A node reconfiguration adaptive packet routing method for delay tolerant network

Abstract

The present application relates to the technical field of delay tolerant network routing, in particular to a node reconfiguration adaptive grouping routing method for delay tolerant network. In view of the problems of frequent grouping overhead and insufficient copy forwarding direction of the invalid history forwarding structure under dynamic topology, a weighted contact graph and a positive point mutual information matrix are constructed in a current time window, dynamic grouping is carried out based on effective contact association, and group head nodes, backup group head nodes and bridge nodes are determined; the node reconfiguration level is determined in combination with the existing state of the last effective node role configuration table, the connection ability of the role node and the change degree of the network structure, and the current effective node role configuration table is formed; and then, message forwarding and copy number updating are performed according to the group to which the destination node belongs, the node role and the remaining copy number, so as to improve the adaptability and effectiveness of message forwarding in the intermittent connectivity scenario.

Description

Technical Field

[0001] This invention relates to the field of delay-tolerant network routing technology, specifically to an adaptive packet routing method for node reconfiguration in delay-tolerant networks. Background Technology

[0002] Delay-tolerant networks are typically used in communication scenarios where mobile nodes are sparse, links are intermittently connected, and end-to-end paths are difficult to maintain continuously. Their routing methods generally achieve message delivery through node storage, carrying, and forwarding. Existing routing methods often select forwarding nodes based on historical node contact frequency, encounter probability, node centrality, social relationships, or message replica control strategies, and perform message delivery or replica forwarding when nodes meet to improve the message arrival rate under intermittent connectivity conditions.

[0003] For scenarios where node movement states and contact relationships change over time, existing methods often struggle to simultaneously accommodate node contact relationships, changes in packet structure, and changes in the state of nodes responsible for forwarding. Frequent adjustments to the forwarding structure can increase computational and maintenance overhead, while relying on historical forwarding structures for extended periods may lead to inappropriate replica allocation and forwarding directions deviating from the destination node's location due to decreased connectivity of critical nodes or changes in cross-packet relationships. Existing methods also frequently rely on relatively simple encounter or role information for determining node reception within the destination node's packet and for cross-packet forwarding priority, making them ill-suited for the dynamic topology of intermittently connected networks. Therefore, an adaptive packet routing method for node reconfiguration in delay-tolerant networks is needed to address these issues. Summary of the Invention

[0004] To address the above problems, this invention provides the following technical solution: an adaptive packet routing method for node reconfiguration in delay-tolerant networks, comprising: S1. Collect node contact relationship information in the current time window, construct a weighted contact diagram, and construct a positive point mutual information matrix from the weighted contact diagram; S2. Based on the positive point mutual information matrix, the nodes are dynamically grouped, and the group head node, backup group head node and bridge node are determined in the dynamic grouping results to form a node role configuration table. S3. Determine the node reconfiguration level based on the existence status of the previous valid node role configuration table, the connectivity of the role nodes in the previous valid node role configuration table in the current time window, and the degree of network structure change determined by the weighted contact graph of the current time window and the weighted contact graph of the previous time window. S4. Perform stable reuse, partial role update or full dynamic regrouping on the node role configuration table according to the node reconfiguration level to obtain the currently effective node role configuration table. S5. Based on the current effective node role configuration table, the group to which the destination node is located, and the remaining number of replicas carried in the message, perform destination group-aware and role-aware message forwarding, and update the message replica count when forwarding message replicas.

[0005] Furthermore, the construction of the positive point mutual information matrix from the weighted contact graph includes: The joint probability of node pairs and the node edge probability are calculated from the contact weights in the weighted contact graph. The point mutual information value of the node pair is calculated based on the joint probability of the node pair and the edge probability of the node; Set the mutual information values ​​of points that are less than zero to zero to obtain the positive mutual information matrix used to characterize the effective contact association of nodes.

[0006] Furthermore, the dynamic grouping of nodes based on the positive point mutual information matrix, and the determination of group head nodes, backup group head nodes, and bridge nodes in the dynamic grouping results, forming a node role configuration table, includes: Landmark selection and spectral embedding are performed on the positive point mutual information matrix to obtain low-dimensional node representations, and the low-dimensional node representations are clustered to obtain node group identifiers; The group head node and the alternate group head node are determined based on at least two of the following: the contact strength of the node within its group, the node degree determined by the weighted contact graph, and the embedding centrality determined by the low-dimensional representation of the node. Bridge nodes are determined based on at least two of the following: cross-group contact strength, cross-group coverage, and cross-group connection ratio between the node and other groups; Write the node group identifier, the group head node, the spare group head node, and the bridge node into the node role configuration table.

[0007] Furthermore, the node role configuration table, the previous effective node role configuration table, and the currently effective node role configuration table all record node identifiers, node group identifiers, role types, and bridge node connection group sets. The role types include group head nodes, backup group head nodes, bridge nodes, and ordinary nodes.

[0008] Furthermore, determining the node reconfiguration level based on the existence status of the previous valid node role configuration table, the connectivity of the role nodes in the previous valid node role configuration table within the current time window, and the degree of network structure change determined by the weighted contact graph of the current time window and the weighted contact graph of the previous time window includes: When there is no previous valid node role configuration table, the node reconfiguration level is determined to be full dynamic regrouping; When a previous valid node role configuration table exists, a previous role node set is formed based on the group head node, spare group head node, and bridge node in it, and the role degradation degree is determined based on the node degree and contact strength of the role nodes in the previous role node set in the current time window. The node reconfiguration level is determined based on the degree of role degradation and the degree of network structure change.

[0009] Furthermore, determining the node reconfiguration level based on the degree of role degradation and the degree of network structure change includes: When the previous role node set is empty or the age of the role table reaches the preset maximum age, the node reconfiguration level is determined to be full dynamic regrouping; When the degree of role degradation is lower than a preset role degradation threshold and the degree of network structure change is lower than a preset change condition, the node reconfiguration level is determined to be stable reuse; When the degree of role degradation reaches the preset role degradation threshold or the degree of network structure change reaches the preset change condition, the node reconfiguration level is determined as local role update or full dynamic regrouping based on the degree of improvement of candidate groups. The degree of improvement of the candidate grouping is determined by the difference between the dynamic grouping result of the current time window and the node grouping recorded in the previous effective node role configuration table in at least one of the following: group size balance, group size entropy, and number of single-node groups.

[0010] Furthermore, the step of performing stable reuse, partial role update, or full dynamic regrouping of the node role configuration table according to the node reconfiguration level to obtain the currently effective node role configuration table includes: When the node reconfiguration level is stable reuse, the previous effective node role configuration table will be used as the currently effective node role configuration table. When the node reconfiguration level is local role update, the node group identifier in the previous effective node role configuration table remains unchanged, and the group head node, backup group head node and bridge node are re-determined based on the weighted contact graph of the current time window to generate the currently effective node role configuration table. When the node reconfiguration level is full dynamic regrouping, the node role configuration table formed after dynamic grouping in the current time window is used as the currently effective node role configuration table.

[0011] Furthermore, the step of performing destination group-aware and role-aware message forwarding based on the currently effective node role configuration table, the group to which the destination node belongs, and the remaining number of replicas carried in the message includes: The node currently holding the message is identified as the current carrying node, and the destination group of the message destination node is determined according to the current effective node role configuration table. When the meeting node is the message destination node, the message is delivered to the message destination node; When the meeting node is not the message destination node, the number of remaining copies carried by the message is greater than one, and the meeting node belongs to the destination group, the meeting node is determined as the message copy receiving node. When the encounter node does not belong to the destination packet, the role types of the encounter node and the current carrying node, as well as the bridge node connection packet set, are read from the current effective node role configuration table. The forwarding priorities of the encounter node and the current carrying node towards the destination packet are determined respectively. When the forwarding priority of the encounter node is higher than that of the current carrying node and the number of remaining copies of the message is greater than one, the encounter node is determined as the message copy receiving node.

[0012] Furthermore, updating the message replica count when forwarding message replicas includes: When it is determined that the message replica receiving node and the current carrying node hold the number of message replicas, and When the value is greater than 1, forward the message replica to the message replica receiving node and update the message replica count of the current carrying node to 1. Round up to the nearest half and update the message replica count of the message replica receiving node to [the nearest integer]. of is rounded down to half; When a node holds only one copy of a message, it will not continue to forward message copies unless the encountered node is the destination node of the message.

[0013] Compared with the prior art, the present invention has the following beneficial effects: By constructing a weighted contact graph and a point-to-point mutual information matrix in the current time window, the contact weights between nodes can be further transformed into effective contact associations. Based on these effective contact associations, dynamic grouping and node role determination are then performed, ensuring that the configuration of group head nodes, backup group head nodes, and bridge nodes corresponds to the node contact status within the current time window. This reduces the mismatch between the forwarding structure and the current contact relationship caused by selecting forwarding nodes solely based on historical encounters or single centrality.

[0014] By combining the existence status of the previous valid node role configuration table, the current connectivity of the role node, and the degree of network structure change, the node reconfiguration level is determined. The current effective node role configuration table is formed by stable reuse, partial role update, or full dynamic regrouping. This allows the node role configuration to be reused when network changes are minor and updated when role nodes degrade or the structure changes. This avoids the maintenance overhead caused by full regrouping in every time window and also reduces the situation where message replicas deviate from the effective forwarding direction due to the long-term use of invalid role configurations.

[0015] By performing destination group-aware and role-aware message forwarding based on the current effective node role configuration table, the destination node's group, and the remaining number of replicas, meeting nodes belonging to the destination group can act as message replica receiving nodes. Furthermore, during cross-group forwarding, the forwarding priority oriented towards the destination group controls the replica handover. Simultaneously, by updating the number of message replicas during message replication and restricting the continued forwarding of single replicas, more message replicas are directed to the destination group or nodes that are more conducive to reaching the destination group, thereby reducing undirected diffusion and invalid replicas occupying link resources. Attached Figure Description

[0016] Figure 1 The method steps provided by this invention are illustrated. Figure 2 The overall flowchart provided for this invention; Figure 3 The flowchart for constructing and dynamically grouping the positive mutual information matrix provided by this invention; Figure 4 A flowchart for generating the node role configuration table provided by this invention; Figure 5 The present invention provides a flowchart for target group perception and role perception forwarding. Figure 6 The node reconfiguration level determination flowchart provided by this invention; Figure 7 The diagram shows a comparison of routing performance provided by this invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] To enable those skilled in the art to better understand the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] Please see Figure 1 and Figure 2 , Figure 1 The method steps provided by this invention are illustrated. Figure 2 This is the overall flowchart provided by the present invention; the present invention provides an adaptive packet routing method for node reconfiguration in delay-tolerant networks, comprising: S1. Collect node contact relationship information in the current time window, construct a weighted contact diagram, and construct a positive point mutual information matrix from the weighted contact diagram.

[0020] Specifically, suppose there are a total of delay-tolerant networks. There are 1 mobile nodes, and the node set is 1. Network runtime interval Divided into There are 3 consecutive time windows, each with a length of:

[0021] No. Each time window is denoted as:

[0022] Within each time window, record contact events between any two nodes. The node contact relationship information includes the number of contacts, the cumulative contact duration, or a combination of both. For nodes... and nodes The contact weight of the current time window can be expressed as:

[0023] in, Represents a node With nodes Contact weights within the current time window. Indicates the number of contacts. Indicates the cumulative duration of contact. and is a non-negative weighting coefficient. This calculation transforms the original contact record into a weighted contact diagram, allowing subsequent grouping to utilize both the frequency of node encounters and the duration of connections after encounters. If only the number of contacts is considered, then... , If only the cumulative duration of contact is considered, it can make , After constructing the weighted contact graph, it is symmetrically processed by setting the diagonal elements to zero and setting negative or non-finite values ​​to zero to avoid the influence of self-loops, outlier sampling values, or unidirectional recording errors on subsequent calculations.

[0024] Furthermore, in one embodiment of the present invention, constructing the positive point mutual information matrix from the weighted contact graph includes: The joint probability of node pairs and the node edge probability are calculated from the contact weights in the weighted contact graph. The point mutual information value of the node pair is calculated based on the joint probability of the node pair and the edge probability of the node; Set the mutual information values ​​of points that are less than zero to zero to obtain the positive mutual information matrix used to characterize the effective contact association of nodes.

[0025] For details, please refer to Figure 3 After constructing the weighted contact graph, the joint probability of node pairs is first calculated based on the contact weights:

[0026] Then calculate the node edge probability:

[0027] Then calculate the point mutual information value of the node pair:

[0028] Finally, the mutual information values ​​of points less than zero are set to zero to obtain the positive point mutual information matrix:

[0029] in, A very small constant greater than zero is used to avoid division by zero. Point mutual information values ​​are used to measure whether the actual contact strength between two nodes is higher than the expected contact strength under random independent conditions; the positive point mutual information matrix only retains positive correlations, which can filter out random weak contacts and invalid negative correlations. After this processing, dynamic grouping is no longer dominated solely by nodes with high contact frequency, but rather focuses more on node contact relationships with stable structural significance.

[0030] S2. Based on the positive point mutual information matrix, the nodes are dynamically grouped, and the group head node, backup group head node and bridge node are determined in the dynamic grouping results to form a node role configuration table.

[0031] For details, please continue reading. Figure 3 Landmark selection and spectral embedding are performed on the positive point mutual information matrix to obtain a low-dimensional representation of the nodes. Landmark selection reduces the computational cost of full matrix embedding, while spectral embedding preserves the contact relationships between nodes. After obtaining the low-dimensional representations of the nodes, clustering is performed on these representations to obtain node group identifiers.

[0032] in, Represents a node Node group identifier, The preset number of groups is used. After dynamic grouping is completed, a group head node and a backup group head node are selected within each group to handle message propagation within the group, and a bridge node is selected to handle cross-group forwarding. Thus, a node role configuration table is generated for the current time window, which serves as a candidate input for subsequent reconfiguration and message forwarding decisions.

[0033] Furthermore, in one embodiment of the present invention, the step of dynamically grouping nodes based on the positive point mutual information matrix, and determining the group head node, backup group head node, and bridge node in the dynamic grouping results to form a node role configuration table includes: Landmark selection and spectral embedding are performed on the positive point mutual information matrix to obtain low-dimensional node representations, and the low-dimensional node representations are clustered to obtain node group identifiers; The group head node and the alternate group head node are determined based on at least two of the following: the contact strength of the node within its group, the node degree determined by the weighted contact graph, and the embedding centrality determined by the low-dimensional representation of the node. Bridge nodes are determined based on at least two of the following: cross-group contact strength, cross-group coverage, and cross-group connection ratio between the node and other groups; Write the node group identifier, the group head node, the spare group head node, and the bridge node into the node role configuration table.

[0034] For details, please refer to Figure 4 To select suitable nodes as group head nodes and alternate group head nodes, the nodes are first calculated. Intra-group contact intensity within its respective group:

[0035] Calculate the node degree again:

[0036] And calculate the embedding centrality based on the low-dimensional representation of the nodes:

[0037] in, Represents a node The center of the group in the embedded space, This is an indicator function, taking the value 1 if the condition is true, and 0 otherwise. Intra-group contact strength represents the total contact between a node and other nodes in the same group; node degree represents the breadth of objects a node contacts; and embedding centrality represents the representativeness of a node within its group structure. Due to the different dimensions and numerical ranges of each indicator, they are normalized as follows:

[0038] in, This represents the set of candidate nodes participating in the current metric comparison. When calculating the group head node score and the backup group head node score, For nodes All candidate nodes within the corresponding group; when calculating the bridge node score. This refers to all candidate nodes within the current group that have cross-group contact relationships; when calculating the role quality of each role node in the previous role node set... The effective nodes in the contact graph are weighted for the current time window. The effective nodes are those that have at least one contact edge with a contact weight greater than zero within the current time window.

[0039] When the difference between the maximum and minimum values ​​of the corresponding indicators in the candidate node set is no greater than If the maximum value of the indicator is greater than zero, then the corresponding normalized indicator of each candidate node is uniformly set to one; if the maximum value of the indicator is zero, then the corresponding normalized indicator of each candidate node is uniformly set to zero.

[0040] After normalization, the group head score can be expressed as:

[0041] in, These are the normalized intra-group contact strength, embedding centrality, node degree, and cache availability, respectively. The weights are non-negative. If cache availability is not used, it can be set as follows: Within each group, the node with the highest group head score is designated as the group head node, and the node with the second highest group head score is designated as the backup group head node. This method ensures that the group head node has strong intra-group contact capabilities while being positioned as centrally as possible in the group structure, thereby improving the reliability of message propagation within the group; the backup group head node provides alternative forwarding capabilities when the group head node's connectivity decreases or when it has not yet encountered another node.

[0042] To select bridge nodes, compute nodes. Cross-group contact strength with other groups:

[0043] Calculate cross-group coverage:

[0044] And calculate the cross-group connection ratio:

[0045] Cross-group contact strength represents the total number of contacts a node has with other group nodes; cross-group coverage represents the number of different groups a node can connect to; and cross-group connection ratio represents the proportion of cross-group contacts in all of a node's contacts. After normalization, the bridge node score can be expressed as:

[0046] in, These are the normalized cross-group contact strength, cross-group coverage, cross-group connection ratio, and cache availability, respectively. The weights are non-negative. When selecting bridge nodes, nodes that have already been identified as group head nodes or backup group head nodes can be excluded, and then selected from candidate nodes with cross-group connections in descending order of bridge node score. This method can prioritize nodes with strong cross-group reachability, ensuring that messages have clear relay objects when propagating across groups and reducing undirected diffusion.

[0047] Furthermore, in one embodiment provided by the present invention, the node role configuration table, the previous effective node role configuration table, and the currently effective node role configuration table all record node identifiers, node group identifiers, role types, and bridge node connection group sets. The role types include group head nodes, backup group head nodes, bridge nodes, and ordinary nodes.

[0048] Specifically, the node role configuration table uses the node identifier as the search key and records the node group identifier, role type, and bridge node connection group set. The node group identifier is used to determine the group where the destination node belongs, the role type is used to determine whether the node undertakes intra-group propagation or cross-group forwarding functions, and the bridge node connection group set is used to indicate other groups that the bridge node can connect to. When a node is not a group head node, backup group head node, or bridge node, its role type is recorded as a normal node. The previous valid node role configuration table is used for subsequent degradation judgment, and the currently effective node role configuration table is used for actual message forwarding. The three role tables use the same fields to ensure consistency in the data read for reconfiguration judgment and routing forwarding.

[0049] S3. Determine the node reconfiguration level based on the existence status of the previous valid node role configuration table, the connectivity of the role nodes in the previous valid node role configuration table in the current time window, and the degree of network structure change determined by the weighted contact graph of the current time window and the weighted contact graph of the previous time window.

[0050] Specifically, in delay-tolerant networks, node connectivity changes over time. A node that was suitable as a group head or bridge node in the previous time window may have lost its connectivity advantage in the current time window. Therefore, this invention does not simply reuse the role table permanently, nor does it unconditionally replace the role table in each time window. Instead, it determines the node reconfiguration level based on whether the previous valid node role configuration table exists, whether the current connectivity of the previous role node has degraded, and whether the network structure has changed. The node reconfiguration levels include stable reuse, partial role update, and full dynamic regrouping. Figure 6 The judgment process is shown.

[0051] Furthermore, in one embodiment of the present invention, determining the node reconfiguration level based on the existence status of the previous valid node role configuration table, the connectivity of the role nodes in the previous valid node role configuration table in the current time window, and the degree of network structure change determined by the weighted contact graph of the current time window and the weighted contact graph of the previous time window includes: When there is no previous valid node role configuration table, the node reconfiguration level is determined to be full dynamic regrouping; When a previous valid node role configuration table exists, a previous role node set is formed based on the group head node, spare group head node, and bridge node in it, and the role degradation degree is determined based on the node degree and contact strength of the role nodes in the previous role node set in the current time window. The node reconfiguration level is determined based on the degree of role degradation and the degree of network structure change.

[0052] Specifically, if a previous valid node role configuration table does not exist, it means there are no reusable historical role configurations in the current time window, and a full dynamic regrouping is directly determined. If a previous valid node role configuration table exists, the group head node, spare group head node, and bridge node are read from the table to form the previous role node set. For sets For nodes in the current time window, calculate the node degree and contact strength in the weighted contact graph:

[0053] Using all valid nodes in the weighted contact graph of the current time window as the normalization range, the node degree and contact strength are normalized separately to obtain the normalized node degree. and normalized contact strength And calculate the character quality accordingly:

[0054] in, To normalize the node degree, To normalize the contact strength, and The weights are non-negative. Node degree reflects how many nodes a character node can currently contact, and contact strength reflects whether the contact is stable enough. Then, the average character quality of the previous set of character nodes is calculated:

[0055] And calculate the degree of character degradation:

[0056] when A higher value indicates that the previous role node still has good connectivity within the current time window, and the risk of continuing to use the original role table is low; when A higher value indicates that the connectivity of the previous node has decreased. Continuing to use it may cause messages to remain on nodes with low connectivity, so an update check needs to be triggered.

[0057] Furthermore, in one embodiment of the present invention, determining the node reconfiguration level based on the degree of role degradation and the degree of network structure change includes: When the previous role node set is empty or the age of the role table reaches the preset maximum age, the node reconfiguration level is determined to be full dynamic regrouping; When the degree of role degradation is lower than a preset role degradation threshold and the degree of network structure change is lower than a preset change condition, the node reconfiguration level is determined to be stable reuse; When the degree of role degradation reaches the preset role degradation threshold or the degree of network structure change reaches the preset change condition, the node reconfiguration level is determined as local role update or full dynamic regrouping based on the degree of improvement of candidate groups. The degree of improvement of the candidate grouping is determined by the difference between the dynamic grouping result of the current time window and the node grouping recorded in the previous effective node role configuration table in at least one of the following: group size balance, group size entropy, and number of single-node groups.

[0058] Specifically, the role table age is used to characterize the number of consecutive time windows in which the node grouping structure in the currently active node role configuration table has been used since the most recent full dynamic regrouping. When the node reconfiguration level is stable reuse or partial role update, the node grouping structure in the previous active node role configuration table is continued, and the role table age is incremented by one; when the node reconfiguration level is full dynamic regrouping, the node grouping structure is updated using the dynamic grouping result of the current time window, and the role table age is reset to zero. If the previous role node set is empty, it means that there is no key role that can be used to evaluate connectivity; if the role table age reaches the preset maximum age, it means that the current node grouping structure has been used for multiple consecutive time windows. Even if the degree of role degradation and the degree of network structure change have not yet reached the corresponding conditions, it may deviate from the current network state due to the continuous accumulation of contact relationship changes. Therefore, the node reconfiguration level is determined to be full dynamic regrouping.

[0059] The degree of change in network structure can be characterized by both the weighted rate of change and the rate of change in network density. To reduce the dominance of the rate of change by maximal edge weights, the weighted contact graph can be processed as follows:

[0060] Next, calculate the weighted rate of change for adjacent time windows:

[0061] The rate of change of network density can be expressed as:

[0062] Among them, the network density of the current time window Represented as:

[0063] Network density in the previous time window Represented as:

[0064] in, This represents the number of undirected contact edges with a contact weight greater than zero in the weighted contact graph for the current time window. This represents the number of undirected contact edges with a contact weight greater than zero in the weighted contact graph of the previous time window. This represents the total number of network nodes. The weighted rate of change reflects the change in the contact strength between nodes in adjacent time windows, while the network density rate of change reflects the change in the number of effective contact edges between adjacent time windows. Combining these two methods can avoid judging the overall stability of the network structure solely based on the connectivity of a few key nodes.

[0065] When the degree of role degradation is below the preset role degradation threshold and the degree of network structure change is below the preset change condition, it indicates that the previous effective node role configuration table still has value and is determined to be stable reuse. When the degree of role degradation reaches the preset role degradation threshold, or the degree of network structure change reaches the preset change condition, the improvement degree of candidate groups is further calculated. The improvement degree of candidate groups is used to determine whether the dynamic grouping results in the current time window are indeed better than the node groups recorded in the previous effective node role configuration table, rather than immediately replacing all groups when there are slight fluctuations. The improvement degree of candidate groups is determined based on the group size balance, group size entropy, and the number of groups per node. Let the node grouping results to be evaluated include... Group 1, the 1st group The number of nodes in each group is The total number of network nodes is The maximum group size ratio of the grouping results is expressed as:

[0066] Group size entropy is expressed as:

[0067] The number of groups per node is expressed as:

[0068] in, This is an indicator function that takes a value of 1 if the condition within the parentheses is true, and zero otherwise.

[0069] Calculate the maximum group size percentage for each node group recorded in the previous valid node role configuration table. Group size entropy and the number of groups per node ; Calculate the percentage of the largest group size for each of the dynamic grouping results within the current time window. Group size entropy and the number of groups per node The degree of improvement of the candidate group is expressed as:

[0070] in, Indicates the degree of improvement of the candidate group. The weights are non-negative. A decrease in the proportion of the largest group size, an increase in group size entropy, and a decrease in the number of groups per node indicate that the dynamic grouping results in the current time window have higher group size balance and lower group fragmentation.

[0071] Improvement of candidate grouping Improved threshold with preset candidate grouping Compare; when: When the dynamic grouping results of the current time window meet the preset full update conditions, the node reconfiguration level is determined to be full dynamic regrouping; when the following conditions are met: If the dynamic grouping results of the current time window do not meet the preset full update conditions, the node reconfiguration level will be determined as a local role update.

[0072] S4. Perform stable reuse, partial role update or full dynamic regrouping on the node role configuration table according to the node reconfiguration level to obtain the currently effective node role configuration table.

[0073] Specifically, once the node reconfiguration level is determined, it is used to determine the role table that is actually effective in the current time window. Stable reuse, partial role update, and full dynamic regrouping are not three independent routing algorithms, but rather three ways in which the same role configuration mechanism takes effect under different degrees of network change. This process reduces role table changes when the network is stable, updates only role nodes when there are local network changes, and uses the full dynamic regrouping result only when there are significant overall network changes.

[0074] Furthermore, in one embodiment of the present invention, the step of performing stable reuse, partial role update, or full dynamic regrouping of the node role configuration table according to the node reconfiguration level to obtain the currently effective node role configuration table includes: When the node reconfiguration level is stable reuse, the previous effective node role configuration table will be used as the currently effective node role configuration table. When the node reconfiguration level is local role update, the node group identifier in the previous effective node role configuration table remains unchanged, and the group head node, backup group head node and bridge node are re-determined based on the weighted contact graph of the current time window to generate the currently effective node role configuration table. When the node reconfiguration level is full dynamic regrouping, the node role configuration table formed after dynamic grouping in the current time window is used as the currently effective node role configuration table.

[0075] Specifically, when the node reconfiguration level is stable reuse, the current time window directly uses the previous valid node role configuration table, and the node group identifier, role type, and bridge node connection group set remain unchanged. When the node reconfiguration level is partial role update, the node group identifier in the previous valid node role configuration table is retained, but the group head score and bridge node score are recalculated using the weighted contact graph of the current time window, and the group head node, spare group head node, and bridge node are updated accordingly. This allows degraded role nodes to be replaced while the grouping structure is still basically valid. When the node reconfiguration level is full dynamic regrouping, the node role configuration table formed after dynamic grouping in the current time window is used as the currently effective node role configuration table, so that the node group identifier and role configuration are refreshed simultaneously.

[0076] S5. Based on the current effective node role configuration table, the group to which the destination node is located, and the remaining number of replicas carried in the message, perform destination group-aware and role-aware message forwarding, and update the message replica count when forwarding message replicas.

[0077] For details, please refer to Figure 5 When a message is created, the destination node identifier and the number of remaining replicas are written to it. When nodes meet, the node currently holding the message becomes the current carrier node and first reads the group to which the destination node belongs from the currently active node role configuration table. If the meeting node is the message's destination node, the message is delivered directly. If the meeting node is not the message's destination node but belongs to the same group as the destination node, and the number of remaining replicas carried by the message is greater than one, then forwarding a copy of the message to the meeting node is allowed. This rule reflects destination group awareness: even if the meeting node is just an ordinary node within the destination group, it may be closer to the destination node than the currently carrier node, and allowing it to receive a copy helps increase the probability of the message propagating after entering the destination area.

[0078] Furthermore, in one embodiment of the present invention, the step of performing destination group-aware and role-aware message forwarding based on the currently effective node role configuration table, the group to which the destination node belongs, and the remaining number of replicas carried in the message includes: The node currently holding the message is identified as the current carrying node, and the destination group of the message destination node is determined according to the current effective node role configuration table. When the meeting node is the message destination node, the message is delivered to the message destination node; When the meeting node is not the message destination node, the number of remaining copies carried by the message is greater than one, and the meeting node belongs to the destination group, the meeting node is determined as the message copy receiving node. When the encounter node does not belong to the destination packet, the role types of the encounter node and the current carrying node, as well as the bridge node connection packet set, are read from the current effective node role configuration table. The forwarding priorities of the encounter node and the current carrying node towards the destination packet are determined respectively. When the forwarding priority of the encounter node is higher than that of the current carrying node and the number of remaining copies of the message is greater than one, the encounter node is determined as the message copy receiving node.

[0079] Specifically, when the encountering node does not belong to the destination packet, it cannot be forwarded directly based solely on the encounter event; otherwise, invalid replicas will be created. In this case, it is necessary to compare the forwarding priorities of the currently carrying node and the encountering node towards the destination packet. Let the destination packet be... For any node First, define five indicator variables: Represents a node Whether it belongs to the target group; if it does, set to 1, otherwise set to 0. Represents a node Whether it is the group head node within the target group; if yes, set to 1, otherwise set to 0. Represents a node Whether it is the spare group head node in the destination group; if yes, set it to 1, otherwise set it to 0. Represents a node Whether it is a bridge node within the destination group; if yes, set to 1, otherwise set to 0. Represents a node Whether it is an external packet bridge node that can connect to the destination packet; if yes, set to 1, otherwise set to 0.

[0080] With the above variables now defined, the node Destination-oriented grouping Forwarding priority is represented as follows:

[0081] in, to The weights are non-negative. In this formula, This indicates whether the node is already in the destination packet. and This reflects the node's intra-group diffusion capability within the target group. This demonstrates the bridging capability within the target group. This demonstrates the router's ability to forward messages from external packets to their destination packets. The router calculates the forwarding priorities of both the currently carrying node and the encountered node; only when the encountered node has a higher forwarding priority and the number of remaining message replicas is greater than one is the encountered node designated as the message replica receiving node. This process ensures that messages are preferentially routed to nodes closer to the destination packet or with greater cross-packet connectivity, rather than being blindly copied at each encounter.

[0082] Furthermore, in one embodiment of the present invention, updating the message replica count when forwarding message replicas includes: When it is determined that the message replica receiving node and the current carrying node hold the number of message replicas, and When the value is greater than 1, forward the message replica to the message replica receiving node and update the message replica count of the current carrying node to 1. Round up to the nearest half and update the message replica count of the message replica receiving node to [the nearest integer]. of is rounded down to the nearest half; When a node holds only one copy of a message, it will not continue to forward message copies unless the encountered node is the destination node of the message.

[0083] Specifically, let the number of message replicas held by the current carrying node before forwarding be... ,and Once the message replica receiving node is identified and forwarding is complete, the replica count is updated as follows:

[0084] in, This indicates the number of replicas of the currently carried node after the update. This indicates the number of message replicas received by the receiving node. A binary distribution method is used to allocate replicas, which limits the total number of replicas while preserving multipath propagation capabilities. When a node holds only one message replica, it will no longer forward message replicas unless the destination node is the node it encounters, thus controlling network overhead.

[0085] In a specific application embodiment, the weight in the group head score can be taken as:

[0086] The weights in the bridge node score can be taken as follows:

[0087] The weights that can be taken in character quality are:

[0088] The weights in the forwarding priority can be:

[0089] In this specific application embodiment, the character degradation threshold can be set to 0.45, the density change threshold to 0.25, the weighted change threshold to 0.45, and the preset maximum age can be set to three time windows. The above values ​​are used to provide a reproducible application configuration; under different node sizes, movement patterns, or caching conditions, the weights and thresholds can be adjusted according to experimental calibration.

[0090] In one experimental embodiment, a delay-tolerant network routing experimental scenario was established using the ONE simulation platform. A graph computation processing program was used to generate weighted contact graphs, point-to-point mutual information matrices, dynamic packet results, and node role configuration tables for each time window. The experimental scenario included 300 network nodes, a total simulation duration of 43,200 seconds, and a warm-up period of 1,000 seconds. The network runtime after warm-up was divided into 30 consecutive time windows, each approximately 1406.67 seconds long. The number of dynamic packets... The parameters were set to 10, the initial number of replicas at message creation was set to 22, the role degradation threshold was set to 0.45, the network density change threshold was set to 0.25, the weighted change threshold was set to 0.45, and the preset maximum age was set to 3 time windows. Experiments were conducted using 10 random seeds from seed6 to seed15, and the average of the 10 experimental results was taken to obtain the delivery rate, average delivery latency, and cost ratio.

[0091] Please see Figure 7 , Figure 7 The diagram shows a comparison of routing performance provided by this invention. Figure 7 In this context, Direct Delivery means direct delivery of routes, Epidemic means flooding routes, PRoPHET means probabilistic prediction routes, EBR means encounter-based routes, and Spray-and-Wait means spray-and-wait routes. Figure 7 (a) is a comparison of delivery rates. A higher delivery rate indicates a higher proportion of messages successfully delivered to the destination node. The delivery rate of the method of this invention is 0.8891, which is higher than that of direct delivery route (0.3642), flood route (0.3156), probability prediction route (0.2728), encounter-based route (0.7130), and spray waiting route (0.8742). Figure 7(b) is a comparison of average delivery delays in seconds. The lower the value, the faster the message is delivered. The average delivery delay of the method of this invention is 2499.82 seconds, which is lower than that of direct delivery route (7155.56 seconds), flood route (3922.44 seconds), probability prediction route (3623.59 seconds), encounter-based route (3364.71 seconds), and spray waiting route (3316.78 seconds). Figure 7 (c) shows the cost ratio comparison; a lower cost ratio indicates less additional forwarding overhead. The cost ratio of the method in this invention is 22.5543, lower than that of flooding routes (421.37) and probabilistic prediction routes (379.65). Figure 7 As can be seen, the present invention controls the role table update by reconfiguring the node, and improves the probability of messages moving to the destination group by using destination group awareness and role awareness forwarding, thereby improving delivery rate and average delivery latency, while avoiding excessive overhead caused by flooding forwarding.

[0092] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0093] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A node reconfiguration adaptive packet routing method for delay-tolerant networks, characterized in that, include: S1. Collect node contact relationship information in the current time window, construct a weighted contact diagram, and construct a positive point mutual information matrix from the weighted contact diagram; S2. Based on the positive point mutual information matrix, the nodes are dynamically grouped, and the group head node, backup group head node and bridge node are determined in the dynamic grouping results to form a node role configuration table. S3. Determine the node reconfiguration level based on the existence status of the previous valid node role configuration table, the connectivity of the role nodes in the previous valid node role configuration table in the current time window, and the degree of network structure change determined by the weighted contact graph of the current time window and the weighted contact graph of the previous time window. S4. Perform stable reuse, partial role update or full dynamic regrouping on the node role configuration table according to the node reconfiguration level to obtain the currently effective node role configuration table. S5. Based on the current effective node role configuration table, the group to which the destination node is located, and the remaining number of replicas carried in the message, perform destination group-aware and role-aware message forwarding, and update the message replica count when forwarding message replicas.

2. The method according to claim 1, characterized in that, The construction of the positive point mutual information matrix from the weighted contact graph includes: The joint probability of node pairs and the node edge probability are calculated from the contact weights in the weighted contact graph. The point mutual information value of the node pair is calculated based on the joint probability of the node pair and the edge probability of the node; Set the mutual information values ​​of points that are less than zero to zero to obtain the positive mutual information matrix used to characterize the effective contact association of nodes.

3. The method according to claim 2, characterized in that, The process involves dynamically grouping nodes based on the positive point mutual information matrix, and determining the group head node, backup group head node, and bridge node from the dynamic grouping results to form a node role configuration table, including: Landmark selection and spectral embedding are performed on the positive point mutual information matrix to obtain low-dimensional node representations, and the low-dimensional node representations are clustered to obtain node group identifiers; The group head node and the alternate group head node are determined based on at least two of the following: the contact strength of the node within its group, the node degree determined by the weighted contact graph, and the embedding centrality determined by the low-dimensional representation of the node. Bridge nodes are determined based on at least two of the following: cross-group contact strength, cross-group coverage, and cross-group connection ratio between the node and other groups; Write the node group identifier, the group head node, the spare group head node, and the bridge node into the node role configuration table.

4. The method according to claim 3, characterized in that, The node role configuration table, the previous effective node role configuration table, and the currently effective node role configuration table all record node identifiers, node group identifiers, role types, and bridge node connection group sets. The role types include group head nodes, backup group head nodes, bridge nodes, and ordinary nodes.

5. The method according to claim 4, characterized in that, The process of determining the node reconfiguration level based on the existence status of the previous valid node role configuration table, the connectivity of the role nodes in the previous valid node role configuration table in the current time window, and the degree of network structure change determined by the weighted contact graph of the current time window and the weighted contact graph of the previous time window includes: When there is no previous valid node role configuration table, the node reconfiguration level is determined to be full dynamic regrouping; When a previous valid node role configuration table exists, a previous role node set is formed based on the group head node, spare group head node, and bridge node in it, and the role degradation degree is determined based on the node degree and contact strength of the role nodes in the previous role node set in the current time window. The node reconfiguration level is determined based on the degree of role degradation and the degree of network structure change.

6. The method according to claim 5, characterized in that, Determining the node reconfiguration level based on the degree of role degradation and the degree of network structure change includes: When the previous role node set is empty or the age of the role table reaches the preset maximum age, the node reconfiguration level is determined to be full dynamic regrouping; When the degree of role degradation is lower than a preset role degradation threshold and the degree of network structure change is lower than a preset change condition, the node reconfiguration level is determined to be stable reuse; When the degree of role degradation reaches the preset role degradation threshold or the degree of network structure change reaches the preset change condition, the node reconfiguration level is determined as local role update or full dynamic regrouping based on the degree of improvement of candidate groups. The degree of improvement of the candidate grouping is determined by the difference between the dynamic grouping result of the current time window and the node grouping recorded in the previous effective node role configuration table in at least one of the following: group size balance, group size entropy, and number of single-node groups.

7. The method according to claim 4, characterized in that, The process of performing stable reuse, partial role update, or full dynamic regrouping of the node role configuration table according to the node reconfiguration level to obtain the currently effective node role configuration table includes: When the node reconfiguration level is stable reuse, the previous effective node role configuration table will be used as the currently effective node role configuration table. When the node reconfiguration level is local role update, the node group identifier in the previous effective node role configuration table remains unchanged, and the group head node, backup group head node and bridge node are re-determined based on the weighted contact graph of the current time window to generate the currently effective node role configuration table. When the node reconfiguration level is full dynamic regrouping, the node role configuration table formed after dynamic grouping in the current time window is used as the currently effective node role configuration table.

8. The method according to claim 4, characterized in that, The step of performing destination group-aware and role-aware message forwarding based on the currently effective node role configuration table, the group to which the destination node belongs, and the remaining number of replicas carried in the message includes: The node currently holding the message is identified as the current carrying node, and the destination group of the message destination node is determined according to the current effective node role configuration table. When the meeting node is the message destination node, the message is delivered to the message destination node; When the meeting node is not the message destination node, the number of remaining copies carried by the message is greater than one, and the meeting node belongs to the destination group, the meeting node is determined as the message copy receiving node. When the encounter node does not belong to the destination packet, the role types of the encounter node and the current carrying node, as well as the bridge node connection packet set, are read from the current effective node role configuration table. The forwarding priorities of the encounter node and the current carrying node towards the destination packet are determined respectively. When the forwarding priority of the encounter node is higher than that of the current carrying node and the number of remaining copies of the message is greater than one, the encounter node is determined as the message copy receiving node.

9. The method according to claim 8, characterized in that, The step of updating the message replica count when forwarding message replicas includes: When it is determined that the message replica receiving node and the current carrying node hold the number of message replicas, and When the value is greater than 1, forward the message replica to the message replica receiving node and update the message replica count of the current carrying node to 1. Round up to the nearest half and update the message replica count of the message replica receiving node to [the nearest integer]. of is rounded down to the nearest half; When a node holds only one copy of a message, it will not continue to forward message copies unless the node it encounters is the destination node of the message.

Images