System and method for optimizing underwater acoustic communication networking

Abstract

The invention discloses an underwater acoustic communication networking optimization system and method, and belongs to the technical field of underwater acoustic communication and networks, the system transmits state data of an underwater acoustic communication network back to the shore for processing, the global optimization capability of the optimization method is improved, the calculation load of communication nodes in the underwater acoustic communication network is reduced, and the network performance of the underwater acoustic communication network is improved. The power consumption of the communication nodes in the underwater acoustic communication network is reduced; based on the returned state data of the underwater acoustic communication network, the cluster head communication nodes and the multi-hop return route are optimized, and the power consumption utilization rate of the communication nodes in the underwater acoustic communication network is improved. In addition, the underwater acoustic communication channel is estimated in combination with an underwater acoustic physical model, so that the method can dynamically adapt to the time and space change characteristics of the underwater acoustic channel.

Description

Technical Field

[0001] This invention relates to the field of underwater acoustic communication and network technology, and more specifically to a system and method for optimizing underwater acoustic communication networks. Background Technology

[0002] Underwater acoustic communication technology uses sound waves as the information carrier and water as the transmission medium. Information is modulated onto sound waves and transmitted to the receiving end through the transmission medium. Underwater acoustic communication is currently the only known method for long-distance underwater communication and can be applied to marine environmental monitoring, diver communication, and communication between ships and underwater vehicles, playing a very important role in marine exploration and related fields.

[0003] Underwater acoustic communication networks are communication networks built upon underwater acoustic communication technology. They enable communication nodes to exchange information over a relatively long period and a large area of ​​sea, effectively improving the acquisition and processing capabilities of marine information. Currently, underwater acoustic communication networks, constructed from interconnected underwater sensors, unmanned submersibles, and surface buoys, are widely used in marine exploration, mining, and national defense. As countries worldwide accelerate their marine development, the demand for underwater sensor networks, underwater acoustic early warning networks, and underwater surveillance systems is becoming increasingly urgent. However, the nodes in underwater acoustic communication networks are battery-powered and difficult to replace, making network lifespan a significant factor limiting the overall lifespan of underwater acoustic communication networks.

[0004] Currently, improving the battery efficiency of underwater acoustic communication nodes mainly involves optimizing the underwater acoustic communication network, specifically including network structure optimization and routing optimization. Network structure optimization divides the entire communication network into clusters for management, reducing redundant communication and lowering energy consumption through hierarchical management. Routing optimization optimizes multiple backhaul routes to reduce energy consumption while ensuring communication reliability. Current proposed solutions often delegate regional decision-making power to the network, such as having cluster nodes determine the cluster head communication node through competition or negotiation. This approach lacks overall network state information, and delegating regional decision-making power to the network may lead to inappropriate decisions. For example, in the competition-based determination method, when one node in the network is determined to be the cluster head communication node, other nodes give up the competition, but there may be nodes that give up the competition with higher remaining energy and are more suitable as the cluster head communication node. Furthermore, delegating regional decision-making power to the network requires communication nodes to execute decision-making algorithms, increasing the nodes' computational energy consumption. The energy consumed in underwater acoustic communication is mainly consumed in the propagation of the acoustic signal. However, the large time-varying and spatial variations of the underwater acoustic channel make estimation based on the amount of data transmitted insufficient. The characteristics of the underwater acoustic communication channel should be included in the optimization factors. Current optimization methods do not consider combining the underwater acoustic propagation physical model to predict propagation loss.

[0005] Therefore, how to provide a system and method for optimizing underwater acoustic communication networks that can avoid the local optimum caused by direct negotiation between underwater communication nodes within a local cluster is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] In view of this, the present invention provides a system and method for optimizing underwater acoustic communication networks. By transmitting the state data of the underwater acoustic communication network back to shore for processing, the global optimization capability of the optimization method is improved while reducing the computational load of communication nodes in the underwater acoustic communication network, thereby reducing the power consumption of communication nodes in the underwater acoustic communication network. Based on the transmitted state data of the underwater acoustic communication network, the cluster head communication nodes and multi-hop backhaul routes are optimized, thereby improving the power consumption utilization rate of communication nodes in the underwater acoustic communication network.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] On the one hand, the present invention provides a system for optimizing underwater acoustic communication networking, including: an underwater acoustic communication network, a surface relay node, an air relay station, and a shore-based base station;

[0009] The underwater acoustic communication network is used to acquire detection data and underwater acoustic communication network status data. The underwater acoustic communication network includes several underwater acoustic communication nodes and sensors. The underwater acoustic communication nodes are used for communication between underwater acoustic communication nodes and communication between underwater acoustic communication nodes and water surface relay nodes.

[0010] The surface relay node is used to communicate with the underwater communication node and the airborne relay station;

[0011] The air relay station is used to communicate with the surface relay node and the onshore base station;

[0012] The onshore base station incorporates an underwater acoustic channel estimation algorithm, a cluster head node decision algorithm, and a routing planning algorithm. The underwater acoustic channel estimation algorithm estimates the channel characteristics of the underwater acoustic communication network using a physical model of underwater acoustic propagation. The cluster head node decision algorithm uses the underwater acoustic communication network state data and channel characteristics to determine the optimal cluster head communication node. The routing planning algorithm uses the underwater acoustic communication network state data to plan the optimal path for the underwater acoustic communication node to transmit back to the cluster head communication node.

[0013] Preferably, the underwater acoustic channel estimation algorithm estimates the channel characteristics of the underwater acoustic communication network using an underwater acoustic propagation physical model, including:

[0014] Calculate the sound velocity profile of the target water area based on the detection data;

[0015] Configure the environment file for the underwater acoustic propagation physical model tool based on underwater acoustic communication network status data, sound velocity profile, and terrain data;

[0016] Run the configured underwater acoustic propagation physical model tool program script to generate ray trajectory diagrams and time-domain impulse response intrinsic ray analysis results;

[0017] Based on the acoustic ray trajectory diagram and intrinsic acoustic ray analysis results, the propagation loss between communication nodes in the underwater acoustic communication network is estimated.

[0018] Preferably, the cluster head node decision algorithm utilizes underwater acoustic communication network state data and channel characteristics to determine the optimal cluster head communication node, including:

[0019] A cluster in the underwater acoustic communication network is modeled as a graph structure G = (V, E), where V represents the set of nodes and E represents the set of edges. The set of nodes in the graph structure corresponds to all communication nodes within the cluster, and the set of edges corresponds to the connection relationships between nodes. The node attributes are defined as the remaining power and geographical coordinates of the communication node, and the edge attributes are the propagation loss between nodes.

[0020] The graph structure is input into a trained graph neural network (GNN) model, which outputs the probability that each node is selected as a cluster head.

[0021] The node with the highest probability is selected as the optimal cluster head communication node.

[0022] Preferably, the loss function of the GNN model is to minimize the expected energy consumption of intra-cluster communication, while simultaneously considering the remaining energy of the node and the average propagation loss from the node to other nodes in the cluster.

[0023] Preferably, the routing planning algorithm uses underwater acoustic communication network state data to plan the optimal path for the underwater acoustic communication nodes to return to the cluster head communication node, including:

[0024] For non-cluster head nodes within a cluster, a deep reinforcement learning (DRL) algorithm is used for route planning.

[0025] Place the DRL agent at the starting underwater acoustic communication node of the path to be planned. Its action space is all adjacent underwater acoustic communication nodes within the communication radius of the starting underwater acoustic communication node.

[0026] The agent selects actions based on its state and explores step by step until it reaches the cluster head node, thus completing the search of a path.

[0027] The agent evaluates the exploration path based on a reward function, the formula for which the reward function is calculated is:

[0028]

[0029] Where r represents the reward received by the agent; H represents the total number of jumps from the starting point to the ending point; TL hB represents the propagation loss at the h-th hop; h This represents the number of transfer services carried by the h-th hop link; κ is a constant.

[0030] Through continuous training, the path with the highest reward value is ultimately determined as the optimal path for each node.

[0031] On the other hand, the present invention provides a method for optimizing underwater acoustic communication networks, comprising the following steps:

[0032] S1: The shore-based base station sends data acquisition commands to the underwater acoustic communication network through air relay nodes and surface relay nodes; after receiving the commands, the surface relay nodes collect the detection data from their own onboard sensors and send data acquisition requests to other underwater acoustic communication nodes; after receiving the requests, the other underwater acoustic communication nodes transmit their own detection data back; the surface relay nodes summarize all detection data and the status data of the underwater acoustic communication network and transmit them back to the shore-based base station through the air relay nodes;

[0033] S2: The shore-based base station estimates the channel characteristics of the underwater acoustic communication network by combining the returned detection data and status data with an underwater acoustic channel estimation algorithm.

[0034] S3: The onshore base station determines the cluster head communication node based on channel characteristics and state data through the cluster head node decision algorithm;

[0035] S4: The shore-based base station uses a routing planning algorithm to plan the optimal path for each underwater acoustic communication node in the cluster to transmit back to the cluster head communication node based on channel characteristics and state data.

[0036] S5: The shore-based base station transmits the cluster head communication node and the optimal path to the underwater acoustic communication network through air relay nodes and surface relay nodes; the underwater acoustic communication network adjusts its network structure according to the cluster head communication node and the optimal path.

[0037] Preferably, in S1, the underwater acoustic communication network status data includes the geographical coordinates, transmission power, and remaining battery power of each underwater acoustic communication node; the detection data includes seawater temperature, salinity, depth, and seabed topography data.

[0038] As can be seen from the above technical solution, compared with the prior art, this invention discloses a system and method for optimizing underwater acoustic communication networks. By transmitting underwater network status data back to the shore base station via surface and air relays, centralized optimization decision-making based on the entire network information is achieved. The shore base station can acquire global status data including topology, node energy, and location, thereby fundamentally overcoming the local optimum problem easily caused by distributed optimization, significantly improving the rationality of cluster head selection and routing planning, and significantly enhancing the overall network transmission performance and energy efficiency. Secondly, this invention places the computationally complex optimization algorithm entirely on the shore base station, while underwater nodes are only responsible for data acquisition and communication, greatly saving their computational energy consumption. The energy of communication nodes is mainly used for communication itself, effectively extending the continuous working time of nodes and the network, solving the problem of energy constraints for underwater nodes. Furthermore, this invention combines real-time collected environmental data such as water temperature and salinity with underwater acoustic physical models (such as Bellhop) to dynamically and accurately estimate channel propagation characteristics, which serve as key inputs to the optimization algorithm. This enables the method to adapt to the strong time- and space-varying characteristics of the underwater acoustic channel, significantly improving the accuracy and environmental robustness of the optimization strategy. Finally, through a data acquisition-return transmission-computation-distribution mechanism, this invention achieves online and dynamic optimization and adjustment of the underwater acoustic network, ensuring that the network always operates in an optimal or near-optimal state, comprehensively improving the reliability, efficiency, and lifespan of the underwater acoustic communication network. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0040] Figure 1 This is a structural schematic diagram provided for the present invention.

[0041] Figure 2 This is a flowchart of the process of the present invention.

[0042] Figure 3 This is a schematic diagram of the horizontal topology distribution of a cluster of underwater acoustic communication nodes.

[0043] Figure 4 This is a schematic diagram of the decision-making process of a cluster head communication node based on GNN.

[0044] Figure 5 This is a schematic diagram of the backhaul path between different cluster head nodes. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] This invention discloses a system for optimizing underwater acoustic communication networks, such as... Figure 1 As shown, it includes: underwater acoustic communication network, surface relay node, air relay station, and shore base station;

[0047] The underwater acoustic communication network is used to acquire detection data and underwater acoustic communication network status data. The underwater acoustic communication network includes several underwater acoustic communication nodes and sensors. The underwater acoustic communication nodes are used for communication between underwater acoustic communication nodes and communication between underwater acoustic communication nodes and water surface relay nodes.

[0048] The surface relay node is used to communicate with the underwater communication node and the airborne relay station; the surface relay node includes, but is not limited to, buoys;

[0049] The airborne relay station is used to communicate with the surface relay node and the onshore base station; the airborne relay node includes, but is not limited to, satellites;

[0050] The onshore base station incorporates an underwater acoustic channel estimation algorithm, a cluster head node decision algorithm, and a routing planning algorithm. The underwater acoustic channel estimation algorithm estimates the channel characteristics of the underwater acoustic communication network, such as propagation loss, using a physical model of underwater acoustic propagation. This algorithm includes normal mode theory and ray theory. The cluster head node decision algorithm uses underwater acoustic communication network state data and channel characteristics to determine the optimal cluster head communication node. The routing planning algorithm uses underwater acoustic communication network state data to plan the optimal path for the underwater acoustic communication node to return to the cluster head communication node.

[0051] Through the surface relay nodes and air relay nodes, the underwater acoustic communication network can communicate bidirectionally with the shore-based base station. The shore-based base station can receive underwater communication network status data, detection data, etc., and utilize the cluster head communication node decision algorithm and routing planning algorithm to calculate the optimal cluster head communication node and multi-hop backhaul path, thereby reducing the energy consumption of the underwater acoustic communication network, improving its energy efficiency, and extending its operating time. Furthermore, in the cluster head node decision and routing planning process, this invention uses the underwater acoustic channel estimation algorithm to estimate the channel, making the algorithm model more accurate and better able to dynamically adapt to changes in the actual environment. The workflow of this invention is as follows: Figure 2 As shown.

[0052] Specifically, in this embodiment, the surface relay station uses a buoy, which is equipped with an underwater acoustic communication device and satellite communication equipment, enabling underwater acoustic communication with other underwater communication stations and transmitting and receiving satellite signals; the underwater communication node is mounted on a submersible mooring. In this embodiment, the airborne relay station is a satellite. The shore-based base station includes a satellite ground receiving station for communicating with the satellite. In this embodiment, the buoy and submersible mooring are equipped with temperature, salinity, and depth profilers (CTD) for detecting the physical parameters of seawater such as temperature, salinity, and depth.

[0053] Furthermore, the underwater acoustic channel estimation algorithm estimates the channel characteristics of the underwater acoustic communication network using an underwater acoustic propagation physical model, including:

[0054] Calculate the sound velocity profile of the target water area based on the detection data;

[0055] Configure the environment file for the underwater acoustic propagation physical model tool based on underwater acoustic communication network status data, sound velocity profile, and terrain data;

[0056] Run the configured underwater acoustic propagation physical model tool program script to generate ray trajectory diagrams and time-domain impulse response intrinsic ray analysis results;

[0057] Based on the acoustic ray trajectory diagram and intrinsic acoustic ray analysis results, the propagation loss between communication nodes in the underwater acoustic communication network is estimated.

[0058] Preferably, the cluster head node decision algorithm utilizes underwater acoustic communication network state data and channel characteristics to determine the optimal cluster head communication node, including:

[0059] A cluster in the underwater acoustic communication network is modeled as a graph structure G = (V, E), where V represents the set of nodes and E represents the set of edges. The set of nodes in the graph structure corresponds to all communication nodes within the cluster, and the set of edges corresponds to the connection relationships between nodes. The node attributes are defined as the remaining power and geographical coordinates of the communication node, and the edge attributes are the propagation loss between nodes.

[0060] The graph structure is input into a trained graph neural network (GNN) model, which outputs the probability that each node is selected as a cluster head.

[0061] The node with the highest probability is selected as the optimal cluster head communication node.

[0062] Preferably, the loss function of the GNN model is to minimize the expected energy consumption of intra-cluster communication, while simultaneously considering the remaining energy of the node and the average propagation loss from the node to other nodes in the cluster.

[0063] Preferably, the routing planning algorithm uses underwater acoustic communication network state data to plan the optimal path for the underwater acoustic communication nodes to return to the cluster head communication node, including:

[0064] For non-cluster head nodes within a cluster, a deep reinforcement learning (DRL) algorithm is used for route planning.

[0065] Place the DRL agent at the starting underwater acoustic communication node of the path to be planned. Its action space is all adjacent underwater acoustic communication nodes within the communication radius of the starting underwater acoustic communication node.

[0066] The agent selects actions based on its state and explores step by step until it reaches the cluster head node, thus completing the search of a path.

[0067] The agent evaluates the exploration path based on a reward function, the formula for which the reward function is calculated is:

[0068]

[0069] Where r represents the reward received by the agent; H represents the total number of jumps from the starting point to the ending point; TL h B represents the propagation loss at the h-th hop; h This represents the number of transfer services carried by the h-th hop link; κ is a constant, and the negative sign in front of it indicates a penalty. If the destination is not reached, a penalty is imposed for each extra step taken, thereby avoiding excessive transfers along the link.

[0070] Through continuous training, the path with the highest reward value is ultimately determined as the optimal path for each node.

[0071] On the other hand, the present invention provides a method for optimizing underwater acoustic communication networks, comprising the following steps:

[0072] S1: The shore-based base station sends data acquisition commands to the underwater acoustic communication network through air relay nodes and surface relay nodes; after receiving the commands, the surface relay nodes collect the detection data from their own onboard sensors and send data acquisition requests to other underwater acoustic communication nodes; after receiving the requests, the other underwater acoustic communication nodes transmit their own detection data back; the surface relay nodes summarize all detection data and the status data of the underwater acoustic communication network and transmit them back to the shore-based base station through the air relay nodes.

[0073] Specifically, in this embodiment, the shore-based base station sends data acquisition commands to the satellite periodically or intermittently. These commands are relayed to the underwater acoustic communication sensor network via satellite relay and the buoy. The buoy system begins collecting detection data from its attached sensors and simultaneously sends data acquisition requests to other underwater acoustic communication devices via its attached underwater communication unit, while waiting to receive detection data returned by these devices. Once all other communication devices have returned detection data, the buoy system transmits the data back to the shore-based base station via its satellite antenna.

[0074] S2: The shore-based base station estimates the channel characteristics of the underwater acoustic communication network by combining the returned detection data and status data with an underwater acoustic channel estimation algorithm.

[0075] Specifically, the onshore base station receives the transmitted status data and probed data from the underwater acoustic communication network, and uses an underwater acoustic channel estimation algorithm to estimate the characteristics of the underwater acoustic channel. The status data of the underwater acoustic communication network includes the geographical coordinates of each station (used to calculate the distance between two stations) and transmission power; the probed data includes seawater temperature, salinity, and depth. These parameters can be used to calculate the sound velocity profile of the target seawater area, denoted by c(h), which is a function of depth h. Sound propagation loss is an important performance indicator of sound wave transmission in water and is closely related to the energy consumption of the communication node. This embodiment uses propagation loss as an example. Propagation loss can be estimated using physical models, including normal mode models, ray acoustic models, and parabolic equations.

[0076] Currently, there are relatively mature algorithm tools for underwater acoustic estimation, such as Bellhop. Bellhop can solve underwater acoustic propagation models based on ray sound and normal modes by appropriately configuring parameters. Bellhop can be used to calculate the characteristics of the underwater acoustic channel. This embodiment takes propagation loss (TL), which is highly correlated with power consumption, as an example; Bellhop can be used to estimate the TL between communication nodes.

[0077] First, configure the Bellhop environment file (.env). Use the parameters obtained from the aforementioned detection to set the Bellhop environment file (.env). These parameters include the geographic coordinates of each communication station, transmission power, h, sound velocity profile c(h), and seabed topographic data. Seabed topographic data can be obtained through sonar scanning or from publicly available databases, such as https: / / www.gebco.net / . After configuring the Bellhop environment file (.env), run the Bellhop program script. The output results, such as ray plotting and time-domain impulse response eigenrays analysis, can be used to estimate the propagation loss from the transmitter to the receiver. Currently, there are stable, open-source Bellhop program scripts available, and development based on theory is also possible.

[0078] However, the accuracy of existing models relies on physical parameters, such as sound velocity profiles. This invention proposes real-time acquisition of key seawater parameters, such as temperature, salinity, and depth (which can be converted to pressure), to provide accurate parameters for physical models. The innovation of this invention lies in the ability to construct a sound velocity profile using the returned seawater temperature, salinity, and depth (which can be converted to pressure). This continuous adjustment of Bellhop's parameters based on measured data improves the accuracy of TL estimation.

[0079] S3: The onshore base station determines the cluster head communication node based on channel characteristics and state data through the cluster head node decision algorithm.

[0080] Specifically, a cluster head node decision algorithm is used to determine the cluster head node for cluster communication. In this embodiment, the decision-making of the cluster head communication node adopts a graph neural network (GNN) algorithm. Based on a predetermined number of clusters, clusters are first divided geographically. A single cluster consists of adjacent underwater acoustic communication nodes. Assuming the horizontal topology distribution of the underwater acoustic communication nodes in a certain cluster is as follows... Figure 3 As shown.

[0081] The network is modeled as a graph structure G = (V, E), where V represents the set of nodes and E represents the set of edges. The attributes of each node include remaining battery power E. b And geographic coordinates, the edge properties are the propagation loss TL between nodes. ij By learning node features through GNN, the probability P of a communication node being selected as a cluster head is output. head The objective function of GNN is:

[0082]

[0083] In equation (1), α and β are weighting factors. The average propagation loss from node i to its neighboring nodes. The average propagation across the entire network (normalized baseline) is used. Meanwhile, to ensure communication reliability, the distance between adjacent nodes is limited to within the range of the communicating nodes, using R... com This is the effective communication radius of the communication node. Based on the above analysis, the node characteristics can be expressed as X = [E...]. b The adjacency matrix of [x,y] is expressed as:

[0084]

[0085] As can be seen from equation (2), when adjacent underwater communication nodes exceed the effective communication range, the adjacency moment is 0, meaning there are no nodes that cannot be connected. Based on describing the underwater communication network as a graph structure, a GNN model is used for training. The loss function L of the GNN is expressed as:

[0086]

[0087] Equation (3) aims to minimize the "expected value" within the entire cluster, i.e., to minimize L by optimizing the loss function. From equation (3), it can be seen that optimization yields the probability of each node in the cluster becoming the cluster head communication node, and the node with the highest probability is selected as the cluster head communication node. From equation (3), it can be seen that due to… Constraint: Given a loss value, if Eb Larger, and When it is smaller, then A larger value indicates that the node is more likely to be selected as the cluster head. To further illustrate, when the node is at the cluster center, its average distance to other nodes in the cluster is smaller, i.e., and... If the value is small and the remaining charge is high, then E b If the energy level is relatively high, the node is more suitable as a cluster head communication node, because the cluster head node is responsible for aggregating and forwarding data from other nodes within the cluster, and therefore requires more spare energy.

[0088] As can be seen from equation (3), the optimization method proposed in this invention includes the remaining energy of the communication node and the propagation loss. Furthermore, the propagation loss is estimated using a physical model based on actual collected data. When the environment changes, it can be evaluated according to the latest conditions, thus dynamically optimizing and adjusting. Therefore, this invention has stronger adaptability to time-varying and spatial changes. The decision-making process of the cluster head communication node based on GNN is as follows: Figure 4 As shown.

[0089] S4: The shore-based base station uses a routing planning algorithm to plan the optimal path for each underwater acoustic communication node in the cluster to transmit back to the cluster head communication node based on channel characteristics and state data.

[0090] Specifically, a routing planning algorithm is used to search for the optimal backhaul path. While GNNs can make cluster head node decisions while considering energy and average communication distance, in intra-cluster communication, nodes far from the cluster head node need to go through multiple hops to transmit data back to the cluster head node. Different backhaul routes exhibit significant differences in signal transmission quality and propagation loss (closely related to the power consumption of the communication nodes). As shown in the figure, assuming cluster head node A, intra-cluster node C needs to transmit data back to node A via hops through nodes D and E, or via an intermediate hop through node B. This embodiment uses Deep Reinforcement Learning (DRL) to optimize the backhaul route of the underwater communication network. By leveraging DRL's automatic policy optimization capabilities, the transmission performance of the underwater network and the power efficiency of the entire system are continuously improved.

[0091] In the DRL model, the state space is the number of nodes, and each cluster is trained independently. In this project, the communication quality of the nodes, as well as information such as temperature, salinity, and depth, are all transmitted back to the shore via satellite. Therefore, the DRL training process can be implemented on land, and the optimization results can be sent to the underwater nodes via simple control commands, avoiding the computational burden on the underwater nodes.

[0092] The state space of the DRL model consists of all nodes within the cluster except the cluster head node. The action space for each node is its neighboring nodes within the communication radius. Each node employs a greedy (ε-greedy) strategy to select actions, simultaneously weighing the quality of the action, the propagation loss along the path, and the number of transmission links that have already traversed that path. Since the temperature, depth, and salinity within the target area have been relayed back to shore via satellite, the propagation loss of the channel can be assessed using the physical model. For clarity, the propagation loss is incorporated into the action reward (or penalty) mechanism. Furthermore, to avoid the "dead zone" effect—where nodes closer to the cluster head node bear an excessive number of relay tasks—load balancing is necessary to reduce communication conflicts caused by these nodes carrying too many relay tasks. Here, the number of relay links carried by B-links is used.

[0093] When an agent starts from a node (i.e., from a certain state) and searches to the cluster head node, assuming it takes H hops, its reward for reaching the destination can be expressed as:

[0094]

[0095] In Equation (4), the first term of the agent is determined by the propagation loss of the entire link. The smaller the cumulative propagation loss, the greater the reward. The second term is the limit on the number of links in the entire link. The larger the maximum number of forwarding links in the entire link, the smaller the reward, which discourages the agent from taking this path because there are intermediate nodes that have already carried too many forwarding tasks. In Equation (4), the agent will receive a small penalty for each step it takes, which allows the agent to reach the cluster head communication node with the fewest possible number of transfers.

[0096] S5: The shore-based base station transmits the cluster head communication node and the optimal path to the underwater acoustic communication network through air relay nodes and surface relay nodes; the underwater acoustic communication network adjusts its network structure according to the cluster head communication node and the optimal path.

[0097] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0098] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A system for optimizing underwater acoustic communication networks, characterized in that, include: Underwater acoustic communication network, surface relay node, air relay station, shore-based base station; The underwater acoustic communication network is used to acquire detection data and underwater acoustic communication network status data. The underwater acoustic communication network includes several underwater acoustic communication nodes and sensors. The underwater acoustic communication nodes are used for communication between underwater acoustic communication nodes and communication between underwater acoustic communication nodes and water surface relay nodes. The surface relay node is used to communicate with the underwater communication node and the airborne relay station; The air relay station is used to communicate with the surface relay node and the onshore base station; The onshore base station incorporates an underwater acoustic channel estimation algorithm, a cluster head node decision algorithm, and a routing planning algorithm. The underwater acoustic channel estimation algorithm estimates the channel characteristics of the underwater acoustic communication network using a physical model of underwater acoustic propagation. The cluster head node decision algorithm uses the underwater acoustic communication network state data and channel characteristics to determine the optimal cluster head communication node. The routing planning algorithm uses underwater acoustic communication network status data to plan the optimal path for underwater acoustic communication nodes to transmit back to the cluster head communication node.

2. The system for optimizing underwater acoustic communication networks according to claim 1, characterized in that, The underwater acoustic channel estimation algorithm estimates the channel characteristics of the underwater acoustic communication network using a physical model of underwater acoustic propagation, including: Calculate the sound velocity profile of the target water area based on the detection data; Configure the environment file for the underwater acoustic propagation physical model tool based on underwater acoustic communication network status data, sound velocity profile, and terrain data; Run the configured underwater acoustic propagation physical model tool program script to generate ray trajectory diagrams and time-domain impulse response intrinsic ray analysis results; Based on the acoustic ray trajectory diagram and intrinsic acoustic ray analysis results, the propagation loss between communication nodes in the underwater acoustic communication network is estimated.

3. The system for optimizing underwater acoustic communication networks according to claim 1, characterized in that, The cluster head node decision algorithm utilizes underwater acoustic communication network state data and channel characteristics to determine the optimal cluster head communication node, including: A cluster in the underwater acoustic communication network is modeled as a graph structure G = (V, E), where V represents the set of nodes and E represents the set of edges. The set of nodes in the graph structure corresponds to all communication nodes within the cluster, and the set of edges corresponds to the connection relationships between nodes. The node attributes are defined as the remaining power and geographical coordinates of the communication node, and the edge attributes are the propagation loss between nodes. The graph structure is input into a trained graph neural network (GNN) model, which outputs the probability that each node is selected as a cluster head. The node with the highest probability is selected as the optimal cluster head communication node.

4. The system for optimizing underwater acoustic communication networks according to claim 1, characterized in that, The loss function of the GNN model is to minimize the expected energy consumption of intra-cluster communication, while simultaneously considering the remaining energy of the node and the average propagation loss from the node to other nodes in the cluster.

5. The system for optimizing underwater acoustic communication networks according to claim 1, characterized in that, The routing planning algorithm uses underwater acoustic communication network state data to plan the optimal path for underwater acoustic communication nodes to transmit back to the cluster head communication node, including: For non-cluster head nodes within a cluster, a deep reinforcement learning (DRL) algorithm is used for route planning. Place the DRL agent at the starting underwater acoustic communication node of the path to be planned. Its action space is all adjacent underwater acoustic communication nodes within the communication radius of the starting underwater acoustic communication node. The agent selects actions based on its state and explores step by step until it reaches the cluster head node, thus completing the search of a path. The agent evaluates the exploration path based on a reward function. Assuming the destination is reached after h jumps, the reward function is calculated using the following formula: Where r represents the reward received by the agent; H represents the total number of jumps from the starting point to the ending point; TL h B represents the propagation loss at the h-th hop; h This represents the number of transfer services carried by the h-th hop link; κ is a constant. Through continuous training, the path with the highest reward value is ultimately determined as the optimal path for each node.

6. A method for optimizing underwater acoustic communication networks, characterized in that, Includes the following steps: S1: The shore-based base station sends data acquisition commands to the underwater acoustic communication network through air relay nodes and surface relay nodes; after receiving the commands, the surface relay nodes collect the detection data from their own onboard sensors and send data acquisition requests to other underwater acoustic communication nodes; after receiving the requests, the other underwater acoustic communication nodes transmit their own detection data back; the surface relay nodes summarize all detection data and the status data of the underwater acoustic communication network and transmit them back to the shore-based base station through the air relay nodes; S2: The shore-based base station estimates the channel characteristics of the underwater acoustic communication network by combining the returned detection data and status data with an underwater acoustic channel estimation algorithm. S3: The onshore base station determines the cluster head communication node based on channel characteristics and state data through the cluster head node decision algorithm; S4: The shore-based base station uses a routing planning algorithm to plan the optimal path for each underwater acoustic communication node in the cluster to transmit back to the cluster head communication node based on channel characteristics and state data. S5: The shore-based base station transmits the cluster head communication node and the optimal path to the underwater acoustic communication network through air relay nodes and surface relay nodes; the underwater acoustic communication network adjusts its network structure according to the cluster head communication node and the optimal path.

7. The method for optimizing underwater acoustic communication networks according to claim 6, characterized in that, In S1, the underwater acoustic communication network status data includes the geographical coordinates, transmission power, and remaining battery power of each underwater acoustic communication node; the detection data includes seawater temperature, salinity, depth, and seabed topography data.

Images