Method for dynamic topology control of underwater sensors

Abstract

The application provides an underwater sensor dynamic topology control method, which comprises the following steps: constructing a first multi-objective optimization function and a constraint set; using a simulated annealing algorithm to globally optimize the first multi-objective optimization function under the constraint of the constraint set, and solving the expected positions of sensor nodes at the moment when all the sensor nodes are deployed; using a meandering ocean current movement model to inversely deduce the actual deployment positions of the sensor nodes based on the expected positions of the sensor nodes and the drift time lengths of the sensor nodes; and deploying the sensor nodes according to the actual deployment positions. By the technical scheme, the nodes do not actively adjust their positions to consume extra energy, but make full use of the characteristics of the nodes moving with ocean currents to realize full connectivity and high coverage of the network.

Description

Technical Field

[0001] This application relates to the field of underwater sensor network technology, specifically to a dynamic topology control method for underwater sensors. Background Technology

[0002] Underwater sensor networks (UWSNs) are widely used in marine resource exploration, marine disaster early warning, and marine environmental data acquisition. Due to the inherent limitations of the underwater environment, underwater acoustic channels suffer from characteristics such as transmission delays, limited available bandwidth, significant multipath effects, and continuous node drift driven by ocean currents. These factors pose significant challenges to the topology control of UWSNs, making efficient and reliable dynamic topology control technology a hot research topic in this field.

[0003] Node deployment and topology optimization are the two core aspects of topology control in underwater sensor networks. Coverage and connectivity are key metrics for evaluating network deployment performance: coverage characterizes the sensor network's ability to perceive and cover the underwater monitoring area; connectivity characterizes the ability of network nodes to communicate with surface convergence nodes. There is an inherent trade-off between the two: pursuing a higher connectivity rate requires shortening the communication distance between nodes, which increases overlapping of sensing coverage and consequently reduces the overall network coverage.

[0004] Existing underwater topology control schemes mainly fall into two categories: The first category involves deployment schemes where nodes actively adjust their positions. These schemes utilize miniature thrusters on the nodes, relying on their own power to move to the target deployment location. While this achieves a certain level of coverage and connectivity, the active movement of nodes incurs significant energy costs. Furthermore, the difficulty in replacing batteries for underwater sensor nodes and the substantial energy consumption significantly shorten the network's lifespan. The second category is static deployment schemes, where nodes are fixed in position after deployment and do not require further adjustment. However, continuous ocean currents can cause sensor nodes to drift passively, leading to broken communication links between nodes, decreased network connectivity, and the creation of coverage blind spots and topology holes. When some nodes run out of energy and fail, static deployment schemes lack corresponding repair methods, further exacerbating the topology hole problem and severely undermining network stability.

[0005] In view of the above, this application is hereby submitted. Summary of the Invention

[0006] This application aims to provide a dynamic topology control method for underwater sensors. It utilizes the global optimization capability of the simulated annealing algorithm to calculate the expected position of each node when the deployment work is completed, and then uses a meandering ocean current movement model to predict the actual deployment position of each node. This avoids the situation where nodes actively adjust their own positions to consume extra energy, but instead makes full use of the characteristics of nodes moving with ocean currents to achieve full connectivity and high coverage of the network.

[0007] This application provides a dynamic topology control method for underwater sensors, including:

[0008] Construct a first multi-objective optimization function and constraint set. The first multi-objective optimization function takes minimizing coverage overlap, maximizing coverage, and maximizing connectivity as optimization sub-objectives at the time when all sensor nodes are deployed.

[0009] The constraint set includes at least the following: First constraint, the network is fully connected when all sensor nodes are deployed; Second constraint, the desired location of each sensor node is within the underwater monitoring area.

[0010] Using the simulated annealing algorithm, under the constraints of the constraint set, the first multi-objective optimization function is globally optimized to obtain the expected position of each sensor node at the time when all sensor nodes are deployed.

[0011] Based on the drift duration of each sensor node, the Meandering Current Mobility Model (MCM) is used to inversely deduce the actual deployment location of each sensor node according to its expected location; and each sensor node is deployed according to its actual deployment location.

[0012] In summary, this application proposes a dynamic topology control method for underwater sensors. It utilizes the global optimization capability of the simulated annealing algorithm to calculate the expected position of each node when the deployment work is completed, and then uses a meandering ocean current movement model to predict the actual deployment position of each node. This avoids the situation where nodes actively adjust their own positions to consume extra energy, but instead makes full use of the characteristics of nodes moving with ocean currents to achieve full connectivity and high coverage of the network. Attached Figure Description

[0013] Figure 1 This is a schematic diagram of an overall network model provided in an embodiment of this application;

[0014] Figure 2 This is a flowchart of a dynamic topology control method for underwater sensors provided in an embodiment of this application;

[0015] Figure 3 This is a schematic diagram of nodes in a fully connected scenario provided in an embodiment of this application;

[0016] Figure 4 This is a schematic diagram of a node in a non-fully connected scenario provided in an embodiment of this application;

[0017] Figure 5 This is a schematic diagram illustrating how the position of a node plane changes continuously at different times due to the influence of meandering ocean currents, as provided in an embodiment of this application. Detailed Implementation

[0018] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0019] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] Example 1

[0021] This embodiment uses an underwater sensor network (UWSN) for nearshore marine monitoring as an application scenario for marine environmental data acquisition and disaster early warning. The overall implementation of this embodiment employs a dynamic topology control method for underwater sensors based on passive drift of nodes due to ocean currents. A schematic diagram of the overall network model is attached. Figure 1 It includes: a convergence node set on the water surface, several sensor nodes deployed underwater, an AUV (Autonomous Underwater Vehicle), a land-based base station, and a satellite; underwater acoustic communication is used between underwater sensor nodes and between sensor nodes and the AUV; the convergence node completes data backhaul with the land-based base station via satellite and electromagnetic waves.

[0022] In this embodiment, the underwater sensor nodes are not equipped with high-thrust propulsion systems, and the nodes themselves do not actively change their spatial position; the positional changes of the nodes rely entirely on passive drift caused by meandering ocean currents; only the AUV has maneuvering capabilities. All sensor nodes have underwater self-localization capabilities and can obtain their own three-dimensional coordinates; the energy of the sensor nodes is only consumed in underwater acoustic communication, sensing sampling, and positioning calculations, and there is no energy overhead from active node movement.

[0023] Example 2

[0024] Based on the above network model, a dynamic topology control method for underwater sensors is proposed. In the sensor node deployment stage, the core objective is to: without relying on the active movement of the sensor nodes themselves, use the meandering ocean current MCM model to infer the actual deployment position of the sensor nodes; after deployment, the sensor nodes passively drift with the ocean current, and when all sensor nodes are deployed, all sensor nodes drift to the desired position obtained by the simulated annealing algorithm; at this time, the network satisfies full connectivity, high coverage, and minimal overlap.

[0025] See Figure 2 The underwater sensor dynamic topology control method specifically includes the following steps:

[0026] S210. Construct the first multi-objective optimization function and constraint set.

[0027] The first multi-objective optimization function takes minimizing coverage overlap, maximizing coverage, and maximizing connectivity as sub-objectives at the time when all sensor nodes are deployed. The optimization object is the desired location of each sensor node at the time when all sensor nodes are deployed.

[0028] The constraint set includes at least the following: First constraint, the network is fully connected when all sensor nodes are deployed, i.e., the connectivity rate is 1.

[0029] The second constraint is that the desired location of each sensor node is within the underwater monitoring area.

[0030] It may also include a third constraint: for the desired position of each sensor node at the time of candidate deployment completion, for each sensor node, there exists a legal actual deployment position within the underwater monitoring area, such that the sensor node can reach the candidate desired position corresponding to that sensor node by starting from that actual deployment position and drifting through the ocean current for the corresponding drift time. That is, for each sensor node, there must exist a legal actual deployment position within the underwater monitoring area; the sensor node can reach the desired position of that node by starting from that actual deployment position and drifting through the ocean current for the corresponding drift time.

[0031] Assumption Let a be a set of underwater sensor nodes, each containing its location and remaining energy information. Then, the first multi-objective optimization function for the node deployment phase is defined as follows:

[0032]

[0033] Indicates overlapping coverage. Indicates coverage rate. This represents the connectivity rate.

[0034] The first multi-objective optimization function is the evaluation metric for the simulated annealing algorithm, used to judge the quality of a set of "desired locations". The smaller the overlap, the higher the coverage, and the higher the connectivity, the better the solution for that set of locations. The first constraint is a hard constraint: at the moment of deployment, all sensor nodes can reach the surface convergence node in a single hop or multiple hops, and there are no isolated islands in the network. If islands appear in the underwater network, the data of the isolated nodes cannot be uploaded to the surface convergence node, and the entire node becomes completely ineffective. Therefore, full connectivity is used as a hard constraint, rather than a normal optimization objective.

[0035] The third constraint restricts the generation of any desired location; the desired endpoint must be physically reachable, requiring a deployment starting point from which the node must drift with the ocean current for a specific duration before reaching the endpoint. If a candidate desired endpoint cannot be reached by any selected deployment starting point within the monitored area, the solution is deemed invalid and discarded. This constraint prevents the SA algorithm from finding mathematically sound but physically unattainable locations due to ocean currents. The third constraint also prevents the desired endpoint from falling outside the monitored area, thus rendering the monitoring meaningless.

[0036] Traditional solutions typically only set optimization targets for coverage and connectivity, neglecting the aforementioned third constraint. The optimal location obtained through optimization is an "ideal static location," failing to consider the physical laws of ocean current drift. Subsequently, nodes either consume significant energy to actively swim to this location or are unable to reach it at all, rendering the theoretically optimal solution unattainable. This embodiment sets "ocean current drift reachability" as a hard constraint, ensuring that all candidate solutions for SA are achievable at the ocean current physical level, thus avoiding the discovery of unachievable theoretically optimal solutions from the outset. Subsequently, nodes only need to be deployed to the deduced initial location and passively drift to the desired endpoint using ocean currents, without consuming any movement energy. Simultaneously, the first constraint, which mandates full connectivity, introduces a natural trade-off: to ensure connectivity for all nodes, the distance between nodes cannot be too great, leading to a certain degree of coverage overlap. Therefore, the sub-objective of "minimizing coverage overlap" is used to minimize coverage overlap under the premise of mandatory full connectivity, thereby increasing the effective monitoring area and resolving the inherent contradiction between "connectivity and coverage."

[0037] Furthermore, the drift duration of each sensor node is determined as follows:

[0038] Based on the planned deployment time of the first sensor node, the total number of sensor nodes, and the serial single-node deployment interval, the deployment completion time of all sensor nodes is calculated. The deployment completion time of all sensor nodes is the time when the last sensor node in the serial deployment process is deployed into the water. For the sensor node with deployment sequence number i, the drift time of sensor node i is the deployment completion time of all sensor nodes minus the planned deployment time of sensor node i.

[0039] In one implementation, a serial deployment mode is adopted, with the AUV deploying each sensor node sequentially, and only one sensor node deployed at a time; the deployment interval between two adjacent sensor nodes is marked as... Assume the total number of sensor nodes is N.

[0040] Set the planned deployment time for sensor node number 1 (i.e., deployment sequence number i=1). The moment when all sensor node deployments were completed. The time when the last sensor node is deployed into the water can be determined using the following formula:

[0041]

[0042] Deploy sensor node with sequence number i, its planned deployment time Represented as:

[0043]

[0044] The drift time of the sensor node with deployment number i .

[0045] For example: Assume N=100, the drift time of the first deployed sensor node with sequence number i=1. The last deployed sensor node with the longest drift time (i=100) has the longest drift time. The deployment is complete the moment it is deployed, with almost no drift time.

[0046] In general, drift time refers to the time a sensor node spends drifting with ocean currents from the moment it is deployed by an AUV until all nodes are deployed. Because deployment is done sequentially, the nodes deployed earlier will drift with ocean currents for a long time; the later the nodes are deployed, the shorter the drift time; the deployment phase ends when the last node is deployed, and the drift time is 0.

[0047] Each node is deployed at a different time, and the duration of its drift due to ocean currents during deployment varies. If all nodes use the same drift duration to calculate their actual deployment position, a significant positional deviation will occur, and the node will not reach the desired endpoint upon deployment completion. This embodiment calculates a unique drift duration for each node based on its deployment sequence number. When the MCM model subsequently reverse-engineers the actual deployment position, each node's actual drift duration is used, ensuring a strict match between the drift time and the actual deployment sequence. This guarantees the physical logic that "after deployment, the node drifts to the desired position upon completion."

[0048] Furthermore, full network connectivity means that all sensor nodes can reach the surface convergence node via a single-hop or multi-hop underwater acoustic communication path. The two endpoints of the communication link are the sensor nodes and / or the surface convergence node. If the Euclidean distance between the two endpoints is less than or equal to the underwater acoustic communication radius, an undirected underwater acoustic communication link is established between them. The connectivity rate is 1 when the network is fully connected. The connectivity rate is the quotient obtained by dividing the reference number by the total number of sensor nodes, where the reference number is the number of sensor nodes that can reach the surface convergence node via a single-hop or multi-hop underwater acoustic communication path.

[0049] The underwater system is a multi-dive acoustic network, which does not require every sensor to be directly connected to the surface convergence node; it allows other intermediate sensors to relay the data; as long as there is a communication path to the convergence node, it is considered a valid connection.

[0050] By adopting a multi-hop connectivity definition, it is not necessary to force all nodes to be less than the communication radius of the water surface convergence node; sensor nodes can be deployed to a larger monitoring area, further improving the overall coverage while ensuring full connectivity.

[0051] For example, refer to Figure 3 In the fully connected scenario shown, the nodes are close together and have significant overlap. Among them, R... c Indicates the distance between adjacent nodes; for comparison, refer to... Figure 4 In the non-fully connected scenario shown, the nodes have less overlap, but some nodes cannot connect to the aggregation node, resulting in network fragmentation.

[0052] Coverage overlap is the sum of the overlapping volumes of the pairwise intersections of the spherical sensing regions of each sensor node. The sensing region of each sensor node is a sphere; coverage overlap refers to the sum of the overlapping volumes of the pairwise intersections of the spherical sensing regions of all sensors, and is the optimization object of the sub-objective, with the optimization direction being to minimize the sum.

[0053] For example, if the Euclidean distance between sensor nodes u and v is known to be d(u, v), and assuming that the detection range of each node is spherical with radius R, then the volume V of the overlapping area covered by the two nodes can be determined by the following formula:

[0054]

[0055] The coverage rate is calculated using a Boolean sensing model, which includes: discretizing the three-dimensional underwater monitoring area into several regular cubic grids; if the center of a cubic grid falls within the spherical sensing range of any sensor node, the cubic grid is determined to be effectively covered; the coverage rate is equal to the ratio of the number of effectively covered cubic grids to the total number of all cubic grids.

[0056] By discretizing continuous three-dimensional water areas into cubic meshes, the abstract underwater coverage performance is transformed into a quantifiable numerical index, allowing the simulated annealing algorithm to quantitatively evaluate the coverage quality of each set of candidate location solutions through numerical computation.

[0057] S220. Using the simulated annealing algorithm, under the constraints of the constraint set, the first multi-objective optimization function is globally optimized to obtain the expected position of each sensor node at the time when all sensor nodes are deployed.

[0058] For example, the initial temperature and cooling coefficient of the simulated annealing algorithm are set, with the initial temperature selected based on the results of underwater network simulation experiments; initial candidate solutions are generated, and constraint set verification is performed on the initial candidate solutions; if the initial candidate solution does not satisfy any constraint in the constraint set, a new initial candidate solution is generated; the initial candidate solution that passes all constraint verifications is used as the current solution for the first iteration; within the iteration loop, neighborhood perturbation is performed on the current solution of the current iteration to generate new candidate solutions; constraint set verification is performed on the new candidate solutions sequentially, and if the new candidate solution does not satisfy any constraint in the constraint set, the new candidate solution is discarded directly, and the next iteration begins; for new candidate solutions that pass all constraint verifications, the new candidate solution is compared with the current iteration... Substituting the current solutions into the first multi-objective optimization function, the difference in objective functions between the two is obtained. Based on the Metropolis criterion: if a new candidate solution is better than the current solution, the new candidate solution is accepted; if a new candidate solution is worse than the current solution, the new candidate solution is accepted with probability based on the acceptance probability related to the current iteration temperature; as the iteration temperature decreases, the probability of accepting a new candidate solution worse than the current solution gradually decreases; when a new candidate solution is accepted, it is updated as the current solution for the next iteration; when a new candidate solution is rejected, the current solution for the current iteration remains unchanged and is used as the current solution for the next iteration; cooling operations are performed according to the cooling coefficient, and the iteration is repeated until the algorithm converges, outputting the optimal expected positions of each sensor node.

[0059] In one implementation, the iteration termination condition is: the optimal objective function no longer shows significant improvement after 50 consecutive iterations, indicating that the algorithm has converged. Execution steps: A set of initial candidate solutions is randomly generated, containing the coordinates of a set of candidate desired endpoints for all N sensor nodes; each initial candidate solution undergoes constraint set verification: only if the first constraint (fully connected network), the second constraint (all desired endpoints fall within the monitored water area), and the third constraint (each node has a reachable actual deployment location) are satisfied, is the initial candidate solution retained; if any constraint is not satisfied, it is discarded, and a new random solution is generated until an initial candidate solution that passes all constraint verifications is obtained, which becomes the current solution for the first iteration.

[0060] Iterative loop: For the current solution, perform neighborhood perturbation: make small-amplitude random offsets to the expected endpoint coordinates of some nodes to generate new candidate solutions; perform constraint set verification on each new candidate solution; if any constraint is not satisfied, discard the new candidate solution and proceed directly to the next iteration without calculating the objective function; if all constraints of the new candidate solution pass verification; substitute the current solution and the new candidate solution into the first multi-objective optimization function respectively, and calculate the difference in objective functions; execute the Metropolis criterion: if the new candidate solution has a better objective function (less overlap, higher coverage, higher connectivity)... If a new candidate solution is better than the current solution, it is accepted unconditionally. If the new candidate solution is worse than the current solution, it is accepted probabilistically according to the acceptance probability related to the current iteration temperature. As the temperature continues to decrease during iteration, the probability of accepting a new candidate solution that is worse than the current solution continues to decrease. If the new candidate solution is accepted, it is assigned as the current solution for the next iteration. If the new candidate solution is rejected, the current solution remains unchanged. The cooling operation is performed according to the cooling coefficient. The iteration is repeated until the iteration convergence termination condition is met. The optimal solution is output, which is the set of expected positions of all sensor nodes at the time of deployment completion.

[0061] Simulated Annealing (SA): A global optimization algorithm that mimics the physical process of cooling during solid metal annealing. Core characteristics: In the early stages of iteration, the high temperature allows for a certain probability of accepting inferior solutions (inferior solutions refer to new candidate solutions that are worse than the current solution), thus escaping local optima. As the temperature decreases, the algorithm becomes increasingly conservative, accepting almost only better solutions, gradually converging to a globally optimal solution. Neighborhood perturbation: Small-scale random modifications are made to existing solutions to generate new candidate solutions. Metropolis criterion: The core judgment rule of the SA algorithm, used to decide whether to accept the modified new solution. Key difference between this embodiment and ordinary SA: Candidate solutions are first subjected to hard constraint verification; those that do not meet the constraints are discarded and do not enter the Metropolis evaluation.

[0062] Underwater sensor deployment is a multi-objective optimization problem with a complex optimization space and numerous local optima. Greedy algorithms and gradient descent can easily get stuck in local optima, resulting in poor deployment quality. Subsurface Sensors (SA) addresses this by allowing for suboptimal solutions in the early stages, thus escaping local optima and finding desirable locations globally that balance coverage overlap, coverage rate, and connectivity.

[0063] In this embodiment, before calculating the objective function, all constraint checks are performed first. Candidate solutions that are physically unrealizable or have disconnected networks are directly filtered out, so as not to waste computing power to calculate the objective function and reduce simulation computing overhead. At the same time, it ensures that all candidate solutions processed in each round of iteration meet the hard boundary conditions of the application scenario.

[0064] S230. Based on the drift duration of each sensor node, the meandering ocean current movement model MCM is used to inversely deduce the actual deployment position of each sensor node according to the expected position of each sensor node.

[0065] In this embodiment, the MCM meandering ocean current movement model adopts a two-dimensional planar drift model; the vertical flow of water is ignored, and the depth of the sensor node remains at the fixed depth set at the time of deployment after deployment, with only the planar coordinates drifting with the ocean current.

[0066] The meandering ocean current movement model (MCM) is configured with model parameters, including wave number, downstream flow phase velocity, and ocean current amplitude function. The input parameters of the ocean current amplitude function include the average current width, modulation amplitude, and modulation frequency. Based on the MCM forward drift simulation, a numerical search is performed to inversely deduce the actual deployment positions of each sensor node. This includes: using the desired position of the sensor node, the corresponding drift duration, and the model parameters as input, searching within the spatial range of the underwater monitoring area to find actual deployment positions that meet the drift reachability conditions. The drift reachability conditions are: inputting the searched actual deployment positions and corresponding drift durations into the meandering ocean current movement model (MCM) for forward drift simulation, the output drift endpoint position matches the desired position.

[0067] MCM forward simulation inputs: initial deployment plane coordinates, drift duration, and all ocean current model parameters; output: drift endpoint plane coordinates after the drift duration.

[0068] This embodiment employs a numerical search method for inverse solving. For sensor node i, the inputs include: the desired endpoint planar coordinates of the node, the node's specific drift duration, and the complete set of MCM ocean current model parameters. Within the planar range of the underwater monitoring area, a numerical search is performed, traversing a large number of candidate initial deployment planar coordinates within the monitoring area. Each candidate initial deployment coordinate is substituted into the MCM for forward drift simulation to obtain the simulated drift endpoint. Drift reachability condition: if the error between the simulated drift endpoint and the desired position obtained from the SA solution is less than a distance threshold, the candidate actual deployment position is determined to be drift reachable. From all candidate coordinates that meet the drift reachability condition, one is selected as the final actual deployment position of the sensor node. The above search is performed on all nodes to obtain a complete set of actual deployment positions.

[0069] The MCM meandering current model is a classic model for simulating meandering ocean currents in nearshore waters; the currents are not simple, uniform, straight-line flows, but exhibit periodic, undulating meandering characteristics. (See appendix.) Figure 5 It shows that at different times, such as 0h, 18h, 36h, and 54h, the position of the node plane changes continuously due to the influence of the meandering ocean current.

[0070] Forward drift simulation: Input the starting point to calculate the ending point; reverse solution (infer the actual deployment location) has no analytical closed-form solution, so numerical search and forward simulation verification are used to achieve this.

[0071] Drift reachability condition: If the endpoint of the forward simulation drift is close enough to the desired position of SA, it means that the deployment starting point is available.

[0072] Traditional MCM models are only used to predict where nodes will drift after deployment (forward). This embodiment uses forward MCM simulation as a verification tool to perform numerical search to deduce the deployment starting point. Subsequently, the AUV deploys sensor nodes one by one according to the actual deployment location obtained from the deduction. After the sensor nodes are deployed into the water, they do not activate their own propulsion and are completely passively driven by ocean currents to drift. When the last node is deployed (deployment completion time), each node has just drifted to the expected position solved by SA. Compared with the existing technology where nodes rely on their own propulsion to actively swim to the target position, this scheme has no energy consumption for sensor node movement, only for communication and sampling power consumption, which significantly extends the single-unit lifespan of the sensor node.

[0073] This paper fully utilizes the movement characteristics of nodes with ocean currents, ensuring that all nodes maintain their optimal deployment state even after moving with the currents during deployment. First, an initial location value is randomly assigned to each node. Then, the SA algorithm is used iteratively to find the optimal desired location for each node, guaranteeing high topology coverage and full connectivity. During algorithm execution, a first objective function needs to be calculated iteratively to find the optimal node location. The full connectivity constraint of the nodes means that the algorithm only accepts solutions with a connectivity rate of 1 in each iteration, which may lead to a decrease in node coverage. However, all nodes can communicate effectively, and by meeting the coverage overlap requirement in the sub-objectives, the maximum coverage is achieved while maintaining full connectivity.

[0074] After calculating the desired locations of the nodes, the actual deployment locations of each node are inferred using MCM based on the node deployment time. It is assumed that the deployment time for each node is constant. ,and Only one node can be deployed within a given time period, so each node has a different drift time after deployment. Assuming the current node's deployment sequence number is i, what is the drift time of this node during the deployment of all 'a' nodes? for:

[0075]

[0076] MCM uses a two-dimensional model to simulate the movement of underwater nodes because the ocean is a rotating, stratified fluid, and most vertical motion can be ignored. Assume the planar coordinates of the node with deployment number i are ( If the node is in time, then the node is in time. The move function within is defined as:

[0077]

[0078] Where k represents the current wavenumber, typically 2π / 7.5; c represents the downstream phase velocity, typically 0.12. This represents the amplitude function of ocean currents.

[0079]

[0080] A represents the average width of the ocean current, typically 1.2. This indicates the modulation amplitude, typically 0.3. This represents the modulation frequency, typically 0.4. Therefore, the time it takes for sensor node i to travel with the ocean current... After the movement, the plane coordinates ( The following formula can be used to determine the result:

[0081]

[0082]

[0083] The above method can be used to deduce the actual deployment location of each node from its optimal expected location, so that even if the nodes move with ocean currents after being deployed according to their actual deployment locations, they can maintain high coverage and full connectivity when all deployment work is completed.

[0084] S240. Deploy each sensor node according to its actual deployment location.

[0085] The AUV navigates to the actual deployment location corresponding to each sensor node and completes the deployment in sequence; once the deployment is complete, the deployment phase ends, and the network is in an optimal topology state with full connectivity and high coverage.

[0086] This application proposes a Dynamic Topology Control Method for Mobile Underwater Sensor Networks (DTCM) based on node mobility. It utilizes the global optimization capability of Simulated Annealing (SA) to calculate the desired position of each node upon completion of deployment. Then, it uses a Meandering Current Mobility Model (MCM) to predict the actual deployment position of each node. This avoids nodes actively adjusting their positions to consume extra energy, instead fully leveraging the node's movement with ocean currents to achieve full network connectivity and high coverage. After node deployment, an AUV periodically collects the node's position and energy information and adopts an efficient topology optimization scheme based on the node's remaining energy, avoiding topology holes and ensuring the continuous and stable operation of the network.

[0087] The goal of DTCM is to achieve optimal topology deployment during the node deployment phase and real-time and efficient optimization of the topology during the topology optimization phase. Therefore, this embodiment introduces the SA algorithm, which has strong global optimization capabilities, to solve the optimization problem and establishes an objective function under various constraints.

[0088] The SA algorithm is a general-purpose optimization algorithm that simulates the annealing process of solids, bringing particles to a low-energy state through cooling. The algorithm's execution consists of two main parts: the annealing process and the Metropolis criterion. During annealing, the algorithm starts with an initial solution and gradually searches for other solutions in the space. Initially, the algorithm sets a high initial temperature and decreases it exponentially. In each iteration, the algorithm accepts a worse solution with a certain probability, allowing it to escape local optima without getting trapped. As the temperature decreases, the probability decreases, leaving increasingly better solutions. Generally, a higher initial temperature yields optimal results, but this depends on the actual convergence speed and application. This paper selects a suitable initial temperature through experiments based on the simulation's actual conditions, ensuring optimal node deployment. The Metropolis criterion is crucial in determining whether to accept a new solution. If the new solution has a better optimization value, it is always accepted; otherwise, a worse solution than the one in the previous iteration is accepted with a certain probability. This probability is related to the initial temperature and gradually decreases over time.

[0089] Example 3

[0090] Based on the above embodiments, the deployment phase only achieves the optimal topology momentarily upon completion. After deployment, the meandering ocean currents continue to drive the sensor nodes to drift. Simultaneously, the sensor nodes continuously engage in underwater acoustic communication, consuming battery power. Over time, the following issues may arise: the distance between nodes increases, leading to communication link breaks; some nodes experience high communication loads and rapid energy depletion; and once nodes run out of energy, they exit the network, creating topological holes, resulting in localized monitoring blind spots and network fragmentation. Therefore, after deployment, periodic topology optimization needs to be performed continuously and permanently.

[0091] Specifically, it also includes: topology optimization, which includes:

[0092] The AUV performs periodic cruises according to a preset cruise path. At each cruise cycle, the AUV collects real-time location information and remaining energy information of all sensor nodes in the network. It determines whether there are sensor nodes in the current network with remaining energy below the minimum energy threshold. When there are sensor nodes with remaining energy below the minimum energy threshold, it performs emergency topology repair, including: the low-energy sensor node sends an emergency repair request to the AUV; after receiving the emergency repair request, the AUV navigates to the location of the low-energy sensor node and deploys the backup sensor node it carries. When there are no sensor nodes with remaining energy below the minimum energy threshold, it performs regular topology optimization, including: constructing a second multi-objective optimization function, which takes maximizing coverage, maximizing connectivity, and minimizing energy balance sub-objectives at the cruise time as optimization sub-objectives; using simulated annealing algorithm to solve the second multi-objective optimization function to obtain the optimal single-point location at the cruise time; deploying backup sensor nodes based on the optimal single-point location to achieve network topology optimization and balance the communication energy consumption of all sensor nodes in the network.

[0093] The energy balance sub-target is constructed based on the real-time position and remaining energy of each sensor node during the cruise; the optimal position of a single point obtained by the energy balance sub-target is biased towards the network region with lower remaining energy.

[0094] When performing the emergency topology repair, if all the backup sensor nodes carried by the AUV have been exhausted, a degraded topology control strategy is implemented, including: marking the sensor node that triggers the lowest energy threshold as a risk node; prioritizing the area where the risk node is located as the optimal location for a single point; and reducing the communication forwarding load of the risk node during network routing scheduling.

[0095] AUVs collect real-time location information and remaining energy information of all sensor nodes in the network, including:

[0096] During cruise, the AUV broadcasts specific packets carrying wake-up commands.

[0097] After receiving a specific packet, the sensor node performs self-localization to obtain its real-time position and reads its remaining energy; the sensor node then reports its real-time position and remaining energy information back to the AUV.

[0098] Specifically, the AUV cruises periodically along a preset fixed cruise path, and performs a complete topology optimization process after each full cruise cycle.

[0099] During each cruise cycle, the AUV broadcasts specific underwater acoustic packets carrying a wake-up command. Each underwater sensor node listens to the underwater acoustic channel; upon receiving the specific packet broadcast by the AUV, it triggers the data acquisition process: the node runs a self-localization algorithm to obtain its current real-time three-dimensional position coordinates; and reads its remaining battery energy. The sensor node encapsulates its real-time position information and remaining energy information into a feedback data packet and reports it to the AUV via underwater acoustic communication. The AUV collects and summarizes the data reported by all sensor nodes in the network to obtain a snapshot of the entire network at the current moment: where each node is now and how much energy it has remaining.

[0100] Specific packets: Dedicated underwater acoustic data packets sent by the AUV, carrying a wake-up command; the sensor node can normally be in a low-power listening state; only when this packet is received will the location reading, energy reading, and reporting operations be triggered, avoiding a large amount of invalid communication energy consumption caused by the sensor node's continuous reporting.

[0101] Ocean currents are time-varying, and node positions change constantly; node energy is dynamically consumed with communication load. AUVs collect network snapshots through periodic cruises, allowing them to grasp the network's true real-time status. Without this step, AUVs would be completely unable to perceive the risks of link breakage or node energy depletion, and thus unable to carry out subsequent topology repairs. Using a broadcast wake-up and passive node reporting mode avoids sensor nodes continuously and actively broadcasting status messages, reducing network-wide underwater acoustic communication overhead and conserving sensor battery energy.

[0102] As the ocean currents continue to drift, even if no nodes immediately run out of battery, the network topology will gradually deteriorate: some areas will have sparse nodes and coverage gaps; in other areas, a few nodes will bear a large number of multi-hop relay forwarding tasks, consuming energy much faster than other nodes. This embodiment uses conventional topology optimization, with coverage, connectivity, and energy balance as joint optimization objectives, to find the optimal location for a single point and deploy backup nodes. This fills the local coverage gaps caused by the drift; and the new nodes share the relay forwarding load of surrounding older nodes, preventing a few local nodes from becoming overloaded and rapidly depleting their energy; thus, it balances the overall network energy consumption and extends the overall network lifespan.

[0103] Due to the difficulty of replacing batteries in underwater sensor nodes, node energy has become a very important indicator. Communication and movement of sensor nodes in UWSNs all contribute to energy loss, with communication and movement being the main factors in energy consumption. Since the method proposed in this embodiment does not involve active node movement, only the energy consumed by communication is considered.

[0104] For example, the second multi-objective optimization function is defined as follows:

[0105]

[0106] Indicates coverage rate. This represents the connectivity rate.

[0107] This represents the energy balance sub-objective. During the topology optimization phase, to ensure the optimal optimization position ol is closer to the node with the minimum energy, this paper uses the position NL of sensor node i as a reference. i and remaining energy NE i The following energy balance function was designed:

[0108]

[0109] This represents the optimal location ol and the sensor node location NL. i The Euclidean distance, where 'a' represents the total number of sensor nodes.

[0110] The solution in this embodiment has the following technical effects:

[0111] High-efficiency communication: Through global optimization of SA and prediction of MCM model, full connectivity and high coverage are guaranteed when nodes are deployed, ensuring high-efficiency communication of underwater sensor network.

[0112] Energy saving: By predicting ocean current trajectories and guiding nodes to drift passively, the nodes are positioned and deployed with "zero energy consumption", which significantly reduces the energy consumption during the node deployment phase from the source.

[0113] Topology Adaptation: A periodic topology optimization mechanism based on AUV cruise was designed to dynamically adjust the network structure according to the real-time collected energy and location information. It can actively identify and repair potential "topology holes" and avoid local network paralysis.

[0114] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. The above descriptions are only preferred embodiments of this application. It should be noted that due to the limitations of written expression, while there are objectively infinite specific structures, those skilled in the art can make several improvements, modifications, or changes without departing from the principles of this invention, and can also combine the above technical features in an appropriate manner. These improvements, modifications, changes, or combinations, or the direct application of the inventive concept and technical solution to other situations without modification, should all be considered within the scope of protection of this application.

Claims

1. A dynamic topology control method for underwater sensors, characterized in that, include: Construct a first multi-objective optimization function and a set of constraints. The first multi-objective optimization function takes minimizing coverage overlap, maximizing coverage, and maximizing connectivity as optimization sub-objectives at the time when all sensor nodes are deployed. The constraint set includes at least: a first constraint, that the network is fully connected when all sensor nodes are deployed; and a second constraint, that the desired location of each sensor node is within the underwater monitoring area. Using the simulated annealing algorithm, under the constraints of the set of constraints, the first multi-objective optimization function is globally optimized to obtain the expected position of each sensor node at the time when all sensor nodes are deployed. Based on the drift duration of each sensor node, a meandering ocean current movement model is adopted, and the actual deployment location of each sensor node is deduced from the expected location of each sensor node. Deploy each sensor node according to the actual deployment location.

2. The underwater sensor dynamic topology control method according to claim 1, characterized in that, The drift duration of each sensor node is determined as follows: Based on the planned deployment time of the first sensor node, the total number of sensor nodes, and the serial single-node deployment interval, the deployment completion time of all sensor nodes is calculated. The deployment completion time of all sensor nodes is the time when the last sensor node in the serial deployment process is deployed into the water. For the sensor node with deployment sequence number i, the drift time of sensor node i is the time when all sensor nodes are deployed minus the planned deployment time of sensor node i, where i is a natural number greater than or equal to 1.

3. The underwater sensor dynamic topology control method according to claim 1, characterized in that, The term "fully connected network" refers to the fact that all sensor nodes can reach the surface convergence node through a single-hop or multi-hop underwater acoustic communication path. The two endpoints of the communication link are the sensor node and / or the surface convergence node. If the Euclidean distance between the two endpoints is less than or equal to the underwater acoustic communication radius, an undirected underwater acoustic communication link is established between them. The connectivity rate is 1 when the network is fully connected. The coverage overlap is the sum of the overlapping volumes of the spherical sensing areas of each sensor node that intersect pairwise; The connectivity rate is the quotient obtained by dividing the reference number by the total number of sensor nodes, where the reference number is the number of sensor nodes that can reach the surface convergence node through a single-hop or multi-hop underwater acoustic communication path; the first constraint requires the connectivity rate to be equal to 1.

4. The underwater sensor dynamic topology control method according to claim 1, characterized in that, The coverage rate is calculated using a Boolean sensing model, which includes: discretizing the three-dimensional underwater monitoring area into several regular cubic grids; if the center of a cubic grid falls within the spherical sensing range of any sensor node, the cubic grid is determined to be effectively covered; the coverage rate is equal to the ratio of the number of effectively covered cubic grids to the total number of all cubic grids.

5. The underwater sensor dynamic topology control method according to claim 1, characterized in that, The simulated annealing algorithm is used to globally optimize the first multi-objective optimization function under the constraints of the set of constraints, and the desired positions of each sensor node at the time when all sensor nodes are deployed are obtained, including: The initial temperature and cooling coefficient of the simulated annealing algorithm are set, and the initial temperature is selected based on the results of underwater network simulation experiments. Generate an initial candidate solution and perform constraint set verification on the initial candidate solution; if the initial candidate solution does not satisfy any constraint in the constraint set, regenerate the initial candidate solution; take the initial candidate solution that passes all constraint verifications as the current solution for the first round of iteration; Within the iteration loop, a neighborhood perturbation is performed on the current solution of the current iteration to generate a new candidate solution; the constraint set is checked sequentially on the new candidate solution; if the new candidate solution does not satisfy any constraint in the constraint set, the new candidate solution is discarded and the next iteration begins; for a new candidate solution that passes all constraint checks, the new candidate solution and the current solution of the current iteration are substituted into the first multi-objective optimization function to obtain the difference in objective functions between the two. Based on the Metropolis criterion: if a new candidate solution is better than the current solution, then the new candidate solution is accepted; if a new candidate solution is worse than the current solution, then the new candidate solution is accepted with a probability based on the acceptance probability related to the current iteration temperature; as the iteration temperature decreases, the probability of accepting a new candidate solution that is worse than the current solution gradually decreases. When a new candidate solution is accepted, the new candidate solution is updated as the current solution for the next iteration. When a new candidate solution is rejected, the current solution for the current iteration remains unchanged and is used as the current solution for the next iteration. The cooling operation is performed according to the cooling coefficient, and the process is iterated until the algorithm converges, outputting the optimal desired position of each sensor node.

6. The underwater sensor dynamic topology control method according to claim 1, characterized in that, The meandering ocean current movement model is a two-dimensional planar drift model; The meandering ocean current movement model is configured with model parameters, including: wave number, downstream flow phase velocity, and ocean current amplitude function; the input parameters of the ocean current amplitude function include average ocean current width, modulation amplitude, and modulation frequency. Numerical search is performed based on the forward drift simulation of the meandering ocean current movement model to reverse-engineer the actual deployment positions of each sensor node. This includes: using the desired position of the sensor node, the corresponding drift duration, and the model parameters as input, searching for actual deployment positions that meet the drift reachability conditions within the spatial range of the underwater monitoring area; wherein, the drift reachability conditions are: inputting the actual deployment positions and corresponding drift durations obtained from the search into the meandering ocean current movement model for forward drift simulation, and the output drift endpoint position matches the desired position.

7. The underwater sensor dynamic topology control method according to claim 1, characterized in that, Also includes: Topology optimization, which includes: The autonomous underwater vehicle performs periodic patrols according to a preset patrol path; During each cruise cycle, the autonomous underwater vehicle collects real-time location information and remaining energy information of all sensor nodes in the network. Determine whether there are sensor nodes in the current network with remaining energy below the minimum energy threshold; When a sensor node has remaining energy below the minimum energy threshold, an emergency topology repair is performed, including: the low-energy sensor node sending an emergency repair request to the autonomous underwater vehicle; after receiving the emergency repair request, the autonomous underwater vehicle sails to the location of the low-energy sensor node and deploys the backup sensor node it carries. When there are no sensor nodes with remaining energy below the minimum energy threshold, perform conventional topology optimization, including: constructing a second multi-objective optimization function, which takes maximizing coverage, maximizing connectivity, and minimizing energy balance sub-objectives at the cruise time as optimization sub-objectives; solving the second multi-objective optimization function using a simulated annealing algorithm to obtain the optimal single-point location at the cruise time; and deploying backup sensor nodes based on the optimal single-point location to achieve network topology optimization and balance the communication energy consumption of all sensor nodes in the network.

8. The underwater sensor dynamic topology control method according to claim 7, characterized in that, The energy balancing sub-target is constructed based on the real-time position and remaining energy of each sensor node during the cruise; the optimal single-point position obtained by the energy balancing sub-target is biased towards the network region with lower remaining energy.

9. The underwater sensor dynamic topology control method according to claim 8, characterized in that, When performing the emergency topology repair, if all the backup sensor nodes carried by the autonomous underwater vehicle have been exhausted, a degraded topology control strategy is implemented, including: marking the sensor node that triggers the lowest energy threshold as a risk node; prioritizing the area where the risk node is located as the optimal location for a single point; and reducing the communication forwarding load of the risk node during network routing scheduling.

10. The underwater sensor dynamic topology control method according to claim 7, characterized in that, The autonomous underwater vehicle collects real-time location information and remaining energy information of all sensor nodes in the network, including: The autonomous underwater vehicle broadcasts specific packets carrying wake-up commands during the cruise process; After receiving a specific packet, the sensor node performs self-localization to obtain its real-time position and reads its remaining energy; the sensor node then reports its real-time position and remaining energy information back to the autonomous underwater vehicle.

Images