Self-adaptive Mesh network architecture construction method for hybrid networking of industrial Internet of Things

Abstract

The invention relates to the technical field of computer network communication, and discloses an adaptive Mesh network architecture construction method for hybrid networking of industrial Internet of Things, which comprises the following steps: when a node in a network receives a data packet to be forwarded, obtaining a data quantum state label packaged in the data packet, the data quantum state label is used for representing potential demand intensity of the data packet for different network service quality dimensions; the node calculates a comprehensive cost for forwarding the data packet for each next-hop neighbor node; the calculation of the comprehensive cost is combined with the matching degree between the data quantum state label and the path capability related to the next-hop neighbor node. According to the invention, accurate and differentiated matching from network resources to service requirements is realized, a solid foundation is provided for guaranteeing the service quality of key services in a complex industrial scene, accurate and selective avoidance of related data streams is realized, and the load balancing capability and the overall throughput of the network are effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of computer network communication technology, and in particular to a method for constructing an adaptive Mesh network architecture for hybrid networking of the Industrial Internet of Things. Background Art

[0002] With the advancement of Industry 4.0 and smart manufacturing, the Industrial Internet of Things (IIoT) has become a critical infrastructure connecting physical production equipment with digital information systems. In complex industrial environments such as large factories, mining areas, and automated warehouses, due to cabling difficulties and high device mobility, the use of self-organizing and self-healing wireless mesh networks (WMNs) as a core transport technology has become a widely adopted deployment solution. With their flexible topology and high network redundancy, WMNs can cost-effectively achieve high-density device coverage over large areas.

[0003] However, data communications in modern industrial applications present unprecedented complexity and diversity. Networks simultaneously transmit a wide range of heterogeneous data streams, including closed-loop control commands for equipment requiring millisecond-level responses, bandwidth-intensive high-definition video surveillance streams, critical safety alarms requiring extremely high transmission success rates, and periodically reported non-real-time device status data. These diverse service data streams impose distinct quality of service (QoS) requirements on the network, including varying priorities for transmission latency, available bandwidth, and data reliability.

[0004] Existing mainstream wireless mesh network routing technologies typically rely on relatively simple physical and link layer metrics when making path selection decisions. For example, some protocols prioritize minimizing the number of transmission hops, while others use link quality metrics such as received signal strength indicator (RSSI) or estimated transmission times to construct routes. While these approaches are universal in design, their fundamental flaw lies in their ignorance of the inherent service requirements of the packets they carry. When making forwarding decisions, the network cannot distinguish between an urgent control command and a routine log file; instead, they are treated identically, forwarding them based solely on a single, universal "best" path criterion.

[0005] This "one-size-fits-all" routing mechanism has exposed its limitations in an increasingly complex industrial environment. It often leads to a mismatch of network resources. For example, a critical data packet that is extremely sensitive to latency may be routed to a reliable but high-latency path due to a slight advantage in a comprehensive link quality indicator, thus missing the optimal response time. Conversely, a file transfer task that requires a large bandwidth may be indiscriminately directed to an area consisting of low-power, narrow-bandwidth nodes. Not only is its own transmission efficiency low, it is also likely to cause congestion and interference to other low-latency services in the area that need to be guaranteed. In addition, when facing network congestion, the adjustment mechanism of existing technologies is often relatively passive and extensive. It lacks the ability to accurately analyze the causes of congestion and differentiate the flow of data, making it difficult to achieve efficient load balancing.

[0006] Therefore, existing technologies urgently need a new method for building industrial Internet of Things networks. This method should be able to transcend the limitations of traditional physical measurements, achieve deep perception of data business intentions, and on this basis establish a set of adaptive network operation mechanisms capable of dynamic learning, intelligent decision-making and precise adjustment, so as to truly meet the diverse and high-performance requirements of future smart industrial applications for network communications. Summary of the Invention

[0007] In response to the shortcomings of the existing technology, the present invention provides a method for constructing an adaptive Mesh network architecture for hybrid networking of the Industrial Internet of Things, which solves the problem that the existing Industrial Internet of Things Mesh network routing decisions lack awareness of the inherent needs of heterogeneous data services, resulting in blind routing selection and low matching between network resources and business needs.

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions: A method for constructing an adaptive Mesh network architecture for hybrid networking of the Industrial Internet of Things, comprising the following steps: S1. When a node in the network receives a data packet to be forwarded, it obtains a data quantum state label encapsulated in the data packet, where the data quantum state label is used to characterize the potential demand intensity of the data packet for different network service quality dimensions; S2. The node calculates a comprehensive cost for forwarding the data packet for each of its next-hop neighbor nodes; the calculation of the comprehensive cost combines the matching degree between the data quantum state label and the path capability associated with the next-hop neighbor node, the path affinity value based on historical transmission experience, and the dissipative field signaling based on the congestion state of the next-hop neighbor node; S3. The node selects a next-hop neighbor node with the lowest cost based on the calculated comprehensive cost to forward the data packet.

[0009] Preferably, the data quantum state label is a multidimensional vector, the dimensions of which include: Delay potential, used to characterize the sensitivity of the data packet to transmission delay; Bandwidth potential, used to characterize the demand for transmission bandwidth of the data packet; Reliability potential, used to characterize the severity of the data packet's requirements for transmission success rate; Aggregation potential is used to characterize the path stability requirement of the data packet as part of a data flow.

[0010] Preferably, the method further comprises the step of updating the path affinity value: When the data packet successfully arrives at the destination node, a confirmation frame is returned along the original route; After receiving the confirmation frame, the node on the path increases the path affinity value stored internally and corresponding to the intention type of the data packet.

[0011] Preferably, the step of updating the path affinity value further comprises: Before increasing the path affinity value, determining whether the number of data packets flowing to the same next-hop neighbor node and having similar data quantum state labels within a preset time window meets the affinity resonance condition; If the affinity resonance condition is met, a larger preset increment is used to increase the path affinity value to accelerate the memorization of efficient paths.

[0012] Preferably, the method further comprises the step of generating the dissipative field signaling: The node monitors its own node capability matrix in real time, which includes the node's processor load and forwarding queue length; When the indicator in the node capability matrix exceeds the preset congestion threshold, the data quantum state label of the data packet in its forwarding queue is analyzed, and a dissipative field signaling that can characterize the dominant intention of the current congestion cause is generated and broadcasted to its neighboring nodes.

[0013] Preferably, the dissipative field signaling is a dissipative field vector consistent with the dimension of the data quantum state label, and the component values ​​of the dissipative field vector are determined by the node analyzing the average distribution of the data quantum state labels of all data packets in its forwarding queue.

[0014] Preferably, the step of calculating the comprehensive cost is specifically: calculating by the following formula: ; in, is the comprehensive cost, is the current node, is the next-hop neighbor node, For data packets; For mismatched measures, is a correction term based on the path affinity value, is a correction term based on the dissipative field signaling.

[0015] Preferably, the dissipative field correction in the comprehensive cost Calculated using the following formula: ; in, is the dissipation influence factor, is the data quantum state label of the data packet, The next hop neighbor node The dissipation field vector corresponding to the broadcast dissipation field signaling, For calculation and The similarity function between them.

[0016] Preferably, the node capability matrix of the node further includes: the node's residual energy and link status information to each neighboring node, wherein the link status information includes available bandwidth, average delay and link reliability.

[0017] The adaptive Mesh network architecture construction system for industrial IoT hybrid networking includes: To address the aforementioned technical issues, this invention provides a novel method and system for constructing an adaptive mesh network architecture for hybrid industrial IoT networking. This solution introduces a mechanism that characterizes intrinsic data requirements and establishes a distributed intelligent decision-making framework that combines historical experience learning with real-time congestion avoidance. This enables the network to self-organize and adaptively match optimal transmission paths for different data types.

[0018] A first aspect of the present invention provides a method for constructing an adaptive Mesh network architecture for hybrid networking of the Industrial Internet of Things.

[0019] In one embodiment, the method includes: when a node in a network receives a data packet to be forwarded, it first obtains a data quantum state label encapsulated in the data packet. This data quantum state label is not a simple priority tag, but rather a structured data that finely characterizes the potential demand intensity of the data packet for different network quality of service dimensions (such as latency, bandwidth, reliability, etc.).

[0020] In a preferred embodiment, the data quantum state tag is a multidimensional vector whose dimensions may include: Delayed Potential , used to characterize the sensitivity of the data packet to transmission delay; Bandwidth potential , used to characterize the demand for transmission bandwidth of the data packet; Reliable potential , used to characterize the severity of the data packet's requirements for the transmission success rate; Aggregation potential , used to characterize the path stability requirement of the data packet as part of a data flow.

[0021] After obtaining the quantum state label of the data, the node will calculate a comprehensive cost for forwarding the data packet for each of its reachable next-hop neighbor nodes. This comprehensive cost is the result of a multi-dimensional, dynamic assessment. Its calculation process breaks through the limitations of traditional reliance on a single physical indicator and innovatively integrates considerations from three dimensions: Intent-Capability Matching: This evaluates the degree of match between the requirements carried by the data quantum state label of a data packet and the actual physical capabilities provided by the path from the current node to the next-hop neighbor node.

[0022] Memory of historical transmission experience: Utilizes a path affinity value distributed and maintained across nodes. This value is the network's memory of historical successful transmission experiences, reflecting the effectiveness of a path in meeting specific data requirements.

[0023] Avoiding real-time congestion: Introducing a dissipative field signaling system that is actively broadcast by potentially congested downstream nodes. This signaling system intelligently indicates the type of data causing congestion, allowing upstream nodes to anticipate and avoid congestion.

[0024] In a specific embodiment, the comprehensive cost Calculated using the following formula: ; in, is the current node, is the next-hop neighbor node, For data packets; For mismatched measures, is a correction term based on the path affinity value, is a correction term based on the dissipative field signaling.

[0025] Finally, the node selects the neighbor node with the lowest cost value as the best next hop based on the comprehensive costs calculated by all next hop neighbor nodes, and forwards the data packet.

[0026] To enable self-organizing learning in the network, the method of the present invention further includes a mechanism for dynamically updating path affinity values. When a data packet successfully reaches its destination, it returns an acknowledgment frame along the original path. Upon receiving this acknowledgment frame, each node along the path increments its internally stored path affinity value corresponding to the packet's intended type.

[0027] To further improve learning efficiency, this update mechanism incorporates the concept of affinity resonance. Within a preset time window, a node determines whether the number of packets with similar data quantum state labels flowing to the same next-hop neighbor node satisfies an affinity resonance condition. If so, the path affinity value is increased by a larger preset increment, thereby achieving rapid convergence and memory enhancement for an efficient and stable transmission path.

[0028] To achieve adaptive network regulation and congestion avoidance, the method of the present invention further includes a mechanism for generating and applying dissipative field signaling. In an embodiment, each node monitors its own node capability matrix in real time. When an indicator in the node capability matrix exceeds a preset congestion threshold, the node analyzes the statistical distribution of the data quantum state labels of the data packets in its congested queue, thereby generating a dissipative field signal that can represent the dominant intention of the current congestion cause and broadcast it to its neighboring nodes.

[0029] In a specific embodiment, the dissipative field correction term in the comprehensive cost is Calculate using the following formula to achieve intelligent avoidance: ; in, is the dissipation influence factor, is the data quantum state label of the data packet, The next hop neighbor node The dissipation field vector corresponding to the broadcast dissipation field signaling, For calculation and This mechanism dynamically increases the forwarding cost of a packet based on its relevance to the downstream node's congestion cause, thereby achieving precise and selective congestion avoidance rather than blanket path suppression.

[0030] A second aspect of the present invention provides an adaptive Mesh network architecture construction system for industrial Internet of Things hybrid networking.

[0031] The system is designed to perform the steps of the aforementioned method, which may include: A data encapsulation module configured to encapsulate a data quantum state tag in a data packet; a routing decision module configured to calculate a comprehensive cost for forwarding to each next-hop neighbor node when a node in the system receives a data packet; a forwarding selection module configured to select a next-hop neighbor node with the lowest cost for forwarding according to the calculated comprehensive cost.

[0032] The routing decision module calculates the comprehensive cost in conjunction with the data quantum state label of the data packet, the path affinity value stored by the node, and the dissipation field signaling received from the neighbor node.

[0033] The present application provides an adaptive Mesh network architecture construction method for industrial Internet of Things hybrid networking. 1、The present application introduces a data quantum state label that can finely represent the internal demand of data services, and matches it with the physical capability of the path for calculation, so that the network node has the ability to deeply perceive the data intent for the first time. Compared with the traditional technology that only relies on physical layer indicators for blind forwarding, the present application can distinguish different service quality demand services from the source, and evaluate the matching degree of intent and capability at each hop, thereby realizing accurate and differentiated matching of network resources to service demand, and providing a solid foundation for guaranteeing the service quality of key services in complex industrial scenarios.

[0034] 2、The present application establishes a negative feedback regulation mechanism based on dissipation field signaling, realizes an active and accurate congestion avoidance, and the upstream node can calculate the similarity between its data packet intent and the dissipation field intent when making routing decisions, to judge the risk of its data packet aggravating congestion, thereby realizing accurate and selective avoidance of related data flow, and effectively improving the load balancing capability and overall throughput of the network.

[0035] 3、The present application distributes the core intelligence such as routing decision, path memory and congestion regulation completely to each network node, and relies on local interaction rules to emerge macroscopic adaptive behavior, thereby constructing a self-organizing network without central controller and with high robustness. This distributed architecture eliminates the inherent single point failure risk and performance bottleneck of centralized scheme, so that the network can adaptively and flexibly adapt to dynamic events such as node failure, link quality fluctuation and service load change, and exhibits stronger environmental adaptability and self-repairing ability. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 The method flowchart of the present application; Figure 2 The system architecture diagram of the present application. DETAILED DESCRIPTION

[0037] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described, obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0038] Embodiments:

[0039] Please refer to the accompanying Figure 1 The embodiment of the present application provides an adaptive Mesh network architecture construction method of an industrial Internet of Things hybrid networking, comprising the following steps: S1, when a node in the network receives a data packet to be forwarded, a data quantum state label encapsulated in the data packet is obtained, and the data quantum state label is used to represent the potential demand intensity of the data packet for different network service quality dimensions; In the embodiment, in order to construct an adaptive Mesh network capable of accurately perceiving its own state and network environment, S1, node initialization and state representation, as a basic step of the method of the present application, its core is to establish a dynamic and real-time self-state description mechanism for each forwarding node in the network.

[0040] In the embodiment, when a physical device, such as a sensor gateway or a wireless router, is started for the first time and intends to join the Mesh network constructed by the present application, it will perform a network initialization process. The process can include, but is not limited to, finding and connecting to one or more specified network coordinators through a broadcast discovery protocol, thereby completing identity authentication, obtaining a unique internal network identification ID in the whole network, and synchronizing basic network configuration parameters and the like. This initialization step ensures that the node has the basic qualification of being uniquely identified and managed in the network, and is the premise of its participation in all subsequent network activities.

[0041] After the initialization is completed, the core of the method enters the stage of continuously and dynamically representing the node's own capability. Specifically, each node will locally and in real time monitor and maintain a structured data set, which is defined as a node capability matrix in the present application. The node capability matrix is not a static configuration parameter, it is the self-cognition basis of the node for distributed and intelligent decision-making, and the accuracy and timeliness of its information is directly related to the final effect of the whole network adaptive adjustment.

[0042] In a specific embodiment, the node capability matrix For example, the node The internal part can include the following multiple dimension key performance indicators, each of which reflects the current state of the node from a specific side: First, in order to characterize the computing resource usage of the node, the node capacity matrix includes the processor load This indicator reflects the busyness of the microcontroller unit or central processing unit inside the node. In a specific implementation, this value can be periodically obtained by monitoring the CPU utilization statistics provided by the node operating system kernel. A high processor load means that the node may not be able to handle complex protocol stack calculations or high-rate packet forwarding in a timely manner, which is crucial for evaluating its ability to handle delay-sensitive services.

[0043] Secondly, in order to intuitively reflect the congestion level of node data forwarding, the node capacity matrix includes the forwarding queue length This metric directly quantifies how many packets are queued in the node's network interface buffer at a specific moment, waiting to be sent. This value can be obtained by querying the network interface driver's status information.

[0044] Furthermore, for nodes deployed at special locations and powered by batteries or other limited energy sources, the node capacity matrix preferably includes the remaining energy This metric is crucial for achieving balanced energy efficiency across the network and extending the lifecycle of specific nodes. When making routing decisions, proactively avoiding nodes with low energy levels can effectively prevent communication interruptions caused by unexpected node offlines. For nodes powered by a stable external power source, this metric can be set to a constant maximum value.

[0045] In addition, to fully evaluate the quality of a potential path, the node capability matrix must also include link status information about the quality of the connection. This information is not a single value, but a collection that records in detail the dynamic characteristics of the wireless or wired link from the current node to each of its neighboring nodes. For example, the link status information can be broken down into the following categories: Available bandwidth This metric is used to assess the data carrying capacity of a link. In a preferred embodiment, it can be calculated by taking a sliding average of the throughput of recently successfully transmitted data, or by actively estimating it by sending a sequence of lightweight probe packets during idle periods, resulting in a value that is closer to the actual bandwidth than the theoretical bandwidth.

[0046] Average latency This metric is key to measuring link responsiveness. It can be accurately obtained by measuring the round-trip time of an acknowledgment frame, or by estimating one-way delay by embedding timestamps in data packets.

[0047] Link reliability This metric is used to assess link stability. It can be quantified by counting the successful packet delivery rate within a recent time window, minus the packet loss rate. Highly reliable links are preferred for transmitting critical control or status data.

[0048] To ensure that the node capability matrix truly reflects the ever-changing industrial environment, the node fully updates all of the aforementioned indicators at a preset, high frequency. This continuous, high-frequency self-state representation ensures that the path capability information used for the subsequent comprehensive cost calculation in step S2 is highly timely and accurate.

[0049] S2. The node calculates a comprehensive cost for forwarding the data packet for each of its next-hop neighbor nodes. This comprehensive cost calculation combines the matching degree between the data quantum state label and the path capability associated with the next-hop neighbor node, the path affinity value based on historical transmission experience, and the dissipative field signaling based on the congestion status of the next-hop neighbor node. In this embodiment, when the method of the present invention executes to step S2: the node receives the data packet to be forwarded and makes routing decisions and forwards based on data intent, historical experience and real-time congestion status, it demonstrates the core intelligence of the present invention that is different from traditional routing mechanisms.

[0050] Specifically, when a forwarding node in the network When a packet Pk is received, the node immediately initiates a distributed, multi-dimensional routing decision process. The goal of this process is to find a forwarding path with the best overall adaptability for the packet among all possible next-hop neighbors.

[0051] The first step in this decision-making process is to parse and obtain the data quantum state label encapsulated in the data packet Pk The data quantum state tag is the key technical feature of the present invention to achieve intention perception. It is assigned at the source of data generation and carries the inherent and potential demand of the data packet for network services. In a preferred embodiment, the tag is a standardized four-dimensional vector Its components are respectively from the four dimensions of latency, bandwidth, reliability and aggregation, and are the logical starting point for all subsequent calculations.

[0052] After obtaining the data quantum state label, the node The core task of the , calculate a comprehensive cost for evaluating the quality of forwarding This comprehensive cost calculation abandons the limitations of a single physical metric and innovatively integrates data intent, path capabilities, historical experience, and real-time congestion warnings to form a comprehensive and dynamic basis for decision-making.

[0053] In a specific embodiment, the comprehensive cost is calculated according to the following formula: ; The three components of this formula each carry different decision-making considerations, and their detailed principles are explained as follows: First, there is a mismatch between the measurement items , which aims to calculate the most basic physical level. To calculate this, the node First, we need to find the neighbor nodes obtained in step S1. Node Capability Matrix and the link state information between them, constructing a path capability vector consistent with the dimension of the data quantum state label For example, the delay capability component of the path capability vector , is based on the neighbor nodes Average latency reported in its NCM , obtained after normalization, its value is inversely proportional to the delay. Similarly, the bandwidth capacity component Available bandwidth Directly proportional.

[0054] After constructing the path capability vector, the mismatch measure can be calculated using the following formula: ; In this formula, It is a system-preset, configurable weight vector that allows network administrators to assign different importance to different dimensions of mismatch according to the overall strategy. For example, for a network with real-time control as its main business, the weight of the delay dimension can be assigned. A higher value. The calculation result of this item intuitively reflects that if the data packet is sent to the node , to what extent their internal needs cannot be met.

[0055] The second is the path affinity modifier , which enables the network to remember and prefer paths that have historically proven to be effective for specific types of services. In the embodiment, its calculation formula is: ; In this formula, Represented by neighbor nodes Maintain and share specific intent types Path affinity value of intent type. It can be preferably defined as the type corresponding to the component with the largest value in the QST vector of the data packet. is an adjustable affinity influencing factor. This mechanism is a direct reflection of the positive feedback learning effect in the subsequent step S3.

[0056] The third is the dissipation field correction term This is the core mechanism for implementing proactive and intelligent congestion avoidance in this invention, giving the network a forward-looking early warning capability. Its calculation formula is as follows: ; In this formula, It is composed of neighboring nodes that may be in a congested state. The dissipative field vector corresponding to the actively broadcast dissipative field signaling, the generation of which will be described in detail in step S4, has the core function of announcing to the upstream node the dominant data intention type that causes its congestion. It is an adjustable dissipation influence factor.

[0057] Cosine similarity function The application here is particularly critical. It does not simply punish all paths leading to congested nodes, but accurately calculates the intention vector of the current data packet. and the dominant intent vector causing downstream congestion This means that if a packet's intent is completely different from the cause of congestion at the downstream node, the calculated value of this item will be low, and the packet will experience little forwarding obstruction. Conversely, if the packet's intent is highly consistent with the cause of congestion, the value of this item will be significantly increased, effectively guiding the packet to avoid congestion points.

[0058] When the node After calculating the above combined costs for all reachable next-hop neighbor nodes, it will perform a deterministic selection operation: ; That is, select the neighbor node with the lowest comprehensive cost As the best next hop for this forwarding. Finally, the node The data packet Pk is passed to the node The connected network interface sends it out, thus completing a complete routing and forwarding operation based on multi-dimensional intelligent decision-making.

[0059] S3. The node selects the next-hop neighbor node with the lowest cost based on the calculated comprehensive cost to forward the data packet.

[0060] In this embodiment, to enable the adaptive mesh network to learn from successful experiences and form effective memories, the method further performs step S3: nodes update network path memories based on successful transmission experiences using a positive feedback method involving a resonance mechanism. This step is the core positive feedback link for achieving network self-organization and path optimization in the present invention. It enables the network to dynamically identify, consolidate, and prioritize transmission paths that have been proven effective for specific business needs.

[0061] In a specific implementation, the triggering of this positive feedback learning mechanism is tied to a successful end-to-end data transmission event. Specifically, when a data packet is successfully delivered to its final destination node after the routing decision process in step S2, the destination node generates and sends a confirmation frame as a success signal. This confirmation frame is not rerouted in the network, but is strictly limited to forwarding hop by hop along the path taken by the original data packet, returning to the data source along the original route. This design ensures that every intermediate forwarding node that contributes to a successful transmission can receive this positive feedback signal without omission, thus providing it with the opportunity to participate in the construction of network memory.

[0062] When any intermediate node on the path When receiving this returned confirmation frame, it will immediately start a process to update the path affinity value stored in its internal storage. The path affinity value is not a general evaluation of the next hop node, but a refined memory that is closely coupled with the intention type of the data packet. Specifically, the node The intent type of the original data packet will first be parsed from the confirmation frame . Subsequently, the node will be specific to that intent type And the next hop direction that generates the confirmation frame is used to enhance the affinity value.

[0063] In the embodiment, the affinity value is updated not by a fixed linear accumulation method, but by introducing a more intelligent affinity resonance mechanism that can perceive the traffic pattern. This mechanism makes the affinity value increment It changes dynamically. The core of this is that the node will perform a resonance condition judgment before updating.

[0064] The judgment logic of affinity resonance condition is: node Will check within the last preset time window Whether the number of data packets forwarded by this node to the same next-hop neighbor node that generates the confirmation frame and with similar data quantum state labels has exceeded a preset resonance threshold .

[0065] According to the judgment result of the resonance condition, the increment of affinity value This will be determined by: ; In this formula, is a regular, smaller base increment that gives a base reward for all successful transfers. The technical purpose of this design is that when a stable, high-traffic transmission channel for a specific business is formed in the network, the method of the present invention can identify the formation of such consensus and adopt a larger increment to , enabling it to quickly stand out in subsequent routing competition.

[0066] To ensure the long-term adaptability of the network and prevent historical path information from being frozen and unable to adapt to new changes in the network environment, the affinity update mechanism also includes an essential affinity decay process. Specifically, all affinity values ​​stored in all nodes will decay periodically over time at a fixed, slow rate.

[0067] Therefore, in a complete update cycle, the final state of an affinity value will be determined by incremental accumulation and time decay. Its complete update formula can be expressed as: ; in, is the affinity value before the update, is the updated value, This is a small, system-preset decay factor. This decay mechanism ensures that the affinity advantage of any path is not permanent. Once a path is no longer used or its performance degrades, its affinity value is gradually forgotten, providing an opportunity for new, better paths to emerge.

[0068] By executing step S3, the present invention establishes a complete, closed-loop, positive-feedback learning system. A successful transmission, through this step, enhances the affinity of a specific path for a specific intent. This enhanced affinity directly influences the comprehensive cost calculation of subsequent packets in step S2, increasing the probability of selecting that path again. This iterative reinforcement process enables the network to automatically generate optimized virtual transmission channels for different service types without manual intervention.

[0069] Please see the attached Figure 2 Another embodiment of the present invention provides an adaptive Mesh network architecture construction system for industrial Internet of Things hybrid networking, including the following steps: a data encapsulation module configured to encapsulate a data quantum state label in a data packet, the data quantum state label being used to represent potential demand intensity of the data packet for different network service quality dimensions; a routing decision module configured to, when a node in the system receives a data packet, calculate a comprehensive cost for forwarding for each next-hop neighbor node of the data packet; the calculation of the comprehensive cost combines a matching degree between the data quantum state label and path capability of the next-hop neighbor node, a path affinity value based on historical transmission experience, and a dissipation field signaling based on a congestion state of the next-hop neighbor node; a forwarding selection module configured to select a next-hop neighbor node with the lowest cost for forwarding the data packet according to the comprehensive cost calculated by the routing decision module.

[0070] The system of the embodiment can be used to execute the method embodiments described above, and has similar principles and technical effects, which will not be described here again.

Claims

1. A method for constructing an adaptive Mesh network architecture for hybrid networking of industrial Internet of Things, characterized by: The following steps are involved: S1. When a node in the network receives a data packet to be forwarded, it obtains a data quantum state label encapsulated in the data packet, where the data quantum state label is used to characterize the potential demand intensity of the data packet for different network service quality dimensions; S2. The node calculates a comprehensive cost for forwarding the data packet for each of its next-hop neighbor nodes; The calculation of the comprehensive cost combines the matching degree between the data quantum state label and the path capability associated with the next-hop neighbor node, the path affinity value based on historical transmission experience, and the dissipative field signaling based on the congestion state of the next-hop neighbor node; S3. The node selects a next-hop neighbor node with the lowest cost based on the calculated comprehensive cost to forward the data packet.

2. The method for constructing an adaptive Mesh network architecture for hybrid networking of industrial Internet of Things according to claim 1 is characterized in that: The data quantum state label is a multidimensional vector, whose dimensions include: Delay potential, used to characterize the sensitivity of the data packet to transmission delay; Bandwidth potential, used to characterize the demand for transmission bandwidth of the data packet; Reliability potential, used to characterize the severity of the data packet's requirements for transmission success rate; Aggregation potential is used to characterize the path stability requirement of the data packet as part of a data flow.

3. The method for constructing an adaptive Mesh network architecture for hybrid networking of industrial Internet of Things according to claim 1 is characterized in that: The method further comprises the step of updating the path affinity value: When the data packet successfully arrives at the destination node, a confirmation frame is returned along the original route; After receiving the confirmation frame, the node on the path increases the path affinity value stored internally and corresponding to the intention type of the data packet.

4. The method for constructing an adaptive Mesh network architecture for hybrid networking of industrial Internet of Things according to claim 3 is characterized in that: The step of updating the path affinity value further comprises: Before increasing the path affinity value, determining whether the number of data packets flowing to the same next-hop neighbor node and having similar data quantum state labels within a preset time window meets the affinity resonance condition; If the affinity resonance condition is met, a larger preset increment is used to increase the path affinity value to accelerate the memorization of efficient paths.

5. The method for constructing an adaptive Mesh network architecture for hybrid networking of industrial Internet of Things according to claim 1 is characterized in that: The method further comprises the step of generating the dissipative field signaling: The node monitors its own node capability matrix in real time, which includes the node's processor load and forwarding queue length; When the indicator in the node capability matrix exceeds the preset congestion threshold, the data quantum state label of the data packet in its forwarding queue is analyzed, and a dissipative field signaling that can characterize the dominant intention of the current congestion cause is generated and broadcasted to its neighboring nodes.

6. The method for constructing an adaptive Mesh network architecture for hybrid networking of industrial Internet of Things according to claim 5 is characterized in that: The dissipative field signaling is a dissipative field vector consistent with the dimension of the data quantum state label. The component values ​​of the dissipative field vector are determined by the node analyzing the average distribution of the data quantum state labels of all data packets in its forwarding queue.

7. The method for constructing an adaptive Mesh network architecture for hybrid networking of industrial Internet of Things according to claim 1, characterized in that: The step of calculating the comprehensive cost is specifically: calculating by the following formula: ； in, is the comprehensive cost, is the current node, is the next-hop neighbor node, For data packets; For mismatched measures, is a correction term based on the path affinity value, is a correction term based on the dissipative field signaling.

8. The method for constructing an adaptive Mesh network architecture for hybrid networking of industrial Internet of Things according to claim 7, characterized in that: Dissipative field correction in the comprehensive cost Calculated using the following formula: ； in, is the dissipation influence factor, is the data quantum state label of the data packet, The next hop neighbor node The dissipation field vector corresponding to the broadcast dissipation field signaling, For calculation and A function of the similarity between .

9. The method for constructing an adaptive Mesh network architecture for hybrid networking of industrial Internet of Things according to claim 5, characterized in that: The node capability matrix of the node also includes: the remaining energy of the node and link status information to each neighboring node, and the link status information includes available bandwidth, average delay and link reliability.

10. An adaptive Mesh network architecture construction system for hybrid networking of industrial Internet of Things, according to the method for constructing an adaptive Mesh network architecture for hybrid networking of industrial Internet of Things according to any one of claims 1 to 9, characterized in that: include: A data encapsulation module is configured to encapsulate a data quantum state label in a data packet, wherein the data quantum state label is used to characterize the potential demand intensity of the data packet for different network service quality dimensions; a routing decision module configured to calculate a comprehensive cost for forwarding the data packet for each of its next-hop neighbor nodes when a node in the system receives the data packet; the calculation of the comprehensive cost combines the matching degree between the data quantum state label and the path capability associated with the next-hop neighbor node, the path affinity value based on historical transmission experience, and the dissipative field signaling based on the congestion state of the next-hop neighbor node; The forwarding selection module is configured to select a next-hop neighbor node with the lowest cost according to the comprehensive cost calculated by the routing decision module, so as to forward the data packet.

Images