Fault-tolerant routing method based on double-layer security state model

Abstract

The invention belongs to the technical field of network communication and distributed computing, and discloses a fault-tolerant routing method based on a double-layer security state model, which comprises the following steps of: 1, selecting position information of two different fault-free nodes in a large-scale cluster distributed network as a source node and a target node, dividing the position information of the source node and the target node into two parts; 2, constructing a double-layer safety state model; 3, based on the constructed double-layer safety state model, evaluating the safety states of the source node and the target node; 4, checking the hierarchical relationship of the nodes; and step 5, constructing an initial path from the source node to the target node, verifying path fault tolerance in the cluster, and returning to a final fault-tolerant path from the source node to the target node. According to the method, accurate evaluation of node and link states in the network is realized, data transmission path selection is optimized, the fault-tolerant capability and transmission efficiency of the network are comprehensively improved, and support is provided for efficient and safe communication in a complex dynamic environment.

Description

Technical Field

[0001] The invention belongs to the technical field of network communication and distributed computing, and in particular relates to a fault-tolerant routing method based on a double-layer security state model. Background Art

[0002] With the widespread application of large-scale distributed systems in data centers, supercomputing networks, and cloud computing platforms, network reliability and efficiency have gradually become the key bottlenecks of system performance. However, existing network topologies and routing mechanisms have significant deficiencies when facing complex dynamic failure scenarios. In the network topology, we generally call servers nodes, two servers connected to the same switch adjacent nodes, and two adjacent nodes are connected to form a link. Direct node or link failures are caused by a variety of reasons such as hardware damage, resource overload, and network attacks. Traditional routing algorithms usually rely on fixed paths or hop count optimization, lack the ability to flexibly respond to dynamic failures, and can easily lead to transmission interruptions or performance degradation. At the same time, modern network topologies such as The complexity and multi-dimensional characteristics of ,Fat Tree, etc. further increase the computational complexity of path selection.,Since these algorithms are often difficult to adapt to the dynamic changes of node states,,they only focus on local optimization, resulting in poor overall network performance.

[0003] In addition, existing fault-tolerant routing algorithms generally lack refined assessments of node and link security, and are mostly based on binary classifications of "normal" or "faulty", failing to provide a reliable risk quantification mechanism to intelligently avoid high-risk areas. When faced with multi-point failures or complex failure scenarios, the responsiveness and dynamic adjustment capabilities of these algorithms are insufficient. At the same time, as the scale of the network expands, many algorithms have poor adaptability in large-scale distributed systems and are difficult to achieve effective optimization in multi-dimensional topological structures, limiting their application in data centers and supercomputing environments. Therefore, there is an urgent need for a new fault-tolerant routing algorithm that can achieve dynamic security status assessment, real-time path adjustment, and high scalability. Summary of the invention

[0004] In order to solve the current large-scale cluster distributed network In order to solve the security and reliability problems of the network during data transmission, the present invention proposes a fault-tolerant reasoning method based on a double-layer security state model. The method proposes a double-layer security state model and its application in a fault-tolerant routing algorithm. And the multi-layer security status assessment mechanism can realize the accurate assessment of the node and link status in the network, optimize the selection of data transmission paths, comprehensively improve the network's fault tolerance and transmission efficiency, and provide support for efficient and secure communications in complex dynamic environments.

[0005] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:

[0006] The present invention is a fault-tolerant routing method based on a double-layer security state model, which is applied to a large-scale cluster distributed network and specifically includes the following steps:

[0007] Step 1: Select the location information of two different fault-free nodes in the large-scale cluster distributed network as the source node and the target node , the source node and the target node The location information is divided into two parts;

[0008] Step 2: construct a two-layer security state model, wherein the two-layer security state model includes an intra-cluster security state model and an inter-cluster security state model;

[0009] Step 3: Evaluate the source node based on the constructed two-layer security status model and the target node The security status of the source node is calculated Unsafe factor and the source node The insecurity factor of the link between its adjacent points , and find the source node The neighboring point with the smallest objective function;

[0010] Step 4: Check the node hierarchy and determine the source node and the target node Are they in the same cluster? If the source node and the target node In the same cluster, the cluster content error path search method is called to find the source node in the cluster. To the target node The fault-tolerant path is stored in the path collection in the cluster. If the source node and the target node If the nodes are not in the same cluster, call the inter-cluster fault-tolerant path search method to find the inter-cluster slave source node. To the target node The fault-tolerant path is stored in the inter-cluster path set. middle.

[0011] Step 5: Build from the source node To the target node The initial path , verify the path fault tolerance within the cluster and return to the source node To the target node The final fault-tolerant path .

[0012] A further improvement of the present invention is that: the security model in the cluster obtains the source node To the target node The security information on the path is the node status and link status. The specific method for obtaining the security information on the path includes the following steps:

[0013] Step 2.1.1: The in-cluster security state model takes a single cluster as the basic unit and converts the source nodes in the large-scale cluster into , target node , source node Neighbor nodes and target nodes The neighbor nodes are regarded as the core components of large-scale cluster distributed networks;

[0014] Step 2.1.2: Source nodes within a large-scale cluster , target node , source node Neighbor nodes and target nodes The security status of the neighbor nodes of the cluster is classified, and all source nodes in the cluster are obtained based on the security status of all the source nodes in the cluster. , target node , source node Neighbor nodes and target nodes Security information of neighboring nodes;

[0015] A further improvement of the present invention is that in step 2.1.2, the security state model in the cluster is obtained by the source node in the same cluster. , target node , source node Neighbor nodes and target nodes The real-time status evaluation of neighbor nodes and links is used to convert the source node , target node , source node Neighbor nodes and target nodes The neighbor node states are divided into:

[0016] Secure nodes: All connections are normal and there is no direct connection to unsafe nodes;

[0017] Mildly unsafe node: a node that is connected to a completely unsafe node or a failed link, but still has a usable path;

[0018] Highly insecure nodes: nodes that are connected to two or more completely insecure nodes or failed links, or nodes that have only one secure link left around them;

[0019] Completely unsafe node: all connections are down;

[0020] The link status is divided into:

[0021] Healthy link: a link where both nodes are safe nodes;

[0022] Dangerous link: a link where at least one of the two ends is a slightly or highly unsafe node;

[0023] Broken link: A link that is physically or logically unable to transmit data.

[0024] A further improvement of the present invention is that: the inter-cluster security state model obtains the source node The first part to the target node The first part of the security information of the nodes on the first part of the path is the node status and link status of the independent node. The specific method for obtaining the information includes the following steps:

[0025] Step 2.2.1, the inter-cluster security state model takes a single cluster as the basic unit, and regards each cluster as an independent node in a large-scale cluster distributed network;

[0026] Step 2.2.2: Source nodes in large-scale clusters , target node , source node Neighbor nodes and target nodes The node status of the neighbor nodes is classified to classify the source nodes in the large-scale cluster , target node , source node Neighbor nodes and target nodes The link status between neighbor nodes;

[0027] Step 2.2.3: Based on the node status and link status of the large-scale cluster in step 2.2.2, classify the node status between clusters in the large-scale cluster distributed network, and obtain the connection status of all nodes in the cluster based on the node security status.

[0028] A further improvement of the present invention is that in step 2.2.2, the inter-cluster security state model regards the cluster as an independent node in a high-level topology in a large-scale cluster distributed network, and divides the independent node state into:

[0029] Secure nodes: All connections are normal and there is no direct connection to unsafe nodes;

[0030] Mildly unsafe node: a node that is connected to a completely unsafe node or a failed link, but still has a usable path;

[0031] Highly insecure nodes: nodes that are connected to two or more completely insecure nodes or failed links, or nodes that have only one secure link left around them;

[0032] Completely unsafe node: all connections are down;

[0033] The link status is divided into three categories:

[0034] Healthy link: a link where both nodes are safe nodes;

[0035] Dangerous link: a link where at least one of the two ends is a slightly or highly unsafe node;

[0036] Broken link: A link that is physically or logically unable to transmit data.

[0037] A further improvement of the present invention is that: Step 3 calculates the source node Unsafe factor , source node The insecurity factor of the link between its adjacent points , and find the source node The neighboring point with the minimum objective function specifically includes the following steps:

[0038] Step 3.1: Calculate the source node The insecurity factor value:

[0039]

[0040] in, Represents the source node The number of failed connections, Represents the source node The total number of connections, Represents the source node The number of adjacent unsafe nodes, Represents the source node The total number of adjacent nodes, Represents the weight factor, which is used to balance the influence of connection status and adjacency status;

[0041] Step 3.2, source node Its adjacent points The insecurity factor of the link Values:

[0042]

[0043] in, , Respectively represent the source nodes at both ends of the link Its adjacent points The insecurity factor The value of , They represent weight factors, representing the contribution of the two end nodes to the link security status;

[0044] Step 3.3: Establish the current source node The objective function Minimum neighbors:

[0045]

[0046] in, Represents the source node The neighbor nodes of Represents the source node The insecurity factor value, Represents the source node Its adjacent points The insecurity factor of the link value.

[0047] A further improvement of the present invention is that in step 4, a cluster content error path search method is called to search for the source node in the cluster. To the target node The fault-tolerant path is as follows: Initialize the path set , set the path collection Used to store the generated path. If the source node or target node Completely unsafe, returns no path, if the source node To the target node If it is a directly adjacent safe or slightly unsafe node, then directly connect to the source node To the target node , find the source node All adjacent points of , if there is a safe or slightly unsafe node, then calculate the source node Unsafe factor And select the neighboring point with the smallest objective function Update source node , repeatedly search for the source node All adjacent points of , until reaching the destination node Or all candidate nodes are unavailable, return to the source node To the target node A collection of paths , store the path nodes into the path collection within the cluster .

[0048] A further improvement of the present invention is that: in step 4, the inter-cluster fault-tolerant path search method is called to search for the inter-cluster slave node To the target node The fault-tolerant path is as follows:

[0049] Initialize and create an empty path collection , set the starting node to the source node , at the source node Neighboring points , filter out the nodes that match the target node A set of safe or slightly unsafe nodes in the same cluster or across clusters, calculating each adjacent node The insecurity factor , select the neighboring point with the lowest objective function As the next hop node, repeat the adjacent point The selection process gradually expands the path until it reaches the target node Or determine that there is no path, after each path extension, for each source node in the path To the target node The security of the connection is verified. If a high-risk connection is found, it is dynamically adjusted to a suboptimal path. In path extension, it is checked whether there are repeated nodes or loops in the path. If a loop is detected, it goes back to the previous step to reselect the path to ensure that the final path has no redundant jumps. When the path spans multiple clusters, the cross-cluster connection is dynamically adjusted according to the objective function, and the inter-cluster link with the lowest objective function is selected and returned from the source node. To the target node A collection of paths , to ensure transmission efficiency and security in cross-cluster scenarios, store path nodes in the inter-cluster path set .

[0050] A further improvement of the present invention is that: the step 5 constructs To the target node The initial path Specifically:

[0051] If the source node and the target node Located in the same cluster, construct the initial path A set of paths within the cluster ;

[0052] If the source node and the target node Not in the same cluster, the path starts from the source node , and then include the collection path The node in the path, the end point of the path is the target node ;

[0053] For the initial path For each pair of consecutive nodes in the cluster, check whether they are in the same cluster and whether they are both safe or slightly unsafe nodes. If the above conditions are met, call the cluster content wrong path search method to generate a subpath between the two consecutive nodes and replace the initial path with this subpath The direct connection in the source node ensures the security and fault tolerance of the final path, and the path back to the source node ensures the security and fault tolerance of the path. To the target node The final fault-tolerant path ,The path ensures that the node state is safe and non-duplicate under given fault tolerance conditions.

[0054] The beneficial effects of the present invention are:

[0055] The present invention significantly improves the network fault tolerance: the present invention introduces a double-layer security state model and an insecurity factor The evaluation mechanism significantly improves the fault tolerance of the network; the two-layer model of the present invention can quickly detect unsafe areas when a node or cluster fails, and The quantitative indicators accurately locate the fault points and high-risk paths. When some nodes or links fail, the fault-tolerant algorithm will automatically bypass the high-risk paths. The path with the highest value is selected, and a more secure backup path is selected, thereby ensuring a high success rate of data transmission. In a large-scale distributed network, the path obstacle avoidance mechanism of the present invention can minimize the impact of failures and improve the overall risk resistance of the network by at least 30%. Even in complex failure scenarios, the present invention can achieve efficient data forwarding and avoid global network paralysis.

[0056] The present invention has a more intelligent routing decision mechanism: compared with the traditional routing algorithm based only on path length optimization, the fault-tolerant algorithm of the present invention fully considers the real-time security status of nodes and links. The algorithm is dynamically calculated during data transmission. The system can calculate the value of the node and give priority to the safe path, and automatically adjust the routing strategy in the high-risk area to avoid potential failure threats. In addition, the fault-tolerant algorithm can comprehensively evaluate the nodes and paths in different security states. For mildly unsafe nodes, the system can judge them according to their specific conditions instead of simply eliminating them, making rational use of network resources while ensuring transmission security. Especially in the scenario of multiple path alternatives, the algorithm can quickly calculate and determine the optimal path to improve the overall network efficiency.

[0057] The present invention can improve network resource utilization and reliability: The present invention improves network resource utilization to a new level through refined classification of node and link status. Mildly insecure nodes can still be used strategically, while completely insecure nodes are quickly isolated to avoid affecting the overall stability of the network. This strategy not only optimizes network resource allocation, but also reduces the pressure on other nodes through a load balancing mechanism, thereby extending the service life of the equipment and reducing network operation and maintenance costs. At the same time, the two-layer security state model can effectively avoid the centralized exhaustion of global network resources and improve the long-term reliability of the network.

[0058] The present invention has strong adaptability and broad application prospects: the two-layer security state model and fault-tolerant routing algorithm of the present invention are highly adaptable and can be widely used in various network scenarios, including but not limited to data center networks, supercomputing networks and communication systems. Its modular design enables it to be seamlessly integrated into the existing network architecture without the need for large-scale transformation of the underlying network. In addition, for different topologies such as augmented cubic networks or other layered distributed architectures, the algorithm can be flexibly adjusted to adapt its characteristics. In future large-scale distributed computing environments and high-performance network scenarios, such as 5G core networks, edge computing, and cloud computing platforms, the present invention can provide an effective fault-tolerant solution, providing stable and reliable transmission guarantees for these highly complex networks, while meeting user needs for low latency and high stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a process model diagram of the double-layer security state model of the present invention.

[0060] Figure 2 It is a flow chart of the fault-tolerant routing method of the present invention.

[0061] Figure 3 It is a flow chart of the cluster content misrouting method of the present invention.

[0062] Figure 4 It is a flow chart of the inter-cluster fault-tolerant routing method of the present invention.

[0063] Figure 5 It is a comparison chart of transmission success rate in a small-scale test of the present invention.

[0064] Figure 6 It is a comparison chart of average path lengths in small-scale tests of the present invention. DETAILED DESCRIPTION

[0065] The following will disclose the embodiments of the present invention with drawings. For the purpose of clear description, many practical details will be described together in the following description. However, it should be understood that these practical details should not be used to limit the present invention. That is to say, in some embodiments of the present invention, these practical details are not necessary.

[0066] like Figure 1 As shown, the present invention is a fault-tolerant routing method based on a two-layer security state model, which is applied to a large-scale cluster distributed network. The fault-tolerant routing method specifically includes the following steps:

[0067] Step 1: Select the location information of two different fault-free nodes in the large-scale cluster distributed network as the source node and the target node , the source node and the target node The location information of is divided into two parts. The node is represented as , ,in and Indicates the first part of the node address, and Indicates the second part.

[0068] Step 2: Construct a two-layer security state model, which includes an intra-cluster security state model and an inter-cluster security state model. The intra-cluster security state model consists of an intra-cluster node state classification layer, an intra-cluster link state classification layer, and an insecurity coefficient calculation layer. The intra-cluster link state classification layer calculates the insecurity coefficient of each source node in the cluster. and the target node The insecurity coefficient calculation layer of the security state model in the cluster is to calculate the current source node and the target node The insecurity factor and the current source node The insecurity factor of the link between its adjacent points , and find the current source node The adjacent point with the smallest objective function provides a basis for path selection. The inter-cluster security state model consists of a cluster scanning layer, an inter-cluster node state classification layer, an inter-cluster link state classification layer, and an insecurity coefficient calculation layer. The inter-cluster node state classification layer calculates the source node of a large-scale cluster distributed network after passing through the cluster scanning layer. , target node , source node Neighbor nodes and target nodes The security status of the neighboring nodes is used for path selection; the inter-cluster link status classification layer calculates the source node obtained by the large-scale cluster distributed network after the cluster scanning layer , target node , source node Neighbor nodes and target nodes The link security status between neighbor nodes of the cluster; the insecurity coefficient calculation layer of the inter-cluster security status model is to calculate the current source node and the target node The insecurity factor and the current source node The insecurity factor of the link between its adjacent points , and find the current source node The neighboring point with the smallest objective function provides a basis for path selection.

[0069] The specific steps include:

[0070] Step 2.1: First, establish a cluster security status model. The cluster security model obtains the source node To the target node The security information on the path is the node status and link status, including the following steps:

[0071] Step 2.1.1: The in-cluster security state model takes a single cluster as the basic unit and converts the source nodes in the large-scale cluster into , target node , source node Neighbor nodes and target nodes The neighbor nodes are regarded as the core components of large-scale cluster distributed networks;

[0072] Step 2.1.2: Source nodes within a large-scale cluster , target node , source node Neighbor nodes and target nodes The security status of the neighbor nodes of the cluster is classified, and all source nodes in the cluster are obtained based on the security status of all the source nodes in the cluster. , target node , source node Neighbor nodes and target nodes Security information of neighboring nodes.

[0073] According to the source node in the cluster , target node , source node Neighbor nodes and target nodes The security information between any two adjacent nodes among the neighboring nodes is obtained to obtain the security status model within the cluster;

[0074] The intra-cluster security state model can quickly identify potential high-risk nodes and links in the transmission path. , target node , source node Neighbor nodes and target nodes The real-time status evaluation of neighbor nodes and links is used to convert the source node , target node , source node Neighbor nodes and target nodes The neighbor nodes of a node are divided into: safe nodes are nodes with normal connections and no direct connection to unsafe nodes; slightly unsafe nodes are nodes that are connected to a completely unsafe node or a failed link, but there are still available paths; highly unsafe nodes are nodes that are connected to two or more completely unsafe nodes or failed links, or have only one safe connection around them; completely unsafe nodes are nodes with all connections faulty; links are divided into: healthy links are links with safe nodes at both ends; dangerous links are links with at least one of the two ends being a slightly or highly unsafe node; failed links are links that cannot transmit data physically or logically.

[0075] Step 2.2: Establish an inter-cluster security state model. The inter-cluster security state model is obtained from the source node. The first part to the target node The first part of the security information of the nodes on the first part of the path, namely the node status and link status of the independent node, includes the following steps:

[0076] Step 2.2.1, the inter-cluster security state model takes a single cluster as the basic unit, and regards each cluster as an independent node in a large-scale cluster distributed network;

[0077] Step 2.2.2: Source nodes in large-scale clusters , target node , source node Neighbor nodes and target nodes The node status of the neighbor nodes is classified to classify the source nodes in the large-scale cluster , target node , source node Neighbor nodes and target nodes The link status between neighbor nodes;

[0078] Step 2.2.3: Based on the node status and link status of the large-scale cluster in step 2.2.2, classify the node status between clusters in the large-scale cluster distributed network, and obtain the connection status of all nodes in the cluster based on the node security status.

[0079] According to the link status between any two adjacent nodes in the cluster, the inter-cluster security status model is obtained.

[0080] The inter-cluster security status model regards the cluster as a single node in the high-level topology of a large-scale cluster distributed network, and divides the single node into the following categories through real-time status evaluation of the single node and the link: a safe node is a node with normal connections and no direct connection to an unsafe node; a slightly unsafe node is a node that is connected to a completely unsafe node or a failed link, but still has an available path; a highly unsafe node is a node that is connected to two or more completely unsafe nodes or failed links, or has only one safe link left around it; a completely unsafe node is a node where all connections have failed; links are divided into three categories: a healthy link is a link where both ends are safe nodes; a dangerous link is a link where at least one of the two ends is a slightly or highly unsafe node; and a failed link is a link that cannot transmit data physically or logically.

[0081] Step 2.3: According to the node and link status classification in steps 2.1 and 2.2, the node and link status of the source node, the target node, and its adjacent nodes are evaluated to determine which status they meet. Then, the data transmission path between clusters is dynamically adjusted. The inter-cluster security status model not only focuses on the risks within a single cluster, but also comprehensively considers the status and security of the inter-cluster links to ensure that data can be transmitted efficiently and securely across clusters.

[0082] Step 3: Evaluate the source node based on the constructed two-layer security status model and the target node The security status of the source node is calculated Unsafe factor and the source node The insecurity factor of the link between its adjacent points , and find the source node The neighboring point with the smallest objective function. Specifically includes the following steps:

[0083] Step 3.1: Calculate the source node The insecurity factor value:

[0084]

[0085] in, Represents the source node The number of failed connections, Represents the source node The total number of connections, Represents the source node The number of adjacent unsafe nodes, Represents the source node The total number of adjacent nodes, Represents the weight factor, which is used to balance the influence of connection status and adjacency status;

[0086] Step 3.2, source node Its adjacent points The insecurity factor of the link Values:

[0087]

[0088] in, , Respectively represent the source nodes at both ends of the link Its adjacent points The insecurity factor The value of , They represent weight factors, representing the contribution of the two end nodes to the link security status;

[0089] Step 3.3: Establish the current source node The objective function Minimum neighbors:

[0090]

[0091] in, Represents the source node The neighbor nodes of Represents the source node The insecurity factor value, Represents the source node Its adjacent points The insecurity factor of the link value.

[0092] Step 4: Check the node hierarchy and determine the source node and the target node Are they in the same cluster? If the source node and the target node In the same cluster, the cluster content error path search method is called to find the source node in the cluster. To the target node The fault-tolerant path is stored in the path collection in the cluster. If the source node and the target node If the nodes are not in the same cluster, call the inter-cluster fault-tolerant path search method to find the inter-cluster slave source node. To the target node The fault-tolerant path is stored in the inter-cluster path set. middle.

[0093] The specific steps include:

[0094] Step 4.1: If the source node and the target node In the same cluster, such as Figure 2 As shown, the cluster content wrong path search method is called to find the source node between clusters To the target node The fault-tolerant path is as follows: Initialize the path set , set the path collection Used to store the generated path. If the source node or target node Completely unsafe, returns no path, if the source node To the target node If it is a directly adjacent safe or slightly unsafe node, then directly connect to the source node To the target node , find the source node All adjacent points of , if there is a safe or slightly unsafe node, then calculate the source node Unsafe factor And select the neighboring point with the smallest objective function Update source node , repeatedly search for the source node All adjacent points of , until reaching the destination node Or all candidate nodes are unavailable, return to the source node To the target node A collection of paths , store the path nodes into the path collection within the cluster .

[0095] Step 4.2: If the source node and the target node Not in the same cluster, such as Figure 3 As shown, the inter-cluster fault-tolerant path search method is called to find the inter-cluster slave node To the target node The fault-tolerant path is as follows:

[0096] Initialize and create an empty path collection , set the starting node to the source node , at the source node Neighboring points , filter out the nodes that match the target node A set of safe or slightly unsafe nodes in the same cluster or across clusters, calculating each adjacent node The insecurity factor , select the neighboring point with the lowest objective function As the next hop node, repeat the adjacent point The selection process gradually expands the path until it reaches the target node Or determine that there is no path, after each path extension, for each source node in the path To the target node The security of the connection is verified. If a high-risk connection is found, it is dynamically adjusted to a suboptimal path. In path extension, it is checked whether there are repeated nodes or loops in the path. If a loop is detected, it goes back to the previous step to reselect the path to ensure that the final path has no redundant jumps. When the path spans multiple clusters, the cross-cluster connection is dynamically adjusted according to the objective function, and the inter-cluster link with the lowest objective function is selected and returned from the source node. To the target node A collection of paths , to ensure transmission efficiency and security in cross-cluster scenarios, store path nodes in the inter-cluster path set .

[0097] Step 5: Build from the source node To the target node The initial path , verify the path fault tolerance within the cluster and return to the source node To the target node The final fault-tolerant path , specifically:

[0098] Step 5.1: If the source node and the target node Located in the same cluster, construct the initial path A set of paths within the cluster ;

[0099] Step 5.1: If the source node and the target node Not in the same cluster, the path starts from the source node , and then include the collection path The node in the path, the end point of the path is the target node .

[0100] Step 6: Verify the path fault tolerance within the cluster and return the final fault tolerance path , for the initial path For each pair of consecutive nodes in the cluster, check whether they are in the same cluster and whether they are both safe or slightly unsafe nodes. If the above conditions are met, call the cluster content wrong path search method to generate a subpath between the two consecutive nodes and replace the initial path with this subpath The direct connection in the source node ensures the security and fault tolerance of the final path, and the path back to the source node ensures the security and fault tolerance of the path. To the target node The final fault-tolerant path ,The path ensures that the node state is safe and non-duplicate under given fault tolerance conditions.

[0101] In order to verify the present invention, the present invention carried out the following experiment:

[0102] 1. Experimental environment

[0103] Python 3.8

[0104] Processor: Intel i7-9750H

[0105] Memory 16GB

[0106] 2. Purpose of the experiment

[0107] This experiment is a small-scale test experiment, and the topology is , the total number of topological nodes is 112, and the total number of topological links is 392. It is mainly used to perceive the transmission success rate and average path length of the fault-tolerant routing method based on the double-layer security state model under different fault links, and to conduct a small-scale test of the algorithm.

[0108] 3. Experimental Configuration

[0109] Experimental topology: The total number of topological nodes is 112, and the total number of topological links is 392. Structure. The network bandwidth is 100Gbps and the link delay is 1us.

[0110] exist In the topology, the results of 100 experiments were averaged under the background of setting different numbers of faulty links. In each experiment, a certain number of links were randomly selected as faulty links, and the source server and the target server were randomly selected. The fault-tolerant routing method based on the two-layer security state model proposed in this invention was compared with the routing methods under the local information security model (LSM) and the conditional information security model (CLSM). The experimental results are shown in Figure 2. Figure 4 , Figure 5 As shown. Figure 4-5It can be seen that the proposed fault-tolerant routing method can ensure the successful transmission of data to a large extent and has a shorter average path length in most cases.

[0111] The present invention proposes a two-layer security state model and its application in fault-tolerant routing algorithm. And the multi-layer security status assessment mechanism can realize the accurate assessment of the node and link status in the network, optimize the selection of data transmission paths, comprehensively improve the network's fault tolerance and transmission efficiency, and provide support for efficient and secure communications in complex dynamic environments.

[0112] The above description is only an embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A fault-tolerant routing method based on a two-layer security state model, characterized in that: Applied to large-scale cluster distributed networks, the fault-tolerant routing method specifically includes the following steps: Step 1: Select the location information of two different fault-free nodes in the large-scale cluster distributed network as the source node and the target node , the source node and the target node The location information is divided into two parts; Step 2: construct a two-layer security state model, wherein the two-layer security state model includes an intra-cluster security state model and an inter-cluster security state model; Step 3: Evaluate the source node based on the constructed two-layer security status model and the target node The security status of the source node is calculated Unsafe factor and the source node The unsafe factor of the link between its adjacent points , and find the source node The neighboring point with the smallest objective function; Step 4: Check the node hierarchy and determine the source node and the target node Are they in the same cluster? If the source node and the target node In the same cluster, call the cluster content error path search method to find the source node in the cluster To the target node The fault-tolerant path is stored in the cluster path collection. If the source node and the target node If the nodes are not in the same cluster, call the inter-cluster fault-tolerant path search method to find the inter-cluster slave source node. To the target node The fault-tolerant path is stored in the inter-cluster path set. middle; Step 5: Build from the source node To the target node The initial path , verify the path fault tolerance within the cluster and return to the source node To the target node The final fault-tolerant path .

2. The fault-tolerant routing method based on a two-layer security state model according to claim 1, characterized in that: The security model in the cluster obtains the source node To the target node The security information on the path is the node status and link status. The specific method for obtaining the security information on the path includes the following steps: Step 2.1.1: The in-cluster security state model takes a single cluster as the basic unit and converts the source nodes in the large-scale cluster distributed network into , target node , source node Neighbor nodes and target nodes The neighbor nodes are regarded as the core components of large-scale cluster distributed networks; Step 2.1.2: Source nodes within a large-scale cluster distributed network , target node , source node Neighbor nodes and target nodes The security status of neighbor nodes is classified, and all source nodes in the cluster are obtained based on the security status of all source nodes in the large-scale cluster distributed network. , target node , source node Neighbor nodes and target nodes Security information of neighboring nodes.

3. The fault-tolerant routing method based on a two-layer security state model according to claim 2, characterized in that: In step 2.1.2, the in-cluster security state model is obtained by using the source node in the same cluster. , target node , source node Neighbor nodes and target nodes The real-time status evaluation of neighbor nodes and links is used to convert the source node , target node , source node Neighbor nodes and target nodes The neighbor node states are divided into: Secure nodes: All connections are normal and there is no direct connection to unsafe nodes; Mildly unsafe node: a node that is connected to a completely unsafe node or a failed link, but still has a usable path; Highly insecure nodes: nodes that are connected to two or more completely insecure nodes or failed links, or nodes that have only one secure link left around them; Completely unsafe node: all connections are down; The link status is divided into: Healthy link: a link where both nodes are safe nodes; Dangerous link: a link where at least one of the two ends is a slightly insecure node or a highly insecure node; Broken link: A link that is physically or logically unable to transmit data.

4. The fault-tolerant routing method based on a two-layer security state model according to claim 1, characterized in that: The inter-cluster security state model is obtained from the source node The first part to the target node The first part of the security information of the nodes on the first part of the path is the node state of the independent node and the link state of the independent node. The specific method for obtaining the information includes the following steps: Step 2.2.1, the inter-cluster security state model takes a single cluster as the basic unit, and regards each cluster as an independent node in a large-scale cluster distributed network; Step 2.2.2: Source nodes in large-scale cluster distributed networks , target node , source node Neighbor nodes and target nodes The node status of neighbor nodes is classified to classify the source node in the large-scale cluster distributed network. , target node , source node Neighbor nodes and target nodes Classify the link status between neighbor nodes; Step 2.2.3: Based on the node status and link status of the large-scale cluster in step 2.2.2, classify the node security status between clusters in the large-scale cluster distributed network, and obtain the connection status of all nodes in the cluster based on the node security status.

5. The fault-tolerant routing method based on a double-layer security state model according to claim 4, characterized in that: In step 2.2.2, the inter-cluster security state model regards the cluster as an independent node in a high-level topology in a large-scale cluster distributed network, and divides the node state of the independent node into the following categories through real-time status evaluation of the nodes and links of the independent node: Secure nodes: All connections are normal and there is no direct connection to unsafe nodes; Mildly unsafe node: a node that is connected to a completely unsafe node or a failed link, but still has a usable path; Highly insecure nodes: nodes that are connected to two or more completely insecure nodes or failed links, or nodes that have only one secure link left around them; Completely unsafe node: all connections are down; The link status is divided into three categories: Healthy link: a link where both nodes are safe nodes; Dangerous link: a link where at least one of the two ends is a slightly insecure node or a highly insecure node; Broken link: A link that is physically or logically unable to transmit data.

6. The fault-tolerant routing method based on a double-layer security state model according to claim 1, characterized in that: Step 3 calculates the source node Unsafe factor , source node The unsafe factor of the link between its adjacent points , and find the source node The neighboring point with the minimum objective function specifically includes the following steps: Step 3.1: Calculate the source node The insecurity factor value: ； in, Represents the source node The number of failed connections, Represents the source node The total number of connections, Represents the source node The number of adjacent unsafe nodes, Represents the source node The total number of adjacent nodes, Represents the weight factor, which is used to balance the influence of connection status and adjacency status; Step 3.2, source node Its adjacent points The insecurity factor of the link Values: ； in, , Respectively represent the source nodes at both ends of the link Its adjacent points The insecurity factor The value of , They represent weight factors, representing the contribution of the two end nodes to the link security status; Step 3.3: Establish the current source node The objective function Minimum neighbors: ； in, Represents the source node The neighbor nodes of Represents the source node The insecurity factor value, Represents the source node Its adjacent points The insecurity factor of the link value.

7. The fault-tolerant routing method based on a double-layer security state model according to claim 6, characterized in that: In step 4, call the cluster content error path search method to find the source node in the cluster. To the target node The fault-tolerant path is as follows: Initialize the path set , set the path collection Used to store the generated path. If the source node or target node Completely unsafe, returns no path, if the source node To the target node If it is a directly adjacent safe node or slightly unsafe node, it will be directly connected to the source node To the target node , find the source node All adjacent points of , if there is a safe node or a slightly unsafe node, then calculate the source node Unsafe factor And select the neighboring point with the smallest objective function Update source node , repeatedly search for the source node All adjacent points of , until reaching the destination node Or all candidate nodes are unavailable, return to the source node To the target node A collection of paths , store the path nodes into the path collection within the cluster .

8. The fault-tolerant routing method based on a double-layer security state model according to claim 6, characterized in that: In step 4, call the inter-cluster fault-tolerant path search method to find the inter-cluster slave source node To the target node The fault-tolerant path is as follows: Initialize and create an empty path collection , set the starting node to the source node , at the source node Adjacent points , filter out the nodes that match the target node A set of secure nodes or slightly insecure nodes in the same cluster or across clusters, calculating each adjacent node The insecurity factor , select the neighboring point with the lowest objective function As the next hop node, repeat the adjacent point The selection process gradually expands the path until it reaches the target node Or determine that there is no path, after each path extension, for each source node in the path To the target node The security of the connection is verified. If a high-risk connection is found, it is dynamically adjusted to a suboptimal path. In path extension, it is checked whether there are repeated nodes or loops in the path. If a loop is detected, it goes back to the previous step to reselect the path to ensure that the final path has no redundant jumps. When the path spans multiple clusters, the cross-cluster connection is dynamically adjusted according to the objective function, and the inter-cluster link with the lowest objective function is selected and returned from the source node. To the target node A collection of paths , to ensure transmission efficiency and security in cross-cluster scenarios, store path nodes in the inter-cluster path set .

9. The fault-tolerant routing method based on a double-layer security state model according to claim 1, characterized in that: Step 5 builds the source node To the target node The initial path Specifically: Step 5.1: If the source node and the target node Located in the same cluster, construct the initial path A set of paths within the cluster ; Step 5.2: If the source node and the target node Not in the same cluster, the path starts from the source node , and then contains the set of inter-cluster paths The node in the path, the end point of the path is the target node ; Step 5.3: Initial path For each pair of consecutive nodes in the cluster, check whether they are in the same cluster and whether they are both safe or slightly unsafe nodes. If the above conditions are met, call the cluster content wrong path search method to generate a subpath between the two consecutive nodes and replace the initial path with this subpath The direct connection in the source node ensures the security and fault tolerance of the final path, and the path back to the source node ensures the security and fault tolerance of the path. To the target node The final fault-tolerant path ,The path ensures that the node state is safe and non-duplicate under given fault tolerance conditions.

Images