Important label distribution identification method and system based on incremental Cuckou filter

Abstract

The invention discloses an important label distribution identification method and system based on an incremental Cuckou filter, and the method comprises the steps: constructing a large-scale RFID system model based on important label distribution, and carrying out the problem definition on the basis. Aiming at the bottleneck that the performance of a traditional filter is obviously reduced when the number of important tags is large, the invention innovatively designs a novel incremental Cuckou filter, and provides an efficient important tag distribution identification method based on the filter. According to the method, interference of common tags can be effectively filtered out, and a unique time slot is allocated to each important tag for response, so that rapid identification of important tag distribution in a large-scale RFID system covered by multiple readers is realized, and real-time monitoring and management of upper-layer applications on the important tags are effectively supported.

Description

Technical Field

[0001] The present invention relates to the field of tag recognition, and in particular to a method and system for identifying the distribution of important tags based on an incremental cuckoo filter. Background Art

[0002] RFID (Radio Frequency Identification) technology is a wireless radio frequency identification technology that uses the spatial coupling transmission characteristics of radio frequency signals to identify objects attached with tags. Its non-contact, multi-target batch identification, non-line-of-sight, and repeatable read / write characteristics make it widely used in fields such as inventory counting, human-computer interaction, and object tracking.

[0003] In a large-scale RFID system with a massive tag set covered by multiple readers, the upper-layer application usually focuses on a subset of tags rather than all of them, and these tags are called important tags. For example, in a large logistics warehouse, the warehouse manager may only need to track the tags of a certain batch of high-value goods, and these tags are regarded as important tags, while the tags of other goods are ordinary tags. Similarly, in a medical institution, the hospital may be more concerned about the tags of certain high-demand drugs or equipment, and has a lower priority for the tags of conventional drugs. In this case, communicating with important tags in real time and efficiently is the core task of important tag management. Efficient tag management depends on real-time communication with these important tags, so it is necessary to quickly identify their distribution, and be collaboratively managed by relevant readers while avoiding interference from ordinary tags.

[0004] Existing tag recognition protocols or tag distribution recognition protocols usually require all tags (including ordinary tags) to participate in data transmission, resulting in time delay and energy consumption. In contrast, the tag polling method broadcasts the IDs of important tags in turn by the reader to activate the target tags, which is a more efficient solution, but its inherent ID leakage and cloning risks may cause important tags to be lost and unable to be detected in time, resulting in economic losses. In addition, the tag distribution recognition protocol based on the traditional cuckoo filter significantly degrades in performance in high-density important tag scenarios, especially in systems where the number of important tags is large or the proportion is extremely high, and the performance is particularly poor. Summary of the Invention

[0005] Object of the Invention: The technical problem to be solved by the present invention is to provide a method and system for identifying the distribution of important tags based on an incremental cuckoo filter, which can quickly identify the distribution of important tags without transmitting any tag IDs, aiming at the deficiencies of the prior art.

[0006] To achieve the above object of the invention, the technical solution of the present invention is as follows:

[0007] An important tag distribution recognition method based on an incremental cuckoo filter, comprising the following steps:

[0008] In an RFID system including multiple readers, important tags, and ordinary tags, the reader constructs a cuckoo filter CF based on the important tag subset and constructs an incremental cuckoo filter ICF based on CF. The ICF consists of two layers. In the first layer, the reader stores the difference △F of the continuous fingerprints of the tags i and the hash equation index h used when the important tag is successfully inserted into CF h ; in the buckets of the second layer, the hash equation indexes used when the important tag is successfully inserted into CF are stored; for empty buckets, the reader fills them with a random useless hash equation index. If there is no useless hash equation index, the reader constructs a vector EI to record the indexes of the empty buckets, where the important tags are defined as the set of tags that need to be monitored by the upper-layer application, and the ordinary tags are defined as other tags outside the important tags in the system;

[0009] The reader broadcasts a query request including the ICF and its construction parameters in the system. After each tag receives this query request, it calculates and detects whether it can pass the ICF according to the received ICF construction parameters. If it can pass, it remains active; otherwise, it remains silent and no longer participates in the important tag distribution recognition;

[0010] Each reader performs the important tag recognition work within a single coverage area during its scheduling time. All readers recognize in parallel. After all the schedulings are completed, the readers exchange information with each other based on the scheduling algorithm to complete the important tag distribution recognition in multiple coverage areas.

[0011] Further, for an important tag k j , the calculation formulas for the tag fingerprint and the hash equation index inserted into CF are as follows:

[0012] F j =H(ID k ,s f ) mod 2 d

[0013]

[0014] where F j is the fingerprint of the tag, is the index of h candidate buckets, H(·) is the hash function, and the reader broadcasts h hash seeds to the tag to simulate h hash functions; s f and s h are the hash seeds used to calculate the fingerprint and the candidate buckets respectively, and d are the length of the second layer of the incremental filter ICF and the length of the tag fingerprint inserted into ICF respectively.

[0015] Further, the reader constructs a traditional cuckoo filter CF based on the subset of important tags, including:

[0016] For an important tag, the reader checks h buckets in the CF. If there are empty buckets among the h buckets, the reader inserts the important tag into the CF by putting the ID of the important tag into any empty bucket; otherwise, the reader randomly selects one of the h buckets and replaces the existing tag in the bucket with the important tag. The replaced important tag will be re-shifted in the same insertion step. This process continues until all important tags are successfully inserted or the maximum number of iterations is reached. If the construction fails, the reader increases the length of the CF and reconstructs the CF until all important tags are successfully inserted. For each non-empty bucket, the ID of the important tag is replaced with the fingerprint of the tag, and the empty buckets are not processed for the time being.

[0017] Further, the unused hash equation refers to the equation that is not used for inserting important tags in the set where h is the number of hash equations for successfully constructing the CF. When there are no unused hash equations.

[0018] Further, the ICF and its construction parameters broadcast by the reader in the system are where △d and △h represent the maximum fingerprint difference and the number of bits required for the hash equation index respectively, and h u is the index of the unused hash equation.

[0019] Further, after receiving the query request, each tag calculates and detects whether it can pass the ICF according to the received ICF construction parameters, including:

[0020] Check each bucket in the first layer of the ICF. For the i-th bucket, check whether the bucket is empty, including two cases:

[0021] h u When it is an empty set, check whether i is in the vector EI. If it exists, the i-th bucket is considered empty;

[0022] h u When it is not an empty set, calculate the hash value using the hash equation index stored in the bucket, denoted as If the value stored in the -th bucket in the second layer is h u , then the bucket is considered empty;

[0023] In both cases, if the i-th bucket is an empty bucket, the tag will skip it; otherwise, the tag will compare with the Compare the values stored in each bucket. If there is no match, skip to the next bucket. If there is a match, calculate its fingerprint and compare the fingerprint with the sum of the last d bits of the elements in the previous i buckets in the first layer of the ICF. If they match, increment its success counter p k and continue the matching. Otherwise, skip this bucket; the search continues until h times of checks have been performed or all buckets have been checked; if p k = 1, the tag passes the search and remains active; otherwise the tag is silenced.

[0024] Furthermore, the slot index setting method for each tag passing through the ICF is as follows:

[0025] Let SI denote the slot index of the important tag. The important tag records the index of the bucket in the first layer when passing through the ICF, denoted as p j , and set SI = p j ; only the tags that successfully pass through the ICF will be assigned a unique slot index. Tags that match multiple buckets will not pass through the ICF and remain silent; or each important tag calculates its slot index by counting the number of non-empty buckets before the bucket it matches in the second layer.

[0026] An important tag distribution recognition system based on an incremental cuckoo filter is applied to an RFID system including multiple readers, important tags, and ordinary tags.

[0027] The reader is configured to:

[0028] Construct a cuckoo filter CF based on the important tag subset, and construct an incremental cuckoo filter ICF based on CF. The ICF consists of two layers. In the first layer, the reader stores the difference △F of the continuous fingerprints of the tags i and the hash equation index h used when the important tag is successfully inserted into CF h ; in the second layer, the hash equation index used when the important tag is successfully inserted into CF is stored in the bucket; for an empty bucket, the reader fills it with a random useless hash equation index. If there is no useless hash equation index, the reader constructs a vector EI to record the index of the empty bucket, where the important tag is defined as the set of tags that need to be monitored by the upper-layer application, and the ordinary tag is defined as other tags outside the important tags in the system;

[0029] Broadcast a query request containing the ICF and its construction parameters in the system;

[0030] Execute the important tag recognition work in a single coverage area within its own scheduling time. All readers perform parallel recognition. After all scheduling executions are completed, the readers exchange information based on the scheduling algorithm to complete the important tag distribution recognition in multiple coverage areas.

[0031] The tag is configured to:

[0032] Receive the query request containing the ICF and its construction parameters sent by the reader, calculate and detect whether it can pass the ICF according to the received ICF construction parameters. If it passes, it remains active; otherwise, it remains silent and no longer participates in the important tag distribution recognition.

[0033] Beneficial effects: In response to the challenge of important tag distribution recognition in large-scale RFID systems, the present invention proposes an innovative method based on an incremental cuckoo filter, which effectively distinguishes important tags from ordinary tags. Compared with existing methods, the present invention has the following significant advantages: First, a new type of incremental cuckoo filter is constructed. Compared with the traditional cuckoo filter, the incremental filter stores the fingerprint difference of important tags and the index of the hash equation used to successfully insert into the filter through a double-layer structure, significantly reducing the false positive probability and improving the space efficiency; Second, an efficient important tag distribution recognition protocol is designed based on this filter. While filtering the interference of ordinary tags, this protocol assigns a unique time slot to each important tag and stores the hash function index used to calculate the positions and fingerprints of each tag. By detecting the status of these time slots, the distribution of important tags can be quickly recognized without transmitting the tag ID, thus significantly reducing resource waste and enhancing privacy security. In addition, this method can still maintain good performance in the scenario of a high proportion of important tags, and the performance degradation speed is significantly lower than that of existing methods. Brief Description of the Drawings

[0034] Figure 1 It is a flowchart of an important tag distribution recognition method based on an incremental cuckoo filter;

[0035] Figure 2 It is a construction diagram of an incremental cuckoo filter. Detailed Embodiments

[0036] The technical solutions of the present invention will be further described below with reference to the accompanying drawings.

[0037] The present invention proposes an important tag distribution recognition method based on an incremental cuckoo filter, as Figure 1 shown, including the following steps.

[0038] Step 1, create a large-scale RFID system model composed of multiple readers, important tags, and ordinary tags.

[0039] In the present invention, RFID system parameters for important tag distribution recognition are defined: important tags and ordinary tags. Important tags are defined as the set of tags that need to be monitored by upper-layer applications, also known as key tags; ordinary tags are defined as other tags in the system except important tags.

[0040] Establish a mathematical model of the RFID system based on the identification of the important tag distribution: The set of readers is represented as where m is the number of readers; the set of important tags is represented as K = {k 1 , …, k x}, x is the number of important tags; the set of ordinary tags is represented as T = {t 1 , …, t y}, y is the number of ordinary tags; the topological structure of the readers is represented as a graph where ε represents the set of edges connecting two neighboring readers. In this model, if two or more readers have at least one common covered tag, they are called neighboring readers to each other. The reader its neighboring reader set and non-neighboring reader set are denoted as Γ(R i ) and Υ(R i ) respectively.

[0041] Step 2, Define the important tag distribution identification problem in a large-scale RFID system covered by multiple readers according to the system model in Step 1.

[0042] First, represent the set of important tags covered by the reader R i (1 ≤ i ≤ m) as K i , and |K i ∩K j | ≥ 0. Then, the important tag distribution identification problem in a large-scale RFID system covered by multiple readers can be defined as: identifying the values of K i (1 ≤ i ≤ m). According to K i , it can be determined which important tags each reader covers and which readers cover each important tag.

[0043] The above Step 1 and Step 2 are executed on the backend computer, and the following Step 3 is executed on the reader. The CF of the present invention is constructed on the reader side.

[0044] Step 3, Design a method for identifying the important tag distribution based on the cuckoo filter according to the system model in Step 1 and the important tag distribution identification problem in Step 2.

[0045] The specific method is as follows:

[0046] Step 3-1, Design a new method for constructing a cuckoo filter, making it have both filtering and time slot allocation functions, and construct the cuckoo filter ICF based on this method and the set of important tags.

[0047] Step 3-1-1, the reader constructs a traditional Cuckoo Filter CF based on the important tags. For an important tag, the reader calculates h candidate bucket indices through the formula and checks the status of h buckets in the CF, where is the length of the second layer of the ICF. If there are empty buckets among the h buckets, the reader inserts the important tag into the CF by putting the ID of the important tag into any empty bucket. Otherwise, the reader randomly selects one of the h buckets and replaces the existing important tag in the bucket with this important tag, and the replaced important tag will be re-shifted with the same insertion steps. This process continues until all important tags are successfully inserted or the maximum iteration is reached. If the construction fails, the reader increases the length of the CF and reconstructs the CF until all important tags are successfully inserted. Each non-empty bucket uses the formula H(ID k , s f ) mod 2 d to store the fingerprint of the important tag, and the empty buckets are not processed in this step.

[0048] Step 3-1-2, the reader constructs the ICF based on the CF. The ICF consists of two layers. Referring to Figure 2 , to insert x important tags, the reader first sorts the fingerprints of the important tags in ascending order, denoted as F 1 <F 2 ≤…≤F x .

[0049] Step 3-1-3, the reader calculates the differences between consecutive fingerprints, denoted as △F i =F i -F i-1 , and then obtains the set of fingerprint differences {△F 1 , △F 2 ,…, △F x}, where △F 0 =0.

[0050] Step 3-1-4, the parameters of the two-layer ICF are set respectively. In the i-th bucket of the first layer, the reader stores △F i and the hash equation index h h used when the i-th important tag is successfully inserted into the CF. Therefore, the length of the first layer is In the second layer, the i-th bucket stores the hash equation index used to successfully insert the important tag into the i-th bucket of the CF. For empty buckets, the reader fills them with a random useless hash equation index, denoted as h u . A useless hash equation refers to an equation that has not been used to insert any important tag in the set . However, when When there are no unused hash equations, the reader constructs a vector EI to record the indices of the empty buckets.

[0051] Step 3-2, filtering ordinary tags: The reader broadcasts the ICF and its construction parameters. All tags in the system calculate and check whether they can pass the ICF according to the received ICF construction parameters; if they can pass, they remain active, otherwise, they remain silent, that is, they no longer participate in the subsequent important tag distribution recognition. The initial state of each tag in the system is silent. Tags that pass the ICF are activated and are active tags. Tags that cannot pass the ICF continue to remain silent. Since the ICF is constructed based on important tags, ordinary tags cannot pass the ICF.

[0052] Filtering ordinary tags specifically includes the following steps:

[0053] Step 3-2-1, the reader broadcasts the ICF and its construction parameters in the system Among them, and are the lengths of the first and second layers of the ICF respectively, s 1-h represents h hash seeds used to calculate the candidate bucket indices, △d and △h represent the maximum fingerprint difference and the number of bits required for the hash equation index respectively. Additionally, if there are no useless hash equations, then set h u to be Otherwise, set EI to be are the lengths of the first and second layers of the ICF respectively, s 1-h represents h hash seeds used to calculate the candidate bucket indices, which are s 1 , s 2 , …, s h .

[0054] Step 3-2-2, after receiving this query request, each tag determines whether it can pass the ICF according to the following query process. For an important tag, it first checks each bucket in the first layer of the ICF. For the i-th bucket The important tag checks whether the bucket is empty. This check involves two cases, including:

[0055] (a) When h u is an empty set, then the important tag checks whether i is in the vector EI. If it exists, it is considered that the i-th bucket is empty.

[0056] (b) When h u is not an empty set, then the important tag uses the hash equation index stored in the bucket and the formula in Step 3-1-1 to calculate the bucket index, denoted as If the value stored in the -th bucket in the second layer is hu , then this bucket is considered empty.

[0057] In both cases, if a bucket is empty, the important tag skips it; otherwise, the important tag will be compared with the value stored in the th bucket in the second layer. If there is no match, the important tag jumps to the next bucket in the first layer. If there is a match, the important tag uses the formula H(ID k , s f ) mod 2 d in Step 3-1-1 to further calculate its fingerprint, and compare it with the sum of the last d bits of the elements in the first i buckets in the first layer. If they match, the important tag increments its success counter o k (initialized to 0) and continues the match. Otherwise, it skips this bucket. The search continues until h times of checks have been performed or all buckets have been checked. If p k = 1, the important tag passes the search; otherwise, it is silenced. This method helps filter out false positive tags that match multiple buckets. It should be noted that the reader resolves fingerprint conflicts during the ICF construction to ensure that each important tag is mapped to only one bucket.

[0058] Step 3-2-3, if a tag fails to pass the ICF, it updates its flag to 1 before remaining silent, indicating that it is an ordinary tag. Otherwise, it remains active and calculates its slot index for the identification phase.

[0059] Step 3-2-4, calculate the slot index of the important tag for the identification phase. Let SI denote the slot index of the important tag. The important tag records the index of the bucket in the first layer when passing the ICF, denoted as p j , and sets SI = p j . It should be noted that only tags that successfully pass the ICF will be assigned a unique slot index, because tags that match multiple buckets will fail to pass the ICF and thus remain silent. Alternatively, each important tag can calculate its slot index by counting the number of non-empty buckets before the bucket it matches in the second layer. Both methods ensure that each key tag has a unique slot index, but may assign different values. For simplicity, the first method can be used.

[0060] Step 3-3, the reader works in parallel to identify the slot tag distribution in the area covered by only a single reader, and identify the important tag distribution in the area covered by multiple readers based on the traditional scheduling algorithm.

[0061] Specifically, it includes the following steps:

[0062] Step 3-3-1: The reader works in parallel to identify important tags within a single coverage area. Reader R i Lets the tags virtually execute the Aloha protocol, and constructs an array E from all the tags within the single coverage area of the system and the time slots mapped by the false positive tags through CF i , E i Each element in is the status of the corresponding time slot: empty time slot, the element is '0'; single time slot, the element is '1'; collision time slot, the element is 'c'. The reader constructs an array A based on the true responses of the tags, i.e., the true time slot status i .

[0063] Step 3-3-2: Compare array E i with A i , R i can identify the important tags within its single coverage area.

[0064] Step 3-3-3: Identify the tag distribution within multiple coverage areas based on the existing reader scheduling algorithm.

[0065] Each reader executes Steps 3-3-1 and 3-3-2 within its scheduling period to complete the identification of important tags in the single coverage area. After the scheduling is completed, each reader determines the coverage of each important tag through information exchange, thereby realizing the identification of the important tag distribution.

[0066] In Step 3, the solution process for the number of hash equations h, the length f of the first layer of ICF 1 , the length f of the second layer of ICF 1 and the length of the maximum fingerprint difference △d is as follows. Since there are false positives in the incremental cuckoo filter (ICF), there may still be some ordinary tags remaining active and participating in subsequent identification after filtering, which may cause the reader to fail to identify important tags mapped to the same hash bucket. Considering the worst case, that is, each false positive ordinary tag causes an important tag to be unidentifiable, the following constraint conditions are set:

[0067]

[0068] where I fn is the false negative index, n fp and n k are the numbers of false positive ordinary tags and unrecognized important tags, respectively, and α is a given index threshold determined by application requirements.

[0069] In each round of identification, the goal is to maximize the number of important tags successfully identified per unit time, thereby minimizing the total identification time as much as possible. The identification of important tags is achieved through three types of time slot pairs: <1,1>, <1,0> and <c,0>. E 11 , E10 and E c0 are respectively expressed as the expected values of the number of time slot pairs <1, 1>, <1, 0> and <c, 0>. Let t i be the execution time. Therefore, the number of important tags θ i recognized per unit time is expressed as

[0070]

[0071] where

[0072] According to the number of important tags θ i recognized per unit time, the set K i (1 ≤ i ≤ m) of important tags covered by the reader R i should have a size of θ i ×t i .

[0073] The reader constructs an ICF with n k important tags, and then broadcasts the ICF and the parameters used to construct the ICF. Each tag performs one lookup, and only those tags that pass the ICF can respond to the reader in their assigned time slots according to their time slot indices.

[0074] Analyze the false positive probability of the incremental cuckoo filter For a bucket in the first layer of the ICF, the probability that a tag not inserted into the bucket still successfully matches (i.e., false positive) is at most because the tag must match the d-bit fingerprint calculated from the fingerprint difference △d of the corresponding bucket. According to the structure of the incremental cuckoo filter, each tag is checked at most h times during the lookup process. Let h c be the number of checks, where 1 ≤ h c ≤ h. The probability of a successful match after comparing h c times is:[[]]

[0075]

[0076] By comparing the expected and actual time slot states, the reader determines the time slot pairs (e.g., <1, 1>, <1, 0> and <c, 0>). Since the ICF assigns a unique time slot index to each important tag, the time slot frame size f is also n k . When a bucket in the ICF is filled with local important tags without any false positive ordinary tags, a <1, 1> time slot pair appears. Let p 11 be the probability of <1, 1> occurring, then there is Therefore, the expectation of the recognized important tags in <1, 1> is

[0077] The value of d will be given in the subsequent analysis. Given d, the cardinality estimation method can be used to calculate and values:

[0078]

[0079] where m 11 is the number of <1,1> time slot pairs calculated by the reader. In addition, the value of can be calculated:

[0080]

[0081] When a time slot is inserted with only one non-locally important tag and no false positive tags, the time slot pair <1,0> appears. The probability of the appearance of <1,0> can be deduced and the expected value can be further calculated

[0082] When a bucket is inserted with a non-locally important tag and mapped by one or more false positive ordinary tags, the time slot pair <c,0> appears. Similarly, the probability p of the appearance of <c,0> can be deduced c0 as Therefore, it can be deduced that

[0083]

[0084] For clarity, d is retained in the expressions of E 11 , E 10 , E c0 , and their values can be calculated using the known parameters n , n k , n t , n 11 , n c0 and α.

[0085] According to the construction of ICF, f 1 = n k . Given that the numerator E 11 + E 10 + E c0 of θ depends only on d when other parameters are fixed, and since f 2 is related to h, the denominator t depends on h and d. To maximize θ, the optimal h that minimizes the h-related term in t needs to be found first, and then d is optimized with this h. To insert n k important tags into ICF, the constraint f 2 ≥ n k is required. For the buckets in the second layer, the probability that no tag is mapped to it using the h hash function result is The expected value of the empty bucket is Therefore, it can be deduced that:

[0086]

[0087] The relationship between h and f can be deduced from the above formula: 2 relationship:

[0088]

[0089] By changing f 2 from n k to 1.5n k , the corresponding values of h and Denote the total number of bits in t related to h and f 2 as B(), that is:

[0090]

[0091] where e() is the number of bits required to transmit EI to the tag. If then Otherwise, e() = 0. Determine the optimal h by minimizing B(), so as to minimize the number of bits related to h transmitted by the reader within time t. The optimal value of h is 3. As f 2 increases, both h and gradually decrease. Correspondingly, when decreases, B() will also decrease. However, when remains unchanged, B() may change. For example, when h = 3 and h = 4, both make but B() is smaller when h = 3, because when h = 4, holds, and additional e() bits are required. The optimal value of h, 3, is independent of the number of tags used to construct the ICF. To determine the optimal f 2 , it is necessary to solve the equation by finding the intersection point of and using numerical techniques such as the bisection method to obtain That is, when n k = 500, the optimal value of f 2 is 532.

[0092] To optimize d, determine the minimum fingerprint length min(d) to meet the recognition accuracy, and calculate the expected value of △d. Substitute △d and other parameters into the equation Solve for d numerically to maximize θ. According to the calculation formula of p fI , the number of false positive tags of the ICF is:

[0093] ​

[0094] According to the constraint conditions The minimum fingerprint length is:

[0095]

[0096] Analyze the expected value of △d, which is given by Given. To calculate this formula, first determine It can be transformed into the problem of finding the expected value of the length of the longest segment when a rope of length 2d is randomly divided into n k segments. Let represent the ordered positions of these cuts, and △F i = F i - F i-1 represent the length of the i-th segment. The probability that any specific subset of k segments simultaneously exceeds a given threshold is Furthermore, applying the principle of inclusion - exclusion, when c1 = c2 = … = c k = z, the probability that the length of the longest segment exceeds z is:

[0097]

[0098] where The expected value of is:

[0099]

[0100]

[0101] In the formula, is the harmonic number of n k . Approximate as lnb + γ, where γ≈0.5772156649... is the Euler - Mascheroni constant. Then the expected value of the maximum value △d of the fingerprint difference of the important tag is:

[0102]

[0103] According to the above - mentioned parameter analysis, in order to optimize d to maximize θ, the optimal value of d is set when taking the maximum value of θ. Let the initial value of d be min(d).

[0104] The present invention also provides an important - tag distribution recognition system based on an incremental cuckoo filter, which is applied to an RFID system including multiple readers, important tags, and ordinary tags

[0105] The reader is configured to:

[0106] Construct a Cuckoo Filter (CF) based on a subset of important tags, and construct an Incremental Cuckoo Filter (ICF) based on the CF. The ICF consists of two layers. In the first layer, the reader stores the difference △F of the continuous fingerprints of the tags i and the hash equation index h used when the important tags are successfully inserted into the CF h ; In the buckets of the second layer, store the hash equation indexes used when the important tags are successfully inserted into the CF; For empty buckets, the reader fills them with a random useless hash equation index. If there is no useless hash equation index, the reader constructs a vector EI to record the indexes of the empty buckets, where the important tags are defined as the set of tags that need to be monitored by the upper-layer application, and the ordinary tags are defined as other tags in the system except the important tags;

[0107] Broadcast a query request containing the ICF and its construction parameters in the system;

[0108] Execute the important tag recognition work in a single coverage area within its own scheduling time. All readers recognize in parallel. After all the scheduling is completed, the readers exchange information with each other based on the scheduling algorithm to complete the important tag distribution recognition in multiple coverage areas.

[0109] The tag is configured to:

[0110] Receive the query request containing the ICF and its construction parameters sent by the reader, calculate and detect whether it can pass the ICF according to the received ICF construction parameters. If it can pass, it remains active; otherwise, it remains silent and no longer participates in the important tag distribution recognition.

[0111] The present invention provides a method and system for important tag distribution recognition based on an incremental cuckoo filter. The above content is only the preferred implementation manner of the present invention, and there are various possibilities for the methods and ways to implement this technical solution. It should be noted that for those of ordinary skill in the art, without departing from the basic principle of the present invention, various improvements and optimizations can be made to it, and these improvements and optimizations should also be regarded as the protection scope of the present invention. In addition, each component not clearly defined in this embodiment can be realized by the prior art.

Claims

1. A method for identifying important label distribution based on incremental cuckoo filter, characterized in that: The following steps are involved: In an RFID system including multiple readers, important tags and ordinary tags, the reader constructs a cuckoo filter CF based on a subset of important tags, and constructs an incremental cuckoo filter ICF based on CF. ICF consists of two layers. In the first layer, the reader stores the difference △F of the continuous fingerprints of the tag. i And the hash function index h used to successfully insert the important tags into CF h ; The second-layer bucket stores the hash function index used to successfully insert important tags into the CF; for empty buckets, the reader fills them with a random useless hash function index. If there is no useless hash function index, the reader constructs a vector EI to record the index of the empty bucket, where important tags are defined as the set of tags that the upper-layer application needs to monitor, and ordinary tags are defined as other tags other than important tags in the system; The reader broadcasts a query request containing ICF and its construction parameters in the system. After receiving this query request, each tag calculates and detects whether it can pass the ICF based on the received ICF construction parameters. If it passes, it remains active, otherwise it remains silent and no longer participates in the important tag distribution identification; Each reader performs the important tag identification work in a single coverage area within its scheduling time, and all readers identify in parallel. After all scheduling is completed, the readers exchange information with each other based on the scheduling algorithm to complete the distribution identification of important tags in multiple coverage areas.

2. According to the method of claim 1, for an important label k j , the calculation formula of the label fingerprint and the hash equation index inserted into CF is as follows: F j =H(ID k ,s f )mod 2 d in, F j is the fingerprint of the tag, is the index of h candidate buckets, H(·) is the hash function, and the reader broadcasts h hash seeds to the tag to simulate h hash functions; s f and h They are the hash seeds used to calculate fingerprints and candidate buckets, respectively. and d are the length of the second layer of the incremental filter ICF and the length of the label fingerprint inserted in ICF, respectively.

3. The method according to claim 1, characterized in that The reader constructs a traditional cuckoo filter CF based on a subset of important tags, including: For an important tag, the reader checks the h buckets in the CF. If any of the h buckets is empty, the reader inserts the important tag into the CF by putting the ID of the important tag into any empty bucket. Otherwise, the reader randomly selects one of the h buckets and replaces the important tag already in the bucket with the important tag. The replaced important tag will be re-shifted with the same insertion steps. This process continues until all important tags are successfully inserted or the maximum iteration is reached. If the construction fails, the reader increases the length of the CF and rebuilds the CF until all important tags are successfully inserted. For each non-empty bucket, the ID of the important tag is replaced with the fingerprint of the tag. Empty buckets are not processed for the time being.

4. The method according to claim 1, characterized in that: Useless hash function refers to the set There is no equation for inserting important tags in , where h is the number of hash equations that successfully construct CF. , there are no unused hash equations.

5. The method according to claim 2, characterized in that: The ICF and its construction parameters broadcast by the reader in the system are Where △d and △h represent the maximum fingerprint difference and the number of bits required for the hash equation index, respectively. u It is a useless hash function index.

6. The method according to claim 1, characterized in that After receiving the query request, each tag calculates and checks whether it can pass the ICF based on the received ICF construction parameters, including: Check each bucket of the first layer of ICF. For the i-th bucket, check whether the bucket is empty, including two cases: h u If it is an empty set, check whether i is in the vector EI. If it exists, the i-th bucket is considered empty; h u When it is not an empty set, the hash value is calculated using the hash function index stored in the bucket, denoted as If the second layer The value stored in the bucket is h u , then the bucket is considered empty; In both cases, if the i-th bucket is empty, the label will skip it, otherwise the label will With the second layer If there is no match, jump to the next bucket. If there is a match, calculate its fingerprint and compare it with the sum of the last d bits of the elements in the previous i buckets in the first layer of ICF. If they match, increase its success counter p. k and continue matching, otherwise, skip the bucket; the search continues until h times or all buckets have been checked; if p k =1, the tag passes the lookup and remains active; otherwise the tag is silenced.

7. The method according to claim 1, characterized in that The time slot index setting method for each label passing through ICF is as follows: Let SI represent the time slot index of the important tag, and the index of the first-layer bucket when the important tag record passes through ICF, denoted as p j , and set SI = p j ; Only tags that successfully pass ICF will be assigned a unique time slot index, and tags that match multiple buckets will not pass ICF and remain silent; or each important tag calculates its time slot index by calculating the number of non-empty buckets before the bucket it matches in the second layer.

8. An important tag distribution identification system based on incremental cuckoo filter, applied to an RFID system including multiple readers, important tags and ordinary tags, characterized in that: The reader is configured to: A cuckoo filter CF is constructed based on a subset of important tags, and an incremental cuckoo filter ICF is constructed based on CF. ICF consists of two layers. In the first layer, the reader stores the difference △F of the continuous fingerprints of the tag i And the hash function index h used to successfully insert the important tags into CF h ; The second-layer bucket stores the hash function index used to successfully insert important tags into the CF; for empty buckets, the reader fills them with a random useless hash function index. If there is no useless hash function index, the reader constructs a vector EI to record the index of the empty bucket, where important tags are defined as the set of tags that the upper-layer application needs to monitor, and ordinary tags are defined as other tags other than important tags in the system; Broadcast a query request containing the ICF and its construction parameters in the system; The important tags in a single coverage area are identified within their own scheduling time, and all readers identify in parallel. After all scheduling is completed, the readers exchange information with each other based on the scheduling algorithm to complete the distribution identification of important tags in multiple coverage areas. The tag is configured as: Receive the query request containing ICF and its construction parameters sent by the reader, calculate and detect whether it can pass the ICF based on the received ICF construction parameters, and remain active if it passes, otherwise remain silent and no longer participate in the important tag distribution identification.

Images