Adaptive MAC (Media Access Control) power transmission line ad hoc network communication enhancement method and related equipment

Abstract

The invention provides an adaptive MAC power transmission line ad hoc network communication enhancement method and related equipment, and the method comprises the steps: obtaining the channel state data and network environment data of a mesh network node, and obtaining the type of a communication link between the mesh network node and other nodes; determining a plurality of service quality index values of each communication link type based on the channel state data; based on the network environment data, screening pre-stored actions of communication between mesh network nodes and other nodes to obtain screening actions; based on the network environment data, each communication link type and a plurality of service quality index values corresponding to the communication link type, determining a target action corresponding to each communication link type in all screening actions by using a pre-constructed agent; and the mesh network node communicates with other nodes corresponding to each communication link type according to the target action corresponding to the communication link type. The technical problem of poor communication stability of the wireless ad hoc network in the prior art is solved.

Description

Technical Field

[0001] This application relates to the field of communication technology, and in particular to an adaptive MAC method and related equipment for enhancing communication in power transmission line self-organizing networks. Background Technology

[0002] Remote areas often experience harsh environments such as strong winds, sandstorms, blizzards, and large temperature differences, severely impacting the stability of wireless and satellite communications. These factors can lead to long transmission channel delays, low signal-to-noise ratios, and even complete connection loss, affecting communication stability. Especially in extreme weather conditions, wireless signals are easily interfered with, resulting in degraded signal quality, unstable data transmission, and random bit errors and high packet loss rates. Traditional congestion control protocols struggle in such environments, potentially impacting communication stability and efficiency due to frequent rate reductions and retransmissions. Therefore, it is urgent to expand the coverage of communication networks and enhance communication transmission capabilities to ensure the stable operation of power transmission lines and fundamentally solve the communication stability challenges faced during power line inspections. Summary of the Invention

[0003] In view of this, the purpose of this application is to propose an adaptive MAC method for enhancing communication in transmission line ad hoc networks and related equipment, so as to overcome all or part of the shortcomings of the prior art.

[0004] To achieve the above objectives, this application provides an adaptive MAC-based communication enhancement method for transmission line ad hoc networks, applied to each mesh network node in a wireless ad hoc network. The method includes: acquiring channel state data and network environment data corresponding to the mesh network node; acquiring communication link types between the mesh network node and other nodes communicating with it; determining multiple quality of service (QoS) index values ​​corresponding to each communication link type based on the channel state data; filtering pre-stored communication actions between the mesh network node and other nodes communicating with it based on the network environment data to obtain at least one filtering action; determining a target action corresponding to each communication link type from all filtering actions using a pre-built intelligent agent based on the network environment data, each communication link type, and its corresponding multiple QoS index values; and the mesh network node communicating with other nodes corresponding to each communication link type according to the target action corresponding to the communication link type.

[0005] Optionally, the step of determining the target action corresponding to each communication link type based on the network environment data, each communication link type and its corresponding multiple service quality index values, using a pre-built intelligent agent among all filtering actions includes: for each communication link type, searching for at least one predetermined action corresponding to the communication link type among all filtering actions; and using the intelligent agent to determine the target action among at least one predetermined action corresponding to the communication link type based on the network environment data and the multiple service quality index values ​​corresponding to the communication link type.

[0006] Optionally, determining the target action from at least one predetermined action corresponding to the communication link type using the agent based on the network environment data and multiple service quality index values ​​corresponding to the communication link type includes: determining an initial predetermined action probability distribution corresponding to the communication link type using the agent based on the multiple service quality index values ​​corresponding to the communication link type; generating a heuristic score corresponding to the communication link type using the agent and a pre-trained quantization model based on the network environment data; adjusting the bias of the initial predetermined action probability distribution using the agent based on the heuristic score to obtain a target predetermined action probability distribution; and randomly determining the target action corresponding to the communication link type from the predetermined actions corresponding to the target predetermined action probability distribution.

[0007] Optionally, the step of generating a heuristic score corresponding to the communication link type based on the network environment data, using the agent and a pre-trained quantization model, includes: determining the performance data corresponding to each action of the communication link type using the quantization model based on the network environment data; and determining the heuristic score corresponding to the communication link type using the agent based on each performance data corresponding to the communication link type.

[0008] Optionally, determining multiple service quality indicator values ​​corresponding to each communication link type based on the channel state data includes: for each predetermined service quality indicator corresponding to each communication link type, determining an initial service quality indicator value corresponding to the predetermined service quality indicator and a weight corresponding to the predetermined service quality indicator based on sub-data in the channel state data corresponding to the predetermined service quality indicator; and determining the service quality indicator value corresponding to the predetermined service quality indicator corresponding to the communication link type based on the initial service quality indicator value and the weight.

[0009] Optionally, determining the heuristic score corresponding to the communication link type using the agent based on each performance data corresponding to the communication link type includes: determining the heuristic score corresponding to the communication link type using a reward function in the agent based on each performance data corresponding to the communication link type.

[0010] Optionally, determining the performance data corresponding to each action of the communication link type based on the network environment data and using the quantization model includes: for each action corresponding to the communication link type, inputting the first sub-data in the network environment data corresponding to the action into a pre-trained throughput quantization model, determining and outputting the theoretical throughput value corresponding to the action through the throughput quantization model; inputting the second sub-data in the network environment data corresponding to the action into a pre-trained latency quantization model, determining and outputting the theoretical latency value corresponding to the action through the latency quantization model; inputting the third sub-data in the network environment data corresponding to the action into a pre-trained packet loss rate quantization model, determining and outputting the theoretical packet loss rate value corresponding to the action through the packet loss rate quantization model; and using the theoretical throughput value, theoretical latency value, and theoretical packet loss rate value corresponding to the action as the performance data corresponding to the action.

[0011] Based on the same inventive concept, this application also provides an adaptive MAC transmission line ad hoc network communication enhancement device, applied to each mesh network node in a wireless ad hoc network, comprising: an acquisition module configured to acquire channel state data and network environment data corresponding to the mesh network node, and to acquire the communication link type between the mesh network node and other nodes communicating with it; a first determination module configured to determine multiple service quality index values ​​corresponding to each communication link type based on the channel state data; a filtering module configured to filter pre-stored communication actions between the mesh network node and other nodes communicating with it based on the network environment data, obtaining at least one filtering action; a second determination module configured to determine the target action corresponding to each communication link type using a pre-built intelligent agent among all filtering actions based on each communication link type and its corresponding multiple service quality index values; and a communication module configured to allow the mesh network node to communicate with other nodes corresponding to each communication link type according to the target action corresponding to the communication link type.

[0012] Optionally, the second determining module is further configured to: for each communication link type, search for at least one predetermined action corresponding to the communication link type among all filtering actions; and determine the target action using the agent among at least one predetermined action corresponding to the communication link type based on the network environment data and multiple service quality index values ​​corresponding to the communication link type.

[0013] Optionally, the second determining module is further configured to: determine an initial predetermined action probability distribution corresponding to the communication link type using the agent based on multiple service quality index values ​​corresponding to the communication link type; generate a heuristic score corresponding to the communication link type using the agent and a pre-trained quantization model based on the network environment data; adjust the initial predetermined action probability distribution by bias using the agent based on the heuristic score to obtain a target predetermined action probability distribution; and randomly determine a target action corresponding to the communication link type from the predetermined actions corresponding to the target predetermined action probability distribution.

[0014] Optionally, the second determining module is further configured to: determine the performance data corresponding to each action of the communication link type based on the network environment data and using the quantization model; and determine the heuristic score corresponding to the communication link type based on each performance data corresponding to the communication link type using the agent.

[0015] Optionally, the first determining module is further configured to: for each predetermined service quality indicator corresponding to each communication link type, determine the initial service quality indicator value corresponding to the predetermined service quality indicator and the weight corresponding to the predetermined service quality indicator based on the sub-data in the channel state data corresponding to the predetermined service quality indicator; and determine the service quality indicator value corresponding to the predetermined service quality indicator corresponding to the communication link type based on the initial service quality indicator value and the weight.

[0016] Optionally, the second determining module is further configured to: determine the heuristic score corresponding to the communication link type based on each performance data corresponding to the communication link type using the reward function in the agent.

[0017] Optionally, the second determining module is further configured to: for each action corresponding to the communication link type, input the first sub-data in the network environment data corresponding to the action into a pre-trained throughput quantization model, determine and output the theoretical throughput value corresponding to the action through the throughput quantization model; input the second sub-data in the network environment data corresponding to the action into a pre-trained latency quantization model, determine and output the theoretical latency value corresponding to the action through the latency quantization model; input the third sub-data in the network environment data corresponding to the action into a pre-trained packet loss rate quantization model, determine and output the theoretical packet loss rate value corresponding to the action through the packet loss rate quantization model; and use the theoretical throughput value, theoretical latency value, and theoretical packet loss rate value corresponding to the action as the performance data corresponding to the action.

[0018] Based on the same inventive concept, this application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor implements the method described above when executing the computer program.

[0019] Based on the same inventive concept, this application also provides a non-transitory computer-readable storage medium that stores computer instructions for causing a computer to perform the method described above.

[0020] As can be seen from the above, the adaptive MAC transmission line ad hoc network communication enhancement method and related equipment provided in this application include: acquiring channel state data and network environment data corresponding to mesh network nodes; acquiring communication link types between the mesh network node and other nodes communicating with it; determining multiple service quality index values ​​corresponding to each communication link type based on the channel state data; filtering pre-stored communication actions between the mesh network node and other nodes communicating with it based on the network environment data to obtain at least one filtering action; determining the target action corresponding to each communication link type using a pre-built intelligent agent among all filtering actions based on the network environment data, each communication link type, and its corresponding multiple service quality index values; and the mesh network node communicating with other nodes corresponding to each communication link type according to the target action corresponding to the communication link type. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in this application or related technologies, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart illustrating the adaptive MAC-based transmission line ad hoc network communication enhancement method according to an embodiment of this application. Figure 2 This is an architecture diagram of a dense power transmission line wireless ad hoc network communication enhancement scenario according to an embodiment of this application; Figure 3 This is a flowchart illustrating the adaptive MAC mechanism combination decision-making process in an embodiment of this application. Figure 4 This is a schematic diagram of the structure of the adaptive MAC transmission line self-organizing network communication enhancement device according to an embodiment of this application; Figure 5 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with specific embodiments and the accompanying drawings.

[0024] It should be noted that, unless otherwise defined, the technical or scientific terms used in the embodiments of this application should have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms "first," "second," and similar terms used in the embodiments of this application do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are only used to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0025] As described in the background section, in recent years, dense power transmission channels in remote areas, as an important component of the "West-to-East Power Transmission" project, have gradually become the main channels for transmitting new energy from large-scale wind and solar power bases in Northwest China. They bear the key task of promoting clean power consumption and energy development, and have a significant impact on energy security. However, remote areas suffer from harsh environments such as strong winds, sandstorms, blizzards, and large temperature differences, severely affecting the stability of wireless and satellite communications. These factors can lead to long delays, low signal-to-noise ratios, or even complete loss of connection in transmission channels, impacting power line inspection and maintenance. Especially under extreme weather conditions, wireless signals are easily interfered with, resulting in decreased signal quality, unstable data transmission, and random bit errors and high packet loss rates. Traditional congestion control protocols are difficult to handle in such environments, and may further reduce communication stability and efficiency due to frequent speed reduction and retransmission. Therefore, it is urgent to expand the coverage of communication networks and enhance communication transmission capabilities to ensure the stable operation of transmission lines and fundamentally solve the communication stability problems faced during power line inspection.

[0026] A Wireless Ad Hoc Network (WAN) is a dynamic, decentralized wireless communication network composed of mobile nodes that can autonomously form a network and transmit data without relying on pre-existing infrastructure. Multiple nodes in a WAN typically communicate through a shared communication medium. The Multiple Access Control Protocol (MAC) is a protocol that manages access to the shared medium among multiple nodes in the network. It defines how users share and compete for access to the medium to avoid collisions and conflicts, thereby achieving reliable, fair, and efficient data transmission. Currently, traditional WAN MAC protocols are commonly used in dense power transmission lines under extreme environments such as strong winds, sandstorms, and blizzards. However, in harsh environments such as sandstorms, wireless channels face severe interference, manifesting as increased signal attenuation, significant multipath effects, and enhanced channel noise. This leads to a sharp decline in the performance of traditional MAC protocols (such as CSMA / CA and TDMA) in terms of latency, throughput, and reliability.

[0027] The existing technologies have the following problems: (1) Static parameter configuration: relying on fixed parameter configuration or a small number of protocol switching settings leads to rigid channel access strategies and makes it impossible to dynamically adapt to rapid fluctuations in channel conditions. (2) Centralized decision-making bottleneck: some intelligent solutions rely on central node coordination, which makes it difficult to cope with node failures or link interruptions in harsh environments. (3) Deficiencies in link consideration: most existing intelligent solutions are applied to single-type nodes and do not consider the differentiated needs of node links in real self-organizing networks (such as mobile terminals needing low-latency access and fixed terminals needing high reliability). Adopting a unified MAC strategy will lead to an imbalance in resource competition, reduce the overall efficiency of the network, and affect the low communication stability of wireless self-organizing networks.

[0028] In addition, existing technologies face two major challenges: first, how to deeply integrate environmental disturbances with communication status to achieve quantitative modeling of channel quality; and second, how to dynamically optimize the combination of MAC mechanisms (backoff algorithm, transmission power, and time slot allocation) through autonomous decision-making by distributed intelligent agents in densely distributed and high-interference environments, while ensuring low signaling overhead and global resource balance.

[0029] In view of this, embodiments of this application propose an adaptive MAC method for enhancing communication in transmission line ad hoc networks, referring to... Figure 1 Applied to each mesh network node in a wireless ad hoc network, it includes the following steps: Step 101: Obtain channel status data and network environment data corresponding to the mesh network node, and obtain the communication link type between the mesh network node and other nodes communicating with it.

[0030] In this step, this application adopts a fully distributed architecture, applied to each mesh node in a wireless ad hoc network in a dense power transmission line scenario, enabling each mesh node to possess autonomous computing and decision-making capabilities. Each mesh node is deployed with an intelligent agent, which acts as an independent unit with computing and decision-making capabilities. The agent on each mesh node independently perceives its surrounding local network environment, senses channel state changes caused by extreme environmental changes in real time, and autonomously selects the optimal MAC mechanism combination for each communication link it is responsible for (including Mesh-Mesh links with other mesh network nodes and Mesh-Terminal links with terminal nodes). It dynamically executes different MAC strategy combinations, effectively adapting to complex wireless transmission environments and achieving continuous enhancement and dynamic adaptation of communication mechanisms. This not only reduces the computational burden on the central node and avoids network-wide paralysis caused by single-point failures, but also enables rapid responses based on local data, improving the network's reaction speed and adaptability.

[0031] A wireless ad hoc network consists of multiple mesh network nodes and several other nodes. For example... Figure 2 As shown, Figure 2This is an architecture diagram of a dense power transmission line wireless ad hoc network communication enhancement scenario according to an embodiment of this application. Other nodes are terminal nodes, which are divided into fixed terminals and mobile terminals. Fixed terminals are typically installed on the towers of the power transmission line, such as vibration sensors, tilt sensors, and temperature sensors. They can collect various data from the towers and surrounding environment in real time and stably, providing basic information for power transmission line status monitoring. Mobile terminals, represented by drones, have the advantage of flexibility and mobility, allowing them to move quickly within a wide power transmission line area to perform detailed detection and data collection on specific areas or equipment, effectively compensating for the limitations of the monitoring range of fixed terminals.

[0032] Mesh network nodes communicate with each other. Obtaining channel state data and network environment data for each node is crucial. Channel state data reflects the quality of the communication links between nodes, while network environment data reflects the quality of the overall network communication. The type of communication link plays a decisive role in the communication actions between nodes in a mesh network. Different types of links differ in bandwidth, latency, reliability, and coverage. These characteristics directly affect how nodes perform data transmission, resource allocation, security, and adapt to different application scenarios. Therefore, obtaining the communication link type between a mesh network node and other nodes it communicates with lays the data foundation for subsequently determining the communication actions between nodes.

[0033] Communication link types include Mesh-Mesh links and Mesh-terminal links. Mesh-terminal links further include Mesh-mobile terminal links and Mesh-fixed terminal links. Different communication links have different characteristics. Mesh-Mesh links have moderate to poor topology stability, may involve multiple hops, require attention to relay scheduling, and nodes typically have strong processing and coordination capabilities. Mesh-terminal links, especially for mobile terminals, experience rapid dynamic changes and frequent link interruptions, such as in drone communication, resulting in high latency and jitter, necessitating fast access and low retransmission overhead. For fixed terminals, the topology is fixed, data is periodic, small in volume, and low-speed, but high reliability is required, such as in tower sensors, where strong link predictability is crucial.

[0034] Step 102: Based on the channel state data, determine multiple service quality index values ​​corresponding to each communication link type.

[0035] In this step, the channel state data comprehensively and meticulously reflects the actual state of the communication link during operation. By deeply analyzing this channel state data, multiple quality of service (QoS) indicators can be accurately determined for each communication link type. These QoS indicators include throughput, latency, packet loss rate, received signal strength, and neighbor node values ​​on ad hoc network links and service links. These QoS indicators reflect disturbances caused by extreme environmental changes in real time.

[0036] Construct a state space using multiple quality of service (QoS) metrics for each communication link type. Its dimensions include: throughput (T) t ): Current average throughput (Mbps); latency (D) t Average end-to-end latency (ms); Packet loss rate (L) t ): Link packet loss rate (%); Received signal strength (RSSI) t : Received Signal Strength Indicator (RSSI) (dBm), reflecting the severity of the environment; neighbor nodes (V t : Average speed (m / s) of neighboring mobile terminals within the coverage area, reflecting the trend of topology changes.

[0037] Step 103: Based on the network environment data, filter the communication actions between the pre-stored mesh network nodes and other nodes communicating with them to obtain at least one filtered action.

[0038] In this step, a MAC mechanism combination strategy library is designed to address the differentiated needs of the three link types in wireless ad hoc networks (Mesh-Mesh link type, Mesh-mobile terminal link type, and Mesh-fixed terminal link type). This library includes combinations of various common mechanisms such as Free Access (FA), Time Slot Access (SA), Backoff (BO), and Handshake (HS), each combination constituting an "action." The MAC mechanism combination strategy library contains pre-stored communication actions between mesh network nodes and other communicating nodes. While the library contains a relatively large number of communication actions, not all actions are suitable for communication between current nodes. Poor external environments can affect the stability of network data, such as strong signal interference, limited bandwidth, and large latency fluctuations. Some pre-stored actions may be feasible in ideal network environments but may fail to execute properly or cause communication interruptions under current adverse conditions.

[0039] By filtering pre-stored communication actions between mesh network nodes and other communicating nodes based on network environment data, at least one filtering action is obtained. This allows actions unsuitable for the current network conditions to be eliminated, retaining only those that can operate stably even in poor network conditions. Furthermore, in poor network environments, network resources are relatively scarce, such as limited bandwidth. Continuing to use all pre-stored communication actions without filtering could lead to a significant waste of resources on unnecessary or inefficient actions. By filtering all pre-stored actions, interference from poor-performing communication actions is reduced, while those with superior inherent performance characteristics are retained, which is beneficial for improving the stability and efficiency of subsequent inter-node communication.

[0040] It should be noted that, in order to differentiate the actions in the MAC mechanism combination policy library, the actions in the MAC mechanism combination policy library are divided into four stages according to the MAC protocol lifecycle, and each stage provides multiple candidate actions for the agent to choose from. For example, as shown in Table 1, Table 1. Correspondence between MAC operation phases and mechanism actions

[0041] Step 104: Based on the network environment data, each communication link type and its corresponding multiple service quality index values, the target action corresponding to each communication link type is determined by a pre-built intelligent agent in all filtering actions.

[0042] In this step, network environment data reflects the communication performance of each communication link type, as different link types have unique physical characteristics and transmission capabilities. Simultaneously, each link type corresponds to multiple Quality of Service (QoS) indicators, reflecting the actual performance of the communication link from different dimensions. The pre-built agent, trained and learned from extensive data, can deeply understand network environment data and the complex relationships between different link types and QoS indicators. Based on this detailed information, it accurately selects the target action that best matches the characteristics and current performance status of each link type from all filtering actions, ensuring the accuracy of the determined target action. Furthermore, it eliminates the need to determine the target action from all actions; instead, it selects it from all filtered actions. The number of filtered actions is relatively small, and their performance is relatively superior, which is beneficial for improving the communication stability and efficiency between nodes, thereby enhancing the communication stability and efficiency of the wireless ad hoc network.

[0043] Furthermore, the agent possesses real-time perception and rapid decision-making capabilities. It can continuously monitor changes in the link's quality of service indicators and adjust target actions promptly based on this dynamic information. The agent can quickly reselect a target action with stronger anti-interference capabilities from the selected actions to ensure communication stability. This dynamic adjustment capability enables the network to maintain efficient operation in complex and ever-changing environments, greatly enhancing the network's adaptability and robustness.

[0044] This application proposes an adaptive MAC mechanism enhancement method based on the Proximal Policy Optimization (PPO) algorithm. This method dynamically optimizes MAC mechanism parameter combinations (such as backoff window, transmission power, and access mode) based on the PPO algorithm, enabling autonomous coordination and dynamic response among nodes. This method is specifically designed for communication transmission across links in wireless ad hoc network scenarios, and is particularly adaptable to the dynamic channel changes caused by complex natural environments along power transmission lines, such as strong winds, sandstorms, snowstorms, and large temperature differences. The core idea is to embed a PPO-based reinforcement learning agent into each mesh network node. This agent can autonomously perceive the channel state and dynamically optimize the combination of MAC layer mechanisms, allowing the agent to concurrently and independently select and optimize MAC policies for different links. This process aims to maximize system throughput while ensuring communication robustness. The specific implementation process includes key steps such as network model initialization, state space design, action space design, reward function design, policy update mechanism, and online deployment and updating, which together constitute a complete adaptive MAC mechanism optimization closed loop.

[0045] Step 105: The mesh network node communicates with other nodes corresponding to each communication link type according to the target action corresponding to the communication link type.

[0046] In this step, each mesh network node needs to communicate with other nodes corresponding to each communication link type. For each other node corresponding to a specific communication link type, the mesh network node communicates with that node according to the target action corresponding to that communication link type. By ensuring a high degree of compatibility between the target action and the communication link type, and avoiding low communication efficiency or frequent failures due to action-link mismatch, the communication stability between nodes is improved, thereby enhancing the communication stability of the wireless ad hoc network.

[0047] By deploying independent intelligent agents at each mesh network node and dynamically optimizing the MAC mechanism at each stage using reinforcement learning algorithms, node-level autonomous decision-making is achieved. This method utilizes locally sensed real-time extreme environmental disturbance information to collaboratively adjust multi-node strategy combinations, overcoming the latency and rigidity problems of traditional centralized control. It significantly improves communication stability and network throughput efficiency in scenarios with dense power transmission lines, such as strong winds, sandstorms, and snowstorms. To achieve full-process collaborative control of the adaptive MAC mechanism, such as... Figure 3 As shown, Figure 3 This is a flowchart illustrating the adaptive MAC mechanism combination decision-making process in an embodiment of this application. This application constructs a complete closed-loop mechanism encompassing "environmental state awareness—channel state quantification modeling—policy generation and selection—mechanism execution." Through environment-driven channel state awareness and distributed node autonomous policy optimization, the throughput, real-time performance, and link reliability of dense transmission lines under extreme environments are significantly improved.

[0048] This application aims to address the problems of rigid channel access strategies, insufficient fusion of environmental awareness and communication status, and low efficiency of distributed collaborative decision-making in existing wireless ad hoc network MAC protocols under harsh environments such as dense power transmission lines. To address the issues of parameter rigidity, insufficient environmental awareness, and low global coordination efficiency in traditional wireless ad hoc network MAC protocols under harsh environments such as strong winds, sandstorms, and blizzards, this application utilizes a fully distributed multi-agent reinforcement learning architecture and dynamic channel state quantization modeling to achieve adaptive optimization of the communication mechanism. This solves the response lag problem caused by traditional solutions relying on fixed parameters or centralized decision-making, significantly improving the stability and real-time performance of the communication link. The results of this application achieve synergistic optimization of throughput, real-time performance, and reliability, providing an enhanced solution for high-reliability communication in dense power transmission line wireless ad hoc networks under extremely harsh environments.

[0049] The above scheme obtains channel state data and network environment data corresponding to mesh network nodes, and acquires the communication link types between the mesh network nodes and other nodes communicating with them. Based on the channel state data, multiple service quality index values ​​corresponding to each communication link type are determined. Based on the network environment data, pre-stored communication actions between the mesh network nodes and other nodes communicating with them are filtered to obtain at least one filtering action. Based on the network environment data, each communication link type and its corresponding multiple service quality index values ​​are used by a pre-built agent to determine the target action corresponding to each communication link type from all filtering actions. The mesh network nodes communicate with other nodes corresponding to each communication link type according to the target action corresponding to the communication link type.

[0050] In some embodiments, the step of determining the target action corresponding to each communication link type based on the network environment data, each communication link type and its corresponding multiple service quality index values, using a pre-built intelligent agent among all filtering actions includes: for each communication link type, searching for at least one predetermined action corresponding to the communication link type among all filtering actions; and using the intelligent agent to determine the target action among at least one predetermined action corresponding to the communication link type based on the network environment data and the multiple service quality index values ​​corresponding to the communication link type. In this embodiment, the number of all screening actions is relatively large. If the agent is directly asked to determine the target action based on the service quality index value among all screening actions, the agent would need to perform a large number of calculations and comparisons, which would consume a lot of computing resources and time, resulting in low processing efficiency. To avoid the above problems, this application pre-constructs the communication actions corresponding to each communication link type, associates the communication link type with its corresponding actions, and stores them in the MAC mechanism combination strategy library. Based on the communication link type, the agent searches for at least one predetermined action corresponding to the communication link type in the MAC mechanism combination strategy library, which improves the efficiency of obtaining at least one predetermined action corresponding to the communication link type. The predetermined actions are pre-set for different communication link types and have a high degree of adaptability to specific link types. Network environment data can reflect the communication performance of each communication link type. Different communication link types have unique physical characteristics and transmission capabilities. Based on network environment data and multiple service quality index values ​​corresponding to the communication link type, the agent determines the target action from at least one predetermined action corresponding to the communication link type. By determining the target action, the search scope is significantly narrowed from all filtered actions to a predetermined range of actions related to the link type. This greatly reduces the amount of data that the agent needs to process, lowers computational complexity, and thus significantly improves processing efficiency. This allows the agent to determine the target action more quickly, meeting the real-time requirements of the communication network.

[0051] For example, at least one predetermined action corresponding to the Mesh-Mesh link type is shown in Table 2, at least one predetermined action corresponding to the Mesh-mobile terminal link type is shown in Table 3, and at least one predetermined action corresponding to the Mesh-fixed terminal link type is shown in Table 4.

[0052] Table 2 Candidate Mesh-Mesh Link Mechanisms

[0053] Table 3. Mesh-Mobile Terminal Link Mechanism Candidates

[0054] Table 4. Mesh-fixed terminal link mechanism candidates

[0055] In some embodiments, determining the target action from at least one predetermined action corresponding to the communication link type using the agent based on the network environment data and multiple service quality index values ​​corresponding to the communication link type includes: determining an initial predetermined action probability distribution corresponding to the communication link type using the agent based on the multiple service quality index values ​​corresponding to the communication link type; generating a heuristic score corresponding to the communication link type using the agent and a pre-trained quantization model based on the network environment data; adjusting the bias of the initial predetermined action probability distribution using the agent based on the heuristic score to obtain a target predetermined action probability distribution; and randomly determining the target action corresponding to the communication link type from the predetermined actions corresponding to the target predetermined action probability distribution.

[0056] In this embodiment, multiple service quality index values ​​corresponding to the communication link type reflect the current actual performance status and service requirements of the link from different dimensions. By analyzing the multiple service quality index values ​​corresponding to the communication link type, the agent assigns a reasonable probability to each predetermined action corresponding to the communication link type. The probability magnitude reflects the degree of matching between the action and the current link, resulting in an initial predetermined action probability distribution corresponding to the communication link type. This initial predetermined action probability distribution includes at least one predetermined action corresponding to the communication link type and the probability of each predetermined action. Actions with higher probabilities have a greater advantage in meeting the link's service quality requirements.

[0057] Based on network environment data, this paper utilizes intelligent agents and pre-trained quantization models to generate heuristic scores corresponding to different communication link types. These heuristic scores characterize the real-time performance of each communication link. In complex and dynamically changing communication network environments, the performance of different communication link types is correlated with network environment data. Traditional static rules or manual configurations are difficult to accurately match actual needs. Therefore, by collecting network environment data and using intelligent agents combined with pre-trained quantization models, heuristic scores for different communication link types can be dynamically generated. This process combines data-driven objective analysis with adaptive decision-making by intelligent algorithms, overcoming the subjective limitations of human experience and avoiding the dependence of a single model on static assumptions. This provides a scientific basis for link selection, resource allocation, or path planning, ultimately achieving a comprehensive improvement in communication efficiency, reliability, and resource utilization.

[0058] Based on heuristic scores, an agent adjusts the bias of the initial predetermined action probability distribution to obtain the target predetermined action probability distribution. This target predetermined action probability distribution includes at least one target predetermined action corresponding to the communication link type and the probability of each target predetermined action. By determining the target predetermined action probability distribution, the number of at least one predetermined action corresponding to the communication link type is adjusted. In complex and dynamic communication networks or decision-making scenarios, to optimize the agent's action selection strategy, the agent can dynamically adjust the initial predetermined action probability distribution based on previously generated heuristic scores (reflecting the real-time superiority or inferiority of different communication links). Specifically, the agent uses the heuristic scores as weights, combined with environmental feedback and learning mechanisms, to selectively strengthen or suppress the probability of selecting each action in the initial probability distribution. For example, it can increase the probability of high-scoring actions and decrease the probability of low-scoring actions, thereby generating a target predetermined action probability distribution that better reflects the current environmental state. This process integrates data-driven heuristic evaluation with the adaptive learning capabilities of intelligent agents, preserving the rationality of the initial strategy while achieving precise optimization of decisions through real-time feedback, ultimately improving the system's response efficiency, resource utilization, and task success rate in dynamic environments.

[0059] The target predetermined action probability distribution shows a relatively high degree of matching between all predetermined actions and the current link. After the agent obtains the target predetermined action probability distribution through heuristic score adjustment, it randomly determines the target action corresponding to the communication link type from the predetermined actions corresponding to the target predetermined action probability distribution. This ensures a high degree of adaptability between the decision and the current network state, and avoids getting trapped in local optima by introducing randomness, thus enhancing the system's fault tolerance to sudden anomalies. Ultimately, a balance is achieved between efficiency and robustness, improving the quality and reliability of communication task completion.

[0060] The Cartesian product space of all combinations of mechanisms, i.e., A = {a1, a2, ..., a...} n}, where each a i This represents a specific set of MAC action strategies. At any given time, the agent independently selects the most suitable configuration for a given communication pair (Mesh-Mesh or Mesh-Terminal). The agent's output is a composite action that simultaneously selects an optimal combination of mechanisms for all communication link types.

[0061] In some embodiments, generating a heuristic score corresponding to the communication link type based on the network environment data, using the agent and a pre-trained quantization model, includes: determining performance data corresponding to each action of the communication link type using the quantization model based on the network environment data; and determining a heuristic score corresponding to the communication link type using the agent based on each performance data corresponding to the communication link type.

[0062] In this embodiment, based on network environment data, a quantization model is used to determine the performance data corresponding to each action of the communication link type. This application pre-trains a quantization model whose core purpose is to directly and deterministically quantify the specific communication quality changes resulting from any MAC mechanism combination action, helping the agent understand and learn the specific enhancement indicators and degrees of each decision, thereby accurately associating abstract actions with measurable performance indicators.

[0063] The input to this quantization model is as follows: Action parameters : Describes the action parameter vector of the selected MAC mechanism. Network state determined by network environment data. : A state vector describing the current local network environment. The output of this quantization model is a quantized performance feature vector. This vector represents the intrinsic performance of the action in that state. A quantitative model of the impact of MAC mechanism combinations on communication quality is constructed, directly transforming the discrete actions of the agent into quantified performance characteristics, providing an intrinsic value assessment for decision-making. Using this quantification result, the exploration direction of the agent is intelligently guided during training, realizing intelligent evolution from inefficient random trial and error to efficient guided learning, significantly accelerating policy convergence speed and improving final performance.

[0064] Based on the performance data corresponding to each communication link type, an intelligent agent determines the heuristic score for that link type. This comprehensive data set reflects the actual operational status of the link at different times and under different environments. Subsequently, relying on an intelligent agent with autonomous learning and reasoning capabilities, this performance data undergoes in-depth analysis and comprehensive processing: the agent combines historical data patterns with real-time network characteristics to quantitatively evaluate the performance of each communication link type and generate corresponding heuristic scores. These scores not only provide an intuitive numerical expression of the link's current performance but also implicitly contain a comprehensive judgment on the link's reliability, efficiency, and applicability. The heuristic scores output by the agent provide crucial information for subsequent link selection, resource allocation, or strategy adjustments, helping the system achieve efficient and stable communication performance in dynamic network environments.

[0065] In some embodiments, determining multiple service quality indicator values ​​corresponding to each communication link type based on the channel state data includes: for each predetermined service quality indicator corresponding to each communication link type, determining an initial service quality indicator value corresponding to the predetermined service quality indicator based on sub-data in the channel state data corresponding to the predetermined service quality indicator, and a weight corresponding to the predetermined service quality indicator; and determining the service quality indicator value corresponding to the predetermined service quality indicator corresponding to the communication link type based on the initial service quality indicator value and the weight. In this embodiment, when constructing the state space of the reinforcement learning agent, the following key Quality of Service (QoS) metrics are comprehensively considered to fully characterize the dynamic characteristics of the wireless channel. For each predetermined QoS metric corresponding to each communication link type, an initial QoS metric value is determined based on sub-data in the channel state data corresponding to the predetermined QoS metric. The initial QoS metric value includes throughput, latency, packet loss rate, RSSI, and neighbor nodes.

[0066] Throughput is divided into two sub-metrics: terminal throughput and inter-mesh throughput.

[0067] Terminal throughput, which is the sum of data flows from all terminal nodes covered by the mesh network nodes, is calculated using the following formula: Formula 1 in, For terminal throughput, U i Refers to the current node The set of terminals covered, D ki (t) refers to the terminal To the node The amount of data successfully sent and received confirmation. The length of the time window used to measure terminal throughput.

[0068] Inter-mesh throughput, which is the sum of data flows from neighboring mesh network nodes that communicate with it, is calculated using the following formula: Formula 2 in, For the throughput between meshes, N i Pointer node The set of neighboring mesh network nodes, D ji (t) refers to the node Successfully sent to the node And the amount of data received was confirmed by it. The length of the time window used to measure the throughput between meshes.

[0069] The initial delay is calculated using the following formula: Formula 3 in, For the initial delay, t1 represents the time when the receiving node receives the data packet, and t2 represents the time when the sending node receives the ACK (acknowledgment character) message.

[0070] To avoid significant jitter from a single measurement, the node records the delay of the most recent n measurements, and then the average of these delays is used as the final measured delay, denoted as Delay. A,B The final delay is calculated using the following formula: Formula 4 Among them, delay A,B This represents the i-th measurement value among the most recent n measurements of delay. For example, n can be 3.

[0071] Given a time window W, record the packet loss rate between nodes, and calculate the packet loss rate using the following formula: Formula 5 in, The packet loss rate is nt, where nt represents the number of packets sent. r This indicates the number of data packets received.

[0072] RSSI measures the received signal power and calculates the received signal strength using the following formula: RSSI=-10n log10(d) + R, Formula 6 Where RSSI is the signal strength value, n is the path loss exponent, d is the distance, and R is the reference signal strength at 1 meter.

[0073] Each mesh network node will record the address and speed information of mobile terminal nodes within its range, and calculate the average speed of neighbor nodes using the following formula: Formula 7 in, Here, vk is the speed of the k-th neighbor node, and N is the current value of the table. v The total number of neighboring nodes recorded in the Table v This is a table of neighbor node speed information.

[0074] The aforementioned metrics (throughput, latency, packet loss rate, RSSI, and average speed of neighboring nodes) collectively constitute the state vector S of the reinforcement learning agent. t This provides intelligent agents with the environmental information needed for decision-making.

[0075] In the current environment, communication quality fluctuates frequently due to extreme natural factors such as strong sandstorms, blizzards, and large temperature differences. Therefore, QoS (Quality of Service) is the primary consideration for evaluating network data transmission quality. To fully reflect the impact of extreme natural environments on wireless communication link quality and to provide a decision-making basis for subsequent reinforcement learning agents, this application employs the entropy weight method to objectively define the weights of key indicators constituting the state space. The basic principle of the entropy weight method is: the greater the dispersion (i.e., the greater the variance) of an indicator's data sequence, the smaller its information entropy, indicating that the indicator provides more effective information and should therefore be assigned a higher weight; conversely, if the value change of an indicator is very small, its information entropy is large, indicating that it provides less information and should be assigned a lower weight.

[0076] (1) First, standardization is performed using the extreme value processing method: For throughput and RSSI, higher values ​​are generally better. The following formula is used to standardize throughput and RSSI: Formula 8 For latency and packet loss rate, lower values ​​are better. The following formula is used to standardize latency and packet loss rate: Formula Nine in, It is the first The first sample The original values ​​of each indicator and They are the first The minimum and maximum values ​​of the index in all samples, the i-th Each sample is a set of QoS indicators collected by the agent within a single decision cycle.

[0077] (2) Calculate the number using the following formula. The first sample The proportion of each indicator value: Formula 10, where, The above proportions are given.

[0078] (3) Calculate the entropy value of each indicator using the following formula: Formula 11 in, Let j be the entropy value of the j-th index, a constant. And stipulates that when hour, .

[0079] (4) Calculate the coefficient of difference for each indicator using the following formula: , Formula 12, in, The coefficient of variation, The larger the value, the more significant the th... The more information a single indicator contains, the better.

[0080] (5) Normalize the difference coefficients using the following formula to obtain the final weight of each indicator: Formula Thirteen.

[0081] The output at this point represents the optimal QoS indicator weights for the current network state, ensuring that the reinforcement learning agent focuses on the most critical performance bottlenecks in the current network. These determined weights can then be directly applied to the reward function. By employing an objective QoS indicator weight determination method based on entropy weighting, network performance bottlenecks can be accurately located.

[0082] In some embodiments, determining the heuristic score corresponding to the communication link type using the agent based on each performance data corresponding to the communication link type includes: determining the heuristic score corresponding to the communication link type using a reward function in the agent based on each performance data corresponding to the communication link type.

[0083] In this embodiment, this application designs a multi-objective reward function based on link type differences, effectively reducing link collision probability and redundant signaling overhead, and optimizing channel resource utilization. The reward function is pre-designed according to the actual needs and performance indicators of the communication network, and it can quantitatively evaluate the performance of action combinations in terms of bandwidth, latency, and reliability. Based on multiple service quality index values ​​corresponding to each communication link type, the heuristic score corresponding to the communication link type is determined using the reward function in the agent. Through the evaluation mechanism of the reward function, it is ensured that the finally determined target action combination has reliable performance guarantees, providing users with high-quality communication services.

[0084] When designing the reward function, the communication states of other nodes within the range are considered to avoid local optima or greedy selection behavior caused by over-reliance on local states during reinforcement learning. In the wireless ad hoc network adopted in this application, each mesh network node acts as a reinforcement learning agent, handling not only relay link communication with neighboring mesh network nodes but also link scheduling with the covered terminal nodes (including mobile terminal devices and immobile fixed sensors). Different link types differ in the candidate mechanism stage; therefore, the reward function design should construct independent performance feedback mechanisms for each link type to support the agent in simultaneously outputting three types of actions, optimizing the MAC mechanism combination of the corresponding links.

[0085] The reward function is a link-type separated multi-objective reward function structure, meaning that each agent generates three types of actions at each training time step, and rewards are calculated separately for each of the three link types. At time step... When the agent observes the state And consider taking an action. At that time, the objective weight vector determined by the entropy weight method will be... Calculate the reward value The specific formula for the Mesh-to-Mesh link is as follows: Formula Fourteen in, To award points, The normalized throughput, The normalized delay, The normalized packet loss rate. The value is a normalized RSSI (Received Signal Strength Indicator) value. The weight corresponding to throughput, The weights corresponding to the delay. The weights corresponding to the packet loss rate. The weights corresponding to RSSI.

[0086] For the other two types of links (Mesh – mobile terminals, Mesh – fixed terminals), the reward function structure is completely identical, except that the neighbor set is replaced with the set of terminals covered by each link. Ultimately, each mesh network node agent will independently evaluate the communication performance of the three types of links at each time step, obtaining: R mesh t Neighbor Mesh communication link reward; R mobile t Mobile terminal link reward; R fixed tFixed terminal link reward.

[0087] In another embodiment provided in this application, the PPO algorithm is used to optimize the agent's policy, and a quantization model is invoked to guide the agent's exploration, thereby improving training convergence speed and training accuracy. The following steps are performed: a. Data Collection: At the beginning of each training cycle, the agent collects a series of empirical data by interacting with the environment, forming a trajectory. In this process, an action is selected at each time step. The process employs a guidance mechanism, in which... The communication status at the current moment is determined by environmental perception and is reflected by the service quality index value; For communication actions between nodes; As a reward point; This is the communication state determined by environmental perception at the next moment.

[0088] b. Policy Optimization: Using the collected trajectory data, the agent updates its policy network parameters by optimizing the PPO shearing agent objective function. : Formula 15 in, For the proxy objective function in the Proximal Policy Optimization (PPO) algorithm, For the empirical expectation operator, The probability ratio between the old and new strategies. This refers to the estimate of the dominance function obtained through methods such as generalized dominance estimation (GAE). The clipping threshold hyperparameter, clipping factor, controls the update magnitude. `clip` represents the clipping operation function, which compares the probability ratio... Restricted to the interval [1- ,1+ ]Inside.

[0089] c. Predictive guidance mechanism: Candidate action evaluation: in a given state Before making a final action choice, the agent first considers its current policy. Identify a subset of high-probability candidate actions. .

[0090] Generating heuristic scores: for candidate sets Each candidate action in The agent calls in the quantization model Enter the current status and actions Quickly obtain a quantized performance feature vector ,in, The predicted throughput is calculated using a theoretical model based on the selected MAC mechanism combination and the current network state. The predicted latency is calculated using a theoretical model based on the selected MAC mechanism combination and the current network state. The predicted packet loss rate is calculated using a theoretical model based on the selected MAC mechanism combination and the current network state.

[0091] Subsequently, the objective weight vector was determined using the entropy weight method. This performance feature vector is then transformed into a single heuristic score. : Formula Sixteen Among them, the score This represents a quantitative assessment of the intrinsic value of the action. This is the throughput after normalization. The normalized delay is... The packet loss rate is the normalized value. The weight corresponding to throughput, The weights corresponding to the delay. The weights corresponding to the packet loss rate.

[0092] Biased Sampling Distribution: The agent uses these heuristic scores to adjust (bias) the action sampling probabilities of its original policy. The guided action probability distribution is calculated. : Formula 17 in, It is a "temperature" hyperparameter used to control the guidance intensity. The higher the value, the more the exploration behavior tends to favor actions that the quantification model deems to have superior performance characteristics; conversely, the lower the value, the closer it is to original random exploration. This represents the initial action probability distribution.

[0093] Final action selection: Ultimately, the agent chooses from this intelligently guided new probability distribution. Randomly sample an action To execute.

[0094] d. Online deployment and updates: After system deployment, the agent continues to perform periodic online learning and policy fine-tuning to adapt to channel quality fluctuations caused by environmental degradation or topology changes over long periods, thereby achieving adaptive MAC mechanism selection and optimization.

[0095] The agent in this application can perform real-time evaluation and optimization through a reward function, ensuring the accuracy of the target action corresponding to each communication link type. By using a link-type-separated multi-objective reward function based on the PPO algorithm, a performance feedback mechanism is independently designed for each type of link. Combined with online policy updates and periodic fine-tuning, differentiated dynamic optimization is achieved in multi-link, heterogeneous terminal scenarios. This reduces link conflicts and resource waste while improving overall network efficiency in complex environments with dense power transmission lines.

[0096] In some embodiments, determining the performance data corresponding to each action of the communication link type based on the network environment data and using the quantization model includes: for each action corresponding to the communication link type, inputting a first sub-data from the network environment data corresponding to the action into a pre-trained throughput quantization model, determining and outputting the theoretical throughput value corresponding to the action through the throughput quantization model; inputting a second sub-data from the network environment data corresponding to the action into a pre-trained latency quantization model, determining and outputting the theoretical latency value corresponding to the action through the latency quantization model; inputting a third sub-data from the network environment data corresponding to the action into a pre-trained packet loss rate quantization model, determining and outputting the theoretical packet loss rate value corresponding to the action through the packet loss rate quantization model; and using the theoretical throughput value, theoretical latency value, and theoretical packet loss rate value corresponding to the action as the performance data corresponding to the action. In this embodiment, for each action corresponding to a communication link type, the first sub-data from the network environment data corresponding to the action is input into a pre-trained throughput quantization model. The throughput quantization model then determines and outputs the theoretical throughput value corresponding to the action. Throughput quantization model For CSMA / CA (Carrier Sense Multiple Access / Collision Avoidance) mechanisms: throughput is primarily affected by the collision probability. According to the Bianchi model, the collision probability... It is the number of competing nodes. and competition window size (From the selected action) The function of ). The probability of successful transmission. It can be represented as Therefore, the final throughput characteristic can be quantified as: Formula 18 in, It is the length of the data packet payload. , , These are the average times for successful transmission, collision, and channel idle for one time slot, respectively.

[0097] For TDMA (Time Division Multiple Access) mechanisms: its throughput characteristics are a direct function of resource allocation; if the action... Assigned to a certain link There are 1 time slot, and the total number of time slots per frame is 1 Each time slot is [duration] Then its theoretical throughput is: Formula Nineteen.

[0098] The second sub-data from the network environment data corresponding to the action is input into a pre-trained latency quantization model. The latency quantization model determines and outputs the theoretical latency value corresponding to the action. Latency quantization model The latency mainly consists of queuing latency, channel access latency, and transmission latency. For the CSMA / CA mechanism, channel access latency primarily depends on the backoff process. Therefore, the average access latency can be modeled as a function of the number of backoff slots and the number of collisions, and... and Relevant. For the TDMA mechanism: access latency mainly depends on the distance between the arrival time of the data packet and the next allocated time slot. On average, the access latency is approximately half the length of a TDMA frame.

[0099] The third sub-data from the network environment data corresponding to the action is input into a pre-trained packet loss rate quantization model. The packet loss rate quantization model determines and outputs the theoretical packet loss rate corresponding to the action. (Packet Loss Rate Quantization Model) Packet loss is the main collision. and channel error This is caused by... The quantitative characteristics of the total packet loss rate are: Formula 20, in which, The predicted packet loss rate caused by packet collisions. This represents the predicted packet loss rate caused by transmission errors in the wireless channel.

[0100] The theoretical values ​​of throughput, latency, and packet loss rate corresponding to an action are collectively used as the performance data for that action. Throughput quantization models, latency quantization models, and packet loss rate quantization models belong to the category of quantization models. Through quantization models, the agent can evaluate any candidate action at each decision point. It performs rapid performance change assessment. It takes an action and the current state as input and immediately obtains a quantified performance feature vector. , , Furthermore, during the training phase, this quantification result can be used to intelligently guide the agent's exploration direction, making it more inclined to choose actions with superior intrinsic performance characteristics, thereby improving learning efficiency.

[0101] It should be noted that the method in this embodiment can be executed by a single device, such as a computer or server. The method can also be applied in a distributed scenario, where multiple devices cooperate to complete the task. In such a distributed scenario, one of these devices may execute only one or more steps of the method in this embodiment, and the multiple devices will interact with each other to complete the method described.

[0102] It should be noted that the above description describes some embodiments of this application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in a different order than that shown in the above embodiments and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0103] Based on the same inventive concept, and corresponding to any of the above embodiments, this application also provides an adaptive MAC transmission line self-organizing network communication enhancement device.

[0104] refer to Figure 4 The adaptive MAC transmission line ad hoc network communication enhancement device, applied to each mesh network node in a wireless ad hoc network, includes: The acquisition module 10 is configured to acquire channel status data and network environment data corresponding to the mesh network node, and to acquire the communication link type between the mesh network node and other nodes communicating with it.

[0105] The first determining module 20 is configured to determine multiple service quality index values ​​corresponding to each communication link type based on the channel state data.

[0106] The filtering module 30 is configured to filter the communication actions between pre-stored mesh network nodes and other nodes communicating with them based on the network environment data, and obtain at least one filtering action.

[0107] The second determining module 40 is configured to determine the target action corresponding to each communication link type in all filtering actions based on the network environment data, each communication link type and its corresponding multiple service quality index values, using a pre-built intelligent agent.

[0108] The communication module 50 is configured to allow the mesh network nodes to communicate with other nodes corresponding to each communication link type in accordance with the target action corresponding to the communication link type.

[0109] Using the aforementioned apparatus, channel state data and network environment data corresponding to mesh network nodes are acquired, as well as communication link types between the mesh network node and other nodes communicating with it. Based on the channel state data, multiple quality of service (QoS) index values ​​corresponding to each communication link type are determined. Based on the network environment data, pre-stored communication actions between the mesh network node and other nodes communicating with it are filtered to obtain at least one filtering action. Based on the network environment data, each communication link type and its corresponding multiple QoS index values ​​are used by a pre-built intelligent agent to determine the target action corresponding to each communication link type from all filtering actions. The mesh network node communicates with other nodes corresponding to each communication link type according to the target action corresponding to the communication link type.

[0110] In some embodiments, the second determining module 40 is further configured to, for each communication link type, search for at least one predetermined action corresponding to the communication link type among all filtering actions; and, based on the network environment data and multiple service quality index values ​​corresponding to the communication link type, use the agent to determine the target action among at least one predetermined action corresponding to the communication link type.

[0111] In some embodiments, the second determining module 40 is further configured to: determine an initial predetermined action probability distribution corresponding to the communication link type using the agent based on multiple service quality index values ​​corresponding to the communication link type; generate a heuristic score corresponding to the communication link type using the agent and a pre-trained quantization model based on the network environment data; adjust the initial predetermined action probability distribution by bias using the agent based on the heuristic score to obtain a target predetermined action probability distribution; and randomly determine a target action corresponding to the communication link type from the predetermined actions corresponding to the target predetermined action probability distribution.

[0112] In some embodiments, the second determining module 40 is further configured to determine, based on the network environment data, the performance data corresponding to each action of the communication link type using the quantization model; and to determine, based on each performance data corresponding to the communication link type, the heuristic score corresponding to the communication link type using the agent.

[0113] In some embodiments, the first determining module 20 is further configured to, for each predetermined service quality indicator corresponding to each communication link type, determine an initial service quality indicator value corresponding to the predetermined service quality indicator and a weight corresponding to the predetermined service quality indicator based on sub-data in the channel state data corresponding to the predetermined service quality indicator; and determine the service quality indicator value corresponding to the predetermined service quality indicator corresponding to the communication link type based on the initial service quality indicator value and the weight.

[0114] In some embodiments, the second determining module 40 is further configured to determine a heuristic score corresponding to the communication link type based on each performance data corresponding to the communication link type using a reward function in the agent.

[0115] In some embodiments, the second determining module 40 is further configured to, for each action corresponding to the communication link type, input a first sub-data from the network environment data corresponding to the action into a pre-trained throughput quantization model, determine and output the theoretical throughput value corresponding to the action through the throughput quantization model; input a second sub-data from the network environment data corresponding to the action into a pre-trained latency quantization model, determine and output the theoretical latency value corresponding to the action through the latency quantization model; input a third sub-data from the network environment data corresponding to the action into a pre-trained packet loss rate quantization model, determine and output the theoretical packet loss rate value corresponding to the action through the packet loss rate quantization model; and use the theoretical throughput value, theoretical latency value, and theoretical packet loss rate value corresponding to the action as the performance data corresponding to the action.

[0116] For ease of description, the above devices are described in terms of function, divided into various modules. Of course, in implementing this application, the functions of each module can be implemented in one or more software and / or hardware.

[0117] The apparatus of the above embodiments is used to implement the corresponding adaptive MAC transmission line ad hoc network communication enhancement method in any of the foregoing embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0118] Based on the same inventive concept, corresponding to any of the above embodiments, this application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the adaptive MAC transmission line ad hoc network communication enhancement method as described in any of the above embodiments.

[0119] Figure 5This embodiment illustrates a more specific hardware structure of an electronic device. The device may include a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, memory 1020, input / output interface 1030, and communication interface 1040 are interconnected internally via the bus 1050.

[0120] The processor 1010 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0121] The memory 1020 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program code is stored in the memory 1020 and is called and executed by the processor 1010.

[0122] The input / output interface 1030 is used to connect input / output modules to realize information input and output. The input / output modules can be configured as components in the device (not shown in the figure) or externally connected to the device to provide corresponding functions. Input devices may include keyboards, mice, touch screens, microphones, various sensors, etc., and output devices may include displays, speakers, vibrators, indicator lights, etc.

[0123] The communication interface 1040 is used to connect a communication module (not shown in the figure) to enable communication between this device and other devices. The communication module can communicate via wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0124] Bus 1050 includes a pathway for transmitting information between various components of the device, such as processor 1010, memory 1020, input / output interface 1030, and communication interface 1040.

[0125] It should be noted that although the above-described device only shows the processor 1010, memory 1020, input / output interface 1030, communication interface 1040, and bus 1050, in specific implementations, the device may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the above-described device may only include the components necessary for implementing the embodiments of this specification, and not necessarily all the components shown in the figures.

[0126] The electronic devices described in the above embodiments are used to implement the corresponding adaptive MAC transmission line ad hoc network communication enhancement method in any of the foregoing embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0127] Based on the same inventive concept, corresponding to the methods of any of the above embodiments, this application also provides a non-transitory computer-readable storage medium storing computer instructions for causing the computer to execute the adaptive MAC transmission line ad hoc network communication enhancement method as described in any of the above embodiments.

[0128] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0129] The computer instructions stored in the storage medium of the above embodiments are used to cause the computer to execute the adaptive MAC transmission line ad hoc network communication enhancement method as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0130] Based on the same concept, corresponding to the methods of any of the above embodiments, this application also provides a computer program product, including computer program instructions, which, when run on a computer, cause the computer to execute the adaptive MAC transmission line ad hoc network communication enhancement method as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0131] It should be noted that the embodiments of this application can also be further described in the following ways: It is understood that before using the technical solutions of the various embodiments in this disclosure, users will be informed of the type, scope of use, and usage scenarios of the personal information involved in an appropriate manner, and user authorization will be obtained.

[0132] For example, upon receiving a user's active request, a prompt message is sent to the user to explicitly inform them that the requested operation will require the acquisition and use of the user's personal information. This allows the user to independently choose, based on the prompt message, whether to provide personal information to the software or hardware such as electronic devices, applications, servers, or storage media performing the operations of this disclosed technical solution.

[0133] As an optional but not limited implementation, in response to a user's active request, sending a prompt message to the user can be done via a pop-up window, where the prompt message can be presented in text format. Furthermore, the pop-up window can also include a selection control allowing the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0134] It is understood that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation of this disclosure. Other methods that comply with relevant laws and regulations may also be applied to the implementation of this disclosure.

[0135] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of this application is limited to these examples; under the concept of this application, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of the embodiments of this application as described above, which are not provided in detail for the sake of brevity.

[0136] Additionally, to simplify the description and discussion, and to avoid obscuring the embodiments of this application, the well-known power / ground connections to integrated circuit (IC) chips and other components may or may not be shown in the provided drawings. Furthermore, the apparatus may be shown in block diagram form to avoid obscuring the embodiments of this application, and this also takes into account the fact that the details of the implementation of these block diagram apparatuses are highly dependent on the platform on which the embodiments of this application will be implemented (i.e., these details should be fully understood by those skilled in the art). While specific details (e.g., circuits) have been set forth to describe exemplary embodiments of this application, it will be apparent to those skilled in the art that the embodiments of this application can be implemented without these specific details or with variations thereof. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0137] Although this application has been described in conjunction with specific embodiments thereof, many substitutions, modifications, and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may be used with the embodiments discussed.

[0138] The embodiments of this application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of this application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the embodiments of this application should be included within the protection scope of this application.

Claims

1. A method for enhancing communication in an adaptive MAC-based transmission line ad hoc network, characterized in that, Each mesh network node used in a wireless ad hoc network includes: Obtain channel status data and network environment data corresponding to the mesh network node, and obtain the communication link type between the mesh network node and other nodes communicating with it; Based on the channel state data, multiple service quality index values ​​corresponding to each communication link type are determined; Based on the network environment data, the communication actions between pre-stored mesh network nodes and other nodes communicating with them are filtered to obtain at least one filtered action; Based on the network environment data, for each communication link type and its corresponding multiple service quality index values, a pre-built intelligent agent is used to determine the target action corresponding to each communication link type in all filtering actions. The mesh network nodes communicate with other nodes corresponding to each communication link type according to the target action corresponding to the communication link type.

2. The method according to claim 1, characterized in that, Based on the network environment data, each communication link type and its corresponding multiple service quality index values, the pre-built intelligent agent determines the target action corresponding to each communication link type from all filtering actions, including: For each communication link type, find at least one predetermined action corresponding to the communication link type among all filtering actions; Based on the network environment data and multiple service quality index values ​​corresponding to the communication link type, the agent determines the target action from at least one predetermined action corresponding to the communication link type.

3. The method according to claim 2, characterized in that, The process of determining the target action using the agent within at least one predetermined action corresponding to the communication link type, based on the network environment data and multiple service quality index values ​​corresponding to the communication link type, includes: Based on multiple service quality index values ​​corresponding to the communication link type, the agent determines the initial predetermined action probability distribution corresponding to the communication link type. Based on the network environment data, the agent and a pre-trained quantization model are used to generate a heuristic score corresponding to the communication link type. Based on the heuristic score, the agent is used to bias and adjust the initial predetermined action probability distribution to obtain the target predetermined action probability distribution; The target action corresponding to the communication link type is randomly determined from the predetermined actions corresponding to the predetermined action probability distribution of the target.

4. The method according to claim 3, characterized in that, The step of generating a heuristic score corresponding to the communication link type based on the network environment data, using the agent and a pre-trained quantization model, includes: Based on the network environment data, the performance data corresponding to each action of the communication link type is determined using the quantization model. Based on the performance data corresponding to each communication link type, the agent determines the heuristic score corresponding to the communication link type.

5. The method according to claim 1, characterized in that, The determination of multiple quality of service (QoS) index values ​​corresponding to each communication link type based on the channel state data includes: For each predetermined quality of service index corresponding to each communication link type, based on the sub-data in the channel state data corresponding to the predetermined quality of service index, the initial quality of service index value corresponding to the predetermined quality of service index and the weight corresponding to the predetermined quality of service index are determined. Based on the initial service quality index value and weight corresponding to the predetermined service quality index, the service quality index value corresponding to the predetermined service quality index of the communication link type is determined.

6. The method according to claim 4, characterized in that, The step of determining the heuristic score corresponding to each communication link type using the agent based on each performance data corresponding to the communication link type includes: Based on the performance data corresponding to each communication link type, the heuristic score corresponding to the communication link type is determined using the reward function in the agent.

7. The method according to claim 4, characterized in that, The process of determining the performance data for each action corresponding to the communication link type using the quantization model based on the network environment data includes: For each action corresponding to the communication link type, the first sub-data in the network environment data corresponding to the action is input into the pre-trained throughput quantization model, and the theoretical throughput value corresponding to the action is determined and output through the throughput quantization model. The second sub-data in the network environment data corresponding to the action is input into a pre-trained latency quantization model, and the latency quantization model is used to determine and output the theoretical latency value corresponding to the action. The third sub-data in the network environment data corresponding to the action is input into the pre-trained packet loss rate quantization model, and the theoretical value of the packet loss rate corresponding to the action is determined and output through the packet loss rate quantization model. The theoretical throughput, theoretical latency, and theoretical packet loss rate corresponding to the action are used together as the performance data corresponding to the action.

8. A communication enhancement device for self-organizing power transmission lines with adaptive MAC, characterized in that, Each mesh network node used in a wireless ad hoc network includes: The acquisition module is configured to acquire channel status data and network environment data corresponding to the mesh network node, and to acquire the communication link type between the mesh network node and other nodes communicating with it. The first determining module is configured to determine multiple service quality index values ​​corresponding to each communication link type based on the channel state data; The filtering module is configured to filter the communication actions between pre-stored mesh network nodes and other nodes communicating with them based on the network environment data, and obtain at least one filtering action; The second determining module is configured to determine the target action corresponding to each communication link type in all filtering actions based on the network environment data, each communication link type and its corresponding multiple service quality index values, using a pre-built intelligent agent. The communication module is configured to allow the mesh network nodes to communicate with other nodes corresponding to each communication link type in accordance with the target action corresponding to the communication link type.

9. The apparatus according to claim 8, characterized in that, The second determining module is further configured to: For each communication link type, at least one predetermined action corresponding to the communication link type is searched among all filtering actions; based on the network environment data and multiple service quality index values ​​corresponding to the communication link type, the target action is determined by the agent among at least one predetermined action corresponding to the communication link type.

10. The apparatus according to claim 9, characterized in that, The second determining module is further configured to: Based on multiple service quality index values ​​corresponding to the communication link type, the agent determines the initial predetermined action probability distribution corresponding to the communication link type; based on the network environment data, the agent and a pre-trained quantization model generate a heuristic score corresponding to the communication link type; based on the heuristic score, the agent adjusts the bias of the initial predetermined action probability distribution to obtain a target predetermined action probability distribution; and randomly determines the target action corresponding to the communication link type from the predetermined actions corresponding to the target predetermined action probability distribution.

11. The apparatus according to claim 10, characterized in that, The second determining module is further configured to: Based on the network environment data, the performance data corresponding to each action of the communication link type is determined using the quantization model; based on each performance data corresponding to the communication link type, the heuristic score corresponding to the communication link type is determined using the agent.

12. The apparatus according to claim 8, characterized in that, The first determining module is further configured to: For each predetermined quality of service index corresponding to each communication link type, based on the sub-data in the channel state data corresponding to the predetermined quality of service index, the initial quality of service index value corresponding to the predetermined quality of service index and the weight corresponding to the predetermined quality of service index are determined. Based on the initial service quality index value and weight corresponding to the predetermined service quality index, the service quality index value corresponding to the predetermined service quality index of the communication link type is determined.

13. The apparatus according to claim 8, characterized in that, The second determining module is further configured to: Based on the performance data corresponding to each communication link type, the heuristic score corresponding to the communication link type is determined using the reward function in the agent.

14. The apparatus according to claim 13, characterized in that, The second determining module is also configured to: For each action corresponding to the communication link type, the first sub-data in the network environment data corresponding to the action is input into a pre-trained throughput quantization model, and the theoretical throughput value corresponding to the action is determined and output through the throughput quantization model; the second sub-data in the network environment data corresponding to the action is input into a pre-trained latency quantization model, and the theoretical latency value corresponding to the action is determined and output through the latency quantization model. The third sub-data in the network environment data corresponding to the action is input into the pre-trained packet loss rate quantization model, and the theoretical value of the packet loss rate corresponding to the action is determined and output through the packet loss rate quantization model. The theoretical throughput, theoretical latency, and theoretical packet loss rate corresponding to the action are used together as the performance data corresponding to the action.

15. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1 to 7.

16. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to perform the method described in any one of claims 1 to 7.

Images