Air-sea cross-domain network routing protocol method based on fuzzy logic and q-learning optimization

Abstract

The application discloses a kind of air-sea cross-domain network routing protocol methods based on fuzzy logic and learning optimization, belong to air-sea cross-domain network communication technical field, including: each node in network is regarded as independent intelligent agent, and candidate forwarding set is constructed by maintaining neighbor node information to narrow action space;Introduce fuzzy inference system, design reward function by comprehensively remaining energy factor, node advance factor and link quality factor three factors, update value by combining algorithm iteration, select optimal next hop node to complete data packet forwarding, and set reselection mechanism for forwarding failure.The present application significantly improves the data packet delivery rate, reduces end-to-end delay, while prolonging the network life cycle, suitable for high reliability data transmission in complex air-sea cross-domain environment.

Description

Technical Field

[0001] This invention belongs to the field of air-sea cross-domain network communication technology, specifically relating to a method based on fuzzy logic and... Learn optimized air and sea cross-domain network routing protocol methods. Background Technology

[0002] Air-Sea Cross-Domain Network (ASCDN) is a three-dimensional network system that integrates various heterogeneous nodes such as aerial drones, surface ships, and underwater sensors. It is widely used in marine environmental monitoring, underwater resource exploration, and maritime search and rescue. In ASCDN, data is generated from underwater sensor nodes, undergoes multi-hop forwarding through these underwater nodes, and is finally transmitted to the aerial nodes via surface buoy nodes. Due to the unique characteristics of the marine environment, the communication link quality between nodes is extremely unstable, and node energy is limited. Therefore, designing an efficient and reliable routing protocol is a core challenge for ASCDN.

[0003] Traditional routing protocols are mainly divided into two categories: proactive routing and on-demand routing. Neither can adapt to the complex environment of cross-domain networks in the air and sea. Proactive routing protocols (such as DSDV and OLSR) maintain a global routing table by periodically exchanging routing information. However, in an underwater environment, frequent broadcasts consume a lot of energy and are difficult to adapt to dynamically changing link quality. On-demand routing protocols (such as AODV and DSR) only establish routes when needed, but the route discovery process may lead to high end-to-end latency. In addition, these traditional routing protocols usually select paths based solely on distance or hop count, ignoring the remaining energy of nodes and real-time link quality. This can easily lead to some nodes prematurely exhausting their energy or choosing links with poor quality, resulting in packet loss and retransmission, thereby reducing network delivery rate and increasing latency.

[0004] In recent years, reinforcement learning has been introduced into routing protocol design, where nodes act as agents to learn optimal forwarding strategies through interaction with the environment. However, existing methods based on... The learning-based routing protocol still has significant shortcomings: First, the reward function is overly simplified, considering only single factors such as distance and hop count, failing to comprehensively reflect the nonlinear coupling relationship between multiple factors such as node energy, link quality, and transmission distance, resulting in low accuracy in routing decisions; second, The learning action space is too large, the algorithm converges slowly, and it cannot adapt to the dynamic changes in the underwater network topology; thirdly, there is a lack of effective forwarding failure response mechanism, which can easily cause data packet loss when the link is interrupted, further reducing the reliability of transmission.

[0005] Therefore, there is an urgent need for an efficient routing method that can comprehensively consider multiple factors, converge quickly, and be highly adaptable, in order to solve the problems of high latency and low delivery rate caused by unreasonable transmission paths and frequent link interruptions in the existing ASCDN. Summary of the Invention

[0006] To address the shortcomings of the existing technologies, the purpose of this invention is to propose a method based on fuzzy logic and... We learn an optimized air-sea cross-domain network routing protocol method to achieve multi-factor fusion routing decisions, reduce the reinforcement learning action space, accelerate algorithm convergence, improve packet delivery rate, reduce end-to-end latency, and extend network lifetime, thus adapting to complex and dynamic air-sea cross-domain communication environments.

[0007] To achieve the above-mentioned objectives, this invention is applied to a heterogeneous node air-sea cross-domain network composed of underwater sensors, submersibles, buoys, UAVs, surface vessels, and AUVs, treating each node as an independent intelligent agent. The technical solution adopted by this invention is as follows:

[0008] Step 1: Each node maintains a local neighbor information table and periodically collects the status information of neighboring nodes;

[0009] Step 2: When the current node sends a data packet to the destination node, the current node constructs a candidate forwarding set based on the neighbor information table and multiple filtering conditions. ;

[0010] Step 3: Based on the current node and candidate forwarding set For each neighboring node, three factors are defined, and a fuzzy inference system is used to calculate the immediate reward value for performing a forwarding action; the three factors include the node's remaining energy factor. Node forward factor and link quality factor The fuzzy reasoning system comprises three stages: fuzzification, fuzzy rule reasoning, and defuzzification.

[0011] Step 4: After completing the construction of the candidate forwarding set and the calculation of the immediate reward value, based on The algorithm updates the current node's information to its neighboring nodes based on the immediate reward value. Value, and select The neighbor node with the largest value is used as the next-hop forwarding node;

[0012] Step 5: The current node forwards the data packet to the selected next-hop forwarding node and updates the status information in the data packet;

[0013] Step 6: The current node forwards the data packet to the next-hop forwarding node and then confirms the forwarding result. If the forwarding is successful, the process ends; if the forwarding fails, the next-hop forwarding node is removed, and a new node is selected. The node with the largest value forwards the message until the forwarding is successful or the candidate forwarding set is empty.

[0014] Preferably, step 1 is as follows:

[0015] After each node starts up, it periodically broadcasts Hello messages while listening for Hello messages from surrounding nodes, establishes and maintains a local neighbor information table, and continuously collects and updates the status information of neighboring nodes in real time. When a node listens for any data packet, it extracts the node's status information from it and completes the dynamic update of the neighbor information table.

[0016] Preferably, the process of constructing the candidate forwarding set in step 2 is as follows:

[0017] The multi-dimensional filtering conditions include neighbor node set, remaining energy set, depth set, and set merging;

[0018] Neighbor set: Set the current node Within the communication radius Each neighbor node is considered as a set of neighbor nodes. , The formula for representing is:

[0019]

[0020] in, Indicates the communication range One neighboring node;

[0021] Remaining energy set: Filtering values ​​greater than the remaining energy ratio threshold The current node is times the current node. Neighboring nodes with remaining energy Surplus energy constitutes the surplus energy set , The formula for expressing it is:

[0022]

[0023] in, Indicates the current node The remaining energy set, Representing neighboring nodes The remaining energy set;

[0024] Depth set: Selecting shallower neighbor nodes to form a depth set The formula is:

[0025]

[0026] in, Indicates the current node depth set, Representing neighboring nodes Depth set;

[0027] Set merging: Select neighboring nodes with higher remaining energy and shallower depth within the communication radius to form a candidate forwarding set. , The defining formula is:

[0028]

[0029] in, This represents neighbor nodes that meet the multi-dimensional filtering criteria;

[0030] like If it is an empty set, then increase the threshold for the ratio of remaining energy. Or relax the depth requirements.

[0031] Preferably, step 3 is implemented as follows:

[0032] The node's remaining energy factor Let be the ratio of the remaining energy of the current node to its initial energy, expressed as: ,in, This represents the remaining energy of the current node; The initial energy of the current node;

[0033] The node forward factor The expression is ,in, Indicates the destination node. Indicates the current node to the destination node distance, Representing neighboring nodes to the destination node distance, Indicates the current node Maximum communication radius;

[0034] The link quality factor For the current node to neighboring nodes The data packet transmission success rate is calculated based on the underwater acoustic channel model, and its expression is: ,in, This indicates that an m-bit data packet is located at a distance of... The transmission success rate;

[0035] Fuzzification: Using a triangular membership function, the remaining energy factors of the nodes are respectively... Node forward factor and link quality factor The average numerical value is mapped to a fuzzy linguistic variable, specifically: the residual energy factor. The fuzzy linguistic variables are low, medium, and high, and the node forward factor is... The fuzzy linguistic variables are small, medium, and large, and the link quality factor is... The fuzzy linguistic variables are poor, average, and excellent;

[0036] Fuzzy rule reasoning: A fuzzy rule base containing 27 IF-THEN rules is constructed based on expert experience. The fuzzy rule base includes all combinations of fuzzy linguistic variables of three factors. Fuzzy rule reasoning is performed on the fuzzy rule base to obtain the fuzzy linguistic result of the reward value. The fuzzy linguistic result of the reward value is very low, low, relatively low, medium, relatively high, high, and very high.

[0037] Defuzzification: The centroid method is used to defuzzify the fuzzy language results of the reward value, converting them into precise instant reward values. The formula for calculating instant reward value is:

[0038]

[0039] in, This represents the candidate values ​​for the reward. This represents the membership function.

[0040] Preferably, in step 4 The formula for updating the value is:

[0041] After completing the construction of the candidate forwarding set and the calculation of the immediate reward value, based on The algorithm updates the current node's information to its neighboring nodes based on the instantaneous reward value calculated by the fuzzy inference system. The value and the selection of the next-hop forwarding node, the The formula for updating the value is:

[0042] ;

[0043] in, Indicates the learning rate, This represents the immediate reward value calculated using a fuzzy logic reasoning system. Indicates the loss factor. Indicates the new after the update value, Indicates the old version before the update. value, Represented as the maximum value, Indicates the next-hop forwarding node;

[0044] In the initial state, build a local The table, initially all All values ​​are set to 0; After the value update is complete, select the candidate forwarding set. The neighbor node with the largest value is selected as the next-hop forwarding node.

[0045] Preferably, step 5 specifically includes the following:

[0046] The current node encapsulates the data packet according to a preset data packet format, which includes the following fields: data packet ID, source node address, destination node address, and next-hop forwarding node address. The data packet is encapsulated with the following information: value, initial energy, remaining energy, current node location information, destination node location information, and data payload. After encapsulation, the data packet is forwarded to the selected next-hop forwarding node.

[0047] Preferably, step 6 specifically includes the following:

[0048] After the current node forwards the data packet to the selected next-hop forwarding node, it enters the forwarding result confirmation phase, which proceeds as follows: Immediately after sending the data packet, the current node starts a timer, waiting for an confirmation message from the next-hop forwarding node. If an confirmation message is received from the next-hop forwarding node before the timer expires, the forwarding is considered successful, and the current node's forwarding process ends. If no confirmation message is received after the timer expires, the forwarding is considered a failure, the next-hop forwarding node is removed from the candidate forwarding set, and a new node is selected from the candidate forwarding set. The second-best node with the largest value is used as the new next-hop forwarding node for forwarding. This process is repeated until success is achieved or the candidate forwarding set is empty. If the candidate forwarding set is empty (there are no available next-hop forwarding nodes), the current data packet is discarded.

[0049] Preferably, the method further includes a neighbor monitoring and information update process executed throughout the entire process:

[0050] When any node listens to any data packet sent by a neighboring node and extracts the status information of the data packet, it updates the information in the local neighbor information table in real time. If the node is not the destination node, it acts as a relay node and repeats the process of steps 2 to 6 to forward the data packet until it reaches the destination node. If the node is the destination node, the current node is the selected next-hop forwarding node.

[0051] Preferably, the system includes multiple heterogeneous nodes, each node acting as an independent intelligent agent, and each node has the following built-in functional modules:

[0052] Neighbor information collection module: used to periodically broadcast Hello messages and receive feedback messages from neighboring nodes, collect status information of neighboring nodes, and establish and maintain a local neighbor information table in real time;

[0053] Candidate forwarding set construction module: used to filter nodes from the neighbor information table and construct a candidate forwarding set when a node needs to forward data packets, based on communication radius, remaining energy threshold and depth conditions;

[0054] Fuzzy reasoning reward calculation module: It has a built-in triangular membership function and a 27-rule fuzzy rule library, which is used to fuzzify, reason about, and defuzzify the remaining energy factor, node forward factor, and link quality factor to calculate the accurate instant reward value.

[0055] Learning Decision Module: Used to maintain local... The table is iteratively updated based on the immediate reward value according to the update formula. Value, and select The neighbor node with the largest value is selected as the optimal next-hop forwarding node.

[0056] Packet forwarding module: It is used to encapsulate data packets, fill in fields and send them, start a timer to listen for confirmation messages, and execute node removal and suboptimal node reselection logic when forwarding fails.

[0057] Compared with existing technologies, this invention proposes a method based on fuzzy logic and... The study explores an optimized air-sea cross-domain network routing protocol method, the advantages of which are:

[0058] The beneficial effects of this invention are as follows:

[0059] 1. By introducing a fuzzy inference system to design a reward function, the three key factors of node remaining energy, forward factor and link quality are comprehensively considered, which can more accurately evaluate the quality of forwarding actions and avoid the problem of unreasonable paths caused by a single factor.

[0060] 2. The construction of the candidate forwarding set has been narrowed down. The learning action space accelerates the algorithm's convergence speed, while energy and depth filtering eliminate nodes with poor forwarding potential in advance, improving the efficiency of routing decisions.

[0061] 3. Adopt The learning framework allows each node to learn independently without requiring global information, exhibiting excellent distributed adaptability and the ability to cope with dynamic changes in the underwater environment.

[0062] 4. It has a reselection mechanism when forwarding fails, which improves the reliability of data transmission, reduces packet loss caused by link interruption, thereby effectively improving the data packet delivery rate and reducing end-to-end latency, and extending the network life cycle. Attached Figure Description

[0063] Figure 1This is a diagram illustrating the air-sea cross-domain network scenario of the air-sea cross-domain network in this invention.

[0064] Figure 2 This invention is based on fuzzy logic and A schematic diagram of the optimized air-sea cross-domain network routing protocol process;

[0065] Figure 3 This is a schematic diagram of the fuzzy logic system in this invention;

[0066] Figure 4 This is a schematic diagram illustrating the principle of node forward factor calculation in this invention;

[0067] Figure 5 This is a graph showing the input-output membership functions of the fuzzy logic system in this invention.

[0068] Figure 6 This is a schematic diagram of the data packet format in this invention. Detailed Implementation

[0069] The technical solutions of the embodiments of this application will be further clearly and completely described below with reference to the accompanying drawings. It should be noted that the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0070] To make the inventive objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be further described in detail below with reference to the accompanying drawings: In order to better understand the above-mentioned objectives, features, and advantages of this invention, the advantages of this invention will be further illustrated below by comparing the embodiments with the accompanying drawings and specific implementation methods.

[0071] This invention proposes a method based on fuzzy logic and This paper studies and explains in detail the optimized air-sea cross-domain network routing protocol method.

[0072] This embodiment is applied to a heterogeneous node air-sea cross-domain network composed of underwater sensors, submersible buoys, buoys, drones, surface vessels, and AUVs, treating each node as an independent intelligent agent, such as... Figure 1 and Figure 2 As shown, the details are as follows:

[0073] Step 1: Each node maintains a local neighbor information table and periodically collects the status information of neighboring nodes;

[0074] Specifically, step 1 is as follows:

[0075] After each node starts up, it periodically broadcasts Hello messages while simultaneously listening for Hello messages from surrounding nodes, establishing and maintaining a local neighbor information table; it also continuously collects and updates the status information of neighboring nodes in real time. This status information includes the neighboring node's ID, initial energy, remaining energy, location coordinates, historical link quality, and most recently updated... This ensures that the neighbor node data obtained by the node is accurate and up-to-date; when the node listens to any data packet (including Hello messages, data packets, etc.), it extracts the node's status information from it and completes the dynamic update of the neighbor information table.

[0076] Step 2: When the current node sends a data packet to the destination node, the current node constructs a candidate forwarding set based on the neighbor information table and multiple filtering conditions. This eliminates neighboring nodes with poor forwarding performance, thus narrowing down the scope. The algorithm's action space helps increase the probability of selecting nodes with strong forwarding potential during training, thereby improving training efficiency and forwarding reliability.

[0077] Specifically, the process of constructing the candidate forwarding set in step 2 is as follows:

[0078] The multi-dimensional filtering conditions include neighbor node set, remaining energy set, depth set, and set merging;

[0079] Neighbor set: Set the current node Within the communication radius Each neighbor node is considered as a set of neighbor nodes. , The formula for representing is:

[0080]

[0081] in, Indicates the communication range One neighboring node;

[0082] Remaining energy set: Filtering values ​​greater than the remaining energy ratio threshold The current node is times the current node. Neighboring nodes with remaining energy Surplus energy constitutes the surplus energy set This ensures that the selected nodes have sufficient energy for data transmission, preventing forwarding interruptions or network fragmentation due to premature power depletion. The formula for expressing it is:

[0083]

[0084] in, Indicates the current node The remaining energy set, Representing neighboring nodes The remaining energy set;

[0085] Depth set: Selecting shallower neighbor nodes to form a depth set That is, neighboring nodes The depth is less than the current node If the depth (i.e., closer to the water surface) is greater, then neighboring nodes are retained. The depth is determined, otherwise it is discarded, as expressed by the formula:

[0086]

[0087] in, Indicates the current node depth set, Representing neighboring nodes Depth set;

[0088] Set merging: Select neighboring nodes with higher remaining energy and shallower depth within the communication radius to form a candidate forwarding set. , The defining formula is:

[0089]

[0090] in, This represents neighbor nodes that meet the multi-dimensional filtering criteria;

[0091] like If the set is empty (i.e., there are no neighboring nodes that satisfy all conditions), then the remaining energy ratio threshold is increased. Alternatively, the depth requirement could be relaxed to ensure that the data forwarding process is not interrupted.

[0092] Step 3: Based on the current node and candidate forwarding set For each neighboring node, three factors are defined, and a fuzzy inference system is used to calculate the immediate reward value for performing a forwarding action; the three factors include the node's remaining energy factor. Node forward factor and link quality factor The fuzzy reasoning system comprises three stages: fuzzification, fuzzy rule reasoning, and defuzzification.

[0093] Specifically, step 3, as Figure 3 As shown, the specific implementation is as follows:

[0094] The node's remaining energy factor This is the ratio of the current node's remaining energy to its initial energy, reflecting the node's energy endurance. The expression is: ,in, This represents the remaining energy of the current node; The initial energy of the current node;

[0095] The node forward factor To measure how close a neighboring node is spatially to the destination node relative to the current node, and to guide data packets toward the target, the expression is: ,in, Indicates the destination node. Indicates the current node to the destination node distance, Representing neighboring nodes to the destination node distance, Indicates the current node Maximum communication radius;

[0096] Assuming a water surface buoy node The position coordinates are That is, the destination node. underwater relay node The position coordinates are That is, a forwarding node. Underwater relay node The position coordinates are This is the relay node. underwater relay node A neighboring node within the transmission range Received from underwater relay node Data packets;

[0097] Target node With forwarding nodes The distance between them can be expressed as:

[0098]

[0099] relay node With the target node The distance between them can be expressed as:

[0100]

[0101] Then forwarding node To relay node The forward factor expression can be represented as:

[0102]

[0103] in, Indicates the maximum communication radius of the node;

[0104] The distance relationship between nodes is as follows Figure 4 As shown in the figure, the relay node and nodes All are forwarding nodes The neighboring nodes; the destination node in the graph Centered on the circle, destination node To relay node Draw an arc with the straight-line distance as the radius. This arc is marked with a red dashed line and is parallel to the straight line. Intersect at one point (marked in red); forwarding node The length of the line segment to the intersection point (the solid red line in the diagram) is the forwarding node. The forward factor before normalization; similarly, the length of the blue solid line corresponds to the node. The forward factor before normalization; it can be seen that the larger the forward factor from the forwarding node to the relay node, the closer its spatial location is to the destination node, which helps to reduce the number of transmission hops, reduce end-to-end latency, and improve overall forwarding efficiency.

[0105] The link quality factor For the current node to neighboring nodes The data packet transmission success rate, reflecting the reliability of the link, is calculated based on the underwater acoustic channel model, and its expression is: ,in, This indicates that the mbits data packets are at a distance of The transmission success rate;

[0106] Specifically, the attenuation of the underwater acoustic channel model consists of two parts: propagation spread loss and absorption loss, calculated as follows:

[0107]

[0108] In the formula, This represents the total attenuation. Indicates the distance the data packet was transmitted. This represents the energy diffusion factor, which is typically taken as 2. It is the carrier frequency. Indicates the energy absorption coefficient. Indicates the geometric diffusion coefficient;

[0109] in, The calculation formula is:

[0110]

[0111] As can be seen from the above formula, the magnitude of the energy absorption coefficient is only related to the signal frequency;

[0112] Furthermore, considering that underwater noise is mainly divided into four categories: turbulence noise, ship noise, wave noise, and thermal noise, respectively, we can use: , , and The total environmental noise power spectral density is expressed as follows: The calculation formula is:

[0113]

[0114] Calculate a certain distance by combining underwater acoustic channel attenuation and environmental noise. Signal-to-noise ratio at the receiving end The calculation formula is:

[0115]

[0116] In the formula, This refers to the transmitting power of the underwater acoustic transducer. This represents the noise bandwidth at the receiving end.

[0117] Therefore, when a given transmission distance An optimal working center frequency can be found. This allows the transmission power to be appropriately reduced, thereby saving energy, while maintaining a certain signal-to-noise ratio.

[0118] For underwater acoustic communication, the modulation model uses binary phase shift keying (BPSK). Based on the Rayleigh channel model, the two phase distances can be estimated. Bit error rate of inter-node underwater acoustic communication and transmission byte transmission success rate The calculation formula is as follows:

[0119]

[0120] In the formula, Indicates the bit error rate;

[0121] Blurring: such as Figure 5 As shown, for the node's remaining energy factor Node forward factor and link quality factor All three factors are mapped to fuzzy linguistic variables using a triangular membership function, specifically: residual energy factor. The fuzzy linguistic variables are low, medium, and high, and the node forward factor is... The fuzzy linguistic variables are small, medium, and large, and the link quality factor is... The fuzzy linguistic variables are poor, medium, and excellent, realizing the conversion from precise quantities to fuzzy quantities;

[0122] Fuzzy rule reasoning: A fuzzy rule base containing 27 IF-THEN rules is constructed based on expert experience. The fuzzy rule base includes all combinations of fuzzy linguistic variables of three factors (composed of 3 factors and 3 linguistic variables). Fuzzy rule reasoning is performed on the fuzzy rule base to obtain the fuzzy linguistic result of the reward value. The fuzzy linguistic result of the reward value is very low, low, relatively low, medium, relatively high, high, and very high, which comprehensively reflects the quality of the forwarding action.

[0123] The fuzzy rule base was determined by expert experience and consists of 27 rules, as shown in Table 1:

[0124] Table 1. Fuzzy Logic Rule Base for Human Experience Reward Values

[0125]

[0126] Defuzzification: The centroid method is used to defuzzify the fuzzy linguistic results of the reward value, converting them into precise instantaneous reward values. ,for The value update provides a quantitative basis; the formula for calculating the instant reward value is as follows:

[0127]

[0128] in, This represents the candidate values ​​for the reward. This represents the membership function.

[0129] Step 4: After completing the construction of the candidate forwarding set and the calculation of the immediate reward value, based on The algorithm updates the current node's information to its neighboring nodes based on the immediate reward value. Value, and select The neighbor node with the largest value is used as the next-hop forwarding node;

[0130] Specifically, in step S4 The formula for updating the value is:

[0131] After completing the construction of the candidate forwarding set and the calculation of the immediate reward value, based on The algorithm updates the current node's information to its neighboring nodes based on the instantaneous reward value calculated by the fuzzy inference system. The value and the selection of the next-hop forwarding node, where, The value represents the "long-term benefit score" for selecting a neighbor as the next hop. The formula for updating the value is:

[0132]

[0133] in, This represents the learning rate. This represents the immediate reward value calculated using a fuzzy logic reasoning system. This represents the discount factor. Indicates the new after the update value, Indicates the old version before the update. value, Represented as the maximum value, Representing neighboring nodes The next hop forwarding node;

[0134] In the initial state, build a local The table, initially all All values ​​are set to 0, indicating that no forwarding experience has been accumulated yet; After the value update is complete, select the candidate forwarding set. The neighbor node with the largest value is used as the next-hop forwarding node to achieve the optimal routing decision.

[0135] Step 5: When a node forwards a data packet to the selected next-hop forwarding node, it updates the status information in the data packet.

[0136] Specifically, step 5 includes the following:

[0137] The current node encapsulates the data packet according to the preset data packet format. Encapsulate the data packet, filling in the packet ID, source address, destination address, and next-hop address, and then set the current... Values, Nodes The remaining energy, location information, etc., are written into the corresponding fields of the data packet, along with the actual data payload; the data packet format is as follows: Figure 6 As shown; the packet format for routing forwarding includes the following:

[0138] Packet ID: A globally unique identifier used to distinguish different packets and prevent duplicate processing;

[0139] Source node address: The original node address that generated this data packet;

[0140] Next-hop node address: The address of the node selected as the next hop during the transmission of the data packet;

[0141] Destination node address: The address of the final destination node to which the data packet is to be delivered;

[0142] Value: The candidate neighbor nodes calculated by the current node. The value is used to decide the next hop;

[0143] Remaining Energy: The current remaining energy of neighboring nodes, reflecting their ability to continue working;

[0144] Location information: The geographical location information of the current neighboring nodes, used to determine the node's location in the network and its relative position with other nodes;

[0145] Distance from the buoy: The distance between the current node and the buoy, used to determine whether the data packet can move forward effectively;

[0146] Data payload: The actual data content to be transmitted;

[0147] node The communication module forwards data packets to the selected next-hop forwarding node.

[0148] Step 6: The current node forwards the data packet to the next-hop forwarding node and then confirms the forwarding result. If the forwarding is successful, the process ends; if the forwarding fails, the next-hop forwarding node is removed, and a new node is selected. The node with the largest value is forwarded until forwarding is successful or the candidate forwarding set is empty;

[0149] Specifically, step 6 includes the following:

[0150] After the current node forwards the data packet to the selected next-hop forwarding node, it enters the forwarding result confirmation phase, which proceeds as follows: Immediately after sending the data packet, the current node starts a timer, waiting for an confirmation message from the next-hop forwarding node to verify whether the data packet was successfully delivered. If an confirmation message is received from the next-hop forwarding node before the timer expires, the forwarding is considered successful, and the current node's forwarding process ends. If no confirmation message is received after the timer expires, the forwarding is considered a failure, the next-hop forwarding node is removed from the candidate forwarding set, and a new node is selected from the candidate forwarding set. The second-best node with the largest value is used as the new next-hop forwarding node for forwarding. This process is repeated until success or the candidate forwarding set is empty. If the candidate forwarding set is empty (there are no available next-hop forwarding nodes), the current data packet is discarded to avoid invalid transmission.

[0151] Specifically, the method also includes a neighbor monitoring and information update process executed throughout the entire process:

[0152] When any node listens to any data packet sent by a neighboring node (including data packets sent by a neighboring node to other nodes and Hello messages broadcast by a neighboring node), it extracts the status information of the data packet and updates the remaining energy, location, link quality, and other information in the local neighbor information table in real time to ensure that the neighbor information always reflects the node's current real status; if the node is not the destination node, it acts as a relay node and repeats the process of steps 2 to 6 to forward the data packet until it reaches the destination node; if the node is the destination node, the current node is the selected next-hop forwarding node.

[0153] This embodiment also provides an air-sea cross-domain network routing system for implementing the method of the embodiment. The system consists of heterogeneous nodes of underwater sensors, underwater buoys, buoys, UAVs, surface vessels, and AUVs. Each node is an independent intelligent agent, and each node has a built-in neighbor information collection module, a candidate forwarding set construction module, and a fuzzy inference reward calculation module. The learning decision-making module and the data packet forwarding module are implemented in hardware by embedding a chip with corresponding software programs. The functions of each module are as follows:

[0154] Neighbor information collection module: used to periodically broadcast Hello messages and receive feedback messages from neighboring nodes, collect status information of neighboring nodes, and establish and maintain a local neighbor information table in real time;

[0155] Candidate forwarding set construction module: Reads neighbor information table data, automatically calculates candidate forwarding sets according to communication radius, remaining energy threshold, and depth conditions; if the candidate set is empty, it automatically adjusts the threshold and re-filters.

[0156] Fuzzy inference reward calculation module: Built-in triangular membership function and 27 fuzzy rule base, input Then, a fuzzification, rule-based reasoning, and defuzzification process is performed to output a precise, real-time reward value.

[0157] Learning decision module: Uses an array structure to maintain local... The table is automatically updated iteratively based on the instant reward value according to a preset formula. Values, automatically filtered The neighbor node with the largest value is selected as the optimal next-hop forwarding node.

[0158] Packet forwarding module: Write a packet encapsulation program to fill in fields according to a preset format; implement packet sending through the underwater acoustic communication module; write timer and listening programs to implement forwarding failure judgment; write node removal and reselection programs to automatically execute reselection logic when forwarding fails.

[0159] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0160] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method based on fuzzy logic and A learning-optimized routing protocol method for air-sea cross-domain networks is applied to a heterogeneous node air-sea cross-domain network composed of underwater sensors, submersibles, buoys, UAVs, surface vessels, and AUVs. Its key feature is that... The method includes the following steps: Step 1: Each node maintains a local neighbor information table for neighboring nodes and periodically collects the status information of neighboring nodes; Step 2: When the current node sends a data packet to the destination node, the current node constructs a candidate forwarding set based on the neighbor information table and a multi-dimensional set. ; Step 3: Based on the current node and candidate forwarding set For each neighboring node in the algorithm, three factors are defined; based on these factors, a fuzzy inference system is used to calculate the immediate reward value for performing a forwarding action; the three factors include the node's remaining energy factor. Node forward factor and link quality factor The fuzzy inference system comprises three stages: fuzzification, fuzzy rule inference, and defuzzification. The node's remaining energy factor Let be the ratio of the remaining energy of the current node to its initial energy, expressed as: ,in, This represents the remaining energy of the current node. The initial energy of the current node; The node forward factor The expression is ,in, Indicates the destination node. Indicates the current node to the destination node distance, Representing neighboring nodes to the destination node distance, Indicates the current node Maximum communication radius; The link quality factor For the current node to neighboring nodes The data packet transmission success rate is calculated based on the underwater acoustic channel model, and its expression is: ,in, This indicates that an m-bit data packet is located at a distance of... The transmission success rate; Step 4: Based on The algorithm updates the current node's information to its neighboring nodes based on the immediate reward value. Value, and select The neighbor node with the largest value is used as the next-hop forwarding node; Step 5: The current node forwards the data packet to the selected next-hop forwarding node and updates the status information in the data packet; Step 6: After forwarding the data packet to the next-hop forwarding node, the current node confirms the forwarding result. If the forwarding is successful, the current forwarding ends; if the forwarding fails, the next-hop forwarding node is removed, and a new node is selected. The neighbor node with the largest value is used as the new next-hop forwarding node for forwarding until forwarding is successful or the candidate forwarding set is empty.

2. A method based on fuzzy logic and... as described in claim 1 The method for learning and optimizing air-sea cross-domain network routing protocols is characterized by, Step 1 is as follows: After each node starts up, it periodically broadcasts Hello messages, listens for Hello messages from surrounding nodes, establishes and maintains a local neighbor information table, and continuously collects and updates the status information of neighboring nodes in real time.

3. A method based on fuzzy logic and... as described in claim 1 The method for learning and optimizing air-sea cross-domain network routing protocols is characterized by, The process of constructing the candidate forwarding set in step 2 is as follows: Step 2.1 Design a multi-dimensional set, including a neighbor node set, a remaining energy set, and a depth set; Neighbor set: Set the current node Within the communication radius Each neighbor node is considered as a set of neighbor nodes. , The formula for representing is: , in, Indicates the communication range One neighboring node; Remaining Energy Set: Select a neighboring node whose remaining energy is at least equal to that of the current node. Remaining energy The remaining energy set consists of a number of neighboring nodes. , The formula for expressing it is: ， in, Indicates the current node The remaining energy, Representing neighboring nodes The remaining energy; Depth set: Select the shallower neighbor nodes to form the depth set. The formula is: , in, Indicates the current node depth, Representing neighboring nodes Depth set; Step 2.2 Set Merging: Take the intersection of the neighbor node set, the remaining energy set, and the depth set to form the candidate forwarding set. , The defining formula is: , in, This represents a neighbor node that satisfies a multi-dimensional set.

4. A method based on fuzzy logic and... as described in claim 1 The method for learning and optimizing air-sea cross-domain network routing protocols is characterized by, In step 3, the specific process of the fuzzy inference system is as follows: Fuzzification: Using a triangular membership function, the remaining energy factors of the nodes are respectively... Node forward factor and link quality factor The average numerical value is mapped to a fuzzy linguistic variable, specifically: the residual energy factor. The fuzzy linguistic variables are low, medium, and high, and the node forward factor is... The fuzzy linguistic variables are small, medium, and large, and the link quality factor is... The fuzzy linguistic variables are poor, average, and excellent; Fuzzy rule reasoning: Based on fuzzy linguistic variables with three factors, a fuzzy rule base containing 27 IF-THEN rules is constructed. Fuzzy rule reasoning is performed on the fuzzy rule base to obtain fuzzy linguistic results. The fuzzy rule base includes all permutations and combinations of fuzzy linguistic variables with three factors. The fuzzy linguistic results are: very low, low, relatively low, medium, relatively high, high, and very high. Defuzzification: The centroid method is used to defuzzify the fuzzy language results, converting them into precise instant reward values. The formula for calculating instant reward value is: ， in, Represents the membership function. This represents the candidate values ​​for the reward.

5. A method based on fuzzy logic and... The method for learning and optimizing air-sea cross-domain network routing protocols is characterized by, In step 4, the value is determined after the candidate forwarding set is constructed and the instant reward value is calculated, based on... The algorithm updates the current node's information to its neighboring nodes based on the instantaneous reward value calculated by the fuzzy inference system. The value and the selection of the next-hop forwarding node, the The formula for updating the value is: ， in, Indicates the learning rate, Indicates the instant reward value. Indicates the loss factor. Indicates the new after the update value, Indicates the old version before the update. value, Represented as the maximum value, Indicates the next-hop forwarding node; In the initial state, build a local The table, initially all All values ​​are set to 0; After the value update is complete, select the candidate forwarding set. The neighbor node with the largest value is selected as the next-hop forwarding node.

6. A method based on fuzzy logic and... The method for learning and optimizing air-sea cross-domain network routing protocols is characterized by, Step 5 is specifically as follows: The current node encapsulates the data packet according to a preset data packet format, which includes the following fields: data packet ID, source node address, destination node address, and next-hop forwarding node address. The data packet is encapsulated with the following information: value, initial energy, remaining energy, current node location information, destination node location information, and data payload. After encapsulation, the data packet is forwarded to the selected next-hop forwarding node.

7. A method based on fuzzy logic and... The method for learning and optimizing air-sea cross-domain network routing protocols is characterized by, Step 6 is specifically as follows: After the current node forwards the data packet to the selected next-hop forwarding node, it enters the forwarding result confirmation stage. The process is as follows: The current node starts a timer immediately after sending the data packet and waits for the confirmation message returned by the next-hop forwarding node. If the confirmation message from the next-hop forwarding node is received before the timer expires, the forwarding is considered successful, and the forwarding of the current node is completed. If no acknowledgment message is received before the timer expires, the forwarding is considered a failure. The next-hop forwarding node is removed from the candidate forwarding set, and a new node is selected from the candidate forwarding set. The neighbor node with the largest value is used as the new next-hop forwarding node for forwarding. This process is repeated until forwarding is successful or the candidate forwarding set is empty.

8. A method based on fuzzy logic and... The method for learning and optimizing air-sea cross-domain network routing protocols is characterized by, The method also includes a continuous neighbor monitoring and information update process: When any node listens to any data packet sent by a neighboring node and extracts the status information of the data packet, it updates the information in the local neighbor information table in real time. If the node is not the destination node, it acts as a relay node and repeats the process of steps 2 to 6 to forward the data packet until it reaches the destination node. If the node is the destination node, the current node is the selected next-hop forwarding node.

9. A cross-domain air-sea network routing system implementing the method of any one of claims 1-7, characterized in that, The system comprises multiple heterogeneous nodes, each of which has the following built-in functional modules: Neighbor information collection module: used to periodically broadcast Hello messages and receive feedback messages from neighboring nodes, collect status information of neighboring nodes, and establish and maintain a local neighbor information table in real time; Candidate forwarding set construction module: used to filter nodes from the neighbor information table and construct a candidate forwarding set when a node needs to forward data packets, based on communication radius, remaining energy threshold and depth conditions; Fuzzy reasoning reward calculation module: It has a built-in triangular membership function and a 27-rule fuzzy rule library, which is used to fuzzify, reason by rules and defuzzify the remaining energy factor, node forward factor and link quality factor, and calculate the accurate instant reward value. Learning Decision Module: Used to maintain local... The table is iteratively updated based on the immediate reward value according to the update formula. Value, and select The neighbor node with the largest value is selected as the optimal next-hop forwarding node. Packet forwarding module: It is used to encapsulate data packets, fill in fields and send them, start a timer to listen for confirmation messages, and execute node removal and suboptimal node reselection logic when forwarding fails.

Images