Dynamic topology ad hoc network multipath forwarding method and system

Abstract

The invention relates to the technical field of wireless communication, and particularly discloses a multipath forwarding method and system for a dynamic topology ad hoc network. The method comprises the following steps: acquiring node states and link connection information, and generating a connection graph and a state sensing graph; link elements are extracted based on the graph structure, a link value function model is constructed, and a scoring graph is generated; candidate paths are screened through the scoring graph, and a path set structure is calculated; setting a multi-path forwarding priority according to the path set and the service type, and completing channel scheduling and resource allocation; executing multi-path concurrent data forwarding and detecting conflicts, and updating a forwarding record; and finally, adjusting link function parameters based on the conflict information to realize adaptive updating of the model. According to the method, by introducing the path priority scoring model and the optimal scheduling mechanism, the routing scheduling precision and the multi-path forwarding efficiency of the ad hoc network in a dynamic environment are remarkably improved, and the method has relatively high topology adaptive capacity and resource allocation optimization capacity.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication technology, and in particular to a dynamic topology self-organizing network multi-path forwarding method and system. Background Art

[0002] With the rapid development of scenarios such as the Internet of Things (IoT), intelligent transportation, and emergency communications, wireless ad hoc networks (Ad Hoc Networks), characterized by rapid deployment and adaptability, are increasingly being used in infrastructure-free or highly dynamic scenarios. However, traditional Ad Hoc Networks, often based on static routing mechanisms and fixed link scoring models, lack the ability to dynamically respond to node mobility, link fluctuations, and service heterogeneity. This leads to problems such as path congestion, link instability, and forwarding conflicts in practical applications.

[0003] Especially in environments with frequent changes in network topology and concurrent transmission of multiple paths, traditional path selection strategies cannot effectively integrate multi-dimensional factors such as path hop count, link stability, service type, and conflict status, resulting in low channel resource allocation efficiency and unreasonable path priority setting, thus affecting the forwarding efficiency and service quality of the overall network.

[0004] While some existing routing algorithms (such as AODV and the enhanced OLSR protocol) attempt to incorporate link quality awareness, they still lack a unified functional structure that integrates initial scoring, path interference propagation, historical conflict modulation, and Bellman optimal scheduling. Therefore, a multi-path forwarding method and system for self-organizing networks is urgently needed, capable of implementing a linkage mechanism between path priority scoring and resource optimization scheduling under conditions of dynamic topology changes and path conflict interference. This approach can improve routing decision accuracy and scheduling adaptability in complex application environments. Summary of the Invention

[0005] The present invention provides a dynamic topology ad hoc network multi-path forwarding method and system to solve the problem of how to build a path priority scoring model based on the number of path hops, service urgency and conflict level under the conditions of dynamic topology changes and multi-path concurrent interference, and to realize a function closed loop of channel resource allocation in combination with the Bellman optimal scheduling mechanism, so as to improve the routing scheduling accuracy and multi-path forwarding efficiency in the ad hoc network environment.

[0006] In order to solve the above technical problems, the present invention provides a multi-path forwarding method for a dynamic topology ad hoc network, comprising: Obtain node status information and link connection information, generate connection graph data and state perception graph structure; Extract link elements from the connection graph data and state perception graph structure and build a link value function model to generate a link score graph structure; Based on the link score graph structure and connection graph data, perform link value accumulation calculation and candidate path set screening to generate a path set structure; Based on the path set structure and service type information, perform multi-path forwarding priority configuration setting, channel scheduling and resource allocation, and generate a scheduling configuration structure; Based on the results of the scheduling score, the minimization path optimization model is introduced to define the scheduling path Scheduling resource costs; Based on the scheduling configuration structure and the path set structure, it performs multi-path concurrent data forwarding, monitors concurrent conflict events, generates conflict detection data, and updates multi-path forwarding records. Based on the multi-path forwarding records and conflict detection data, the deviation adjustment of the link value function parameters is performed, the link value function parameters are updated, and the model adaptive update is completed.

[0007] Furthermore, the steps of obtaining node status information and link connection information and generating connection graph data and a state perception graph structure include: Obtain the connection status, movement trajectory and adjacency information of each node in the network, perform cleaning and synchronization processing, and obtain node status information; Based on the node status information, the relationship between link establishment, disconnection and hop count is extracted to generate link connection information; Node status information and link connection information are jointly modeled to generate connection graph data and state-aware graph structure.

[0008] Furthermore, the steps of extracting link elements from the connection graph data and the state perception graph structure and constructing a link value function model to generate a link scoring graph structure include: Obtain connection graph data and state-aware graph structure to extract channel quality, hop count, and link stability between nodes; Standardize channel quality, hop count, and link stability to build a link value function model; The link value function model is calculated pair by pair to generate a link score graph structure.

[0009] Furthermore, based on the link score graph structure and the connection graph data, link value accumulation calculation and candidate path set screening are performed to generate the path set structure, including the following steps: Obtain the link score graph structure and connection graph data, and extract the path set that meets the reachability constraints; Perform link value accumulation calculation and sorting on the path set to screen the candidate path set; The candidate path set is verified for connectivity and deduplication to generate a path set structure.

[0010] Furthermore, the step of performing multi-path forwarding priority configuration setting based on the path set structure and service type information includes: Get the path set structure and current business type information, extract the path number and business urgency information.

[0011] Furthermore, the step of performing multi-path forwarding priority configuration setting includes: The path number is matched with the business urgency information to generate a multi-path forwarding priority configuration setting.

[0012] Furthermore, the step of generating a multi-path forwarding priority configuration setting includes: Based on the multi-path forwarding priority configuration setting, channel scheduling and resource allocation are completed, and a scheduling configuration structure is generated.

[0013] Furthermore, the steps of performing multi-path concurrent data forwarding and monitoring concurrent conflict events include: Obtain the scheduling configuration structure and path set structure, and perform multi-path concurrent data forwarding; Monitor channel interference and conflict status during multi-path concurrent data forwarding and generate conflict detection data; Based on the conflict detection data, the path load status and channel conflict log are recorded, and the multipath forwarding record is updated.

[0014] Furthermore, the steps of performing deviation adjustment of link value function parameters and completing model adaptive update include: Obtain multipath forwarding records and conflict detection data, extract path load information and conflict event results; Based on the path load information and the conflict event results, the deviation adjustment of the link value function parameters is performed; The adjusted link value function parameters are applied to the link value function model to complete the model adaptive update.

[0015] A dynamic topology self-organizing network multi-path forwarding system, applied to any of the above-mentioned dynamic topology self-organizing network multi-path forwarding methods, comprising: The state perception module is used to obtain node state information and link connection information, and generate connection graph data and state perception graph structure; The link modeling module is used to extract link elements from the connection graph data and the state perception graph structure and construct a link value function model to generate a link scoring graph structure; The path screening module is used to perform link value accumulation calculation and candidate path set screening based on the link score graph structure and connection graph data to generate a path set structure; Priority configuration module, used to perform multi-path forwarding priority configuration setting, channel scheduling and resource allocation based on the path set structure and service type information, and generate a scheduling configuration structure; A forwarding execution module is used to execute multi-path concurrent data forwarding based on the scheduling configuration structure and the path set structure, monitor concurrent conflict events, generate conflict detection data, and update multi-path forwarding records; The model update module is used to perform deviation adjustment of link value function parameters based on multi-path forwarding records and conflict detection data, update link value function parameters, and complete model adaptive update.

[0016] The key innovations of the present invention include: (1) Innovation in the construction method of the path priority scoring model: It integrates three indicators: the number of path hops, business urgency and conflict level, and iteratively generates path scores through logarithmic modulation function and PageRank-like propagation structure, thus realizing a refined sorting mechanism in multi-path concurrent scheduling scenarios.

[0017] (2) Innovation in inter-path interference modeling and propagation mechanism: By constructing an interference matrix through inter-path shared links and historical conflict data, the limitation of the traditional path scoring model that cannot consider the impact between paths is expanded.

[0018] (3) Innovation in the design of scheduling optimization structure: Combining the Bellman dynamic programming strategy with the path scoring function, an operational channel action set and resource cost function are introduced to complete the closed-loop mapping between path scoring and resource actions.

[0019] The following are its main beneficial effects: (1) Improving the accuracy of dynamic routing decisions: By building a multi-factor priority scoring model that integrates the number of path hops and business urgency, the system can dynamically evaluate the value of paths under conditions of frequent topology changes, avoiding decision imbalances caused by relying solely on link length or historical load.

[0020] (2) Enhanced multi-path scheduling flexibility: The proposed path interference propagation mechanism combined with the conflict history penalty factor can accurately identify potential conflict relationships between paths, improve the spatial separation and concurrency efficiency of path selection, and significantly reduce the probability of data transmission conflicts.

[0021] (3) Achieve a closed loop for resource allocation optimization: The channel scheduling cost function is constructed in combination with the Bellman equation, so that a direct function mapping is established between the path scoring results and the channel configuration actions, ensuring that a resource binding strategy that minimizes scheduling costs is implemented in resource-constrained scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A schematic diagram of a multi-path forwarding method for a dynamic topology ad hoc network provided in an embodiment of the present application; Figure 2 This is a structural block diagram of the dynamic topology self-organizing network multi-path forwarding system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0023] Example 1: Reference Figure 1 , is a flow chart of a multi-path forwarding method for a dynamic topology ad hoc network provided by an embodiment of the present invention. The flow chart may include at least steps S100-S600: S100, obtaining node status information and link connection information, generating connection graph data and state perception graph structure; S200, extracting link elements from the connection graph data and the state perception graph structure and constructing a link value function model to generate a link scoring graph structure; S300: Based on the link score graph structure and the connection graph data, perform link value accumulation calculation and candidate path set screening to generate a path set structure; S400, based on the path set structure and service type information, perform multi-path forwarding priority configuration setting, channel scheduling and resource allocation, and generate a scheduling configuration structure; S500, based on the scheduling configuration structure and the path set structure, perform multi-path concurrent data forwarding, monitor concurrent conflict events, generate conflict detection data, and update multi-path forwarding records; S600: Based on the multi-path forwarding records and conflict detection data, perform deviation adjustment of link value function parameters, update link value function parameters, and complete model adaptive update.

[0024] Step S100 at least includes steps S110-S130: S110 , obtaining the connection status, movement trajectory and adjacency relationship information of each node in the network, performing cleaning and synchronization processing, and obtaining node status information.

[0025] Specifically, the "connection status information" includes the activation status of the node, whether the communication module is turned on, the current network access status, etc., which are reported regularly through the local status monitoring program of the node device and collected uniformly by the central scheduling module.

[0026] The "mobile trajectory information" is based on the inertial measurement unit, positioning chip (such as a GNSS module) and motion estimation model carried by the node, recording its continuous position change data in two-dimensional or three-dimensional space; this data can be compressed using a sliding window mechanism to improve processing efficiency.

[0027] The "adjacency information" is obtained through the mechanism of each node periodically broadcasting adjacency detection messages. When each node receives a broadcast message from another node, it records the neighbor node's identifier, received signal strength, relative distance and timestamp to form a temporal adjacency table.

[0028] After the above three types of data are collected, a unified timestamp alignment operation needs to be performed, that is, by referring to the unified network time base, the asynchronous problem caused by the delay in reporting the status of each node is eliminated.

[0029] Furthermore, outlier elimination and interpolation repair processing are performed on incomplete or noisy trajectory data or abnormal jump adjacency information to obtain structured node status information, which will be used to identify link establishment and disconnection events in S120.

[0030] S120 , based on the node status information, extract the relationship between link establishment, disconnection and hop count, and generate link connection information.

[0031] Specifically, based on the acquired node status information, the system first performs a time-series matching analysis on the adjacency record sequences of any two nodes. If, within a certain time window, two nodes are marked as mutually adjacible and their signal strength is above a preset access threshold, a valid link is established within that window, representing a link establishment event. If, within multiple consecutive time windows, a previously connected node pair has no adjacency records or its signal strength falls below the threshold, a link disconnection event is identified.

[0032] Furthermore, for all nodes in the network, an adjacency graph is constructed based on the adjacency table. Based on this, methods such as breadth-first traversal are used to calculate the hop count between nodes. Hop count is the number of relay nodes required to transmit data from the source node to the destination node under the current topology. It is an important basic indicator for path calculation and link value assessment.

[0033] Through the above operations, a link connection information structure is generated. This structure records the link status (connected / disconnected), link duration, link frequency, hop level, and other elements between each node pair in the network. This link connection information serves as the key input for the joint modeling in the next sub-step.

[0034] S130 , jointly modeling the node status information and the link connection information to generate connection graph data and a status perception graph structure.

[0035] The node status information and link connection information have a multi-dimensional coupling relationship in structure: the former contains the attributes of the node itself (such as status, location, and adjacency set), and the latter reflects the structure and interaction behavior between node pairs.

[0036] Specifically, each node is represented as a graph node element in a connection graph. The valid links recorded in the link connection information are mapped to edge elements in the graph. Each edge is assigned a weight attribute, including but not limited to the link's lifetime, mean signal strength, and hop count. The resulting connection graph data is stored using an adjacency matrix, edge list, or compressed sparse structure to improve operational efficiency in large-scale ad hoc networks.

[0037] Furthermore, a state-aware graph structure is constructed based on the connection graph data. This structure achieves a dynamic attribute fusion representation of the connection graph by associating each node with its time series state information and each edge with its time series stability indicator. The specific modeling method uses a data structure based on an attribute graph, where nodes contain state vectors and edges contain multi-dimensional link features. All attributes are updated in real time over a sliding window.

[0038] The connection graph data and state-aware graph structure serve as the unified input foundation for subsequent steps: link quality extraction (S210), link function model construction (S220), and path screening (S310). The link historical stability, node hop count, and connection frequency in the state-aware graph structure are directly extracted as link quality factors in S210 and input into the link value function model in S220.

[0039] This step constructs graph-structured basic data that integrates the dynamic attributes of nodes and links through multi-dimensional node state perception and link event modeling. This provides high-precision, updateable data source input for subsequent link quality assessment and path selection, adapts to the self-organizing network environment with high-speed topology changes, and improves the routing system's responsiveness to structural evolution and the accuracy of path screening.

[0040] Step S200 at least includes steps S210-S230: S210 , obtaining connection graph data and state perception graph structure, and extracting channel quality, hop count, and link stability between nodes.

[0041] Specifically, the connection graph data is a graph structure constructed in the previous step based on node state information and link connection information. It records link-related attributes such as connection frequency, communication latency, and average signal strength corresponding to each graph edge. The state-aware graph structure, by binding it to the time dimension, provides indicators of link dynamic characteristics such as the number of establishments, disconnection duration, and reconnection frequency.

[0042] In this step, the system first traverses all edge structures in the connection graph data and reads the attached attribute set for each edge structure. The "channel quality" is estimated based on the average received signal strength, packet loss rate, and interference detection during node communication. The system preferentially selects signal strength data with small fluctuations and high sampling frequency over multiple time windows in the past and uses this as a representative channel quality indicator. The "hop count" is directly taken from the node hop count relationship table constructed based on the adjacency table in S120, reflecting the shortest path length between the source node and the target node.

[0043] Link stability is calculated using time-series metrics such as link disconnection frequency and average duration recorded in the state perception graph. Specifically, if a node pair maintains continuous communication with no abnormal interruptions over a period of time, its link stability is rated high; if frequent connection establishment and disconnection occur, its link stability is rated low.

[0044] The above three link elements are uniformly extracted and encapsulated into a link feature triple structure, providing unified data input for constructing the link value function model in the next sub-step.

[0045] S220: Standardize the channel quality, hop count, and link stability to build a link value function model.

[0046] After obtaining the link feature triple structure, in order to achieve a unified evaluation of cross-dimensional indicators, this step performs a normalization process on all link elements. The normalization process includes range normalization and direction adjustment operations.

[0047] Specifically, the channel quality is affected by device differences and environmental interference between different node pairs, so normalized mapping is performed by taking the maximum and minimum value intervals within the sliding time window; The hop count is a discrete integer indicator. The system divides it into grades according to the fluctuation range of the average network hop count and converts it into a continuous scoring factor. Link stability was originally a non-quantifiable attribute and needed to be converted into a numerical indicator through a weighted calculation method based on communication continuity and link disconnection events.

[0048] After standardization is completed, the system introduces a link value function modeling module to perform a weighted combination of the above three standardized elements.

[0049] The weight configuration is preset based on the type of network segment the link is in: For edge region node pairs, the hop count factor has a higher weight; For high-load paths, the stability factor dominates; For core forwarding nodes, channel quality is the primary consideration.

[0050] The resulting link value function model is a three-dimensional linear combination model with real-time update capabilities. It can output the current link value score for any given node pair and has interpretability and generalization capabilities. This model will be used for batch calculation in the next step.

[0051] S230: Calculate the link value function model pair by pair to generate a link score graph structure.

[0052] After completing the construction of the link value function model, this step sequentially performs value score calculation operations on all graph edge structures in the connection graph data to form a link score graph structure.

[0053] Specifically, for any node pair, the system calls the link feature triplet, inputs it into the aforementioned link value function model, and outputs the link score value for the current node pair. The score result is written as a weight into the weight field of the corresponding graph edge structure, and the original connection graph data is updated to the link score graph structure.

[0054] Compared to the original connection graph data, the link scoring graph structure no longer contains only link existence and connectivity information, but also includes a scoring indicator field. This structure serves as an important input for path set screening and path value accumulation calculation in module S300. Its scoring value is used in the aggregation and sorting operation in S320, which in turn influences the path optimization results.

[0055] Furthermore, to improve the efficiency of path screening, this step compresses and stores the scoring graph structure in the form of a sparse adjacency list, and establishes a fast indexing mechanism to ensure the linear scalability of path calculation efficiency in subsequent large-scale node environments.

[0056] This module constructs a callable link value function and generates a quantitative scoring graph structure through unified extraction and normalized modeling of multi-dimensional link elements, providing a unified measurement basis for path optimization and channel scheduling, and enhancing the dynamic adaptability and precise evaluation capabilities of path calculation.

[0057] Step S300 at least includes steps S310-S330: S310: Obtain the link score graph structure and connection graph data, and extract a path set that meets the reachability constraint.

[0058] Specifically, the system first reads the link scoring graph structure, which consists of scoring edges between node pairs, with edge weights representing the transmission quality of the current link. Combined with the topological connection relationship of each node in the connection graph data, the system establishes the initial conditions for graph traversal.

[0059] Subsequently, a path expansion operation is initiated for each source node based on the path discovery mechanism. Using a depth-first search strategy with reachability constraints, multi-hop path expansion is performed along the scored edges in the graph, constructing a set of paths from the source node to all reachable destination nodes. These reachability constraints include, but are not limited to, a maximum hop count threshold, node non-repetition restrictions, and loop detection mechanisms to ensure that the path is physically feasible and conforms to the transmission logic.

[0060] During the execution process, the system binds a path number to each path and records its node sequence, edge sequence and hop count information to initially form a set of path structures for use in the next stage of value calculation and screening operations.

[0061] S320: Perform link value accumulation calculation and sorting processing on the path set to screen the candidate path set.

[0062] After obtaining the preliminary path set structure, this step accumulates the full link value of each path. Specifically, the system reads the score value of each edge in the path from the link score graph structure and performs a cumulative calculation to obtain the total link value of the path. The "total link value" represents the comprehensive performance evaluation of data transmission on the path.

[0063] To ensure that the selected paths have high transmission value, the system sorts all paths based on their total value in descending order, that is, the paths with the highest total value are prioritized.

[0064] Then, based on the set candidate path number threshold, the paths with the highest scores are selected to form the candidate path set. If there are paths with the same score and the same number of hops, the system uses the score of the weakest link in the reference path as a secondary ranking indicator for hierarchical screening.

[0065] The above processing ensures that the path subset with the best value can be quickly located in the huge set of reachable paths, providing the basis for efficient path selection for the subsequent resource scheduling module.

[0066] S330: Perform connectivity verification and deduplication processing on the candidate path set to generate a path set structure.

[0067] In order to further ensure the validity and uniqueness of path selection, this step performs structural integrity verification and redundancy removal on the candidate path set.

[0068] First, the system traverses the candidate path set and verifies for each path whether its node sequence contains broken hops or indirect node pairs. By querying the connection graph data, the system confirms whether all consecutive nodes in the path have valid edges in the graph structure. If a path is found to contain broken links, missing edges, or logical loops, the path is marked as abnormal and removed.

[0069] Next, the system performs a duplicate structure recognition operation on the remaining paths. Specifically, the system hashes the node sequence of the path and compares it with the set of existing path hash values. If a completely identical path encoding is found, only the path with the better score is retained, and the redundant paths with poorer scores are removed.

[0070] Ultimately, the system outputs a path set structure that is legal, optimally scored, and free of redundancy. Each path in this structure records the path number, node sequence, cumulative link score, and hop count information, and serves as the standard form of the path set structure input to the S400 module.

[0071] Through the structural extraction, scoring evaluation and legitimacy processing of the path set in this step, the transformation from a dynamic link scoring graph to a path set structure with global optimality is achieved, providing high-quality and stable path input for subsequent multi-path forwarding strategy generation, effectively reducing routing redundancy, improving path reliability, and enhancing the overall forwarding efficiency of the system in a dynamic self-organizing network environment.

[0072] Step S400 at least includes steps S410-S430: S410: Obtain the path set structure and current service type information, and extract the path number and service urgency information.

[0073] Specifically, the system first obtains the unique identification number of each path in the path set structure, the corresponding node sequence and the cumulative link score, and at the same time obtains the service type distribution data within the scheduling period. The service type information contains multiple dimensions, including real-time performance, reliability, packet size, retransmission tolerance, etc. The system integrates and maps them into the urgency score value of the service required by the path. The urgency score is recorded as , the number of path hops is , the system performs the priority initial value distribution calculation based on this, the formula is as follows: ; in, : No. The initial priority score of the path, in normalized ratio; : No. The urgency score of the service associated with each path is weighted based on service real-time performance, packet loss tolerance, and other attributes. : No. The hop length of the path, that is, the number of relays between nodes; : Index variable, sums all paths in the path set; : The total number of paths in the current path set structure, which is a positive integer; : A logarithmic mapping function of the path hop count, used to modulate the nonlinear gain effect of the hop count. This value serves as the basic initialization input for subsequent inter-path impact modeling.

[0074] S420: Match the path number with the service urgency information to generate a multi-path forwarding priority configuration setting.

[0075] After the initial score is generated, the system constructs an inter-path impact matrix based on the shared link relationship between the path structures, defining the forwarding impact weights between paths due to factors such as channel sharing and conflict history. The system then introduces an iterative priority scoring model based on improved PageRank, and uses the following formula to iteratively update the path priority: ; in, : No. The path in Priority score value in the round iteration; : Scoring retrospective coefficient, the value range is , control the proportion of initial scores; : No. The path to The influence weight of each path is derived from the link overlap and conflict history between paths; : No. The path in Scoring in round iterations; : No. Initial scores of paths (from formula ①); 、 : Path collection index variable.

[0076] Eventually converges to a stable distribution The scoring result is used as the basic value input for path scheduling priority.

[0077] S430 : Based on the multi-path forwarding priority configuration setting, complete channel scheduling and resource allocation, and generate a scheduling configuration structure.

[0078] The system obtains the final priority score of the path Finally, the path historical conflict level is comprehensively considered , path hop count And the priority scoring results, the scheduling score function is constructed as follows: ; in, : No. The comprehensive scheduling score of each path is used for channel scheduling sorting; : Three independent scoring weight coefficients, satisfying unit unity; : No. The final stable priority score of the path (obtained after the convergence of formula ②); : No. the historical conflict level of each path; : No. The hop length indicator of each path comes from the path set structure; : The path index in the path set, which is consistent with formulas ① and ②.

[0079] The historical conflict level expression is as follows: ; in, This is the historical conflict level in formula ③; is the indicator function; Indicates the path In time Whether a channel conflict occurs.

[0080] Based on the scheduling score, the system introduces the Bellman minimization path optimization model to define the scheduling path The scheduling resource cost is as follows: ; in, : No. The resource scheduling cost value of the path under the current channel action selection; :path Optional channel action set, usually including: channel number, time slot duration, transmission power level, etc.; :path In the selected action The real-time resource consumption under the condition can be calculated by combining the interference degree, channel utilization and time slot proportion; : Discount factor for future costs, valued at , control the long-term resource consumption impact; : The optimal resource scheduling value of the associated path after the current path, forming Bellman recursion; :Indicates the first Path index; : The set of channel actions available in the current scheduling period of the system.

[0081] Finally, the system generates a scheduling configuration structure based on the optimal value path sequence of formula ④, which includes fields such as path number, allocated channel, occupied time slot and scheduling priority.

[0082] Step 500 at least includes steps S510-S530: S510: Obtain a scheduling configuration structure and a path set structure, and execute multi-path concurrent data forwarding.

[0083] Specifically, the system first calls the scheduling configuration structure generated in the previous step S430 to extract the control parameters of each path, such as the scheduling time slot, allocated channel number, transmission power level, and scheduling priority. This scheduling configuration structure is then combined with the path set structure generated in S330 to form a complete concurrent forwarding task table.

[0084] The system then activates the scheduling control module based on the scheduling clock. During each scheduling cycle, concurrent data forwarding operations are performed on the assigned paths according to the scheduling priority order. These forwarding operations are driven by the node's local cache queue and the link transmission protocol, and specifically include payload data loading, modulation and coding settings, physical channel activation, and acknowledgment feedback.

[0085] To meet high throughput requirements, the system uses multithreading or asynchronous I / O for concurrent forwarding. A confirmation callback function is set for each path forwarding operation, which is then used for subsequent conflict detection and synchronization with the result feedback structure. Forwarding events are written to the forwarding behavior log in real time during the scheduling cycle, recording each successful and failed forwarding attempt for a path-channel pair.

[0086] S520: Monitor channel interference and conflict status during multi-path concurrent data forwarding, and generate conflict detection data.

[0087] This step starts the channel conflict monitoring mechanism in real time during the concurrent forwarding process, and determines the interference level and conflict events in the forwarding channel. The system combines the following three types of information for conflict monitoring: Channel monitoring module output: Based on the local physical layer monitor of each node, it records the overlapping signals, burst noise, power imbalance and other phenomena in the scheduling channel; Abnormal reception confirmation record: If the ACK signal from the target node is not obtained within the specified reception window, the system records it as a potential conflict or packet loss event; Conflict identification criterion model: The system calls a statistical model based on a sliding window to perform abnormal identification and processing on the number of consecutive failures, average received signal-to-noise ratio, and bit error rate changes.

[0088] When the set conflict condition threshold is met, the system marks the path corresponding to the channel as a conflict event within the time period. The system constructs a conflict detection data structure based on the scheduling cycle, which records the following: Conflict path number; The timestamp of the conflict; Corresponding channel number and frequency band; Interference intensity level; Conflict retry count and final processing status.

[0089] The above conflict detection data will serve as the core input content for S530 module path load statistics and conflict log updates.

[0090] S530: Based on the conflict detection data, record the path load status and channel conflict log, and update the multipath forwarding record.

[0091] After completing the conflict detection data collection, the system enters the path forwarding status archiving phase, performing load status statistics and conflict log archiving operations on all paths in this scheduling cycle.

[0092] First, the system summarizes the total packet forwarding volume, number of successful receipts, and number of failures for each path in the current cycle. Based on the transmission success rate and retry rate, the system constructs path load status indicators, including: Actual throughput; Average number of retransmissions; Average load per unit time; Channel resource utilization efficiency.

[0093] Next, the system combines collision detection data to construct a collision log, marking each path with its collision frequency, collision level, and duration. The collision log also includes the system's response measures after each collision, such as channel migration, rescheduling delay, and power adjustment.

[0094] The system binds the path load status to the channel conflict log structure, writes it into the multipath forwarding record table, and stores it in the path scheduling status history database. This database serves as an important input for adjusting the parameters of the link value function model in the subsequent step S600, driving the model to continuously optimize link scoring and path selection strategies under dynamic network changes.

[0095] Through the design and implementation of this module, the system achieves a closed-loop connection between path scheduling and actual concurrent forwarding in a complex ad hoc network environment, significantly improving the following performance aspects: Enhanced system robustness: In dynamic channel environments, fast adjustments are achieved by combining scheduling priorities with real-time conflict feedback. Improve forwarding efficiency: Through path load monitoring and channel conflict detection, effectively optimize resource utilization and avoid channel idling or resource contention; Supports model adaptive updates: Conflict detection data and load records provide data support for subsequent link function model adjustments, improving the accuracy of the scoring system; Reduce network congestion: Conflict identification and load backtracking mechanisms can reduce the probability of interference caused by multi-path concurrency and ensure the timeliness of high-priority services.

[0096] In summary, this module is the core execution link in the invention patent, ensuring the efficient connection between the upper-layer path scoring structure and the lower-layer physical channel scheduling, forming a closed-loop link scheduling system from scoring optimization to execution feedback.

[0097] Step S600 at least includes steps S610-S630: S610: Acquire multi-path forwarding records and conflict detection data, and extract path load information and conflict event results.

[0098] Specifically, the system first extracts metrics such as the cumulative forwarded data volume, the number of packets actually completed, the forwarding delay range, and the number of retransmissions for each path during the current scheduling period from the multipath forwarding record structure, and categorizes these metrics as path load information. This path load information includes, but is not limited to, average bandwidth utilization per time slot, packet processing rate, and workload levels of path relay nodes.

[0099] Simultaneously, the system retrieves a list of channel conflict events detected during the scheduling period from the conflict detection data, extracts the conflict path, conflict type (e.g., channel overlap, interference amplification, synchronization loss), and corresponding timestamp, and aggregates this information into a conflict event result set. Furthermore, based on the path number, the system maps path load information to conflict event results to construct a path performance feedback table. This feedback table will be used in subsequent steps to drive the deviation correction mechanism of the link value function model.

[0100] The source of the path load information depends on the data forwarding status record completed in S510, while the conflict event result comes directly from the conflict detection result set constructed in S520. Both serve as the core input of the S600 module, ensuring that the data source for model modification is real-time and dynamically effective.

[0101] S620: Execute deviation adjustment of link value function parameters according to the path load information and the conflict event result.

[0102] In this step, based on the aforementioned path performance feedback table, the system traces back the scoring functions of each link segment involved in each path in the link scoring graph structure and extracts the various weight parameters in the corresponding link value function model. Combined with the feedback data, the system performs the following three types of deviation measurement operations: First, if a link segment appears repeatedly on multiple heavily loaded paths, and the average transmission delay along the corresponding path is significantly higher than the model's previous estimate, the system will deem the link's stability parameter overestimated. In this case, the weight parameter corresponding to the link's stability factor will be adjusted downward, reflecting its actual low availability.

[0103] Second, if the path associated with a link segment frequently appears in conflict event results, especially in channel interference-related conflict records, the system deems the link to have high channel contention or dense interference sources. In this case, the influence weight of its channel quality factor is reduced, and its trust score reflected in the model is updated.

[0104] Third, if a link segment appears in multiple paths with low hop counts but high conflict frequencies, the system will combine the hop count impact factor with the historical conflict level to establish a weighted coefficient correction model and dynamically update the hop count factor weight to prevent it from being overestimated by the nonlinear modulation in the original model.

[0105] All parameter adjustments are based on the multi-cycle feedback mean calculation within a sliding time window to prevent model overfitting caused by single-cycle incidents. The system also records all parameter adjustment processes and results, constructing a model parameter evolution log structure to provide application records for subsequent S630.

[0106] The parameter structure of the link value function model is derived from the standardized weight combination model defined in step S220. The parameter symbols were applied and instantiated in step S230. The adjustment performed in S620 is to substantially modify the aforementioned weight parameters based on the feedback structure obtained in step S500, thereby completing the primary task of updating the model core.

[0107] S630: Apply the adjusted link value function parameters to the link value function model to complete the model adaptive update.

[0108] After completing the parameter deviation adjustment, the system will re-inject the new weight configuration into the link value function model and perform model version overwriting and index reconstruction operations. Specifically, it includes the following three sub-operations: First, the system performs a consistency check between the adjusted weight factor set and the original model weight structure, ensuring structural alignment of the model dimensions and semantic consistency of the factors. Once this is confirmed, the new parameter set replaces the static configuration in the original model structure, generating a Link Value Function model with an updated version identifier.

[0109] Next, the system performs a full validation operation on the updated model, recalculates the scores for all link segments in the connectivity graph data, and constructs a new link score graph structure based on the updated results. This structure serves as the basic input data for the path screening and channel scheduling modules (i.e., S300 and S400) in the next scheduling cycle, ensuring that the resulting path set fully reflects the current forwarding feedback effect.

[0110] Finally, the system structures and encapsulates the path numbers, conflict levels, load metrics, and parameter change records involved in this adaptive update, storing them in the model evolution history module. This record structure includes timestamp indexing, parameter change trend tracking, and version rollback mechanisms, providing support for subsequent model stability monitoring and reverse restoration of abnormal updates.

[0111] The model update task completed in this step constitutes a key closed loop from actual forwarding feedback to self-repair of the scoring mechanism, indicating that the link value function has the ability to learn, modify, and evolve model behavior, significantly enhancing the intelligent level of path selection and the robustness of system scheduling in the dynamic topology self-organizing network environment.

[0112] This implementation step achieves the full-cycle adaptive capability of the link value function model by constructing a dynamic feedback link from data forwarding feedback to scoring function parameter updates. By integrating path load information and conflict event records, the system can accurately identify static parameter deviations in the scoring model and implement factor-by-factor corrections based on dynamic behavior results. Ultimately, the system completes the automatic iterative optimization of the link scoring mechanism, ensuring that the path selection and resource scheduling logic are tightly coupled with the actual transmission status, improving the reliability, recoverability, and self-learning capabilities of the entire system in a high-speed topology change environment, and providing core intelligent evolution mechanism support for the present invention.

[0113] Example 2: Figure 2 FIG. 1 shows a structural block diagram of a dynamic topology self-organizing network multi-path forwarding system according to an embodiment of the present invention. Figure 2 As shown, the structure may include: The state perception module 10 is used to obtain the state information and link connection information of each node in the network, and generate connection graph data and a state perception graph structure. Specifically, the state perception module 100 includes a node data acquisition unit and a graph structure modeling unit. The node data acquisition unit is used to periodically collect the connection state information, movement trajectory information, and adjacency relationship information of each communication node, and complete cleaning, synchronization, and completion processing; the graph structure modeling unit is used to construct connection graph data and a dynamic state perception graph structure based on the link establishment and disconnection events between nodes, combined with the node hop count relationship, and provide multi-dimensional structure input for the subsequent link modeling module.

[0114] The link modeling module 20 extracts link quality factors from the connection graph data and the state-aware graph structure, constructs a link value function model, and generates a link scoring graph structure. This module comprises a link feature extraction unit, an indicator normalization unit, and a value modeling unit. The link feature extraction unit extracts key link factors such as inter-node channel quality, hop count, and link stability; the indicator normalization unit normalizes and directs link quality indicators across different dimensions; and the value modeling unit constructs a real-time link value function model based on a multi-factor weighted structure and performs score calculations on all node pairs, forming a link scoring graph structure.

[0115] The path screening module 300 is used to perform link value accumulation calculations and candidate path set screening based on the link scoring graph structure and connection graph data, thereby generating a path set structure. The path screening module 300 comprises a path extraction unit, a path scoring unit, and a structure verification unit. The path extraction unit performs multi-hop path traversal based on the scoring graph to construct a path set that satisfies reachability constraints. The path scoring unit performs link value accumulation calculations and ranking on each path to screen a set of candidate paths. The structure verification unit further performs path legitimacy verification, loop detection, and redundancy removal, outputting a path set structure that provides input to the priority configuration module.

[0116] Priority configuration module 40 is used to perform multi-path forwarding priority configuration settings, channel scheduling, and resource allocation based on the path set structure and service type information, generating a scheduling configuration structure. The priority configuration module includes a service perception unit, a priority scoring unit, and a resource allocation unit. The service perception unit is used to obtain the type and urgency information of various service requests within the current scheduling cycle and complete the mapping with the path structure. The priority scoring unit constructs a multi-round iterative scoring function based on the number of path hops, service urgency, inter-path interference relationships, and historical conflict levels, and outputs the path scheduling priority. The resource allocation unit combines the scheduling scoring results and implements channel allocation and transmission resource scheduling based on the Bellman optimal model. It generates a scheduling configuration structure and defines the channel resource binding relationship between each path within a specific time slice.

[0117] The forwarding execution module 50 is used to execute multipath concurrent data forwarding based on the scheduling configuration structure and the path set structure, monitor concurrent collision events, generate collision detection data, and update multipath forwarding records. This module comprises a data forwarding execution unit, a collision monitoring unit, and a record update unit. The data forwarding execution unit executes multipath parallel forwarding operations within each scheduling cycle according to the scheduling configuration structure, transmitting data packets according to the allocated channels and time slots. The collision monitoring unit uses a real-time channel monitoring mechanism and interference matching rules to identify channel interference, path collisions, and retransmission events during the forwarding process and generate collision detection data. The record update unit stores metrics such as path transmission rate, number of retransmissions, and channel utilization in a structured manner to form a multipath forwarding record structure, which provides input for the subsequent model update module.

[0118] Model update module 60 is used to adjust the deviation of link value function parameters based on multipath forwarding records and conflict detection data, update the link value function parameters, and complete the adaptive update of the link scoring model. This module includes a feedback analysis unit, a parameter correction unit, and a model reconstruction unit. The feedback analysis unit classifies and analyzes data such as path load information, conflict event frequency, and channel occupancy trends to identify the source of the model's current evaluation error. The parameter correction unit dynamically adjusts the stability weight, channel quality factor, and hop contribution weight in the model based on the actual link transmission performance and score deviation to achieve factor calibration of the link value function. The model reconstruction unit redeploys the updated parameters to the original link value function model, generating a new round of scoring graph structure for the next round of calculations in the path screening module, forming a complete closed-loop scoring-forwarding-update mechanism.

[0119] Compared with the prior art, this embodiment provides the following beneficial effects: Strong dynamic adaptability: By introducing the state perception module and path screening module, the path selection basis can be quickly updated in an environment where the network topology frequently changes, achieving efficient routing strategy reconstruction.

[0120] Improved forwarding efficiency: Through the multipath priority scoring mechanism in the priority configuration module, the system can dynamically allocate channel resources according to the urgency of the service, effectively alleviating multipath conflicts and improving forwarding rates and packet success rates.

[0121] Enhanced system self-learning capabilities: Relying on the model update module, the system can adjust the link value function parameters according to actual forwarding records and conflict detection results, realize model adaptive update, and improve the accuracy and robustness of path scoring.

[0122] The module collaborative structure is clear: the system modules transmit data in a closed loop, forming a complete state perception - path modeling - scheduling forwarding - model feedback link, which effectively ensures the continuity and stability of the system operation.

[0123] Wide applicability: The system can be widely applied to various dynamic self-organizing communication scenarios such as highway vehicle networks, urban wireless sensing networks, and emergency rescue networks, and has strong generalization capabilities and practical deployment value.

[0124] In summary, this embodiment provides an ad hoc network forwarding mechanism that highly integrates perception, modeling, decision-making, and updating capabilities in a dynamic topology environment, and has extremely high technical innovation and engineering practicality.

[0125] Obviously, the embodiments described above are only some of the embodiments of the present application, rather than all of the embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application has been described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions described in the aforementioned specific embodiments, or to make equivalent replacements for some of the technical features therein. Any equivalent structure made using the contents of the present application specification and the accompanying drawings, directly or indirectly used in other related technical fields, is also within the scope of patent protection of the present application.

Claims

1. A multi-path forwarding method for a dynamic topology ad hoc network, characterized in that: The following steps are involved: Obtain node status information and link connection information, generate connection graph data and state perception graph structure; Extract link elements from the connection graph data and state perception graph structure and build a link value function model to generate a link scoring graph structure; Based on the link score graph structure and connection graph data, perform link value accumulation calculation and candidate path set screening to generate a path set structure; Based on the path set structure and service type information, perform multi-path forwarding priority configuration setting, channel scheduling and resource allocation, and generate a scheduling configuration structure; Based on the results of the scheduling score, the minimization path optimization model is introduced to define the scheduling path Scheduling resource costs; Based on the scheduling configuration structure and the path set structure, it performs multi-path concurrent data forwarding, monitors concurrent conflict events, generates conflict detection data, and updates multi-path forwarding records. Based on the multi-path forwarding records and conflict detection data, the deviation adjustment of the link value function parameters is performed, the link value function parameters are updated, and the model adaptive update is completed.

2. The multi-path forwarding method for a dynamic topology ad hoc network according to claim 1, wherein: The steps of obtaining node status information and link connection information and generating connection graph data and state perception graph structure include: Obtain the connection status, movement trajectory and adjacency information of each node in the network, perform cleaning and synchronization processing, and obtain node status information; Based on the node status information, the relationship between link establishment, disconnection and hop count is extracted to generate link connection information; Node status information and link connection information are jointly modeled to generate connection graph data and state-aware graph structure.

3. The multi-path forwarding method for a dynamic topology ad hoc network according to claim 1, characterized in that: The steps of extracting link elements from the connection graph data and the state perception graph structure and constructing a link value function model to generate a link score graph structure include: Obtain connection graph data and state-aware graph structure to extract channel quality, hop count, and link stability between nodes; Standardize channel quality, hop count, and link stability to build a link value function model; The link value function model is calculated pair by pair to generate a link score graph structure.

4. The multi-path forwarding method for a dynamic topology ad hoc network according to claim 1, wherein: Based on the link score graph structure and the connection graph data, link value accumulation calculation and candidate path set screening are performed to generate the path set structure. The steps include: Obtain the link score graph structure and connection graph data, and extract the path set that meets the reachability constraints; Perform link value accumulation calculation and sorting on the path set to screen the candidate path set; The connectivity of the candidate path set is verified and duplicate removal is performed to generate a path set structure.

5. The multi-path forwarding method for a dynamic topology ad hoc network according to claim 1, characterized in that: The steps of performing multi-path forwarding priority configuration setting based on the path set structure and service type information include: Get the path set structure and current business type information, extract the path number and business urgency information.

6. The multi-path forwarding method for a dynamic topology ad hoc network according to claim 1, characterized in that: The steps to configure multi-path forwarding priority include: The path number is matched with the business urgency information to generate a multi-path forwarding priority configuration setting.

7. The multi-path forwarding method for a dynamic topology ad hoc network according to claim 6, characterized in that: The steps for generating a multipath forwarding priority configuration setting include: Based on the multi-path forwarding priority configuration setting, channel scheduling and resource allocation are completed, and a scheduling configuration structure is generated.

8. The multi-path forwarding method for a dynamic topology ad hoc network according to claim 1, characterized in that: The steps of performing multi-path concurrent data forwarding and detecting concurrent conflict events include: Obtain the scheduling configuration structure and path set structure, and perform multi-path concurrent data forwarding; Monitor channel interference and conflict status during multi-path concurrent data forwarding and generate conflict detection data; Based on the conflict detection data, the path load status and channel conflict log are recorded, and the multipath forwarding record is updated.

9. The multi-path forwarding method for a dynamic topology ad hoc network according to claim 1, characterized in that: The steps of performing the deviation adjustment of the link value function parameters and completing the model adaptive update include: Obtain multipath forwarding records and conflict detection data, extract path load information and conflict event results; Based on the path load information and the conflict event results, the deviation adjustment of the link value function parameters is performed; The adjusted link value function parameters are applied to the link value function model to complete the model adaptive update.

10. A dynamic topology self-organizing network multi-path forwarding system, applied to the dynamic topology self-organizing network multi-path forwarding method according to any one of claims 1 to 9, characterized in that: include: The state perception module is used to obtain node state information and link connection information, and generate connection graph data and state perception graph structure; The link modeling module is used to extract link elements from the connection graph data and the state perception graph structure and construct a link value function model to generate a link scoring graph structure; The path screening module is used to perform link value accumulation calculation and candidate path set screening based on the link score graph structure and connection graph data to generate a path set structure; Priority configuration module, used to perform multi-path forwarding priority configuration setting, channel scheduling and resource allocation based on the path set structure and service type information, and generate a scheduling configuration structure; A forwarding execution module is used to execute multi-path concurrent data forwarding based on the scheduling configuration structure and the path set structure, monitor concurrent conflict events, generate conflict detection data, and update multi-path forwarding records; The model update module is used to perform deviation adjustment of link value function parameters based on multi-path forwarding records and conflict detection data, update link value function parameters, and complete model adaptive update.

Images