prefixes and push the two prefixes onto the stack. (4.2) If the 𝑏𝑢+1 bit is 1, then cut two branches of 4-ary tree; only use 𝑆 𝑏1 𝑏2 , . . . , 𝑏𝑢−1 01 and 𝑆 𝑏1 𝑏2 , . . . , 𝑏𝑢−1 11 as two new prefixes and push them onto the stack. (4.3) If the 𝑏𝑢+1 bit is a collided bit, then generate four new branches. The reader uses 𝑆𝑏1 𝑏2 , . . . , 𝑏𝑢−1 00, 𝑆𝑏1 𝑏2 , . . . , 𝑏𝑢−1 01, 𝑆𝑏1 𝑏2 , . . . , 𝑏𝑢−1 10, and 𝑆𝑏1 𝑏2 , . . . , 𝑏𝑢−1 11 as new prefixes and pushes them onto the stack. (5) Repeat step (2)–step (4). (6) Identify tag. (7) Algorithm ends.

Using the five tags (they are 0100, 1001, 0010, 1011, and 0110) as an example from Figure 1, the identifying process of A4PQT is shown in Table 1, and the collision tree pruning process corresponding to the identifying process of the example is shown in Figure 2. When the reader sends an empty string 𝜀, all tags are responding. The reader receives the response of tags and detects a collision according to the Manchester code; the result is XXXX, where X represents a collided bit. The reader uses 00, 01, 10, and 11 as new prefixes according to the principle of pruning. When the query prefix is 00, identify the tag 0010. When the query prefix is 11, there is no tag response, resulting in an idle time slot. When the query prefix is 01, the reader judges that the collided bit is X0 and can cut two branches 01 and 11. When the query prefix is 10, the collided result is X1; this can cut the two branches 00 and 10. It can be seen from Figure 2 that, to identify five tags by using pruning, A4PQT algorithm generates four additional time slots, wherein the collision time slots are three and the idle time slots are one, while as shown in Figure 1, the total additional time slots of binary collision tree are six, and 4-ary collision tree are eight. So A4PQT algorithm can reduce the collision time slots and the idle time slots; it can improve the identification efficiency of RFID system.

4. Performance Analysis In RFID tag identification, the total time slots of an anticollision algorithm to identify all tags are important performance indicators; the less the total time slots, the better the performance of the algorithm. A4PQT algorithm has two kinds of situations in implementation process; one is split in accordance with the 4-ary tree but to cut two branches. The other is split by a completely 4-ary tree. Nodes on 4-ary tree include the root nodes, intermediate nodes, and leaf nodes. The degree of intermediate node is 4 and the degree of leaf node is 0. So the total nodes of a 4-ary tree are 𝑁 = 𝑛0 + 𝑛4 .

(2)

Here 𝑛0 is the number of nodes where the degree is 0 and 𝑛4 is the number of nodes where the degree 4. Since node degree is 4 has 4 children, leaf node has 0 children, and the root node is not a child of the other nodes. Therefore, the total nodes of a 4-ary tree can also be expressed as 𝑁 = 0 ∗ 𝑛0 + 4𝑛4 + 1 = 4𝑛4 + 1.

(3)

𝑛0 + 𝑛4 = 4𝑛4 + 1.

(4)

Then there is

The number of leaf nodes can be determined as follows: 𝑛0 = 3𝑛4 + 1,

(5)

(𝑛0 − 1) . 3

(6)

𝑛4 =

Let 𝑛 be the number of tags to be identified. When pruning condition is satisfied, A4PQT algorithm produces no idle time slots; the number of leaf nodes is equal to the number of tags; namely, 𝑛0 = 𝑛. The number of intermediate nodes is only half the number of intermediate nodes of a full 4-ary tree, namely, 𝑛4󸀠 =

𝑛0 − 1 . 6

(7)

So when meet the pruning condition, the total time slots of A4PQT algorithm are 𝑁 = 𝑛 + 𝑛4󸀠 =

𝑛 + (𝑛 − 1) (7𝑛 − 1) = . 6 6

(8)

When does not meet the pruning condition, each collision time slot in A4PQT algorithm will generate 0–2 idle time slot. If there has 0 idle time slot, the number of leaf nodes is equal to the number of tags; that is, 𝑛0 = 𝑛. According to formula (5), we have 𝑛 = 3𝑛4 + 1.

(9)

Then the total time slots of A4PQT algorithm are 𝑁 = 𝑛0 + 𝑛4 = 𝑛 + 𝑛4 =

(4𝑛 − 1) . 3

(10)

International Journal of Distributed Sensor Networks

5

Table 1: The identifying process of A4PQT. Round R→T T→R Tag 1 Tag 2 Tag 3 Tag 4 Tag 5

1 𝜀 Collided 0100 1001 0010 1011 0110

2 00 Readable

3 01 Collided 0100

4 10 Collided

5 11 Idle

6 0100 Readable 0100

7 0110 Readable

1001

8 1001 Readable

9 1011 Readable

1001

0010 1011

1011

0110

0110

Communication complexity is also a performance index for an anticollision algorithm; it is said to identify all tags required to transmit total number of bits. The communication complexity includes the reader communication complexity and tag communication complexity. Let 𝐶(𝑛) be a communication complexity of A4PQT algorithm to identify n tags, let 𝐶𝑅 (𝑛) represent the communication complexity of reader, let 𝐶𝑇 (𝑛) represent the communication complexity of tag; then there is

5 00

01 2

1

11

10 2

0100

0110

1

1

1001 1011 1

1

𝐶 (𝑛) = 𝐶𝑅 (𝑛) + 𝐶𝑇 (𝑛) .

In A4PQT algorithm, the sum of the reader query prefix length and tag response length is equal to the tag length. Let 𝐿 pre.𝑖 be the query prefix bits length of the reader, let 𝐿 req.𝑖 be the bits length of tag response, and let 𝐿 ID be the bits length of tag ID; then formula (15) can be expressed as

Collision timeslots Readable timeslots Idle

Figure 2: The process of A4PQT for identifying tags.

𝑇

𝐶 (𝑛) = ∑ (𝐿 pre.𝑖 + 𝐿 req.𝑖 ) , If there are 2 idle time slots, it means that one collision node will split two idle nodes; an idle node is also a leaf node, and then the leaf nodes include readable nodes (n tags) and idle nodes. Assuming the number of idle nodes is m, by formula (5), we have 𝑛 + 𝑚 = 3𝑛4 + 1.

(11)

And 𝑚 = 2𝑛4 ; obtain 𝑛4 = 𝑛 − 1; then the total time slots of A4PQT algorithm are 𝑁 = 𝑛 + 𝑚 + 𝑛4 = 𝑛 + 3𝑛4 = 𝑛 + 3 (𝑛 − 1) = 2𝑛 − 3.

(12)

Therefore, the total time slots of A4PQT algorithm are 𝑇=[

(7𝑛 − 1) , 2𝑛 − 3] . 6

(15)

(13)

Throughput is another important performance indicator of an anticollision algorithm, which is the ratio between the number of tags to be identified and the total time slots required to identify them. Throughput reflects the recognition efficiency of the algorithm; the greater the throughput, the higher the recognition efficiency. Let 𝑆 be the identification throughput of A4PQT algorithm; we have 𝑛 𝑛 = . 𝑆= (14) 𝑁 [(7𝑛 − 1) /6, 2𝑛 − 3]

(16)

𝑖=1

where 𝑇 is the total time slots of the A4PQT algorithm. Since 𝐿 ID = 𝐿 pre.𝑖 + 𝐿 req.𝑖 , the communication complexity of the A4PQT algorithm is 𝐶 (𝑛) = [

(7𝑛 − 1) , 2𝑛 − 3] ∗ 𝐿 ID . 6

(17)

5. Simulation and Result To test the performance of A4PQT, we compare our scheme with query tree algorithm (QT), collision tree algorithm (CT) [4], and the improved 4-ary query tree algorithm (I4QTA) [6]. CT eliminates the unnecessary idle time slots based on the basic query tree algorithm. I4QTA adopts a transposition way to reduce idle time slots. When two continuous collision bits occur, tags modify some bits to respond; the reader judges a tag prefix according to the transposed number, but this method also adds an extra cycle. Randomly generating 96 bits tag, the number of tags increases from 0 to 1000, and all results are the average of 100 experiments. In total time slots, Figure 3 shows the total time slots of QT, CT, I4QTA, and A4PQT for identification of all tags. It can be seen that the total time slots of CT are significantly less than that of QT; this is because the CT has no idle time slots. The total time slots of the I4QTA are slightly less than that of

6

International Journal of Distributed Sensor Networks ×107 1.8

3000 2500

1.5 Communication complexity

Total time slots

2000 1500 1000 500

1.2 0.9 0.6 0.3

0 0

100 QT CT

200

300

400 500 600 Number of tags

700

800

900 1000

14QAT A4PQT

Figure 3: Total time slots of the four algorithms to identify tags.

0 0

100 200 300 400 500 600 700 800 900 1000 Number of tags QT CT

14QAT A4PQT

Figure 5: The communication complexity of the four algorithms to identify tags.

1 0.9 0.8

transmitted bits of A4PQT algorithm are always less than that of the other three algorithms. The experimental results show that A4PQT algorithm can significantly reduce the total time slots and the transmitted bits for identifying tags; thus A4PQT algorithm has very fast identification rate and very high throughput; this is because A4PQT algorithm can eliminate some idle time slots.

Throughout

0.7 0.6 0.5 0.4 0.3 0.2

6. Discussion

0.1 0 0

100 200 300 400 500 600 700 800 900 1000 Number of tags QT CT

14QAT A4PQT

Figure 4: Throughput of the four algorithms to identify tags.

CT. I4QTA split by 4-ary tree can reduce collision time slots and adjust its query prefix through the response of tags, which can reduce the idle time slots the 4-ary tree generates. A4PQT algorithm uses an adaptive way to prune 4-ary tree, so it can more effectively eliminate some idle time slots, and the total time slots are less than those of the other three algorithms. Figure 4 shows the throughput of the four algorithms. With the increasing of the number of tags, the throughput of QT is about 34% and that of CT tends to be 50%; the throughput of I4QTA is about 52%, while that of A4PQT tends to be 64%. In communication complexity, Figure 5 shows the total transmitted bits for identifying n tags using the QT, CT, I4QTA, and A4PQT algorithms. As can be seen, the total

QT, CT, I4QAT, and A4PQT all belong to query tree algorithm. QT algorithm is a binary query tree algorithm. The reader sends a query prefix and tags will respond when their IDs have a prefix matching. The collision will occur when multiple tags respond simultaneously; then the reader will update the prefix by adding 0 and 1 in the original prefix. Since the prefix in QT is bitwise incrementally updating, this will cause no tag response when some queries are sent. This will bring idle time slot and increase the query time. CT algorithm is also a binary query tree algorithm. It is an improved version of the QT. The prefix in CT algorithm is updated by jumping; thus it can avoid the idle time slots. CT algorithm performance is also greatly improved and can avoid all idle time slots that the QT has. But it does not avoid the collision time slots. I4QAT is a 4-ary query tree algorithm which can reduce the collision time slots but meanwhile increases the idle time slots. I4QAT uses Manchester code to detect that the collided bit is one bit or two consecutive bits. If there are two consecutive collided bits, the reader adds an additional query to send to tags. When tags receive the query, they change the corresponding two bits. The conversion method is the “00” converted into “1,” “01” converted into “01,” “10” converted into “001,” and “11” converted into “0001”. I4QAT algorithm is to reduce idle

International Journal of Distributed Sensor Networks time slots by using an additional query and it can eliminate the idle time slots. So the performance of I4QAT algorithm is greatly improved. However, I4QAT algorithms rely on additional queries to eliminate idle time slots; thus it also affects the performance of the I4QAT. It is also an additional burden to tags to change two bits. A4PQT algorithm is different from I4QAT algorithm. A4PQT algorithm does not execute additional queries; instead, it assumes there are five tags which use pruning way to eliminate some of the idle time slots. When a collision bit appears, all the invalid branches will be pruned to eliminate all the idle time slots. Even when there are two consecutive collision bits, A4PQT algorithm can still eliminate most of the idle time slots without having to execute additional queries. As shown in our previous performance analysis, the average time slots requested by A4PQT algorithm have been lower than those required by I4QAT algorithm. When the number of tags reaches 500, the average number of time slots requested by A4PQT algorithm will be about 700. Relatively, the average number of time slots requested by I4PQT algorithm will come up to 900. As to some other aspects in terms of performance, A4PQT algorithm also has an obvious advantage over I4QAT algorithm as well as other algorithms.

7

[3]

[4]

[5]

[6]

[7]

[8]

[9]

7. Conclusion In the multitag identification environment, the collision is the major issue affecting the performance of RFID system. Collision resolution strategy based on binary tree can reduce idle time slots but increase collision time slots, while the 4-ary tree resolution strategy is to reduce the collision time slots but increase the idle time slots. In this paper, we propose a novel 4-ary query tree algorithm called adaptive 4-ary pruning query tree (A4PQT) for identifying multiple tags based on the characteristics of collided bit, using pruning way to eliminate some idle time slots of 4-ary tree. The A4PQT algorithm makes use of the advantages of 4-ary tree scheme that has less collision time slots and overcomes shortcomings of its more idle time slots. Both theoretical analysis and simulation results show that A4PQT algorithm can significantly reduce the total time slots for identifying multitag and improve the throughput of identification tags. As evidenced in the performance analysis, the adoption of n-ary tree will increase the idle time slots while the collision time slots are decreased. We envision that the pruning strategy will be an important and effective way to improve the efficiency of identification in the Radio frequency identification (RFID) system. Our further research work will be focused on further study of proper pruning strategy which will eliminate the increased idle time slots and decrease the total time slots.

References [1] J. Shin, B. Jeon, and D. Yang, “Multiple RFID tags identification with m-ary query tree scheme,” IEEE Communications Letters, vol. 17, no. 3, pp. 604–607, 2013. [2] A. Rennane, H. Saadi, R. Touhami, and M. C. E. Yagoub, “A comparative performance evaluation study of the basic binary tree and aloha based anti-collision protocols for passive RFID

[10]

[11]

[12]

system,” in Proceedings of the 24th International Conference on Microelectronics, pp. 1–4, 2012. H. Wu, Y. Zeng, J. Feng, and Y. Gu, “Binary tree slotted ALOHA for passive RFID tag anticollision,” IEEE Transactions on Parallel and Distributed Systems, vol. 24, no. 1, pp. 19–31, 2013. X. Jia, Q. Feng, and L. Yu, “Stability analysis of an efficient anticollision protocol for RFID tag identification,” IEEE Transactions on Communications, vol. 60, no. 8, pp. 2285–2294, 2012. D. R. Hush and C. Wood, “Analysis of tree algorithms for RFID arbitration,” in Proceedings of the IEEE International Symposium on Information Theory, pp. 107–116, 1998. Y. Kim, S. Kim, S. Lee, and K. Ahn, “Improved 4-ary query tree algorithm for anti-collision in RFID system,” in Proceedings of the International Conference on Advanced Information Networking and Applications (AINA ’09), pp. 699–704, May 2009. P. Pupunwiwat and B. Stantic, “Unified Q-ary tree for RFID tag anti-collision resolution,” in Proceedings of the 20th Australasian Database Conference (ADC ’09), Conferences in Research and Practice in Information Technology, pp. 49–58, 2009. H. Gou and Y. Yoo, “A bit collision detection based Hybrid Query Tree protocol for anti-collision in RFID system,” in Proceedings of the 11th IEEE International Conference on Computer and Information Technology (CIT ’11), pp. 158–163, September 2011. Y. Jiang and R. N. Zhang, “An adaptive combination query tree protocol for tag identification in RFID systems,” IEEE Communications Letters, vol. 16, no. 8, pp. 1192–1195, 2012. J. Sung, D. Kim, T. Kim, and J. Choi, “Heuristic query tree protocol: Use of known tags for RFID tag anti-collision,” IEICE Transactions on Communications, vol. E95-B, no. 2, pp. 603– 606, 2012. M.-K. Yeh, J.-R. Jiang, and S.-T. Huang, “Parallel response query tree splitting for RFID tag anti-collision,” in Proceedings of the International Conference on Parallel Processing Workshops (ICPPW ’11), pp. 6–15, September 2011. Y.-H. Chen, S.-J. Horng, R.-S. Run et al., “A novel anti-collision algorithm in RFID systems for identifying passive tags,” IEEE Transactions on Industrial Informatics, vol. 6, no. 1, pp. 105–121, 2010.

International Journal of

Rotating Machinery

Engineering Journal of

Hindawi Publishing Corporation http://www.hindawi.com

Volume 2014

The Scientific World Journal Hindawi Publishing Corporation http://www.hindawi.com

Volume 2014


Distributed Sensor Networks

Journal of

Sensors Hindawi Publishing Corporation http://www.hindawi.com

Volume 2014


Volume 2014


Volume 2014

Journal of

Control Science and Engineering

Advances in

Civil Engineering Hindawi Publishing Corporation http://www.hindawi.com


Volume 2014

Volume 2014

Submit your manuscripts at http://www.hindawi.com Journal of

Journal of

Electrical and Computer Engineering

Robotics Hindawi Publishing Corporation http://www.hindawi.com


Volume 2014

Volume 2014

VLSI Design Advances in OptoElectronics


Navigation and Observation Hindawi Publishing Corporation http://www.hindawi.com

Volume 2014



Chemical Engineering Hindawi Publishing Corporation http://www.hindawi.com

Volume 2014

Volume 2014

Active and Passive Electronic Components

Antennas and Propagation Hindawi Publishing Corporation http://www.hindawi.com

Aerospace Engineering


Volume 2014


Volume 2014

Volume 2014




Modelling & Simulation in Engineering

Volume 2014


Volume 2014

Shock and Vibration Hindawi Publishing Corporation http://www.hindawi.com

Volume 2014

Advances in

Acoustics and Vibration Hindawi Publishing Corporation http://www.hindawi.com

Volume 2014