Maritime area communication system

Abstract

This invention discloses a marine area communication system, relating to the field of communication technology for marine aquaculture platforms. The system includes a multi-level communication platform, a distributed adaptive networking engine, and a cross-domain, cross-media communication gateway. The multi-level communication platform integrates satellite communication modules, surface communication modules, and underwater communication modules, and possesses local edge computing capabilities. The distributed adaptive networking engine communicates with the multi-level communication platform and is used for neighbor discovery, global topology aggregation, dynamic route calculation, and automatic hub switching. The cross-domain, cross-media communication gateway communicates with the multi-level communication platform and the distributed adaptive networking engine, and is used for signal format conversion and intelligent routing between the satellite, surface, and underwater communication domains. This invention's communication system enables cross-domain collaboration between satellite, surface, and underwater communications and possesses self-awareness, self-repair, and self-optimization capabilities.

Description

Technical Field

[0001] This invention relates to the field of communication technology for offshore aquaculture platforms, and particularly to a maritime area communication system. Background Technology

[0002] As marine aquaculture develops towards deeper waters and intelligent technologies, offshore aquaculture platforms face higher demands for stable, efficient, and real-time data communication and remote control. Their technological development has evolved from near-shore wired communication and single wireless spread spectrum to attempts combining various wireless and satellite communication technologies, aiming to overcome the constraints of the complex marine environment on information transmission.

[0003] Currently, communication on offshore aquaculture platforms relies primarily on a combination of technologies. Nearshore areas are exploring the expansion of land-based 4G / LTE networks, while in the open ocean, satellite communication (such as Inmarsat and Iridium) is commonly used to achieve wide-area coverage. Simultaneously, internal and inter-platform networking often utilizes self-built 900MHz / 2.4GHz private wireless networks (such as Wi-Fi and Zigbee), while underwater monitoring mainly relies on underwater acoustic communication technology. These technologies collectively constitute the current mainstream communication solutions for offshore aquaculture platforms.

[0004] However, existing technical solutions have significant shortcomings: Most offshore aquaculture platform networks use fixed topologies and static routing configurations. When the platform moves, equipment fails, new nodes are added, or the network environment changes, the network cannot automatically detect and quickly adjust, requiring manual reconfiguration. Fault recovery often takes hours or even days, seriously affecting the safety of aquaculture production. Furthermore, the network cannot adapt to platform movement, node changes, or link failures, and maintenance is highly dependent on manual intervention, resulting in insufficient system reliability and flexibility. These deficiencies hinder the further development of intelligent offshore aquaculture.

[0005] Therefore, there is an urgent need for a new type of maritime area communication system to solve the problem that the network of offshore platforms cannot automatically sense and adjust. Summary of the Invention

[0006] The main objective of this invention is to propose a maritime area communication system that aims to solve the problem that the network of a maritime platform cannot automatically sense and adjust.

[0007] To achieve the above objectives, the maritime area communication system proposed in this invention includes a multi-level communication platform, a distributed adaptive networking engine, and a cross-domain, cross-media communication gateway. The multi-level communication platform integrates satellite communication modules, surface communication modules, and underwater communication modules, and possesses local edge computing capabilities. The distributed adaptive networking engine communicates with the multi-level communication platform and is used for neighbor discovery, global topology aggregation, dynamic route calculation, and automatic hub switching for the multi-level communication platform. The cross-domain, cross-media communication gateway communicates with both the multi-level communication platform and the distributed adaptive networking engine, and is used for signal format conversion and intelligent routing between the satellite, surface, and underwater communication domains for the multi-level communication platform.

[0008] In one embodiment, the distributed adaptive networking engine is configured to perform the following operations: neighbor discovery operation, which builds and maintains a local adjacency table containing neighbor node status and link quality information by periodically broadcasting and receiving greeting messages; channel quality assessment operation, which calculates a comprehensive link quality score based on the signal-to-noise ratio, end-to-end delay, and packet loss rate of each communication link, and dynamically activates or isolates corresponding links based on the comprehensive link quality score; multi-platform role dynamic election operation, which calculates a hub score based on the remaining energy, communication range, computing power, and network centrality of each platform in the multi-level communication platform, and dynamically elects a master hub and forms a candidate hub pool based on the hub score; hierarchical information aggregation operation, which collects and distributes the entire network topology information in a tree-like aggregation manner by constructing a spanning tree rooted at the master hub; and reinforcement learning route optimization operation, which calculates and continuously optimizes the optimal transmission path for source-target node pairs based on Markov decision processes and Q-learning algorithms.

[0009] In one embodiment, the channel quality assessment operation includes: calculating the signal-to-noise ratio (SNR) using a sliding window average; tracking the trend of end-to-end delay using an exponentially weighted moving average algorithm; calculating the packet loss rate; calculating the comprehensive link quality score for each link using a comprehensive link quality scoring model based on the SNR, the trend of end-to-end delay, and the packet loss rate; automatically isolating the corresponding link and retaining monitoring when the comprehensive link quality score of a single link is lower than a preset threshold; and automatically activating the isolated link when the comprehensive link quality score of the isolated link reaches the preset threshold.

[0010] In one embodiment, the multi-platform role dynamic election operation includes: calculating a hub score based on the remaining energy, communication range, computing power, and network centrality of each platform in the multi-level communication platform; ranking each platform based on the hub score, with the platforms ranking in the top preset proportion becoming candidate platforms, electing the candidate platform ranked first as the main hub, and forming a reserve hub pool with the remaining platforms in the candidate platform; if the main hub fails, the platform ranked first in the reserve hub pool is elected as the new main hub.

[0011] In one embodiment, the hierarchical information aggregation operation includes: constructing a spanning tree rooted at the main hub; wherein each platform corresponds to one node; in the uplink aggregation phase, child nodes send their local adjacency lists to their parent nodes, and the parent nodes in the intermediate layers merge all child node information and continue to pass it uplink so that the main hub obtains a complete global adjacency matrix; each node independently maintains its local adjacency list, and when a neighboring node failure is detected, it uses local alternative adjacencies for forwarding; in the downlink broadcast phase, after the main hub calculates the routing table, it broadcasts it layer by layer to the entire network through the spanning tree structure.

[0012] In one embodiment, the reinforcement learning route optimization operation includes: the master hub constructs the state space, action space, and reward function of a Markov decision process based on the global topology; wherein the reward function is a weighted combination of latency, reliability, and energy consumption; the master hub maintains a Q-value table for each source-target node pair based on the Q-learning algorithm; after aggregating the global topology, the master hub calculates the optimal Q-value and generates a global routing table, and broadcasts the global routing table to the entire network; each node records the actual forwarding effect and periodically provides performance feedback; the master hub incrementally learns and updates the Q-value table based on the performance feedback according to the reward function.

[0013] In one embodiment, the multi-level communication platform includes a hub-level platform, a node-level platform, a mobile-level platform, and a fixed-level platform, with each level of platform communicating and connecting with each other through a distributed adaptive networking engine; and / or, the hub-level platform integrates a satellite communication module, a surface communication module, an underwater communication module, and a 4G / 5G multi-mode receiving unit, and is used for global network topology maintenance, data aggregation, task allocation, and resource scheduling; and / or, the node-level platform is used for data acquisition and inter-node relay forwarding, supporting local edge computing and real-time sensor data processing; and / or, the mobile-level platform uses short-range communication and local cooperation algorithms for area inspection, emergency response, and problem diagnosis; and / or, the fixed-level platform integrates a satellite communication module, a surface communication module, an underwater communication module, and a cellular network multi-mode receiving unit.

[0014] In one embodiment, when the maritime area communication system is started, the platforms at each level are started in sequence according to the priority order of fixed platform, hub platform, node platform and mobile platform, and a distributed adaptive networking engine is loaded.

[0015] In one embodiment, the cross-domain, cross-media communication gateway includes a satellite-to-ground gateway and a surface-to-underwater gateway. The satellite-to-ground gateway is used to convert satellite signals to surface wireless signals for a multi-level communication platform; the surface-to-underwater gateway is used to convert the frequency domain of surface wireless signals to underwater acoustic signals for a multi-level communication platform.

[0016] In one embodiment, the maritime area communication system further includes a converged network management center, which is communicatively connected to the distributed adaptive networking engine and the cross-domain, cross-media communication gateway. The converged network management center is configured to perform the following operations: edge computing offloading and local caching operations, performing local data caching when the 5G link is interrupted and performing data analysis on the mobile edge computing platform; network slicing application operations, applying for differentiated network slices from the 5G core network according to service requirements; protocol conversion and traffic decision operations, converting the private protocol of the distributed adaptive networking engine into the 5G standard protocol, and intelligently routing data streams to the 5G core network, satellite link, or maritime ad hoc network according to service type; and global service orchestration and task allocation operations, calculating an adaptability score based on a global optimization objective using an allocation algorithm, and dynamically allocating control tasks to the platform with the highest score among the multi-level communication platforms for execution based on the adaptability score.

[0017] The technical solution of this invention integrates satellite communication modules, surface communication modules, and underwater communication modules into a multi-level communication platform, enabling various offshore platforms to flexibly support multiple communication methods and providing a foundation for cross-domain communication. Simultaneously, edge computing capabilities allow the platform to process data and execute logical judgments locally, reducing reliance on remote centers and enhancing autonomous operation and rapid response capabilities in abnormal situations such as disconnections. Furthermore, compared to the fixed topology and static routing configurations used in traditional offshore platform networks, this system employs a distributed adaptive networking engine. This engine enables neighbor discovery, global topology aggregation, dynamic route calculation, and automatic hub switching, giving the network self-awareness and self-organization capabilities. The system can automatically discover surrounding nodes, aggregate the entire network topology in real time, and dynamically calculate the optimal route accordingly, thereby maintaining high network availability and adaptive adjustment without manual intervention, significantly reducing operation and maintenance costs. In addition, by setting up a cross-domain and cross-medium communication gateway for signal format conversion and intelligent routing between the three communication domains of satellite, surface and underwater, the adaptation and integration between heterogeneous communication protocols and signal systems are realized. This enables data to intelligently select the optimal transmission path across different communication domains according to business needs and real-time channel conditions, thereby ensuring the reliability and real-time performance of end-to-end transmission.

[0018] Overall, the technical solution of this embodiment integrates a multi-level communication platform, a distributed adaptive networking engine, and a cross-domain and cross-media communication gateway into a maritime area communication system. This system not only realizes the integrated intelligent coordination of satellite, surface, and underwater communications, but also endows the network with powerful self-sensing, self-repair, and self-optimization capabilities, enabling it to automatically adapt to complex marine dynamic environments and network topology changes. Attached Figure Description

[0019] 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 some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0020] Figure 1 A schematic diagram of the system architecture of an embodiment of the maritime area communication system provided by the present invention;

[0021] Figure 2 A schematic diagram of neighbor discovery operation according to an embodiment of the maritime area communication system provided by the present invention;

[0022] Figure 3 A schematic diagram of the channel quality assessment operation process of an embodiment of the maritime area communication system provided by the present invention;

[0023] Figure 4 A schematic diagram of the hub election results of an embodiment of the maritime area communication system provided by the present invention;

[0024] Figure 5 This is a schematic diagram of the node topology of an embodiment of the maritime area communication system provided by the present invention;

[0025] Figure 6 A comparative schematic diagram of tree-like convergence and flood propagation in an embodiment of the maritime area communication system provided by the present invention;

[0026] Figure 7 A schematic diagram of a cross-domain, cross-media converged gateway architecture according to an embodiment of the maritime area communication system provided by the present invention.

[0027] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0028] 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 a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0029] It should be noted that if the embodiments of the present invention involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of the components in a specific posture. If the specific posture changes, the directional indicators will also change accordingly.

[0030] Furthermore, if the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the use of "and / or" or "and / or" throughout the text includes three parallel solutions. For example, "A and / or B" includes solution A, solution B, or a solution where both A and B are satisfied simultaneously. Furthermore, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0031] As marine aquaculture develops towards deeper waters and intelligent technologies, offshore aquaculture platforms face higher demands for stable, efficient, and real-time data communication and remote control. Their technological development has evolved from near-shore wired communication and single wireless spread spectrum to attempts combining various wireless and satellite communication technologies, aiming to overcome the constraints of the complex marine environment on information transmission.

[0032] Currently, communication on offshore aquaculture platforms relies primarily on a combination of technologies. Nearshore areas are exploring the expansion of land-based 4G / LTE networks, while in the open ocean, satellite communication (such as Inmarsat and Iridium) is commonly used to achieve wide-area coverage. Simultaneously, internal and inter-platform networking often utilizes self-built 900MHz / 2.4GHz private wireless networks (such as Wi-Fi and Zigbee), while underwater monitoring mainly relies on underwater acoustic communication technology. These technologies collectively constitute the current mainstream communication solutions for offshore aquaculture platforms.

[0033] However, existing technical solutions have significant shortcomings: Most offshore aquaculture platform networks use fixed topologies and static routing configurations. When the platform moves, equipment fails, new nodes are added, or the network environment changes, the network cannot automatically detect and quickly adjust, requiring manual reconfiguration. Fault recovery often takes hours or even days, seriously affecting the safety of aquaculture production. Furthermore, the network cannot adapt to platform movement, node changes, or link failures, and maintenance is highly dependent on manual intervention, resulting in insufficient system reliability and flexibility. These deficiencies hinder the further development of intelligent offshore aquaculture.

[0034] Therefore, there is an urgent need for a new type of maritime area communication system to solve the problem that the network of offshore platforms cannot automatically sense and adjust.

[0035] To address the aforementioned problems, this invention proposes a maritime area communication system.

[0036] Please see Figure 1 In one embodiment of the present invention, the maritime area communication system includes a multi-level communication platform, a distributed adaptive networking engine, and a cross-domain and cross-media communication gateway. The multi-level communication platform integrates a satellite communication module, a surface communication module, and an underwater communication module, and has local edge computing capabilities. The distributed adaptive networking engine is communicatively connected to the multi-level communication platform and is used for neighbor discovery, global topology aggregation, dynamic route calculation, and automatic hub switching for the multi-level communication platform. The cross-domain and cross-media communication gateway is communicatively connected to both the multi-level communication platform and the distributed adaptive networking engine, and is used for signal format conversion and intelligent routing between the three communication domains of satellite, surface, and underwater for the multi-level communication platform.

[0037] The technical solution of this invention employs a maritime regional communication system. By integrating satellite communication, surface communication, and underwater communication modules into a multi-level communication platform, various maritime platforms can flexibly support multiple communication methods, providing a foundation for cross-domain communication. Simultaneously, edge computing capabilities enable the platform to process data and execute logical judgments locally, reducing reliance on remote centers and enhancing autonomous operation and rapid response capabilities in abnormal situations such as disconnections. Furthermore, compared to the fixed topology and static routing configurations used in traditional maritime platform networks, this system utilizes a distributed adaptive networking engine. This engine enables neighbor discovery, global topology aggregation, dynamic route calculation, and automatic hub switching, giving the network self-awareness and self-organization capabilities. The system can automatically discover surrounding nodes, aggregate the entire network topology in real time, and dynamically calculate the optimal route accordingly. This maintains high network availability and allows for adaptive adjustments without manual intervention, significantly reducing operational costs. In addition, by setting up a cross-domain and cross-medium communication gateway for signal format conversion and intelligent routing between the three communication domains of satellite, surface and underwater, the adaptation and integration between heterogeneous communication protocols and signal systems are realized. This enables data to intelligently select the optimal transmission path across different communication domains according to business needs and real-time channel conditions, thereby ensuring the reliability and real-time performance of end-to-end transmission.

[0038] Overall, the technical solution of this embodiment integrates a multi-level communication platform, a distributed adaptive networking engine, and a cross-domain and cross-media communication gateway into a maritime area communication system. This system not only realizes the integrated intelligent coordination of satellite, surface, and underwater communications, but also endows the network with powerful self-sensing, self-repair, and self-optimization capabilities, enabling it to automatically adapt to complex marine dynamic environments and network topology changes.

[0039] The satellite communication module can utilize systems such as Inmarsat. Currently, satellite, surface, and underwater communication modules are typically deployed as three isolated communication domains in marine aquaculture communication systems. Each domain employs heterogeneous protocol stacks, encoding standards, and modulation methods, lacking a unified adaptation mechanism. When applications require data transmission across multiple communication domains (e.g., end-to-end data transmission between satellite, surface, and underwater), the communication system cannot intelligently assess the quality of each link or dynamically select the optimal transmission path based on real-time channel conditions, leading to frequent link switching failures, high data retransmission rates, and uncontrollable end-to-end latency. Therefore, this invention employs a cross-domain, cross-media communication gateway to achieve signal format conversion and intelligent routing between the satellite, surface, and underwater communication domains. This integrates the three communication domains into a cohesive whole, enabling seamless signal conversion and intelligent routing, ensuring the real-time performance and reliability of critical services.

[0040] Furthermore, it should be noted that the distributed adaptive networking engine is the core innovation of this invention. It transforms the traditional "pre-configured static topology" mode into a "dynamic adaptive self-healing" mode, enabling the marine aquaculture communication network to possess self-sensing, self-adjusting, and self-repairing capabilities. This distributed adaptive networking engine adopts an architecture design of decentralized decision-making, distributed information aggregation, and local intelligent execution. Each platform can independently assess its neighboring environment, while periodically aggregating information to form a global consensus. The entire operation cycle of the distributed adaptive networking engine is divided into three stages: neighbor discovery, global topology aggregation, and routing table calculation and distribution, forming a self-circulating, efficient, and highly fault-tolerant autonomous network system.

[0041] In embodiments of the present invention, the distributed adaptive networking engine is configured to perform the following operations: a neighbor discovery operation, which constructs and maintains a local adjacency table containing neighbor node status and link quality information by periodically broadcasting and receiving greeting messages; a channel quality assessment operation, which calculates a comprehensive link quality score based on the signal-to-noise ratio, end-to-end delay, and packet loss rate of each communication link, and dynamically activates or isolates corresponding links based on the comprehensive link quality score; a multi-platform role dynamic election operation, which calculates a hub score based on the remaining energy, communication range, computing power, and network centrality of each platform in the multi-level communication platform, and dynamically elects a master hub and forms a candidate hub pool based on the hub score; a hierarchical information aggregation operation, which collects and distributes the entire network topology information in a tree-like aggregation manner by constructing a spanning tree rooted at the master hub; and a reinforcement learning route optimization operation, which calculates and continuously optimizes the optimal transmission path for source-target node pairs based on Markov decision processes and Q-learning algorithms.

[0042] It should be noted that the greeting message refers to the HELLO packet, a communication message used in the Open Shortest Path First (OSPF) communication protocol to verify network adjacency relationships. It maintains neighbor relationships through the periodic exchange of Hello messages. Signal-to-noise ratio (SNR) is the ratio of signal power to noise power. End-to-end delay refers to the total time delay in the transmission of data packets from the sender to the receiver. This parameter is composed of four factors: node processing delay, queuing delay, transmission delay, and propagation delay. Its mathematical model can be expressed as the sum of the delays of each link. Packet loss rate is the ratio of the number of data packets lost during network transmission to the total number of packets sent. It is a core indicator for evaluating computer network performance and a key monitoring parameter for Quality of Service (QoS). A Markov Decision Process (MDP) is a mathematical model for sequential decision-making. It simulates stochastic policies and rewards achievable by an agent in environments where the system state possesses the Markov property (a stochastic process is conditionally independent of its past states given its current state). Q-learning is a model-free reinforcement learning algorithm used to solve sequential decision-making problems within Markov Decision Processes (MDPs). Its core objective is to learn an optimal policy that maximizes the cumulative reward during the agent's interaction with the environment.

[0043] In the above embodiments, the distributed adaptive networking engine configures and executes neighbor discovery operations, enabling each node in the network to automatically discover, identify, and continuously track its surrounding neighbor nodes by periodically broadcasting and receiving greeting messages. This dynamically constructs and maintains a local adjacency table containing real-time status and link quality information of neighbor nodes, solving the problem that traditional solutions rely on manual pre-configuration of node information and cannot automatically perceive changes in network members. This lays a dynamic and accurate data foundation for the self-organization of the network.

[0044] Furthermore, in the above embodiments, by performing a channel quality assessment operation, that is, calculating a comprehensive link quality score based on multiple dimensions such as signal-to-noise ratio, end-to-end delay, and packet loss rate, and dynamically activating high-quality links or isolating low-quality links accordingly, the system achieves a key paradigm shift from the traditional "topology connectivity" to "quality availability". It can proactively avoid signal fading and interference caused by the complex maritime environment, effectively improve the reliability and stability of data transmission, and solve the problem of frequent data retransmission and latency surge caused by link quality deterioration in traditional solutions.

[0045] Meanwhile, in the above embodiments, by performing a multi-platform role dynamic election operation, the hub score is calculated based on comprehensive factors such as remaining energy, communication range, computing power and network centrality, and a primary hub and a candidate hub pool are dynamically elected. This endows the network core nodes with automatic optimization and hot backup switching capabilities. When the primary hub fails due to fault or energy depletion, the candidate hub can quickly take over, thereby achieving smooth and rapid recovery of the network core control functions without manual intervention, which greatly improves the robustness and continuous service capability of the network.

[0046] Furthermore, in the above embodiments, by performing hierarchical information aggregation operations—that is, constructing a spanning tree rooted at the main hub and collecting and distributing the entire network topology information in a tree-like aggregation manner—compared to the traditional flooding broadcast mechanism (whose message complexity is O(n^2)), this approach achieves significantly better results. 2 The tree-based aggregation method (with a message complexity of O(n)) greatly reduces the bandwidth consumption and propagation delay of network control messages, improves the efficiency of network-wide state synchronization, and enables the network to quickly form a unified and accurate global view, thus providing a guarantee for efficient dynamic routing calculation.

[0047] Finally, in the above embodiments, by performing reinforcement learning route optimization operations, that is, by modeling based on Markov decision processes and applying the Q-learning algorithm to calculate and continuously optimize the optimal transmission path for source-target node pairs, the system no longer relies on fixed, predefined rules, but can autonomously learn and adapt to complex, time-varying marine channel environments based on historical transmission effects (such as latency, packet loss, and energy consumption), automatically discover and tend towards the forwarding strategy with the best long-term performance, thereby dynamically optimizing the data transmission efficiency and resource utilization of the entire network, and significantly improving the network's intelligent adaptability and overall performance in the face of environmental changes and fluctuations in service demands.

[0048] Please refer to Figure 2 , Figure 2 An example of neighbor discovery and adjacency establishment is given. The neighbor discovery operation can be carried out in the following steps: a. Originally, each platform node is isolated. The platform node broadcasts Hello messages (containing information such as node ID, energy, and communication range) to the surrounding area, while listening for broadcast messages from other nodes; b. After receiving the neighbor message, the node records the neighbor information in the local adjacency table and creates a link quality monitoring record for each neighbor; the node establishes a two-way heartbeat with each neighbor, and periodically sends heartbeat messages to maintain the connection. If there is no response after 3 consecutive heartbeats, the neighbor is considered to be disconnected; c. After the adjacency table is stable, it switches to a convergence structure.

[0049] Please see Figure 3In an embodiment of the present invention, the channel quality assessment operation includes: calculating the signal-to-noise ratio using a sliding window average; tracking the trend of end-to-end delay using an exponentially weighted moving average algorithm; calculating the packet loss rate; calculating the comprehensive link quality score for each link using a comprehensive link quality scoring model based on the signal-to-noise ratio, the trend of end-to-end delay, and the packet loss rate; automatically isolating the corresponding link and retaining monitoring when the comprehensive link quality score of a single link is lower than a preset threshold; and automatically activating the isolated link when the comprehensive link quality score of the isolated link reaches the preset threshold.

[0050] In the above embodiments, by using a sliding window average to calculate the signal-to-noise ratio (SNR), the system can effectively smooth out the drastic fluctuations in instantaneous signal strength caused by environmental factors such as multipath reflections at sea and splashing waves. This results in a more stable and representative channel SNR assessment value, avoiding potential misjudgments based on a single or few samples, and providing reliable basic data for subsequent link quality decisions. Simultaneously, by employing an exponentially weighted moving average algorithm to track end-to-end delay trends, this algorithm assigns higher weight to recent delay data, enabling the system to sensitively capture continuous deterioration or improvement trends in delay, rather than focusing solely on instantaneous delay values. This allows for a more accurate distinction between temporary network congestion and persistent link quality degradation, enhancing the accuracy of the judgment. Combined with the calculation of packet loss rate, the system achieves a quantitative assessment of link reliability. Subsequently, based on the processed signal-to-noise ratio, latency trend, and packet loss rate, a comprehensive link quality score for each link is calculated using a preset weighted formula. This scoring model integrates multiple heterogeneous physical layer and link layer indicators into a unified and comparable quantitative value, providing a core and objective decision-making basis for cross-link performance comparison and intelligent routing selection.

[0051] When the overall score of a single link falls below a preset threshold, the system automatically performs isolation, removing the link from the currently available paths. This proactively avoids problems such as packet loss, retransmission, and increased latency that may be caused by low-quality links, ensuring the immediate effectiveness of business data transmission and overall network throughput efficiency. The isolated link remains in the adjacency list and is monitored, allowing the system to continuously track its recovery status. Once the overall score of the isolated link reaches or exceeds the preset threshold due to improved channel conditions, the system automatically reactivates it as a usable link. This mechanism achieves closed-loop, automated lifecycle management of link status, forming a complete lifecycle management system of "discovery-evaluation-monitoring-isolation-recovery." This enables the network to flexibly adjust its available resources in response to dynamic changes in the marine communication environment, continuously optimizing network connectivity and performance without manual intervention, significantly improving the network's adaptability and the level of intelligent operation and maintenance.

[0052] Compared to traditional mobile ad hoc networks (MANETs), where passive fault detection typically takes 30-60 seconds, the channel quality assessment operation used in the above embodiment, through continuous monitoring and dynamic scoring, allows the system to detect link quality degradation and trigger route adjustments within 5-10 seconds. This represents a 5-10 times faster response time compared to the passive fault detection of traditional MANETs.

[0053] As a feasible implementation method, the sliding window average can use a window size of 20-50 data packets to smooth SNR fluctuations; the attenuation factor α of the exponentially weighted moving average (EWMA) can be set between 0.3 and 0.5, and the attenuation factor is used to control the weight attenuation rate to track the delay trend. Packet loss rate can distinguish between radio fading packet loss and congestion packet loss.

[0054] Additionally, it should be noted that the comprehensive link quality scoring model is a mathematical model, and the calculation formula used in the comprehensive link quality scoring model can be found in the following formula:

[0055]

[0056] In the formula, Q L To provide a comprehensive link quality score; SNR norm To standardize the signal-to-noise ratio, ; NL norm To standardize the delay, The results were obtained using EWMA smoothing: ,in, For the current moment, standardization delay, α The attenuation factor is the weighting coefficient. This is the current delay measurement value. Delay for normalization from the previous moment; L plr For packet loss rate, ; w 1. W 2. w 3 represents the three weighting coefficients, whose values ​​can be found in the following weighting strategy table:

[0057] Weighting strategy table (empirical values):

[0058]

[0059] The preset threshold can be set to 0.6, which is the overall link quality score of a single link. Q L When the overall link quality score falls below a preset threshold of 0.6, the link is isolated; when the overall link quality score of the isolated link is... QL Once the value reaches 0.6 again, the link is activated.

[0060] In an embodiment of the present invention, the multi-platform role dynamic election operation includes: calculating a hub score based on the remaining energy, communication range, computing power, and network centrality of each platform in the multi-level communication platform; ranking each platform based on the hub score, with the platforms ranking in the top preset proportion becoming candidate platforms, electing the candidate platform ranked first as the main hub, and forming a candidate hub pool with the remaining platforms in the candidate platform; if the main hub fails, the platform ranked first in the candidate hub pool is elected as the new main hub.

[0061] In the above embodiments, the multi-platform role dynamic election operation first calculates the hub score of each platform based on four dimensions: remaining energy, communication range, computing power, and network centrality. This transforms the selection of hub nodes from a traditional model that relies on a single indicator or manual designation to a multi-factor, quantitative scientific evaluation. This can more comprehensively and objectively reflect the platform's comprehensive potential as the network core in terms of energy sustainability, communication coverage, data processing capabilities, and topological criticality. As a result, it ensures that the elected main hub is the most competent node for network coordination and management tasks, effectively avoiding network performance bottlenecks or premature failures caused by one-sided evaluations.

[0062] Secondly, the multi-platform role dynamic election operation is based on hub scoring and ranking. Platforms ranked in the top 20% (e.g., the top 20%) are identified as candidate platforms, and the top-ranked platform is elected as the main hub. The remaining candidate platforms automatically form a backup hub pool. This mechanism not only achieves automatic and fair election of the main hub through clear scoring and ranking, eliminating the need for manual designation, but also constructs a "hot backup" system based on capability ranking. The main hub serves as the currently active core, while the nodes in the backup hub pool serve as pre-selected backup forces with the ability to take over, providing pre-organized redundancy guarantees for the continuous operation of the network's core functions.

[0063] Finally, the multi-platform role dynamic election operation elects the platform ranked first in the candidate hub pool as the new primary hub when the primary hub fails. This scheme enables the system to immediately select and activate a new core node from the ready candidate hub pool in emergency situations where the primary hub fails (such as due to energy depletion, communication interruption, or hardware damage). Instead of restarting the time-consuming network discovery and election process, the system can achieve a rapid and smooth transfer of network control. This reduces the network interruption and recovery time, which may take several minutes (e.g., 5-10 minutes) or even longer in traditional schemes, to seconds (e.g., 90-120 seconds). This greatly enhances the network's self-healing ability, service continuity, and overall robustness in the face of core node failures, and significantly reduces the reliance on manual operation and maintenance intervention and the impact of failures on production safety.

[0064] When the main hub fails due to energy depletion (e.g., below 20%), communication interruption (e.g., no response for more than 30 seconds), or reduced centrality, the nodes in the alternative hub pool immediately start competing, and the first-ranked node automatically takes over. It notifies the entire network through a hub transition message and completes a smooth handover within 2-3 cycles (60-90 seconds), achieving seamless switching.

[0065] Among them, hub score Score The calculation formula can be found in the following formula:

[0066]

[0067] In the formula: The percentage of remaining energy. For communication coverage, For the computing power of the node processor, For the betweenness centrality of nodes, α , β , γ , δ All are weighting coefficients (empirical values ​​are 0.3, 0.2, 0.2, and 0.3 respectively).

[0068] Please see Figure 4 , Figure 4 A schematic diagram of the hub election results is provided; in Figure 4In the example, after scoring each platform node, the hub score is S=90.1 for the workboat, S=72.4 for the net cage, S=63.2 for the floating platform, and S=55.8 for the buoy. Therefore, after sorting the platform nodes based on the hub scores, the order is: 1. Workboat, 2. Net cage, 3. Floating platform, 4. Buoy. Thus, the workboat is elected as the main hub, while the net cages and floating platforms with the highest rankings become candidate nodes in the candidate hub pool. The workboat node, as the main hub, broadcasts messages such as "I am the hub" to other nodes, allowing other nodes to confirm the main hub.

[0069] In embodiments of the present invention, the hierarchical information aggregation operation includes: constructing a spanning tree rooted at the main hub; wherein each platform corresponds to one node; in the uplink aggregation phase, child nodes send their local adjacency lists to their parent nodes, and the parent nodes in the intermediate layers merge all child node information and continue to pass it uplink so that the main hub obtains a complete global adjacency matrix; each node independently maintains its local adjacency list to support fault recovery within a 1-hop range, and when a fault is detected in an adjacent node, it uses local alternative adjacencies for forwarding; in the downlink broadcast phase, after the main hub calculates the routing table, it broadcasts it layer by layer to the entire network through the spanning tree structure.

[0070] Please see Figure 5 In the above embodiments, the hierarchical information aggregation operation first constructs a spanning tree rooted at the main hub, mapping all platforms in the network as tree nodes. This optimizes the collection and distribution of network topology information from the traditional, high-overhead "flooding broadcast" mode to a structured "tree aggregation" mode. This structural optimization ensures that the propagation path of control messages (such as topology status and routing tables) is determined and loop-free, thereby avoiding broadcast storms and significantly reducing message complexity from O(n²) to O(n). This greatly reduces network bandwidth consumption and control signaling latency, laying a scalable foundation for the efficient networking of large-scale maritime nodes.

[0071] Secondly, during the uplink aggregation phase, this operation sends the local adjacency list to the parent node through the child nodes, and the intermediate parent node merges all the child node information and continues to pass it up, ultimately enabling the main hub to obtain the complete global adjacency matrix. This bottom-up, layer-by-layer aggregation method achieves efficient collection of the entire network topology state with minimal message interaction, ensuring that the main hub can make core decisions such as routing calculations based on an accurate and timely global view, overcoming the problems of inconsistent state information or slow convergence in traditional distributed algorithms.

[0072] Furthermore, the significant innovation of this operation lies in the introduction of a dual-track maintenance mechanism: on the one hand, efficient aggregation of the global topology is achieved through spanning trees; on the other hand, each node is required to independently maintain its local adjacency table (i.e., neighbor information within a one-hop range). This design allows a node to immediately utilize alternative neighbors in its local adjacency table for data forwarding when it detects a failure in its direct neighbor, without waiting for the main hub to recalculate and issue a global route. This achieves a perfect balance between global path optimization and rapid local fault recovery, significantly shortening the data transmission interruption time caused by single-point link failures and improving the continuity of service transmission.

[0073] Finally, during the downlink broadcast phase, after the main hub calculates the routing table, it broadcasts it layer by layer to the entire network through the constructed spanning tree structure. This ensures that routing update commands can be delivered to every node efficiently and orderly, and utilizes the same spanning tree, eliminating the need to establish additional control channels for command distribution, maintaining network structure consistency, and further reducing control overhead. Overall, this hierarchical information aggregation operation, through the coordination of spanning tree structure, uplink aggregation, local adjacency table maintenance, and downlink broadcasting, constructs a low-overhead, high-efficiency network state management and synchronization mechanism with rapid local self-healing capabilities, significantly improving the intelligence level and overall performance of maritime dynamic network topology management.

[0074] The leaf nodes package their local adjacency lists into topology information packets and send them to their parent nodes. These local adjacency lists may contain information such as neighbor IDs, link quality, and energy status.

[0075] Specifically, please refer to Figure 6 , Figure 6 The tree-shaped node topology graph formed after the hierarchical information aggregation operation according to the above embodiment is given.

[0076] In an embodiment of the present invention, the reinforcement learning route optimization operation includes: the master hub constructs the state space, action space, and reward function of a Markov decision process based on the global topology; wherein, the reward function is a weighted combination of latency, reliability, and energy consumption; the master hub maintains a Q-value table for each source-target node pair based on the Q-learning algorithm; after aggregating the global topology, the master hub calculates the optimal Q-value and generates a global routing table, and broadcasts the global routing table to the entire network; each node records the actual forwarding effect and periodically provides performance feedback; the master hub incrementally learns and updates the Q-value table based on the performance feedback according to the reward function.

[0077] In the above embodiments, reinforcement learning route optimization operations systematically model the complex maritime network routing problem as a sequential decision-making process by constructing the state space, action space, and reward function of a Markov decision process based on the global topology by the main hub. The state space comprehensively characterizes the dynamic environment, including network topology, link quality, and node energy, while the action space contains all feasible end-to-end paths. This provides a standardized mathematical model foundation for intelligent path selection, overcoming the limitations of traditional heuristic or static algorithms in accurately describing and responding to time-varying environmental conditions. Specifically, the operation designs the reward function as a weighted combination of latency, reliability, and energy consumption. This expands the path optimization objective from a single dimension (such as only the shortest hop count) to a flexibly configurable, multi-objective trade-off comprehensive evaluation system. This allows for dynamic adjustment of the optimization focus based on actual business needs (such as low latency for real-time control, high reliability for data acquisition, and energy consumption considerations for mobile platforms), thereby supporting differentiated quality of service assurance.

[0078] Building upon this foundation, by employing the Q-learning algorithm and having the main hub maintain and update the Q-value table for each source-destination node pair, the system can estimate the expected cumulative returns of different paths over long-term operation through iterative learning. Then, based on the aggregated global topology, it calculates the optimal Q-value for all node pairs at once and generates a globally optimal routing table. This mechanism eliminates reliance on preset fixed rules or simple reactions to instantaneous states for routing decisions, instead enabling proactive optimization based on historical experience. It can automatically discover and tend towards transmission strategies with superior long-term performance in complex, non-stationary ocean channel environments. Subsequently, the main hub broadcasts this global routing table to the entire network, ensuring consistency in the forwarding behavior of all nodes and that network traffic is routed according to a unified optimization strategy.

[0079] Subsequently, by having each node record actual forwarding performance (such as latency, packet loss, and energy consumption) and periodically feed back performance data to the main hub, the system constructs a closed-loop feedback loop from decision-making to execution and evaluation. Based on this real-world environmental feedback and service performance data, the main hub incrementally learns and updates the Q-value table according to the aforementioned reward function, enabling the routing strategy to continuously self-correct and self-improve, automatically adapting to changes in network topology, fluctuations in link quality, and migration of service models. This continuous learning and optimization process allows the entire communication system to continuously accumulate experience from actual operation, intelligently improving the efficiency, reliability, and energy efficiency of data transmission across the entire network, significantly enhancing the network's adaptability and overall performance in the face of uncertainties in the marine environment.

[0080] As a feasible implementation, the main hub constructs an MDP based on the global topology, including a state space, action space, and reward function. The state space contains the current network topology, the quality of all links, and the energy status of nodes. The action space comprises all feasible paths from the source to the target. For each source-target node pair, a set of candidate paths is pre-computed using the K-shortest path algorithm (K=3-5) as the action space. The reward function comprehensively considers three dimensions: latency, reliability, and energy consumption, forming a comprehensive evaluation through weighted combinations.

[0081] In reinforcement learning-based route optimization, after constructing the MDP, the master hub maintains a Q-value table for each node pair and continuously updates the Q-values ​​based on actual feedback using the Q-learning algorithm. Then, based on the Q-value table, it selects the path with the highest Q-value for each source-destination pair and generates a global routing table (containing information such as next hop and alternative paths). Subsequently, the master hub broadcasts the routing table layer by layer to the entire network through a spanning tree, and each node extracts relevant routing entries to update its local forwarding table. Each node records the actual forwarding performance (latency, packet loss, energy consumption) and periodically summarizes it into performance feedback messages, which are then reported to the master hub. Finally, the master hub adjusts the Q-value table based on the performance feedback, enabling the system to automatically learn the optimal routing strategy and adapt to environmental changes.

[0082] In embodiments of the present invention, the multi-level communication platform includes a hub-level platform, a node-level platform, a mobile-level platform, and a fixed-level platform. Each level of platform is interconnected via a distributed adaptive networking engine. Alternatively, the hub-level platform integrates a satellite communication module, a surface communication module, an underwater communication module, and a 4G / 5G multi-mode receiving unit. The hub-level platform is used for global network topology maintenance, data aggregation, task allocation, and resource scheduling. Alternatively, the node-level platform is used for data acquisition and inter-node relay forwarding, supporting local edge computing and real-time sensor data processing. Alternatively, the mobile-level platform employs short-range communication and local collaborative algorithms for area inspection, emergency response, and problem diagnosis. Alternatively, the fixed-level platform integrates a satellite communication module, a surface communication module, an underwater communication module, and a cellular network multi-mode receiving unit. The fixed-level platform serves as an access point to the 5G / satellite network, providing long-distance backbone link coverage in offshore areas.

[0083] In the above embodiments, by clearly dividing the multi-level communication platform into hub level, node level, mobile level and fixed level, and interconnecting them through a distributed adaptive networking engine, a heterogeneous network architecture with clear layers, distinct roles and collaborative operation is constructed. This architecture achieves vertical division of labor and horizontal expansion of functions: the hub-level platform, with its integrated satellite, surface, underwater, and 4G / 5G multi-mode communication capabilities and powerful computing power, acts as the "brain" of the network, specializing in global topology maintenance, data aggregation, and resource scheduling, solving the problems of lacking a strong coordinating core and difficulty in achieving unified cross-domain management in traditional solutions; the node-level platform focuses on data acquisition and relay forwarding, and combined with local edge computing capabilities, it can perform real-time preprocessing and analysis of sensor data at the network edge, which not only reduces the backhaul bandwidth pressure but also improves the system's response speed to local events; the mobile-level platform utilizes its mobility and short-range collaboration capabilities to flexibly undertake dynamic tasks such as regional inspection and emergency response, enhancing the network's coverage elasticity and proactive operation and maintenance capabilities; and the fixed-level platform, as a reliable access point to land-based 5G / satellite networks, provides stable long-distance backbone links covering the far sea area, fundamentally ensuring reliable connections between the maritime self-organizing network and the wide area network. The platforms at each level interconnect and dynamically switch roles through a unified networking protocol stack, forming an organic whole that has both centralized optimization capabilities and distributed survivability and functional flexibility. This effectively overcomes the shortcomings of existing technologies, such as single-function platforms, poor interoperability, and difficulty in forming a unified and efficient network system.

[0084] It should be noted that hub-level platforms can include large aquaculture vessels and maintenance vessels, equipped with the most powerful hardware capabilities among all platform levels (such as high-power transmission, multi-band transceiver, and processors with more than 8GB of memory); node-level platforms can include fixed or semi-fixed facilities such as aquaculture cages, floating platforms, and buoys, equipped with medium computing power (such as 4-core CPUs and 4-8GB of memory); mobile platforms can include unmanned surface vessels, drones, and underwater robots (such as remotely operated vehicles (ROVs), autonomous underwater vehicles (AUVs), and full-ocean-depth autonomous remotely operated vehicles (ARVs)), with flexible mobility; fixed-level platforms can include land-based base stations and offshore relay stations, which serve as access points to 5G / satellite networks, providing long-distance backbone link coverage in offshore areas.

[0085] In an embodiment of the present invention, when the maritime area communication system is started, the platforms at each level are started in sequence according to the priority order of fixed platform, hub platform, node platform and mobile platform, and a distributed adaptive networking engine is loaded.

[0086] In the above embodiments, by stipulating that when the maritime area communication system is started, each level of platform is started in strict accordance with the priority order of fixed-level platform, hub-level platform, node-level platform, and mobile-level platform, and a distributed adaptive networking engine is loaded, a systematic process guarantee is provided for the reliable and orderly initialization of the entire network.

[0087] Specifically, the system first activates the fixed-level platforms (land-based base stations / maritime relay stations), enabling the establishment of stable backbone connections with the land-based 5G core network or satellite network. This pre-builds a reliable "backhaul channel" for the subsequent access and remote management of all maritime platforms, ensuring wide-area connectivity from the outset. Next, the hub-level platforms are activated, allowing them to initialize their multi-mode communication hardware and load the networking protocol stack based on the network provided by the fixed-level platforms, quickly assuming network organization and coordination functions within their coverage area. Then, the node-level platforms are activated, enabling them to be discovered and integrated into the network by the ready hub-level platforms, and to begin performing data collection and relay tasks. Finally, the mobile-level platforms are activated, allowing them to accept task scheduling in a stable network environment. This activation sequence, from the stable core to the dynamic edge, effectively avoids potential link contention, role conflicts, and connection failures due to unprepared resources during network initialization. Meanwhile, all platforms load a distributed adaptive networking engine during startup, ensuring that the network has self-awareness and self-organization capabilities from the initialization phase. This lays a solid foundation for the entire system to quickly enter a stable and adaptive operating state, significantly improving the reliability and efficiency of system deployment.

[0088] Please see Figure 7 In embodiments of the present invention, the cross-domain, cross-medium communication gateway includes a satellite-to-ground gateway and a surface-to-underwater gateway. The satellite-to-ground gateway is used to convert satellite signals to surface wireless signals for a multi-level communication platform; the surface-to-underwater gateway is used to convert the frequency domain of surface wireless signals to underwater acoustic signals for a multi-level communication platform.

[0089] In the above embodiments, by including a satellite-to-ground gateway and a surface-to-underwater gateway in the cross-domain, cross-media communication gateway, a proprietary, functionally decoupled hardware implementation path is provided for the deep integration of heterogeneous communication domains. The satellite-to-ground gateway is specifically responsible for the conversion between satellite signals and surface wireless signals. Through the integration of technologies such as Adaptive Coding and Modulation (ACM), it can dynamically adapt to the high latency, variable throughput, and susceptibility to weather interference characteristics of satellite links (Ku / Ka bands), achieving precise adaptation and reliable conversion of signal formats, encoding methods, and protocol stacks. This seamlessly integrates wide-area coverage but complex access satellite communication into near-shore networks. Simultaneously, the surface-to-underwater gateway is specifically responsible for the frequency domain conversion between surface wireless signals (e.g., 2.4 / 5.8 GHz) and underwater acoustic signals (e.g., 8-64 kHz). It employs algorithms such as signal remapping, multipath interference cancellation, and time-varying channel equalization to overcome the significant differences in propagation media and signal attenuation at the water-air interface, achieving connectivity between two distinct physical layer technologies: surface radio frequency and underwater acoustics. The collaborative work of these two gateways has completely broken down the long-standing technical barriers between the three communication domains of satellite, surface radio, and underwater acoustics at the physical and link layers. This enables data to be intelligently selected and smoothly transmitted across different media according to business needs, building an end-to-end unified communication pipeline. This effectively solves the problems of manual bridging, difficult protocol conversion, and inability to guarantee service quality in cross-domain transmission of satellite, surface radio, and underwater acoustic communications on marine platforms.

[0090] The satellite-to-ground gateway is responsible for converting satellite signals to surface radio signals. It dynamically adjusts the coding scheme (from low-rate QPSK 3 / 4 to high-rate 16-QAM 5 / 6) based on channel quality using Adaptive Coding and Modulation (ACM) technology. Simultaneously, it implements flow control and forward error correction (FEC) for the reverse link, ensuring reliability in harsh fading environments. The surface-to-underwater gateway is responsible for frequency domain conversion between radio and acoustic signals. The system uses a signal remapping algorithm to convert the 2.4 / 5.8 GHz modulated signals from the surface into 8-64 kHz FSK or OFDM acoustic signals from underwater. It also performs multipath interference cancellation, time-varying channel equalization, and adaptive power adjustment (to avoid harm to aquatic life).

[0091] The following are specific implementation examples. Converting surface wireless network signals to underwater sensor acoustic signals: The data content of high-frequency surface wireless signals (2.4 / 5.8GHz) can be extracted using a surface-to-underwater gateway and remapped into low-frequency underwater acoustic signals (8-64kHz). Simultaneously, multipath cancellation and time-varying channel equalization are performed to eliminate interference from the complex underwater reflection environment. Converting ground signals from key platforms such as construction vessels to satellite signals: The satellite-to-ground gateway uses Adaptive Coding Modulation (ACM) technology to automatically adjust decoding parameters based on the real-time signal-to-noise ratio, maximizing the decoding success rate; subsequently, transcoding is performed to convert the surface wireless signal to satellite encoding.

[0092] Please see Figure 1 In embodiments of the present invention, the maritime area communication system further includes a converged network management center, which is communicatively connected to the distributed adaptive networking engine and the cross-domain, cross-media communication gateway. The converged network management center is configured to perform the following operations: edge computing offloading and local caching operations, performing local data caching when the 5G link is interrupted, and performing data analysis on the mobile edge computing platform; network slicing application operations, applying for differentiated network slices from the 5G core network according to service requirements; protocol conversion and traffic decision operations, converting the private protocol of the distributed adaptive networking engine into the 5G standard protocol, and intelligently routing data streams to the 5G core network, satellite link, or maritime ad hoc network according to service type; and global service orchestration and task allocation operations, calculating an adaptability score based on a global optimization objective using an allocation algorithm, and dynamically allocating control tasks to the platform with the highest score among the multi-level communication platforms for execution based on the adaptability score.

[0093] In the above embodiments, by setting up a converged network management center and configuring it to perform a series of operations such as edge computing offloading and local caching, network slicing application, protocol conversion and traffic decision-making, and global service orchestration and task allocation, deep and native collaboration between the maritime ad hoc network and the land-based cellular network (5G / 6G) is realized, and an integrated intelligent management and control system of cloud-edge-device is constructed.

[0094] The edge computing offloading and local caching operations enable critical data to be automatically cached locally when network backhaul links (such as 5G) are interrupted. This data is then combined with a mobile edge computing (MEC) platform for real-time analysis (such as anomaly detection), ensuring the continuity of core monitoring and decision-making functions and significantly reducing reliance on stable backhaul links. Furthermore, the mobile edge computing platform can deploy various artificial intelligence models for real-time analysis of critical data, such as MobileNet and SqueezeNet.

[0095] Furthermore, the network slicing application process enables the management center to dynamically apply for dedicated URLLC, eMBB, or mMTC network slices from the 5G core network based on the differentiated Quality of Service (QoS) requirements (such as ultra-low latency, ultra-high reliability, and high bandwidth) of different services, such as aquaculture monitoring, emergency control, and video backhaul. This provides reliable carrier-grade network resources for critical maritime operations. URLLC slices feature ultra-reliable low latency (RTT ≤ 10ms) and reliability ≥ 99.999%, making them suitable for emergency control services. eMBB slices enhance mobile broadband with throughput ≥ 100Mbps, suitable for video surveillance and remote inspection services. mMTC slices support large-scale IoT, supporting ≥ 1000 connections, and are optimized for cost, making them suitable for large-scale sensor data communication services.

[0096] Furthermore, the protocol conversion and traffic decision-making operations efficiently convert the proprietary protocols used by the marine distributed adaptive networking engine into standard 5G protocols (such as NR Sidelink and the 3GPP TS 33.501 certification framework). Based on the real-time nature and importance of data services, it intelligently routes traffic to the optimal transmission path (e.g., emergency alarms use the 5G low-latency link, while routine data uses satellite or the self-organizing network), achieving refined utilization of network resources. Specifically, for critical data (such as aquaculture ecological alarms and water quality anomalies), priority can be given to routing to the 5G core network (low latency <10ms) to support emergency control; for periodic data (such as sensor sampling data), offline processing can be achieved via satellite or the marine self-organizing network (latency is tolerable); and for log data, delayed uploading or local storage is used.

[0097] Finally, the global business orchestration and task allocation operations are based on global optimization goals. The algorithm comprehensively evaluates the energy, distance, load and network quality of each platform, calculates the adaptability score and dynamically allocates control tasks to the optimal execution platform accordingly, thereby realizing unified scheduling and load balancing of computing and communication resources across platforms.

[0098] The synergistic effect of the above operations makes the converged network management center an intelligent hub connecting the maritime dynamic network and the land-based fixed network. It not only solves the problems of data silos and inefficient resource utilization caused by simply splicing the maritime network and the land network, but also enables the maritime regional communication system to have differentiated service guarantees and global resource optimization functions by introducing advanced 5G / 6G features such as network slicing and MEC.

[0099] The above description is merely an exemplary embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural transformations made using the contents of the present invention specification and drawings under the technical concept of the present invention, or direct / indirect applications in other related technical fields, are included within the patent protection scope of the present invention.

Claims

1. A maritime area communication system, characterized in that, include: A multi-level communication platform, which integrates a satellite communication module, a surface communication module and an underwater communication module, and has local edge computing capabilities; A distributed adaptive networking engine is communicatively connected to the multi-level communication platform. The distributed adaptive networking engine is used for neighbor discovery, global topology aggregation, dynamic route calculation, and automatic hub switching for the multi-level communication platform. The automatic hub switching performed by the distributed adaptive networking engine includes the following operations: dynamically electing a primary hub and forming a candidate hub pool based on the remaining energy, communication range, computing power, and network centrality of each platform in the multi-level communication platform. And when the primary hub fails, a platform in the alternative hub pool will be elected as the new primary hub; A cross-domain, cross-medium communication gateway is connected to the multi-level communication platform and the distributed adaptive networking engine. The cross-domain, cross-medium communication gateway is used to perform signal format conversion and intelligent routing between the three communication domains of satellite, surface and underwater for the multi-level communication platform. The cross-domain, cross-medium communication gateway includes a satellite-to-ground gateway and a surface-to-underwater gateway. The satellite-to-ground gateway is used to convert satellite signals to surface wireless signals for the multi-level communication platform. The satellite-to-ground gateway dynamically adjusts the coding method according to the channel quality through adaptive coding and modulation technology to achieve flow control and forward error correction coding of the reverse link. The surface-to-underwater gateway is used to perform frequency domain conversion between surface wireless signals and underwater acoustic signals for the multi-level communication platform. The surface-to-underwater gateway uses a signal remapping algorithm to convert the 2.4 / 5.8 GHz modulated signal on the surface into an 8-64 kHz FSK or OFDM acoustic signal underwater, thereby achieving multipath interference cancellation, time-varying channel equalization, and adaptive power adjustment.

2. The maritime area communication system as described in claim 1, characterized in that, The distributed adaptive networking engine is configured to perform the following operations: The neighbor discovery operation involves periodically broadcasting and receiving greeting messages to build and maintain a local adjacency table containing information on neighbor node status and link quality. The channel quality assessment operation calculates a comprehensive link quality score based on the signal-to-noise ratio, end-to-end delay, and packet loss rate of each communication link, and dynamically activates or isolates the corresponding link based on the comprehensive link quality score. The multi-platform role dynamic election operation calculates the hub score based on the remaining energy, communication range, computing power and network centrality of each platform in the multi-level communication platform, and dynamically elects the main hub and forms a candidate hub pool based on the hub score. The hierarchical information aggregation operation collects and distributes the entire network topology information in a tree-like aggregation manner by constructing a spanning tree rooted at the main hub. The reinforcement learning route optimization operation, based on Markov decision process and Q-learning algorithm, calculates and continuously optimizes the optimal transmission path for source-target node pairs.

3. The maritime area communication system as described in claim 2, characterized in that, The channel quality assessment operation includes: The signal-to-noise ratio is calculated using a sliding window averaging method. An exponentially weighted moving average algorithm is used to track the trend of the end-to-end delay; Calculate the packet loss rate; Based on the signal-to-noise ratio, the trend of the end-to-end delay, and the packet loss rate, the comprehensive link quality score for each link is calculated using a comprehensive link quality scoring model. When the overall link quality score of a single link is lower than a preset threshold, the corresponding link is automatically isolated while monitoring continues. When the overall link quality score of the isolated link reaches a preset threshold, the isolated link is automatically activated.

4. The maritime area communication system as described in claim 2, characterized in that, The multi-platform role dynamic election operation includes: The hub score is calculated based on the remaining energy, communication range, computing power, and network centrality of each platform in the multi-level communication platform. Based on the hub score, each platform is ranked. Platforms ranked in the top preset proportion of the hub score become candidate platforms. The candidate platform ranked first is elected as the main hub, and the remaining platforms in the candidate platforms form the alternative hub pool. If the primary hub fails, the platform with the highest hub score in the candidate hub pool will be elected as the new primary hub.

5. The maritime area communication system as described in claim 2, characterized in that, The hierarchical information aggregation operation includes: Construct a spanning tree rooted at the main hub; wherein each platform corresponds to one node; During the uplink aggregation phase, child nodes send their local adjacency lists to their parent nodes. The parent nodes in the intermediate layer merge all child node information and continue to pass it up, so that the main hub can obtain a complete global adjacency matrix. Each node independently maintains its local adjacency list, and when a neighboring node failure is detected, it uses local alternative adjacencies for forwarding. During the downlink broadcast phase, after the main hub calculates the routing table, it broadcasts it to the entire network layer by layer through a spanning tree structure.

6. The maritime area communication system as described in claim 2, characterized in that, The reinforcement learning route optimization operation includes: The main hub constructs the state space, action space, and reward function of a Markov decision process based on the global topology; wherein the reward function is a weighted combination of delay, reliability, and energy consumption. The main hub maintains a Q-value table for each source-target node pair based on the Q-learning algorithm. After aggregating the global topology, the main hub calculates the optimal Q-value and generates a global routing table, which is then broadcast to the entire network. Each node records the actual forwarding effect and provides performance feedback periodically. The main hub incrementally learns and updates the Q-value table based on the performance feedback and the reward function.

7. The maritime area communication system as described in any one of claims 1 to 6, characterized in that, The multi-level communication platform includes a hub-level platform, a node-level platform, a mobile-level platform, and a fixed-level platform, and each level of platform communicates and connects with each other through the distributed adaptive networking engine. And / or, the hub-level platform integrates the satellite communication module, surface communication module, underwater communication module and 4G / 5G multi-mode receiving unit, and the hub-level platform is used for global network topology maintenance, data aggregation, task allocation and resource scheduling; And / or, the node-level platform is used for data acquisition and inter-node relay forwarding, and supports local edge computing and real-time sensor data processing; And / or, the mobile-level platform employs short-range communication and local collaboration algorithms for regional inspection, emergency response, and problem diagnosis; And / or, the fixed-level platform integrates the satellite communication module, the surface communication module, the underwater communication module, and the cellular network multimode receiver unit.

8. The maritime area communication system as described in claim 7, characterized in that, When the maritime area communication system is started, the platforms at each level are started sequentially according to the priority order of the fixed-level platform, the hub-level platform, the node-level platform, and the mobile-level platform, and the distributed adaptive networking engine is loaded.

9. The maritime area communication system as described in any one of claims 1 to 6, characterized in that, The maritime area communication system also includes a converged network management center, which is communicatively connected to the distributed adaptive networking engine and the cross-domain cross-media communication gateway. The converged network management center is configured to perform the following operations: Edge computing offloading and local caching operations are performed to cache data locally when the 5G link is interrupted and to perform data analysis on the mobile edge computing platform. The network slicing application process involves requesting differentiated network slices from the 5G core network based on business needs. Protocol conversion and traffic decision-making operations convert the private protocol of the distributed adaptive networking engine into the 5G standard protocol, and intelligently route the data flow to the 5G core network, satellite link or maritime ad hoc network according to the service type. The global business orchestration and task allocation operation is based on the global optimization goal. It uses an allocation algorithm to calculate the adaptability score and dynamically allocates the control task to the platform with the highest score among the multi-level communication platforms according to the adaptability score.

Images