A method for identity resolution and search based on fuzzy comprehensive evaluation and replica placement

Abstract

The invention discloses an identification resolution and search method based on fuzzy comprehensive evaluation and replica placement, comprising the steps of: first, establishing a hierarchical Chord ring based on a new key-value pair format; second, collecting load information based on an improved small-world model; third, establishing a node performance comprehensive evaluation factor set and an evaluation set; fourth, establishing a comprehensive evaluation model for node performance and performing evaluation; fifth, establishing hotspot information nodes and load situation collection nodes based on node performance evaluation; sixth, calculating a node set for resource replica placement and updating the value of a key-value pair; and seventh, searching for resources based on a replica placement node set and a functional node. The invention utilizes a node performance evaluation scheme to select a suitable node to store resource replicas, store load conditions and store hotspot information, and utilizes an improved small-world model to obtain the load status of all nodes and store them in the load node, thereby balancing the load and improving the security of a structured P2P network based on a DHT technology.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial Internet identity resolution, and in particular to an identity resolution search method based on fuzzy comprehensive evaluation and replica placement. Background Art

[0002] With the development of industrial manufacturing and the rise of emerging technologies, the field of industrial Internet has ushered in major development opportunities. Industrial enterprises have an increasingly strong demand for industrial software data interaction, and data integration between industrial software is imminent. However, the lack of supporting industrial software has become a shortcoming in my country's march towards becoming a manufacturing power. The difficulty of data circulation and exchange has become increasingly prominent.

[0003] At present, the structured P2P network still has the following problems: the DHT-based P2P network uses a hash algorithm to evenly distribute query tasks and resources, but ignores the load balancing problem caused by the differences between nodes; the p2p network allows nodes to join without bias, which makes the network vulnerable to attacks; due to the existence of hot information, some nodes may be overloaded at a certain moment; mainstream routing algorithms such as Chord use the minimum number of hops as the standard for selecting a path, and in fact there will be a high delay at nodes with heavy loads. Therefore, the present invention proposes an identity resolution search method based on fuzzy comprehensive evaluation and replica placement to solve the problems existing in the prior art. Summary of the invention

[0004] In view of the above problems, the purpose of the present invention is to propose an identity resolution search method based on fuzzy comprehensive evaluation and replica placement. The identity resolution search method based on fuzzy comprehensive evaluation and replica placement uses a node performance evaluation scheme to select appropriate node storage resource replicas, storage load conditions and storage hotspot information, and uses an improved small-world model to obtain the load status of all nodes and store them in the load nodes, thereby balancing the load and improving the security of a structured P2P network based on DHT technology.

[0005] To achieve the purpose of the present invention, the present invention is implemented by the following technical solution: an identification resolution search method based on fuzzy comprehensive evaluation and replica placement, comprising the following steps:

[0006] Step 1: Based on the new key-value pair format and according to the industrial database layer, a Chord ring is established, and then the Chord ring is evenly segmented. Then, a hash table and a routing table are established for each node, and a new key-value pair format is defined. Finally, the current node is judged and updated;

[0007] Step 2: Collect load information based on the improved small-world model, use the NW model to build a virtual topology network overlaid on the current peer-to-peer network, and obtain the load status of each node;

[0008] Step 3: Evaluate the node performance based on the fuzzy comprehensive evaluation model, and establish a node performance comprehensive evaluation factor set and an evaluation set for node performance comprehensive evaluation;

[0009] Step 4: Perform a single factor fuzzy evaluation on each node performance indicator in the node performance comprehensive evaluation factor set to obtain a node performance comprehensive evaluation matrix, then obtain a comprehensive evaluation model of node performance by calculating the matrix, and use the comprehensive evaluation model synthesis operator to determine the comprehensive evaluation score of node performance;

[0010] Step 5: Based on the comprehensive evaluation of node performance, the performance of nodes in the Chord ring is evaluated, and the node with the highest performance is used as the hot information node and load situation collection node;

[0011] Step 6: Calculate the set of nodes where resource replicas are placed based on routing diversity, and update the value of the key-value pair;

[0012] Step 7: Find resources based on the replica placement node set and function nodes.

[0013] A further improvement is that: the nodes in the Chord ring after segmentation in step one and the successor nodes directly before and after are connected by pointers, and the new key-value pair format in step one is [key, value1, value2], key is the identifier of the resource, and when the stored key-value pair is a copy, the node ID where the original resource key-value pair is placed is used as value1, otherwise the node ID where the key-value pair is currently stored is used as value1, and the IP location where the actual information is stored is used as value2; when the value of value1 is the same as the ID value of the current node, it indicates that there is no copy.

[0014] A further improvement is that: the determination and updating of the current node in step 1 includes, when the current node is a super node, in addition to the routing table defined by the original Chord algorithm, an additional routing table containing information of all other super nodes is stored;

[0015] When the current node is a load status storage node, it is necessary to additionally store the load status of all nodes in its subnet;

[0016] When the current node is a hotspot information node, it is necessary to additionally store key-value pairs of the hotspot information that has been queried in its subnet within any period.

[0017] A further improvement is that: the specific method for obtaining the load status in step 2 is:

[0018] A1. Assume that the upper virtual network is initially a regular network with N nodes arranged in a ring. The degree of each node is K, that is, each vertex is connected to its K neighbors, with K / 2 neighbors on each side, and N>>K>>ln(N)>>1;

[0019] A2. For the nodes in the regular network, disconnect the original edges with a probability p for any one of them, randomly select new nodes to reconnect, excluding self-loops and duplicate connections, and the reconnected edges will not be disconnected;

[0020] A3. Repeat A2 until all nodes are traversed to obtain the load status of each node.

[0021] The further improvement lies in that: when establishing the comprehensive evaluation factor set of node performance in step three, a set composed of various factors affecting the performance of functional nodes is comprehensively considered, including hot spot information nodes, load situation collection nodes, and key-value pair replica storage nodes, and the performance index elements in the set all have fuzziness.

[0022] The further improvement lies in that: the method for calculating the matrix in step four is

[0023] B1. Use △X = {△Xi丨i = 1, 2,..., 5} to represent five performance index eigenvalues determined according to the effect evaluation level within the node performance quantization value interval, and construct the membership function μ(x) according to the determination method of membership degree, which is represented by the following formula

[0024]

[0025]

[0026]

[0027] Take the membership degree matrix calculated by the membership function for node performance as the fuzzy comprehensive evaluation matrix;

[0028] B2. Use the AHP (Analytic Hierarchy Process) to obtain the weight vector set corresponding to the node performance evaluation index factor set, then use the consistent matrix method to construct the judgment matrix, and compare the pairwise comparison between the node performance indicators to obtain the factor weight vector;

[0029] B3. Through fuzzy transformation, transform the fuzzy vector to obtain the comprehensive evaluation model of node performance.

[0030] The further improvement lies in that: the node performance indicators in step four include five performance evaluation indicators: the main frequency of the CPU, the cache of the CPU, memory, communication rate, and load status.

[0031] The further improvement lies in that: step six specifically includes

[0032] C1. Calculate the replica location set R i，j where N represents the total number of nodes, B represents the replication factor, d represents the replication degree, R i，j represents the location of the jth replica in the i-th round of calculation, ki，j Indicates the storage location of the original key-value pair based on the hash map. In round i, there is B i-1 The replicas are placed equidistantly on the Chord ring. The set of replica positions generated in the i-th round is

[0033] R i，j =(k i，j , k i，j +s i , k i，j +2s i , …, k i，j +(B i-1 -1)s i )

[0034] C2. Based on the comprehensive evaluation model of node performance in step 4, the replica location set R of each round is i，j The performance of each element in is evaluated, and finally three replica storage locations with the best performance are selected;

[0035] C3. The key-value pair is stored at the nodes corresponding to the calculated three replica storage locations, namely, the key-value pair replica storage nodes, and the value of the key-value pair is updated at the same time.

[0036] The beneficial effects of the present invention are as follows: the present invention introduces a performance evaluation scheme based on a fuzzy comprehensive evaluation model, which provides a basis for selecting appropriate hot information nodes, load situation collection nodes and replica key-value pair storage nodes;

[0037] A performance evaluation scheme based on a fuzzy comprehensive evaluation model is proposed to select hotspot information nodes and load collection nodes, which is conducive to quickly locating hotspot information, while avoiding frequent forwarding query requests, reducing the load of related nodes, and avoiding nodes with high loads during routing, thereby alleviating the delay caused by load imbalance and excessive load on some nodes;

[0038] A performance evaluation scheme based on a fuzzy comprehensive evaluation model is proposed to select key-value pair replica storage nodes. The key-value pair replica storage nodes avoid search failures caused by the loss of stored resources after some nodes are maliciously attacked through the storage of replicas. At the same time, different search paths are provided for searching target resources, which is conducive to reducing load and improving search speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 The present invention is a flowchart of parsing and searching according to an embodiment of the present invention. DETAILED DESCRIPTION

[0040] In order to deepen the understanding of the present invention, the present invention will be further described in detail below in conjunction with examples. The examples are only used to explain the present invention and do not constitute a limitation on the protection scope of the present invention.

[0041] Embodiment 1

[0042] according to Figure 1 As shown, this embodiment provides an identification resolution search method based on fuzzy comprehensive evaluation and replica placement, including the following steps:

[0043] Step 1: Based on the new key-value pair format and according to the industrial database layer, a Chord ring is established, and then the Chord ring is evenly segmented. Then, a hash table and routing table for each node are established, and a new key-value pair format is defined. Finally, the current node is judged and updated, including when the current node is a super node, in addition to the routing table defined by the original Chord algorithm, an additional routing table containing information of all other super nodes is stored;

[0044] When the Chord ring is established, the top-level nodes in Beijing, Shanghai, Guangdong, Hubei, and Chongqing are used to evenly divide the entire Chord ring into five sections. Each national top-level node is used as a super node in a section of the Chord ring and is assigned the smallest m-bit ID value in the current section of the Chord ring. The current section of the Chord ring is managed by the top-level node as a subnet. The secondary nodes are evenly distributed in the Chord ring section where the top-level nodes are located based on the principle of geographical proximity. The national top-level nodes and secondary nodes are used as super nodes.

[0045] When the current node is a load status storage node, it is necessary to additionally store the load status of all nodes in its subnet;

[0046] When the current node is a hotspot information node, it is necessary to additionally store the key-value pairs of the hotspot information that has been queried in its subnet within any period;

[0047] The nodes in the segmented Chord ring are connected to the successor nodes directly before and after through pointers. The new key-value pair format is [key, value1, value2]. Key is the resource identifier. When the stored key-value pair is a copy, the node ID where the original resource key-value pair is placed is used as value1. Otherwise, the node ID currently stored in this key-value pair is used as value1, and the IP location where the actual information is stored is used as value2. When the value of value1 is the same as the ID value of the current node, it means there is no copy.

[0048] Step 2: Collect load information based on the improved small-world model, use the NW model to build a virtual topology network overlaid on the current peer-to-peer network, and obtain the load status of each node. The specific method is as follows:

[0049] A1. Assume that the upper-layer virtual network is initially a regular network with N nodes arranged in a ring, where the degree of each node is K, that is, each vertex is connected to its K neighbors, with K / 2 on each side, and N >> K >> ln(N) >> 1;

[0050] A2. For the nodes in the regular network, disconnect the original edges with a probability p of any one, and randomly select new nodes to reconnect, excluding self-loops and duplicate connections, and the reconnected edges will not be disconnected;

[0051] A3. Repeat A2 until all nodes are traversed to obtain the load status of each node;

[0052] Step 3. Evaluate the node performance based on the fuzzy comprehensive evaluation model, and establish a comprehensive evaluation factor set for node performance and an evaluation set for comprehensive evaluation of node performance;

[0053] When establishing the comprehensive evaluation factor set for node performance, consider the set composed of various factors affecting the performance of functional nodes, including hot information nodes, load situation collection nodes, and key-value pair replica storage nodes. The performance index elements in the set are all fuzzy;

[0054] Step 4. Conduct a single-factor fuzzy evaluation on each node performance index in the comprehensive evaluation factor set for node performance to obtain a comprehensive evaluation matrix for node performance. Then, obtain a comprehensive evaluation model for node performance by calculating the matrix, and use the synthesis operator of the comprehensive evaluation model to determine the comprehensive evaluation score of node performance;

[0055] The method for calculating the matrix is

[0056] B1. Let △X = {△Xi丨i = 1, 2,..., 5} represent five performance index eigenvalues determined according to the effect evaluation level within the node performance quantization value range. Construct a membership function μ(x) according to the method for determining membership degree, which is represented by the following formula

[0057]

[0058]

[0059]

[0060] Take the membership matrix calculated by the membership function for node performance as the fuzzy comprehensive evaluation matrix;

[0061] B2. Use the AHP (Analytic Hierarchy Process) to obtain the weight vector set corresponding to the node performance evaluation index factor set, and then use the consistent matrix method to construct a judgment matrix. Compare the pairwise comparison between each node performance index to obtain the factor weight vector;

[0062] B3, transform the fuzzy vector through fuzzy transformation to obtain a comprehensive evaluation model of node performance;

[0063] Step 5: Based on the comprehensive evaluation of node performance, the performance of nodes in the Chord ring is evaluated, and the node with the highest performance is used as the hot information node and load situation collection node;

[0064] Step 6: Based on routing diversity, calculate the set of nodes where resource replicas are placed and update the values ​​of the key-value pairs, including

[0065] C1. Perform copy location set R i，j , where N represents the total number of nodes, B represents the replication factor, and d represents the replication degree. R i，j represents the position of the jth replica in the i-th round of calculation, k i，j Indicates the storage location of the original key-value pair based on the hash map. In round i, there is B i-1 The replicas are placed equidistantly on the Chord ring. The set of replica positions generated in the i-th round is

[0066] R i，j =(k i，j , k i，j +s i , k i，j +2s i , …, k i，j +(B i-1 -1)s i )

[0067] C2. Based on the comprehensive evaluation model of node performance in step 4, the replica location set R of each round is i，j The performance of each element in is evaluated, and finally three replica storage locations with the best performance are selected;

[0068] C3. Store the key-value pair at the nodes corresponding to the calculated three replica storage locations, i.e., the key-value pair replica storage nodes, and update the value of the key-value pair at the same time.

[0069] Step 7: Find resources based on the replica placement node set and function nodes.

[0070] Embodiment 2

[0071] according to Figure 1 As shown, this embodiment provides an identification resolution search method based on fuzzy comprehensive evaluation and replica placement, including the following steps:

[0072] Step 1: Based on the new key-value pair format and according to the industrial database hierarchy, a Chord ring is established, and then the Chord ring is evenly segmented. Each top-level node is used as a super node in a Chord ring segment, and is assigned the m-bit ID value with the smallest value in the current Chord ring segment. Each secondary node is evenly distributed in the Chord ring segment where each top-level node is located based on the principle of geographical proximity. The enterprise node uses the node's IP address as the key and the consistent hash function SHA-1 as the hash function, and first assigns an m-bit node ID through hash mapping. When storing resource key-value pairs, the resource identifier is used as the key, and is dispersed clockwise on the Chord ring according to the hash value of the key, and is stored on the first successor node in the clockwise direction.

[0073] Then, a hash table and routing table for each node are established, and a new key-value pair format is defined. The new key-value pair format is [key, value1, value2]. The key is the resource identifier. When the stored key-value pair is a copy, the node ID where the original resource key-value pair is placed is used as value1. Otherwise, the node ID currently stored in the key-value pair is used as value1, and the IP location where the actual information is stored is used as value2. When the value of value1 is the same as the ID value of the current node, it means there is no copy.

[0074] Finally, the current node is judged and updated, including when the current node is a super node, in addition to the routing table defined by the original Chord algorithm, an additional routing table containing information of all other super nodes is stored;

[0075] When the current node is a load status storage node, it is necessary to additionally store the load status of all nodes in its subnet;

[0076] When the current node is a hotspot information node, it is necessary to additionally store the key-value pairs of the hotspot information that has been queried in its subnet within any period;

[0077] The nodes in the segmented Chord ring are connected to the immediate successor nodes through pointers. If a node N needs to be deleted, the tail of the linked list of the immediate predecessor node of node N is changed to point to its immediate successor node, and then the head of the linked list of its immediate successor node is changed to point to the immediate predecessor node of node N, and then the node N is deleted from the chord ring; if a node N is to be added between nodes N0 and N1, it is only necessary to change the tail of the linked list of node N to point to the head of node N1, and then change its linked list head to point to node N0;

[0078] Step 2: Collect load information based on the improved small-world model, use the NW model to build a virtual topology network overlaid on the current peer-to-peer network, and obtain the load status of each node. The specific method is as follows:

[0079] A1. Assume that the upper virtual network is initially a regular network with N nodes arranged in a ring. The degree of each node is K, that is, each vertex is connected to its K neighbors, with K / 2 neighbors on each side, and N>>K>>ln(N)>>1;

[0080] A2. For nodes in a regular network, disconnect the original edges with any probability p and randomly select new nodes to reconnect, excluding self-loops and repeated connections, and the reconnected edges will not be disconnected;

[0081] A3. Repeat A2 until all nodes are traversed to obtain the load status of each node;

[0082] Step 3: Evaluate the node performance based on the fuzzy comprehensive evaluation model, and establish a node performance comprehensive evaluation factor set and an evaluation set for node performance comprehensive evaluation;

[0083] Establish the factor set U for comprehensive evaluation of node performance i =(u1, u2, ..., u5), where U1 represents the CPU's main frequency, U2 represents the CPU's cache, U3 represents the memory, U4 represents the communication rate, and U5 represents the load condition. This factor set is a set of various factors that comprehensively consider the performance of functional nodes (i.e., hot information nodes, load condition collection nodes, and key-value pair replica storage nodes). The performance indicator elements in this set have varying degrees of fuzziness;

[0084] The evaluation set V for comprehensive evaluation of node performance is established. The evaluation set V is a set of various performance levels of the evaluated nodes. i =(v1, v2, ..., v5), where the element V j represents the jth evaluation result of node performance. Here, the evaluation set for evaluating node performance is V i =(v1, v2, ..., v5), which respectively represent very good performance, good performance, average performance, poor performance, and very poor performance. It can also be modified according to the needs of specific situations;

[0085] Step 4: Perform a single factor fuzzy evaluation on each node performance indicator in the node performance comprehensive evaluation factor set to obtain a node performance comprehensive evaluation matrix, then obtain a comprehensive evaluation model of node performance by calculating the matrix, and use the comprehensive evaluation model synthesis operator to determine the comprehensive evaluation score of node performance;

[0086] If the i-th node performance index factor in the factor set U is j The membership degree of the jth node performance level element is R i，j , then the single factor evaluation result of the performance index of the i-th node is expressed as R using fuzzy set i =(r1,1 , r 1,2 , …, r 1,5 ). There are 5 performance evaluation indicators including the main frequency of the CPU, the cache of the CPU, the memory, the communication rate, and the load status. A matrix R is formed with the single-factor evaluation sets of each performance evaluation indicator as rows 5*5 , which is the fuzzy comprehensive evaluation matrix of the node performance. The specific calculation method is as follows:

[0087] B1. Let △X = {△Xi | i = 1, 2, …, 5} represent five performance index characteristic values determined according to the effect evaluation level within the node performance quantization value range. The membership function μ(x) is constructed according to the determination method of membership degree and is represented by the following formula

[0088]

[0089]

[0090]

[0091] Take the membership degree matrix calculated by the membership function for node performance as the fuzzy comprehensive evaluation matrix;

[0092] B2. Use the AHP (Analytic Hierarchy Process) to obtain the weight vector set corresponding to the node performance evaluation index factor set. The highest-level decision-making objectives are the hot information node, the load situation collection node, and the key-value pair replica storage node respectively. The factors in the middle layer are the five node performance indicators; then use the consistent matrix method to construct the judgment matrix, and determine the appropriate scale according to the pairwise comparison between the node performance indicators by the 1-9 scale method proposed by Santy to obtain the factor weight vector;

[0093] B3. After determining the single-factor judgment matrix R of the node performance indicators and the factor weight vector A, through fuzzy transformation, the fuzzy vector A on U is changed into the fuzzy vector B on V, that is, B = Thus, the comprehensive evaluation model of the node performance is obtained. Among them is called the comprehensive evaluation synthesis operator. The comprehensive evaluation synthesis operator in this solution selects the general matrix multiplication;

[0094] After the comprehensive evaluation model of the node performance is determined, continue to use the comprehensive evaluation synthesis operator to determine the comprehensive score of the node performance, that is where F is the comprehensive score of the node performance, and S is the score of the corresponding performance indicator in the evaluation set V of the node performance comprehensive evaluation. Here, the score is set as S = (100, 80, 60, 40, 20);

[0095] Step 5: Based on the comprehensive evaluation of node performance, the performance of the nodes in the Chord ring is evaluated. When evaluating the hot information node, since the function of the hot information node is similar to that of the cache, the weight of the CPU main frequency is set to be the highest, that is, the value of ω1 in the weight vector ω is the largest; when evaluating the load node, since the main function of the load node is to record the load status of all nodes in this section of the Chord ring, the weight of the memory is set to be the highest, that is, the value of ω3 in the weight vector ω is the largest. In this way, a node with the highest performance is obtained as the hot information node and the load node respectively;

[0096] Step 6: Based on routing diversity, calculate the set of nodes where resource replicas are placed and update the values ​​of the key-value pairs, including

[0097] C1. Perform copy location set R i，j , where N represents the total number of nodes, B represents the replication factor, and d represents the replication degree. R i，j represents the position of the jth replica in the i-th round of calculation, k i，j Indicates the storage location of the original key-value pair based on the hash map. There are a total of (m+1) rounds of operations. Each of the first m rounds includes (B-1) steps, and the last round includes n=(d-1)mod(B-1) steps. In the i-th round, there are B i-1 The replicas are placed equidistantly on the Chord ring. The set of replica positions generated in the i-th round is

[0098] R i，j =(k i，j , k i，j +s i , k i，j +2s i , …, k i，j +(B i-1 -1)s i )

[0099] C2. Based on the comprehensive evaluation model of node performance in step 4, the replica location set R of each round is i，j The performance of each element in is evaluated, and finally three replica storage locations with the best performance are selected;

[0100] C3. The key-value pair is stored at the nodes corresponding to the calculated three replica storage locations, namely, the key-value pair replica storage nodes, and the value of the key-value pair is updated at the same time.

[0101] Step 7: Find resources based on the replica placement node set and the functional node. Specifically, D1: When the client of the node N server sends a request to the database to query the resource identified as key, the current node N receives the Key value to be queried, and calculates the Key value of the possible replicas based on the Key value, and performs parallel search (the following search process takes the query of one of the Key values ​​as an example);

[0102] D2, the current node N calls out the immediate predecessor node N according to the header and tail of its own linked list p and the direct successor node N s ID, determine whether there is N p .ID≤hash(key)≤N.ID or N.ID≤hash(key)≤N s .ID. If it is satisfied, the key-value pair storage location of the resource is located at node N or node N's direct successor node N s In the key-value pair, you can directly search for it, so that you can get the information you need from the corresponding node according to value2 in the key-value pair;

[0103] D3. If the Key value is not located at the current node N or the direct successor node N of node N s The node N that receives the query request will query its own routing table and forward the query request for this Key value to the secondary node S where the subnet is located. N Department;

[0104] D4, secondary node S N According to the header and footer of its own linked list, it is determined whether the Key value is located in node A or A's successor node. If so, the information required by the user can be found and the node ID is returned to the user's client. If not, step D4 is continued.

[0105] D5, secondary node S N Determine which secondary node manages the subnet area where the key value of this resource is located based on its own routing table, and then send this key value to the corresponding secondary node S;

[0106] D6, the secondary node S sends the Key value to the hotspot information node in the managed subnet area to determine whether the Key value is hotspot information. If so, the search is completed and the information is returned to the node N, otherwise, proceed to step D7;

[0107] D7. The secondary node S uses the Chord algorithm to search and forward the query request according to its own routing table. If it encounters a node B whose load condition is higher than the threshold, it forwards the query request to the node's direct predecessor node B. p , continue searching until the key-value pair corresponding to the Key value is found.

[0108] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for identity resolution and search based on fuzzy comprehensive evaluation and replica placement, characterized in that: It includes the following steps: Step 1: Based on the new key-value pair format and according to the industrial database hierarchy, establish a Chord ring, then evenly segment the Chord ring, and then establish a hash table and a routing table for each node, and define a new key-value pair format. Finally, judge and update the current node; Step 2: Collect load information based on the improved small-world model, use the NW model to construct a virtual topology network covering the current peer-to-peer network, and obtain the load status of each node; Step 3: Evaluate the node performance based on the fuzzy comprehensive evaluation model, and establish a comprehensive evaluation factor set for node performance and an evaluation set for comprehensive evaluation of node performance; Step 4: Conduct a single-factor fuzzy evaluation on each node performance index in the comprehensive evaluation factor set of node performance to obtain a comprehensive evaluation matrix of node performance. Then, obtain a comprehensive evaluation model of node performance by calculating the matrix, and use the synthesis operator of the comprehensive evaluation model to determine the comprehensive evaluation score of node performance; The method for calculating the matrix is B1. Let △X = {△Xi丨i = 1, 2,..., 5} represent five performance index characteristic values determined according to the effect evaluation level within the node performance quantization value range. Construct a membership function μ(x) according to the determination method of membership degree, which is represented by the following formula Take the membership degree matrix calculated by the membership function for node performance as the fuzzy comprehensive evaluation matrix; B2. Use the AHP (Analytic Hierarchy Process) to obtain the weight vector set corresponding to the node performance evaluation index factor set, then use the consistent matrix method to construct a judgment matrix, and compare each pair of node performance indicators to obtain the factor weight vector; B3. Transform the fuzzy vector through fuzzy transformation to obtain a comprehensive evaluation model of node performance; Step 5: Evaluate the performance of the nodes in the Chord ring based on the comprehensive evaluation of node performance, and use the node with the highest performance as the hot information node and the load situation collection node; Step 6: Calculate the resource replica placement node set according to the routing diversity, and update the value of the key-value pair; Step 7: Search for resources based on the replica placement node set and functional nodes.

2. According to claim 1, a method for identity resolution and search based on fuzzy comprehensive evaluation and replica placement is characterized in that: The nodes in the segmented Chord ring in Step 1 are connected by pointers to the directly preceding and succeeding nodes. The new key-value pair format in Step 1 is [key, value1, value2], where key is the identifier of the resource. When the stored key-value pair is a replica, the node ID where the original resource key-value pair is placed is used as value1. Otherwise, the node ID where the current key-value pair is stored is used as value1, and the actual IP location where the information is stored is used as value2; when the value of value1 is the same as the ID value of the current node, it means there is no replica.

3. According to claim 1, the identification resolution and search method based on fuzzy comprehensive evaluation and replica placement is characterized in that: The judgment and update of the current node in Step 1 include that when the current node is a super node, in addition to the routing table defined by the original Chord algorithm, a routing table containing information of all other super nodes needs to be additionally stored; When the current node is a load situation storage node, the load status of all nodes in its subnet needs to be additionally stored; When the current node is a hot information node, the key-value pairs of the hot information queried in its subnet within any period need to be additionally stored.

4. According to claim 1, the identification resolution and search method based on fuzzy comprehensive evaluation and replica placement is characterized in that: The specific method for obtaining the load status in step 2 is: A1. Assume that the upper virtual network is initially a regular network with N nodes arranged in a ring. The degree of each node is K, that is, each vertex is connected to its K neighbors, with K / 2 neighbors on each side, and N>>K>>ln(N)>>1; A2. For nodes in a regular network, disconnect the original edges with any probability p and randomly select new nodes to reconnect, excluding self-loops and repeated connections, and the reconnected edges will not be disconnected; A3. Repeat A2 until all nodes are traversed to obtain the load status of each node.

5. According to claim 1, the identification resolution and search method based on fuzzy comprehensive evaluation and replica placement is characterized in that: When establishing the node performance comprehensive evaluation factor set in step three, a set of factors that affect the performance of functional nodes is considered comprehensively, including hot information nodes, load situation collection nodes and key-value pair replica storage nodes, and the performance indicator elements in the set are all fuzzy.

6. The identification resolution and search method based on fuzzy comprehensive evaluation and replica placement according to claim 1 is characterized in that: The node performance indicators in step 4 include five performance evaluation indicators: CPU main frequency, CPU cache, memory, communication rate and load status.

7. The identification resolution and search method based on fuzzy comprehensive evaluation and replica placement according to claim 1 is characterized in that: The step six specifically includes C1. Perform copy location set R i，j , where N represents the total number of nodes, B represents the replication factor, and d represents the replication degree. R i，j represents the position of the jth replica in the i-th round of calculation, k i，j Indicates the storage location of the original key-value pair based on the hash map. In round i, there is B i-1 The replicas are placed equidistantly on the Chord ring. The set of replica positions generated in the i-th round is R i，j =(k i，j ,k i，j +s i ,k i，j +2s i ,…,k i，j +(B i-1 -1)s i ) C2. Based on the comprehensive evaluation model of node performance in step 4, the replica location set R of each round is i，j The performance of each element in is evaluated, and finally three replica storage locations with the best performance are selected; C3. The key-value pair is stored at the nodes corresponding to the calculated three replica storage locations, namely, the key-value pair replica storage nodes, and the value of the key-value pair is updated at the same time.

Images