CN111027335A

CN111027335A - K-bit continuous collision bit detection split tree RFID label anti-collision algorithm

Info

Publication number: CN111027335A
Application number: CN201911162150.1A
Authority: CN
Inventors: 张莉涓; 赵楠; 陈金勇; 雷磊; 路志勇; 李志林; 沈高青; 蔡圣所
Original assignee: Nanjing University of Aeronautics and Astronautics; CETC 54 Research Institute
Current assignee: Nanjing University of Aeronautics and Astronautics; CETC 54 Research Institute
Priority date: 2019-11-22
Filing date: 2019-11-22
Publication date: 2020-04-17
Anticipated expiration: 2039-11-22
Also published as: CN111027335B

Abstract

The invention discloses an anti-collision algorithm for detecting split-tree RFID labels by k bits of continuous collision bits, which solves the technical problem of poor label identification capability of the existing multi-tree algorithm. The method accurately acquires the dividing condition of collision bits according to Manchester coding, and detects the highest continuous collision bit number k according to the received collision condition; the collision tag is then divided into 2^kIn the subgroup, the obtained continuous collision information is utilized, and then the collision labels are divided into groups with different sizes, so that the condition that a large number of collision time slots exist in the multi-branch tree algorithm is eliminated, and the number of the collision time slots and the number of the transmitted information bits can be effectively reduced. MATLAB simulation proves that the method has better label identification capability.

Description

K-bit continuous collision bit detection split tree RFID label anti-collision algorithm

Technical Field

The invention belongs to the technical field of radio frequency identification, and particularly relates to an anti-collision algorithm for a k-bit continuous collision bit detection split-tree RFID label.

Background

Radio Frequency Identification (RFID) technology is a key technology in the internet of things perception layer. Which identifies the data stored in the tag by backscattering the signal emitted by the reader. A typical RFID system generally consists of a reader and a number of tags, each of which is covered with a Unique Identifier (UID). When multiple tags reply simultaneously, tag collisions will occur because all tags share the same wireless channel. Tag collisions not only reduce bandwidth utilization, but also increase recognition latency.

Existing label collision prevention methods are mainly classified into two types, namely Aloha-based and branch tree-based methods. The Aloha-based method is also referred to as an uncertainty method. The essence of the method is that when a tag collision occurs, a reader requires all tags in the collision to randomly delay a period of time to participate in the next identification again. Due to the random selection time delay, there is a label starvation problem, resulting in incomplete identification of all labels. Current Aloha-based approach research is mainly focused on the Dynamic Frame Slot (DFSA) approach. The DFSA dynamically adjusts the length of the frame to achieve optimal system efficiency. However, before the optimal frame length is approached, the system efficiency of these methods is very low.

The branch tree based approach is a deterministic approach that guarantees that all tags are identified. The current method based on the branch tree adopts a method based on Manchester coding to detect collision information in order to improve the identification efficiency. Such as Collision Tree (CT), Collision Window Tree (CWT), multiple prefix query method (DPPS), MCT (M-ary Collision Tree), and so on. CT uses a collision detection method, detecting the first collision bit divides the collision tag into two subgroups, which, although CT greatly reduces the empty slots, still results in many collision slots, which increases the recognition time.

CwT use binary collision tree and heuristic bit window strategy to reduce the information bits transmitted by the tag end. Compared with CT, CwT can reduce the identification time and the information bit transmitted by the tag end. CwT, however, still require multiple time slots and query commands, which waste the header information in many query commands. DPPS and MCT are more time efficient and transmit fewer information bits by using more collision bit information to divide collision tags into more subgroups. They still take a lot of time to identify all tags.

Disclosure of Invention

The invention aims to provide a k-bit continuous collision bit detection split-tree RFID label anti-collision algorithm aiming at the problems of long identification time period and large number of transmitted information bits in the conventional branch tree algorithm, wherein collision labels are divided into 2 by utilizing continuous multi-bit collision bit information k^kIn the subgroups, the subgroup size of each collision slot is dynamically adjusted. In order to realize the purpose, the invention adopts the following main steps:

step 1: initializing command queue Q: the reader initializes command parameters pre ═ epsilon, k ═ 0, and then sends Query commands Query to all tags;

step 2: receiving a Query command and a label matched with the prefix pre, entering a transmission state, selecting a time slot number g, and waiting for the next Query command if the label not matched with the prefix pre; when g is 0, replying the residual ID information, otherwise, waiting for a Qrep command, and reducing the g value of the tag entering the transmission state by one every time the Qrep command is received, when the g value is 0, replying the residual ID information, otherwise, continuously waiting for the Qrep command;

and step 3: judging the time slot state: after the tags respond in different time slots, the reader detects the collision condition of each time slot, and if only one tag reply is detected, one tag is directly identified; if only one collision bit is detected, two tags can be directly identified due to the uniqueness of the tag ID; if no label reply is detected, the time slot is an empty time slot; if more than two tag replies are detected (namely tag collision occurs), updating the values of pre and k respectively, and pushing (pre, k) into a command queue;

and 4, step 4: judging a command queue: judging whether the command queue is empty, if not, taking out the first element of the queue to form a new command parameter, sending a Query command Query (pre, k), and transposing to the step 2); if the command queue is empty, the entire recognition process ends.

The k-bit continuous collision bit detection split-tree RFID label anti-collision algorithm provided by the invention is realized by MATLAB, and FIG. 4 is a simulation result for verifying the performance of the method. Because the packet size can be continuously adjusted by utilizing the received collision information, the method can effectively reduce the times of collision of the tags and the total times of command sending. The method fully considers 3 important performance indexes of command sending times of the reader, propagation delay and information bit number sent by the reader end and the label end, and compared with other branch tree algorithms, the DQTA method has obvious improvement on performance and is more suitable for label anti-collision protocols.

Drawings

FIG. 1 is a pseudo code diagram of an RFID tag collision avoidance algorithm of the present invention;

FIG. 2 is a system transmission model between a reader and a tag of the present invention;

FIG. 3 illustrates a tag operation after a Query command from a reader is received;

FIG. 4 is a simulation result of verifying the performance of the method of the present invention.

Detailed Description

The invention is described in further detail below with reference to the figures and examples.

First, symbols, functions and commands used in the present invention are defined as shown in table 1.

Table one symbol definition

1. Based on the above conditions, the k-bit continuous collision bit detection split-tree RFID tag anti-collision algorithm provided by the invention comprises the following specific implementation steps:

step 1: at the beginning of the identification process, the reader first initializes pre □, k 0, and broadcasts a Query command Query (pre, k) to start the first frame;

step 2: after receiving the Query command, the tag end executes the following steps with reference to fig. 3:

① matching phase tag first compares its top L_preA bit-length ID and a matching prefix pre. If the matching is successful, the label enters a transmission state and participates in the current frame; otherwise, the matching fails and waits for the next Query command;

② index phase tag check in transmission state Lth_preK bits of information after a bit position, i.e. (d)_p+1...d_p.+k) And converts it into a slot index g.

g＝bin2dec(d_p+1，d_p+2，...，d_p+k) (1)

③ reply phase if g is 0, the tag replies with the rest of ID information (d)_p+k+1...d_L). Otherwise, the tag waits for a Qrep command of the reader, the value of g is reduced by one every time the Qrep command is received, and the tag replies the residual tag ID information to the reader only when the value of g is 0.

And step 3: the reader receives the reply of the tag in each time slot and determines the state of the current time slot. The slot cases are divided into the following four cases:

① Single time Slot if only one tag replies, the reader identifies the tag and replies with a tag ID, ID pre | | dec2bin (g, k) | (d)_p+k+1，d_p+k+2，...，d_L) Where | is the join operation.

② identify two tag slots if the received message contains only one collision bit, because each tag is unique, we can directly identify that one tag is in one slot.e., the remaining ID information received by the reader is "10 x 101", where "x" represents the collision bit, and the two identified tags are:

③ free time slot-if the reader does not receive information in this time slot, it is a free time slot.

④ Collision time Slot when more than two tags reply in a time slot at the same time, the reader detects the position of the first collision bit and the number of consecutive collision bits after this position the reader can obtain the command parameters, for example, if the received information is "1 xx0x 1", the frame parameter can be updated to pre₁＝pre+dec2bin(g，k)+1，k ₁2 and the reader will (pre)₁，k₁) Pushed into the queue.

And 4, step 4: after the current frame is completed with all time slots, the reader detects whether the queue Q is empty, if Q is empty, the whole identification process is finished, otherwise, the reader acquires the parameters (pre, k) from the queue Q and starts the next frame.

2. Simulation and performance comparison of DQTA.

This section compares the performance of DQTA with existing CT and DPPS algorithms through simulations. Taking a maximum value of k of 3 in the simulation, the performance of the DQTA was evaluated in three aspects: (1) average collision time slot number in label identification; (2) the number of information bits is averagely sent by the reader end and the tag end in the tag identification process; (3) average identification time of the tag.

(1) Number of collision time slots

Since the collision slot takes longer than the idle slot, a comparison of the number of collision slots is given in fig. 4A. It should be noted that in DPPS, each slot includes two consecutive collision responses, in effect two slots for a frame, and no Qrep command precedes the second slot. For the sake of fairness we consider each slot state of the two slots as an independent tag reaction. We can see from fig. 4A that the DQTA algorithm requires the least number of collision slots to identify the tag.

(2) Number of information bits to be transmitted

In the three protocols, the frame structure and the information transmitted in each time slot are different. Considering collision slots alone is not sufficient to evaluate the overall performance of these protocols. The recognition time is also highly dependent on the number of transmitted information bits, thus giving a result graph comparing the number of transmitted information bits.

The information transmission bit number of the reader is shown in fig. 4B, and we can see that the DQTA costs the least number of information bits of the reader. Compared with CT and DPPS, the DQTA reader information bit number is reduced by 64% and 35.7%, respectively. This is because DQTA divides colliding labels into multiple packets according to the number of consecutive collision bits. We only need one Query () command and several short Qrep commands in each group. This greatly saves the number of bits of the request command information transmitted by the reader. Although DPPS requires only one Query () command, it requires more frames than DQTA, resulting in more information bits being transmitted by the reader side. CT has more collision slots and no frame structure, so it costs the most number of information bits at the reader side.

The result of the information transmission bit number of the tag end is shown in fig. 4C, and the DQTA spends the least information bit transmission number at the tag end. Since the collision slots are divided into smaller packets in DQTA and a smaller number of collision slots results. Therefore, the number of information transmitted by the label end of the DQTA protocol is minimum.

(3) Identifying time

The simulation result of the average time is shown in fig. 4D, and it can be seen that the DQTA average only needs 1.2ms to identify a tag, while CT and DPPS respectively need 1.43ms and 2.08 ms. The time gain of DQTA compared to DPPS and CT is approximately 16% and 42.3%, respectively.

Details not described in the present application are well within the skill of those in the art.

Claims

1. A K-bit continuous collision bit detection split-tree RFID label anti-collision algorithm is provided, an ID number of each label is appointed to be unique, each label comprises a time slot counter g, and the method comprises the following steps:

step 1: initializing a command queue: the reader initializes command parameters (pre ═ epsilon, k ═ 0), and then sends Query commands Query (pre, k) to all tags; the value of parameter k in Query command determines the frameWhen k is equal to n, there is 2 in the current frameⁿA time slot;

step 2: receiving a command and a label matched with the prefix pre, entering a transmission state, selecting a time slot number g, and waiting for the next Query command if the label not matched with the prefix pre; when g is 0, replying the residual ID information, otherwise, waiting for a Qrep command, and reducing the g value of the tag entering the transmission state by one every time the Qrep command is received, when the g value is 0, replying the residual ID information, otherwise, continuously waiting for the Qrep command;

and step 3: judging the time slot state: after the tags respond in different time slots, the reader detects the collision condition of each time slot, and if only one tag reply is detected, one tag is directly identified; if only one collision bit is detected, two tags can be directly identified due to the uniqueness of the tag ID; if no label reply is detected, the time slot is an empty time slot; if more than two tag replies are detected, updating the values of pre and k respectively, and pushing (pre, k) into a command queue;

and 4, step 4: judging a command queue: judging whether the command queue is empty, if not, taking out the first element of the queue to form a new command parameter, sending a Query command Query (pre, k), and transposing to the step 2; if the command queue is empty, the entire recognition process ends.

2. The k-bit collision avoidance algorithm for the RFID tag with the continuous collision bit detection split tree as claimed in claim 1, wherein after the tag receives the Query command, the specific operations performed by the tag end are as follows:

(1) a matching stage: the tag first compares its top L_preA bit-length ID and a matching prefix pre; if the matching is successful, the label enters a transmission state and participates in the current frame; otherwise, the matching fails and waits for the next Query command;

(2) an indexing stage: tag checking L < th > in transmission state_preK bits of information after the bit position and converting the k bits of information into a slot index g (the value of g is equal to the value of converting the number of bits of k bits into a decimal number);

(3) a recovery stage: if g is 0, the tag replies the rest ID information; otherwise, the tag waits for a Qrep command of the reader, the value of g is reduced by one every time the Qrep command is received, and the tag replies the residual tag ID information to the reader only when the value of g is 0.

3. The k-bit collision-prevention algorithm for detecting the split-tree RFID tags according to claim 1, wherein after the reader receives the information replied by the tags in each time slot, the specific operation steps of the reader end are as follows:

the reader end determines the state of the current time slot and divides the time slot condition into the following four conditions:

① single time slot, if only one tag replies, the reader directly identifies the tag and replies the tag ID;

② can identify two tag slots if the received information contains only one collision bit, because each tag is unique, we can directly identify that one tag is in one slot;

③ free time slot, if the reader does not receive information in this time slot, it is a free time slot;

④ collision time slot, when more than two tags reply simultaneously in a time slot, according to the information decoding received by the current time slot, the reader detects the position of the first collision bit and the number of continuous collision bits after the position, resets the pre to the information before the highest collision, decodes the k to the number of continuous collision bits after the highest collision, and puts the new parameters (pre, k) into the command queue.