Adaptive network optimization system based on multiple communication links

Abstract

The invention discloses a self-adaptive network optimization system based on multiple communication links, which relates to the technical field of network optimization and comprises an address change detection module, a neighbor activation wake-up module, a bidirectional interaction monitoring module, a synchronization state vector construction module, a synchronization completion intelligent prediction module and a data flux self-adaptive regulation and control module. According to the method, a bidirectional sensing mechanism of'active detection + passive monitoring 'is introduced, and a support vector machine model based on a feature state vector is combined to intelligently evaluate the synchronization completion degree of neighbor nodes, so that whether address updating synchronization of a path is completed or not is accurately judged; the synchronization state judgment result is dynamically coupled with the current data transmission flux of the hop node, so that a self-adaptive control strategy of the data transmission behavior is realized, and the problems of data mistaken transmission, link interruption, communication black holes and the like caused by the fact that the path is not synchronized are effectively avoided; and smooth transition of the hop node in the state driving mode during the address change period is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of network optimization, and in particular to an adaptive network optimization system based on multiple communication links. Background Art

[0002] The multi-link adaptive network optimization system integrates multiple communication links (such as cellular networks, Wi-Fi, LoRa, and satellite communications) and intelligently schedules link resources based on real-time network status, application requirements, and environmental changes. By incorporating a multi-path dynamic monitoring unit, a link status awareness unit, a QoS prediction engine, and an adaptive scheduling strategy engine, the system continuously assesses the bandwidth, latency, packet loss rate, energy consumption, and interference of each available communication link, and performs real-time selection and switching between multiple links, thereby optimizing overall communication performance and resource utilization efficiency. In critical application scenarios (such as the Internet of Vehicles, Industrial Internet of Things, telemedicine, and emergency communications), the system ensures high reliability, low latency, and load balancing, improving network robustness and service continuity. It serves as a critical infrastructure for intelligent network self-organization and resilient transmission.

[0003] In a multi-link access environment with multi-hop paths, such as a communication system with a mixed deployment of Wi-Fi relays and cellular links, the communication path is typically composed of multiple hop nodes, including relay nodes, bridge devices, and edge routers, which are used to implement step-by-step data forwarding.

[0004] However, in existing technologies, when a hop node undergoes dynamic reallocation of its network address due to a change in network status, adjacent upstream and downstream nodes in the system often fail to promptly identify and synchronize the hop node's new address, resulting in a lag in path status perception. In this situation, the hop node continues forwarding data based on the original path, while its upstream and downstream nodes continue matching communications based on the now-outdated address information. This leads to inconsistencies between path identification and address mapping information, causing data forwarding anomalies. Specifically, data sent by the hop node becomes difficult for downstream nodes to correctly receive and interpret, ultimately creating communication interruptions along the path, leading to unreachable links or data transmission failures. For control applications that rely on high real-time and high-reliability requirements, such as industrial control, remote control, or intelligent terminal linkage, such communication anomalies can prevent critical control commands from being delivered, potentially freezing responses or causing logical failures in downstream devices. In severe cases, these anomalies can even cause system-level failures, impacting the orderly execution of task processes and the overall system's operational stability.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0006] The purpose of the present invention is to provide an adaptive network optimization system based on multiple communication links. By introducing a two-way perception mechanism of "active detection + passive monitoring" and combining it with a support vector machine model based on a characteristic state vector, the system intelligently evaluates the synchronization completion of neighboring nodes, thereby accurately determining whether a path has completed address update synchronization. Furthermore, the synchronization status determination result is dynamically coupled with the current data transmission flux of the hop node, implementing an adaptive control strategy for data transmission behavior. This effectively avoids problems such as data mistransmission, link interruption, and communication black holes caused by unsynchronized paths, ensures that hop nodes can smoothly transition in a state-driven manner during address changes, safeguards the accessibility of key links, and ensures the operational continuity of the entire network system, thereby resolving the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solutions: an adaptive network optimization system based on multiple communication links, comprising an address change detection module, a neighbor activation wake-up module, a two-way interaction monitoring module, a synchronization state vector construction module, a synchronization completion degree intelligent prediction module, and a data flux adaptive control module; The address change detection module continuously monitors the network status of the jump node to determine whether its network address has changed. When a change event is detected for the address of the jump node, it is determined that the jump node has entered the address reallocation state; The neighbor activation wake-up module, once detecting that a hop node has entered the address reallocation state, immediately sends a lightweight detection packet to its known upstream and downstream neighbor nodes through the hop node to actively wake up the neighbor nodes to recognize its new address; The two-way interactive monitoring module starts the monitoring mechanism after the detection data packet is sent, captures and records the response data packets from upstream and downstream neighbor nodes, and establishes a two-way state link from "active detection" to "passive perception"; The synchronization state vector construction module extracts key features related to the synchronization state of the jump node address after receiving response data from neighbor nodes. Within the monitoring window, it conducts in-depth analysis of the extracted key features and, based on the analysis results, constructs the synchronization state vector of the jump node and its corresponding upstream and downstream neighbor nodes, preliminarily quantifying the synchronization status of the current jump node with its upstream and downstream neighbor nodes. The synchronization completion intelligent prediction module inputs the constructed synchronization state vector into a pre-trained support vector machine (SVM) model, and uses the SVM model to intelligently predict the synchronization completion degree of the jump node and its upstream and downstream neighboring nodes; The data flux adaptive control module obtains the current data transmission flux of the jump point node in the monitoring window, and intelligently controls the actual data transmission flux of the jump point node based on the synchronization completion prediction results output by the support vector machine model and the current data flux status, ensuring that the jump point node will not mistakenly send data due to synchronization lag after address reallocation.

[0008] Preferably, in a multi-link access communication network, continuous monitoring of the network address status of a hop node can generally be divided into the following three steps: The first step is to initialize the address recording mechanism. When the jump node is started or online, the system records its current network address information, including IP address, MAC address, subnet mask and gateway information; The second step is to set monitoring trigger conditions to regularly check the interface configuration status, DHCP lease renewal, IPv6 prefix information, NAT mapping table changes of the hop node, or subscribe to operating system-level network events (such as netlink events or interface watchers); The third step is to compare the current address status with the last recorded status. If the address field is found to have changed (such as IP segment drift, interface rebinding, etc.), it is determined to be an address change event.

[0009] Preferably, the lightweight probe data packet can be an ARP request (applicable to IPv4 environment), an NDP neighbor advertisement (applicable to IPv6 environment), a Router Advertisement monitoring trigger signal (used to assist in address prefix update in IPv6) or a UDP probe frame (applicable to direct link layer communication or device-to-device interaction scenarios that do not rely on a complete IP protocol stack).

[0010] Preferably, after obtaining the response data of the neighboring node, key features related to the synchronization status of the jump node address are extracted, wherein the extracted key features include the density of the number of communication round trips for completing a two-way handshake (such as SYN-ACK, ICMP Echo-Reply, UDP response) between the jump node and the upstream and downstream nodes and the distribution concentration of the destination address matching the new address of the jump node. Under the monitoring window, after an in-depth analysis of the extracted key features, a two-way confirmation density factor and an address confirmation distribution factor are generated respectively. Based on the two-way confirmation density factor and the address confirmation distribution factor, a synchronization state vector of the jump node and its corresponding upstream and downstream neighbor nodes is constructed to preliminarily quantify the synchronization status of the current jump node and the upstream and downstream neighbor nodes.

[0011] Preferably, the synchronization state vector constructed by the two-way confirmation density factor and the address confirmation distribution factor is input into a pre-trained support vector machine model, and the synchronization completion evaluation index is output by the support vector machine model. Based on the synchronization completion evaluation index, the synchronization completion of the jump node and the upstream and downstream neighbor nodes is intelligently predicted.

[0012] Preferably, the actual data transmission flux of the hop node is intelligently controlled based on the synchronization completion prediction result output by the support vector machine model and the current data flux state. The specific steps are as follows: Based on the synchronization completion evaluation index output by the support vector machine model and the original data transmission flux of the current jump node, an inhibition factor is calculated to control the actual transmission behavior. The calculation expression is as follows: ,in: This is an evaluation indicator of the synchronization completion between the hop node and its upstream and downstream neighbor nodes. The closer it is to 1, the more complete the synchronization status is. The raw data transmission flux of the hop node in the current monitoring window; is the maximum reference throughput value allowed (such as the rated bandwidth of the link); As an inhibitory factor; is the flux adjustment strength coefficient, which is used to amplify the suppression sensitivity caused by insufficient synchronization. ; is a nonlinear weighted index that controls the nonlinear enhancement of the degree of suppression of the low synchronization state. ; To ensure the inhibitory factor , the larger the value, the stronger the inhibition; Obtaining inhibitory factors After that, the actual data transmission flux of the jump node is dynamically calculated. The calculation expression is: ,in: is the actual data transmission flux of the hop node; is the minimum guaranteed flux constant ( ), used to maintain the bottom line data output of heartbeat packets or link keepalives to prevent zero throughput from causing connection failure.

[0013] Preferably, in the monitoring window, the specific steps of generating a two-way confirmation density factor after performing an in-depth analysis on the density of the number of round trips of two-way handshakes completed between the hop node and the upstream and downstream nodes are as follows: Within the set monitoring window, the hop node monitors and records the two-way handshake communication events between it and all upstream and downstream neighbor nodes, including but not limited to SYN-ACK pairs, ICMPEcho-Reply pairs, UDP request-response pairs, etc. Each complete two-way handshake communication event is identified as a valid round-trip confirmation event, and a handshake event intensity function is constructed to measure the path interaction intensity. The constructed expression is as follows: ,in: Represents the total intensity of two-way handshake communication events within the monitoring window; N is the number of valid round-trip confirmation events observed within the monitoring window; is the confidence weight coefficient of the i-th two-way handshake communication event, which measures the reliability of the two-way handshake communication event (e.g., based on response delay and data integrity). is an event recognition function. It takes the value 1 when the i-th two-way handshake communication event conforms to the complete two-way response structure (a complete two-way response structure means that during a communication process, the probe request packet sent by the hop node successfully reaches the neighbor node, and the neighbor node returns a response packet with the correct structure, complete protocol layer, and can be accurately parsed based on the new address of the hop node, forming a closed request-response round-trip link). Otherwise, it takes the value 0 (for example, half-handshakes and incomplete responses are ignored). In order to avoid judging the synchronization state only by the number of events, a structural entropy adjustment factor of the two-way handshake communication event cluster distribution is further introduced to measure the distribution concentration of two-way handshake communication events among neighboring nodes. A two-way confirmation density factor is generated based on the structural entropy adjustment factor and the total intensity of two-way handshake communication events. The generation expression of the two-way confirmation density factor is: ,in: is the double confirmation density factor, is the structural entropy adjustment factor, and the calculation expression is: , where: M is the total number of upstream and downstream neighbor nodes identified by the hop node; k is the index variable used to calculate the total number of two-way handshake communication events of all neighbor nodes; It represents the number of valid round-trip confirmation events that occur to the j-th neighbor node within the monitoring window.

[0014] Preferably, in the monitoring window, the specific steps of generating the address confirmation distribution factor after performing an in-depth analysis on the distribution concentration of the new address of the destination address matching jump node are as follows: Within the set monitoring window, all neighbor response packets monitored by the jump node are parsed, the destination address field of each response packet is extracted, and the number of response packets with the destination address equal to the jump node's new address is counted. At the same time, the total number of all neighbor response packets is counted. The destination address matching frequency ratio vector is constructed by the number of response packets with the destination address equal to the jump node's new address and the total number of all neighbor response packets. The constructed expression is: ,in: Indicates the number of response packets whose destination address is exactly the same as the new address of the hop node in the monitoring window; Indicates the number of all neighbor response packets captured within the listening window; is the destination address matching frequency ratio, the value range is ,The closer it is to 1, the more concentrated the destination address matches are, and the closer it is to 0, the more dispersed the address distribution is; In order to enhance the sensitivity of high matching rates and avoid ambiguous judgments when the destination address matching frequency ratio is in the middle range (for example, 0.4-0.7), the hyperbolic tangent exponential enhancement function is introduced to construct the address confirmation distribution factor. The constructed expression is as follows: ,in: is the address confirmation distribution factor, The response enhancement coefficient is used to adjust the slope of the nonlinear mapping (the recommended value is 2~5); It is a concentration sensitivity factor used to control the degree of index increase when the frequency is high (the recommended value is 1.5~2.5); The hyperbolic tangent square function makes the address confirmation distribution factor It rises nonlinearly within the range, is sensitive to high matching but suppresses fluctuations in medium and low matching.

[0015] In the above technical solution, the technical effects and advantages provided by the present invention are: The present invention introduces a two-way perception mechanism of "active detection + passive monitoring" and combines it with a support vector machine model based on a characteristic state vector to intelligently evaluate the synchronization completion of neighboring nodes, thereby accurately determining whether the path has completed address update synchronization. Furthermore, the synchronization state determination result is dynamically coupled with the current data transmission flux of the jump node, realizing an adaptive control strategy for data transmission behavior, effectively avoiding problems such as data mistransmission, link interruption, and communication black holes caused by the failure of path synchronization, and ensuring that the jump node can smoothly transition in a state-driven manner during the address change, thereby ensuring the accessibility of key links and the operational continuity of the entire network system. This mechanism breaks through the traditional method of passively waiting for synchronization completion, and realizes high robustness and low-latency transition control in the path reconstruction process, with significant engineering adaptability and system stability improvement value. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction to the drawings required for use in the embodiments will be given below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0017] Figure 1 Schematic diagram of system modules of the adaptive network optimization system based on multiple communication links of the present invention. DETAILED DESCRIPTION

[0018] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of this disclosure will be thorough and complete and will fully convey the concepts of the example embodiments to those skilled in the art.

[0019] The present invention provides Figure 1 The adaptive network optimization system based on multiple communication links shown in the figure includes an address change detection module, a neighbor activation wake-up module, a two-way interaction monitoring module, a synchronization state vector construction module, a synchronization completion intelligent prediction module, and a data flux adaptive control module; The address change detection module, in a multi-link access communication network, first continuously monitors the network status of the hop node to determine whether its network address has changed. When a hop node address change event is detected, it determines that the hop node has entered the address reallocation state; In a multi-link access communication network, continuous monitoring of the network address status of hop nodes can generally be divided into the following three steps: The first step is to initialize the address recording mechanism. When the jump node is started or online, the system records its current network address information, including IP address, MAC address, subnet mask and gateway information; The second step is to set monitoring trigger conditions to regularly check the interface configuration status, DHCP lease renewal, IPv6 prefix information, NAT mapping table changes of the hop node, or subscribe to operating system-level network events (such as netlink events or interface watchers); The third step is to compare the current address status with the last recorded status. If the address field is found to have changed (such as IP segment drift, interface rebinding, etc.), it is determined to be an address change event.

[0020] Through the above steps, continuous perception and accurate identification of dynamic changes in the network address of the hopping node can be achieved. The address change event detection process of the hopping node is the trigger point of the entire mechanism, ensuring that subsequent detection and control processes are only initiated when the hopping node address undergoes substantial changes, avoiding interference with the stable path and realizing "event-driven" state perception response.

[0021] The neighbor activation wake-up module, once detecting that a hop node has entered the address reallocation state, immediately sends a lightweight detection packet to its known upstream and downstream neighbor nodes through the hop node to actively wake up the neighbor nodes to recognize its new address; These lightweight probe packets can be ARP requests (for IPv4), NDP neighbor advertisements (for IPv6), Router Advertisement snooping triggers (used in IPv6 to assist with address prefix updates), or UDP probe frames (for direct link-layer communication or device-to-device interaction scenarios that do not rely on a full IP protocol stack). The purpose of sending these lightweight probe packets is to prompt neighboring nodes to proactively update their address resolution tables, neighbor caches, or routing association information, thereby enabling rapid identification of the new address status of the hop node. By proactively sending lightweight probe packets, the hop node can quickly prompt upstream and downstream neighbors to recognize and accept its new address, avoiding the need to wait for neighboring devices to passively update, thereby accelerating synchronization and shortening the critical window of path unreachability.

[0022] The two-way interactive monitoring module starts the monitoring mechanism after the detection data packet is sent, captures and records the response data packets from upstream and downstream neighbor nodes, and establishes a two-way state link from "active detection" to "passive perception"; The core function of this step is to establish a dynamic path status feedback channel between the hop node and its adjacent upstream and downstream nodes. By passively monitoring and recording the response information after the detection data packet, it can achieve real-time perception of whether the neighboring node has successfully identified and accepted the new address of the hop node. Compared with the active notification method of one-way broadcast, the addition of a passive monitoring mechanism can significantly improve the accuracy and timeliness of status perception: on the one hand, it can capture whether the updated address information is used in the data packet returned by the neighboring node, thereby verifying whether its address mapping is synchronized; on the other hand, it can also observe the fine-grained communication characteristics of the neighboring node, such as response delay, confirmation behavior, and retransmission mode, to provide high-quality raw data for subsequent status modeling and synchronization completion assessment. Through this "active triggering + passive observation" combination, a more stable, low-overhead, and highly reliable path synchronization status determination basis can be achieved.

[0023] The synchronization state vector construction module extracts key features related to the synchronization state of the jump node address after receiving response data from neighbor nodes. Within the monitoring window, it conducts in-depth analysis of the extracted key features and, based on the analysis results, constructs the synchronization state vector of the jump node and its corresponding upstream and downstream neighbor nodes, preliminarily quantifying the synchronization status of the current jump node with its upstream and downstream neighbor nodes. After obtaining the response data from the neighboring nodes, key features related to the address synchronization status of the hopping node are extracted. The extracted key features include the density of the number of round-trip communication cycles (such as SYN-ACK, ICMPEcho-Reply, and UDP response) between the hopping node and the upstream and downstream nodes, and the distribution concentration of the destination address matching the new address of the hopping node. Within the monitoring window, after an in-depth analysis of the extracted key features, a two-way confirmation density factor and an address confirmation distribution factor are generated. Based on the two-way confirmation density factor and the address confirmation distribution factor, the synchronization state vector of the hopping node and its corresponding upstream and downstream neighbor nodes is constructed to preliminarily quantify the synchronization status of the current hopping node with its upstream and downstream neighbor nodes.

[0024] After the probe packet is sent, if a hop node can frequently complete two-way handshakes with its upstream and downstream neighbors (e.g., TCP's SYN-ACK three-way handshake, ICMP's Echo-Reply round trip, UDP's request-response, etc.), this indicates that the neighboring nodes have successfully recognized and accepted the hop node's new address and are able to establish a valid transmission link at multiple protocol layers. A higher density of two-way communication not only indicates that the neighboring node has synchronized with the new address at the link layer (e.g., MAC-IP binding), network layer (e.g., routing path confirmation), and transport layer (e.g., session establishment), but also indicates that packet filtering, caching mechanisms, and session state machines along the path are in place, ensuring stable transmission and reception capabilities. Conversely, sparse or even absent two-way communication may indicate that the neighboring node has not yet completed the address update and is still identifying the old address, or is experiencing an unsynchronized state due to resolution delays or route drift. Therefore, the density of two-way handshakes is a strongly correlated positive indicator of address synchronization completion and can be used as a basis for determining synchronization status with high confidence.

[0025] In the monitoring window, the specific steps for generating a two-way confirmation density factor after in-depth analysis of the density of the number of round-trip communication times that complete a two-way handshake between the jump node and the upstream and downstream nodes are as follows: Within the set monitoring window, the hop node monitors and records the two-way handshake communication events between it and all upstream and downstream neighbor nodes, including but not limited to SYN-ACK pairs, ICMPEcho-Reply pairs, UDP request-response pairs, etc. Each complete two-way handshake communication event is identified as a valid round-trip confirmation event, and a handshake event intensity function is constructed to measure the path interaction intensity. The constructed expression is as follows: ,in: Represents the total intensity of two-way handshake communication events within the monitoring window; N is the number of valid round-trip confirmation events observed within the monitoring window; is the confidence weight coefficient of the i-th two-way handshake communication event, which measures the reliability of the two-way handshake communication event (e.g., based on response delay and data integrity). is an event recognition function. It takes the value 1 when the i-th two-way handshake communication event conforms to the complete two-way response structure (a complete two-way response structure means that during a communication process, the probe request packet sent by the hop node successfully reaches the neighbor node, and the neighbor node returns a response packet with the correct structure, complete protocol layer, and can be accurately parsed based on the new address of the hop node, forming a closed request-response round-trip link). Otherwise, it takes the value 0 (for example, half-handshakes and incomplete responses are ignored). This step establishes a structural indicator of "total handshake event effectiveness" through event-level monitoring and screening. It is not affected by noise events and avoids the use of simple event counting or time averaging, thereby enhancing the robustness of the indicator to abnormal synchronization scenarios.

[0026] In order to avoid judging the synchronization state only by the number of events, a structural entropy adjustment factor of the two-way handshake communication event cluster distribution is further introduced to measure the distribution concentration of two-way handshake communication events among neighboring nodes. A two-way confirmation density factor is generated based on the structural entropy adjustment factor and the total intensity of two-way handshake communication events. The generation expression of the two-way confirmation density factor is: ,in: is the double confirmation density factor, is the structural entropy adjustment factor, and the calculation expression is: , where: M is the total number of upstream and downstream neighbor nodes identified by the hop node; k is the index variable used to calculate the total number of two-way handshake communication events of all neighbor nodes; represents the number of valid round-trip confirmation events that occur within the monitoring window for the jth neighbor node; This step couples the "total handshake event intensity" with its distribution consistency among neighboring nodes by introducing structural entropy regulation, and constructs a two-way confirmation density factor that takes into account quantity, concentration and quality.

[0027] The bidirectional confirmation density factor, generated by deeply analyzing the density of bidirectional handshake round trips between a hop node and its upstream and downstream nodes within the monitoring window, indicates that a higher value indicates a higher synchronization level between the hop node and its upstream and downstream neighbors. Conversely, a lower value indicates a lower synchronization level between the hop node and its upstream and downstream neighbors. This is because the establishment of a bidirectional handshake event depends on the neighbor node's successful resolution of the hop node's new address and the establishment of a bidirectional communication link. Only after the neighbor node completes key synchronization operations such as MAC-IP mapping updates, path routing table adjustments, and session connection establishment can a complete and responsive request-response link be established. Therefore, a higher bidirectional confirmation density factor indicates that the hop node receives frequent, low-latency, and error-free responses from its neighbors under the new address, reflecting higher path stability and neighbor acceptance of the new address. Conversely, if handshake events are rare or scattered, i.e., a lower bidirectional confirmation density factor, synchronization is insufficient and the link has not yet been reliably confirmed.

[0028] When a hop node undergoes network address reallocation, only after its upstream and downstream neighboring nodes successfully receive the hop node's new address and complete address mapping updates (e.g., ARP / NDP cache or routing table revisions) will the neighboring nodes use the new address as the destination address in subsequent response packets. If the destination addresses of response packets monitored within a time window are highly concentrated at the hop node's new address, this indicates that most neighboring nodes have correctly identified, resolved, and bound the new address to the new address, and the node states along the path are becoming consistent, indicating a high level of overall address synchronization. Conversely, if the destination address distribution is scattered, still containing old addresses or other non-matching fields, this indicates that some neighboring nodes have not yet completed synchronization, and the system is still in an unstable phase of address migration. Therefore, "destination address distribution concentration" can be used as an effective indicator of synchronization status, reflecting the address recognition consistency and synchronization coverage between the hop node and its neighboring nodes.

[0029] In the monitoring window, the specific steps for generating the address confirmation distribution factor after in-depth analysis of the distribution concentration of the new addresses of the destination address matching hop nodes are as follows: Within the set monitoring window, all neighbor response packets monitored by the jump node are parsed, the destination address field of each response packet is extracted, and the number of response packets with the destination address equal to the jump node's new address is counted. At the same time, the total number of all neighbor response packets is counted. The destination address matching frequency ratio vector is constructed by the number of response packets with the destination address equal to the jump node's new address and the total number of all neighbor response packets. The constructed expression is: ,in: Indicates the number of response packets whose destination address is exactly the same as the new address of the hop node in the monitoring window; Indicates the number of all neighbor response packets captured within the listening window; is the destination address matching frequency ratio, the value range is ,The closer it is to 1, the more concentrated the destination address matches are, and the closer it is to 0, the more dispersed the address distribution is; This step constructs a matching frequency ratio to extract the distribution characteristics of whether neighbor nodes actively identify and apply the new address of the jump node, providing basic input factors for subsequent address confirmation distribution factor mapping, and has direct quantitative significance for path identification consistency.

[0030] In order to enhance the sensitivity of high matching rates and avoid ambiguous judgments when the destination address matching frequency ratio is in the middle range (for example, 0.4-0.7), the hyperbolic tangent exponential enhancement function is introduced to construct the address confirmation distribution factor. The constructed expression is as follows: ,in: is the address confirmation distribution factor, The response enhancement coefficient is used to adjust the slope of the nonlinear mapping (the recommended value is 2~5); It is a concentration sensitivity factor used to control the degree of index increase when the frequency is high (the recommended value is 1.5~2.5); The hyperbolic tangent square function makes the address confirmation distribution factor It rises nonlinearly within the range, is sensitive to high matching but suppresses fluctuations in medium and low matching; By nonlinearly amplifying highly concentrated inputs, we achieve enhanced confirmation of new addresses collectively recognized by neighboring nodes. This approach exhibits a pronounced above-threshold surge characteristic: when most responding addresses are synchronized, the address confirmation distribution factor rapidly approaches 1, while otherwise it remains low, effectively distinguishing between critical synchronization and steady-state completion.

[0031] The bidirectional confirmation density factor shows that, within the monitoring window, a larger value of the address confirmation distribution factor, generated by deeply analyzing the distribution concentration of destination addresses matching the new address of the hop node, indicates a higher synchronization status between the hop node and its corresponding upstream and downstream neighbors. Conversely, a lower value indicates a lower synchronization status between the hop node and its corresponding upstream and downstream neighbors. This is because the address confirmation distribution factor is calculated based on the distribution concentration of whether the destination address in the neighbor node response data matches the new address of the hop node within the monitoring window. A larger value of the bidirectional confirmation density factor indicates that most neighbor nodes have used the new address of the hop node as the communication target in their responses, indicating that their internal address mapping tables (such as ARP / NDP caches, routing tables, or application-layer binding information) have been updated, thus achieving unified recognition and dissemination of the hop node's address status among neighbor nodes. Conversely, a lower bidirectional confirmation density factor indicates that there are a large number of destination address offsets or confusion in the response packets, indicating that neighbor nodes have not yet completed synchronization and path identification is still unstable. Therefore, the address confirmation distribution factor can be used as an effective quantitative indicator of the synchronization status between the hop node and its upstream and downstream neighbors. The higher the value, the more consistent the representation and the more complete the synchronization. The lower the value, the more dispersed the representation and the worse the synchronization completion.

[0032] The synchronization completion intelligent prediction module inputs the constructed synchronization state vector into a pre-trained support vector machine (SVM) model, and uses the SVM model to intelligently predict the synchronization completion degree of the jump node and its upstream and downstream neighboring nodes; The synchronization state vector constructed by the two-way confirmation density factor and the address confirmation distribution factor is input into the pre-trained support vector machine model. The support vector machine model outputs the synchronization completion evaluation index, and based on the synchronization completion evaluation index, an intelligent prediction is made on the synchronization completion of the jump node and its upstream and downstream neighbor nodes.

[0033] A pre-trained support vector machine (SVM) classification model is one that has undergone offline learning and parameter optimization based on a large amount of historical sample data before being formally applied to the synchronization completion prediction task. In this solution, the model's goal is to determine whether the current path state has achieved valid address synchronization based on feature vectors (such as the bidirectional confirmation density factor and the address confirmation distribution factor) extracted from the interactions between the hop node and its upstream and downstream neighbors. Specifically, during the training phase, the SVM model first incorporates a labeled dataset containing a large number of communication scenario samples under different network conditions. Each sample consists of multiple synchronization features and is accompanied by a true synchronization state label (such as "synchronized," "unsynchronized," or "partially synchronized"), either manually or programmatically. The model then finds an optimal hyperplane in the feature space that divides samples of different categories by the largest margin, thereby obtaining a discriminant boundary that distinguishes synchronization states. During the training process, the model also automatically determines key parameters such as the kernel function type (e.g., linear kernel, RBF kernel), penalty factor C, and margin softness to optimize the model's generalization and fault tolerance. After training, the model is solidified into a classifier that can be directly used for online inference. Its input is the synchronization state vector extracted between the current hop node and its neighboring nodes, and its output is a synchronization completion assessment indicator (such as a confidence score between 0 and 1 or a multi-class label). During runtime, the system does not need to retrain the model; simply inputting the feature vectors extracted in real time into the pre-trained model enables intelligent judgment of the current link state and accurate prediction of the synchronization state. Therefore, this support vector machine model not only has the ability to highly accurately identify nonlinear feature relationships but also can quickly respond to state fluctuations caused by address hopping, playing a key role in building a stable and intelligent network status awareness mechanism.

[0034] The support vector machine model is not specifically limited here, and can realize the bidirectional confirmation density factor and address confirmation distribution factor Conduct comprehensive analysis to generate synchronization completion evaluation indicators The support vector machine model can be used. In order to implement the technical solution of the present invention, the present invention provides a specific implementation method; Synchronous completion evaluation index The generated expression is: , where 、 Double confirmation density factor and address confirmation distribution factor The preset scaling factor of 、 All are greater than 0. Preset proportional coefficient and Refers to the fixed parameters used to adjust the weights of the two-way confirmation density factor and the address confirmation distribution factor in the process of comprehensively generating the synchronization completion evaluation index. It reflects the importance of different feature dimensions in the final evaluation and belongs to the weight factor in the linear weighted fusion model. In practical applications, these two proportional coefficients can be set in advance through expert experience, training data fitting, system sensitivity analysis, etc. to ensure that the synchronization completion evaluation index output by the model has good discrimination and stability under different network conditions. Since the two indicators have different mechanisms of action in expressing the synchronization status, it is necessary to set a reasonable and This helps to achieve a differentiated expression of feature contributions without introducing complex nonlinear structures into the model, thereby improving the accuracy and practicality of synchronous state assessment. In short, the preset proportional coefficient is a quantitative control tool for the contribution of the components of the synchronous assessment index and an important means to achieve balanced weighting of index fusion.

[0035] The synchronization completion evaluation index shows that, within the monitoring window, the greater the performance value of the two-way confirmation density factor generated after in-depth analysis of the density of the number of round-trip communications that complete two-way handshakes between the hopping node and its upstream and downstream nodes, and the greater the performance value of the address confirmation distribution factor generated after in-depth analysis of the distribution concentration of destination addresses matching the new address of the hopping node, the greater the performance value of the synchronization completion evaluation index generated when the support vector machine model is used to intelligently predict the synchronization completion degree of the hopping node and its upstream and downstream neighboring nodes. In other words, the greater the performance value of the synchronization completion evaluation index generated when the support vector machine model is used to intelligently predict the synchronization completion degree of the hopping node and its upstream and downstream neighboring nodes, the higher the synchronization status of the hopping node and its corresponding upstream and downstream neighboring nodes. Conversely, the lower the synchronization status of the hopping node and its corresponding upstream and downstream neighboring nodes.

[0036] The data flux adaptive control module obtains the current data transmission flux of the hop node in the monitoring window and intelligently controls the actual data transmission flux of the hop node based on the synchronization completion prediction results output by the support vector machine model and the current data flux status. This ensures that the hop node does not send data incorrectly due to synchronization lag after address reallocation. During a monitoring window, a hop node's current data transmission flux (i.e., the total amount of data transmitted per unit time) can be obtained through a variety of existing technologies, including interface-level traffic statistics, protocol stack counters, link-layer packet capture analysis, and software-defined networking (SDN) control plane callbacks. Specifically, kernel-level byte and packet counters on network interfaces (such as Ethernet or wireless) can be used to extract the total number of bytes and packets sent in real time. Timer hook functions embedded within the protocol stack can also be used to aggregate and analyze outgoing IP, UDP, and TCP data frames in real time. For higher precision, traffic monitors or high-performance packet processing engines like DPDK can be deployed to aggregate and analyze actual outgoing packets by time window. For systems using SDN architectures, the controller can query the port flow table entries of forwarding devices (such as OpenFlow's FlowStats) to obtain the current hop node's total transmission volume. All of these methods can be adapted to the monitoring window period, providing a quantitative basis for the hop node's actual packet transmission behavior, thereby supporting dynamic response to flux control policies.

[0037] Based on the synchronization completion prediction results output by the support vector machine model and the current data flux status, the actual data transmission flux of the jump node is intelligently controlled. The specific steps are as follows: Based on the synchronization completion evaluation index output by the support vector machine model and the original data transmission flux of the current jump node, an inhibition factor is calculated to control the actual transmission behavior. The calculation expression is as follows: ,in: This is an evaluation indicator of the synchronization completion between the hop node and its upstream and downstream neighbor nodes. The closer it is to 1, the more complete the synchronization status is. The raw data transmission flux of the hop node in the current monitoring window; is the maximum reference throughput value allowed (such as the rated bandwidth of the link); As an inhibitory factor; is the flux adjustment strength coefficient, which is used to amplify the suppression sensitivity caused by insufficient synchronization. ; is a nonlinear weighted index that controls the nonlinear enhancement of the degree of suppression of the low synchronization state. ; To ensure the inhibitory factor , the larger the value, the stronger the inhibition; This step combines the degree of synchronization deficiency and the proportion of emission pressure into a unified, modulatable signal. A larger suppression factor indicates that the current state is less suitable for maintaining the current flux, and restrictive measures should be implemented. By introducing nonlinear compression and trend amplification mechanisms, excessive intervention in high-synchronization nodes is avoided, while improving control sensitivity in low-synchronization states.

[0038] Obtaining inhibitory factors After that, the actual data transmission flux of the jump node is dynamically calculated. The calculation expression is: ,in: is the actual data transmission flux of the hop node; is the minimum guaranteed flux constant ( ), used to maintain the bottom line data output of heartbeat packets or link keepalive, to prevent zero throughput from causing connection failure; This step maps the suppression factor to the control weight of the original flux, realizing continuous dynamic control of the actual packet sending behavior of the jump node. When the rate automatically approaches 0, the system automatically releases the restriction and restores the normal communication rate. Conversely, when the path synchronization is incomplete, the system automatically compresses the flux output to reduce the risk of link errors or broadcast flooding.

[0039] This step aims to establish a safety valve centered around a closed loop of "prediction-feedback-control." It uses the synchronization completion predictions from the support vector machine model to determine whether the path is fully synchronized. Combined with the node's current data transmission throughput, it assesses link load and potential risks in real time, dynamically adjusting the actual packet transmission intensity of the hop node. When synchronization is insufficient and packet transmission pressure is high, the throughput is proactively tightened to suppress mis-transmissions and broadcast flooding. When synchronization is complete and the link is idle, the restrictions are automatically relaxed to restore normal throughput. This intelligent, adaptive control effectively prevents data from entering "black holes," routing oscillations, or control command loss during the critical transition period after address reallocation, ensuring the continuity of the communication link and the stability of the overall system operation.

[0040] Through this solution, the system achieves real-time awareness of path synchronization status and intelligently coordinates packet transmission behavior after a hop node undergoes network address reassignment, significantly improving communication link stability and control command delivery reliability in multi-link access environments. Specifically, this solution utilizes a bidirectional awareness mechanism combining active detection and passive monitoring, combined with a support vector machine model based on a characteristic state vector to intelligently assess the synchronization completion of neighboring nodes, accurately determining whether a path has completed address update synchronization. Furthermore, the system dynamically couples the synchronization status determination result with the hop node's current data transmission flux, implementing an adaptive control strategy for data transmission behavior. This effectively avoids data mistransmission, link interruptions, and communication black holes caused by incomplete path synchronization. It ensures a smooth, state-driven transition between hop nodes during address changes, safeguarding the reachability of critical links and the operational continuity of the entire network system. This mechanism transcends the traditional passive wait for synchronization completion approach, achieving highly robust and low-latency transition control during path reconstruction, significantly enhancing engineering adaptability and system stability.

[0041] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0042] The above description is merely illustrative of certain exemplary embodiments of the present invention. It goes without saying that those skilled in the art will be able to modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of protection of the claims.

[0043] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. An adaptive network optimization system based on multiple communication links, characterized in that: It includes an address change detection module, a neighbor activation wake-up module, a two-way interactive monitoring module, a synchronization state vector construction module, a synchronization completion intelligent prediction module, and a data flux adaptive control module; The address change detection module continuously monitors the network status of the jump node to determine whether its network address has changed. When a change event is detected for the address of the jump node, it is determined that the jump node has entered the address reallocation state; The neighbor activation wake-up module, once detecting that a hop node has entered the address reallocation state, immediately sends a lightweight detection packet to its known upstream and downstream neighbor nodes through the hop node, actively waking up the neighbor nodes to recognize its new address; The two-way interactive monitoring module starts the monitoring mechanism after the detection data packet is sent, capturing and recording the response data packets from upstream and downstream neighbor nodes, thus establishing a two-way state link from "active detection" to "passive perception"; The synchronization state vector construction module extracts key features related to the synchronization state of the jump node address after receiving response data from neighbor nodes. Within the monitoring window, it conducts in-depth analysis of the extracted key features and, based on the analysis results, constructs the synchronization state vector of the jump node and its corresponding upstream and downstream neighbor nodes, preliminarily quantifying the synchronization status of the current jump node with its upstream and downstream neighbor nodes. The synchronization completion intelligent prediction module inputs the constructed synchronization state vector into a pre-trained support vector machine model, and uses the support vector machine model to intelligently predict the synchronization completion degree of the jump node and its upstream and downstream neighboring nodes; The data flux adaptive control module obtains the current data transmission flux of the jump point node in the monitoring window, and intelligently controls the actual data transmission flux of the jump point node based on the synchronization completion prediction results output by the support vector machine model and the current data flux status.

2. The adaptive network optimization system based on multiple communication links according to claim 1, characterized in that: In a multi-link access communication network, continuous monitoring of the network address status of hop nodes is achieved through the following steps: The first step is to initialize the address recording mechanism to record the current network address information of the jump node when it is started or online, including IP address, MAC address, subnet mask and gateway information; The second step is to set the monitoring trigger conditions to regularly check the interface configuration status, DHCP lease renewal, IPv6 prefix information and NAT mapping table changes of the hop node; The third step is to compare the current address status with the last recorded status. If the address field is found to have changed, it is determined to be an address change event.

3. The adaptive network optimization system based on multiple communication links according to claim 1, characterized in that: A lightweight probe packet is any one of an ARP request, an NDP neighbor advertisement, a Router Advertisement snooping trigger signal, or a UDP probe frame.

4. The adaptive network optimization system based on multiple communication links according to claim 1, characterized in that: After obtaining the response data from the neighboring nodes, key features related to the address synchronization status of the hopping node are extracted. The extracted key features include the density of the number of round-trip communications to complete a two-way handshake between the hopping node and the upstream and downstream nodes, and the distribution concentration of the destination address matching the new address of the hopping node. Under the monitoring window, after an in-depth analysis of the extracted key features, a two-way confirmation density factor and an address confirmation distribution factor are generated respectively. Based on the two-way confirmation density factor and the address confirmation distribution factor, the synchronization state vector of the hopping node and its corresponding upstream and downstream neighbor nodes is constructed to preliminarily quantify the synchronization status of the current hopping node and its upstream and downstream neighbor nodes.

5. The adaptive network optimization system based on multiple communication links according to claim 4, characterized in that: The synchronization state vector constructed by the two-way confirmation density factor and the address confirmation distribution factor is input into the pre-trained support vector machine model. The support vector machine model outputs the synchronization completion evaluation index, and based on the synchronization completion evaluation index, an intelligent prediction is made on the synchronization completion of the jump node and its upstream and downstream neighbor nodes.

6. The adaptive network optimization system based on multiple communication links according to claim 5, characterized in that: Based on the synchronization completion prediction results output by the support vector machine model and the current data flux status, the actual data transmission flux of the jump node is intelligently controlled. The specific steps are as follows: Based on the synchronization completion evaluation index output by the support vector machine model and the original data transmission flux of the current jump node, an inhibition factor is calculated to control the actual transmission behavior. The calculation expression is as follows: ,in: It is an evaluation indicator of the synchronization completion degree between the hop node and its upstream and downstream neighbor nodes; The raw data transmission flux of the hop node in the current monitoring window; is the maximum reference flux value allowed; As an inhibitory factor; is the flux adjustment strength coefficient, which is used to amplify the suppression sensitivity caused by insufficient synchronization. ; is a nonlinear weighted index that controls the nonlinear enhancement of the degree of suppression of the low synchronization state. ; Obtaining inhibitory factors After that, the actual data transmission flux of the jump node is dynamically calculated. The calculation expression is: ,in: is the actual data transmission flux of the hop node; is the minimum guaranteed flux constant, , used to maintain the bottom line data output of heartbeat packets or link keepalive.

7. The adaptive network optimization system based on multiple communication links according to claim 4, characterized in that: In the monitoring window, the specific steps for generating a two-way confirmation density factor after in-depth analysis of the density of two-way handshakes completed between the hop node and the upstream and downstream nodes are as follows: Within the set monitoring window, the hop node monitors and records the two-way handshake communication events between it and all upstream and downstream neighbor nodes. Each complete two-way handshake communication event is identified as a valid round-trip confirmation event. A handshake event intensity function is constructed to measure the path interaction intensity. The constructed expression is as follows: ,in: Represents the total intensity of two-way handshake communication events within the monitoring window; N is the number of valid round-trip confirmation events observed within the monitoring window; is the confidence weight coefficient of the i-th two-way handshake communication event; is an event recognition function, which takes the value 1 when the i-th two-way handshake communication event conforms to the complete two-way response structure, otherwise it takes the value 0; The structural entropy adjustment factor of the two-way handshake communication event cluster distribution is introduced to measure the distribution concentration of two-way handshake communication events among neighboring nodes. The two-way confirmation density factor is generated based on the structural entropy adjustment factor and the total intensity of two-way handshake communication events. The generation expression of the two-way confirmation density factor is: ,in: is the double confirmation density factor, is the structural entropy adjustment factor, and the calculation expression is: , where: M is the total number of upstream and downstream neighbor nodes identified by the hop node; k is the index variable used to calculate the total number of two-way handshake communication events of all neighbor nodes; It represents the number of valid round-trip confirmation events that occur to the j-th neighbor node within the monitoring window.

8. The adaptive network optimization system based on multiple communication links according to claim 4, characterized in that: In the monitoring window, the specific steps for generating the address confirmation distribution factor after in-depth analysis of the distribution concentration of the new addresses of the destination address matching hop nodes are as follows: Within the set monitoring window, all neighbor response packets monitored by the jump node are parsed, the destination address field of each response packet is extracted, and the number of response packets with the destination address equal to the jump node's new address is counted. At the same time, the total number of all neighbor response packets is counted. The destination address matching frequency ratio vector is constructed by the number of response packets with the destination address equal to the jump node's new address and the total number of all neighbor response packets. The constructed expression is: ,in: Indicates the number of response packets whose destination address is exactly the same as the new address of the hop node in the monitoring window; Indicates the number of all neighbor response packets captured within the listening window; is the destination address matching frequency ratio, the value range is ; The hyperbolic tangent exponential enhancement function is introduced to construct the address confirmation distribution factor. The constructed expression is as follows: ,in: is the address confirmation distribution factor, is the response enhancement coefficient, which is used to adjust the slope of the nonlinear mapping; It is a concentrated sensitive factor used to control the degree of frequency index increase; The hyperbolic tangent square function makes the address confirmation distribution factor The range increases nonlinearly.

Images