A method and system for optimizing near-shore communication networks

Abstract

This invention belongs to the field of artificial intelligence and discloses a method and system for optimizing near-shore communication networks. Based on the distance between mobile users and the MEC server, a hybrid path loss model (HPL model) adapted to near-shore communication scenarios is constructed. Based on the HPL model, a near-shore communication network optimization model is constructed by simulating actual communication scenarios of near-shore users. The near-shore communication strategy is dynamically adjusted through the near-shore communication network optimization model to achieve low-latency and low-energy near-shore communication. The LAM algorithm proposed in this invention can effectively reduce the weighted sum of latency and energy consumption in near-shore communication under the HPL model environment.

Description

Technical Field

[0001] This invention belongs to the field of artificial intelligence, and specifically relates to a method and system for optimizing near-shore communication networks. Background Technology

[0002] With the rapid development of the global marine economy and the increasing activities in coastal waters, mobile data traffic in coastal waters is experiencing explosive growth. Coupled with the continuous upgrading of coastal smart terminal devices, users are placing higher demands on data transmission rates and communication quality. Compared to the stable communication environment inland, coastal wireless communication faces two major challenges: the high mobility of mobile platforms (such as ships and floating platforms) and the complex and ever-changing coastal communication environment.

[0003] For the complex and ever-changing nearshore communication environment, existing nearshore communication path loss models are divided into two types: two-ray path loss models and three-ray path loss models. The two-ray path loss model mainly considers the impact of the direct path and surface reflection path on signal transmission. The three-ray path loss model further expands the analysis scope by adding a third path through sea surface reflection.

[0004] In nearshore environments, the high mobility of user terminals is often accompanied by higher data transmission demands and more frequent communication activities, which typically leads to network congestion. To quickly process massive mobile data traffic, academia has focused on applying unmanned aerial vehicles (UAVs) to mobile edge computing (MEC) communication networks. By leveraging UAVs, cloud computing capabilities can be brought down to the edge of nearshore communication networks, providing mobile users with low-latency, high-bandwidth computing services locally. However, the increased network complexity presents greater challenges to optimizing latency and energy consumption in nearshore communications.

[0005] For complex, non-convex optimization problems, the existing technique is the alternating minimization algorithm. It is generally used to solve optimization problems with separable structures. Its basic idea is to decompose the multivariate optimization problem into a series of univariate optimization problems, and then gradually approach the global optimum by alternately optimizing each variable.

[0006] Assume the objective function is The alternating minimization algorithm can be described as follows:

[0007] Step 1: Initialization .

[0008] Step 2: Enter the loop.

[0009] for :

[0010] for arrive :

[0011] fixed For all .

[0012] Solve

[0013] Step 3: If the convergence condition (set manually) is met, stop the iteration; otherwise, return to step 2.

[0014] It is evident that existing near-shore communication path loss models are too simplistic to accurately describe the dynamic changes in path loss as communication distance increases. Due to the complex environment, wide coverage, and dispersed distribution of communication nodes in near-shore communications, traditional centralized optimization methods require processing data from all nodes through a central controller. This results in a large amount of data being transmitted via unstable communication links, increasing the overall processing time and failing to meet the real-time requirements of near-shore communication. Summary of the Invention

[0015] To address the problems of existing technologies, this invention provides a method and system for optimizing near-shore communication networks. Considering the complex and variable near-shore communication environment, a hybrid path loss (HPL) model adapted to near-shore communication scenarios is derived based on the distance between mobile users and the MEC server, more accurately simulating the actual communication scenarios of near-shore users. A near-shore communication network optimization method is proposed, namely a four-layer layered alternating minimization (LAM) algorithm, to iteratively solve the complex non-convex near-shore communication latency and energy consumption joint optimization problem. By dynamically adjusting the near-shore communication strategy through the LAM algorithm, low-latency and low-energy near-shore communication is achieved.

[0016] To achieve the above objectives, the present invention provides the following solution:

[0017] A method for optimizing a near-shore communication network, the method comprising:

[0018] Based on the distance between mobile users and MEC servers, a hybrid path loss model, namely the HPL model, is constructed to adapt to near-shore communication scenarios.

[0019] Based on the HPL model, a near-shore communication network optimization model is constructed to simulate the actual communication scenarios of near-shore users.

[0020] By dynamically adjusting the nearshore communication strategy through a nearshore communication network optimization model, low-latency and low-energy nearshore communication can be achieved.

[0021] Preferred, consider A dual-ray path loss model is used within the specified range. A three-ray path loss model is used within the specified range, specifically:

[0022] If at time t, the straight-line distance between the moving ship k and the unique BS-MEC server with index M is... The two-ray path loss model is used:

[0023] ;

[0024] In the formula, h T and h R They are respectively and Antenna height, Where is the carrier wavelength, and ;

[0025] When the distance is greater than At that time, there exists a third ray from the captured signal in the evaporation pipe layer, therefore the path distance of the three rays is... The loss model is defined as:

[0026] ;

[0027] in, h E This is the effective height of the evaporator pipe.

[0028] Preferably, achieving low-latency and low-energy-consumption nearshore communication by dynamically adjusting nearshore communication strategies through a nearshore communication network optimization model includes:

[0029] The hierarchical alternating minimization algorithm is used to solve the decision problem of mobile user task offloading and MEC server selection, and the Nash equilibrium solution is obtained as the optimal solution of the decision matrix.

[0030] Using the decision matrix as a known quantity, the uplink and downlink power dynamic allocation problem of the task execution is solved by the geometric water injection algorithm, and then the UAV trajectory control problem is solved piece by piece by the continuous convex approximation algorithm.

[0031] Given the decision matrix, the uplink and downlink power allocation for task execution, and the UAV flight trajectory, a standard convex optimization algorithm is used to solve the CPU frequency allocation problem.

[0032] Preferably, the hierarchical alternating minimization algorithm is used to solve the decision problem of mobile user task offloading and MEC server selection, and the Nash equilibrium solution is obtained as the optimal solution of the decision matrix, including:

[0033] ;

[0034] in, Let m be the selection variable for mobile user k, m be the MEC server, and k be the mobile user. This represents the optimal decision matrix for a near-shore communication system.

[0035] Preferably, the method of solving the uplink and downlink power dynamic allocation problem for task execution using the geometric water injection algorithm includes:

[0036] ;

[0037] ;

[0038] Where E is a finite set, For when hour, ;when hour, , Indicates the first Uplink unloading power of the step; Indicates the first The "step depth" of a step; This indicates the amount of data that user k needs to uninstall; Indicates the use of steps below the specified level. The data rate achieved by the power; This represents the optimal number of steps for unloading uplink power.

[0039] Preferably, the method of solving the UAV trajectory control problem piecewise using a continuous convex approximation algorithm includes:

[0040] ;

[0041] in, For the drone's flight path, This represents the choice between near-shore users and MEC servers; Indicates the process Iterative near-shore communication cost matrix; Indicates the number of iterations.

[0042] Preferably, the standard convex optimization algorithm is used to solve the CPU frequency allocation problem, including:

[0043] ;

[0044] in, Represents the cost matrix for near-shore communications; This represents the CPU frequency allocation matrix.

[0045] The present invention also provides a near-shore communication network optimization system, the system being used to implement the aforementioned method, the system comprising: a first construction module, a second construction module, and an adjustment module;

[0046] The first building module is used to build a hybrid path loss model, i.e., the HPL model, adapted to near-shore communication scenarios based on the distance between the mobile user and the MEC server.

[0047] The second building module is used to simulate the actual communication scenarios of near-shore users based on the HPL model and to build an optimization model for the near-shore communication network.

[0048] The adjustment module is used to dynamically adjust the nearshore communication strategy through the nearshore communication network optimization model to achieve low-latency and low-energy nearshore communication.

[0049] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0050] As the transmit-receive distance increases, the signal propagation environment in near-shore channels becomes more complex, and multipath effects become more pronounced. For near-shore LoS transmission, this invention... A dual-ray path loss model is used within the specified range. A three-ray path loss model was used within the range to effectively simulate the signal loss of a moving vessel receiving signals near the sea surface.

[0051] To address the complex, non-convex joint optimization problem of latency and energy consumption in near-shore communication, this invention proposes a method based on task offloading and MEC server selection decisions, and power allocation. and Unmanned aerial vehicle (UAV) trajectory control By performing hierarchical processing of CPU frequency allocation f, extracting features from different sub-problems and adopting appropriate optimization strategies for their features, the efficiency and accuracy of the overall solution can be improved, thereby minimizing the total cost of latency and energy consumption for near-shore users to perform communication computing. Attached Figure Description

[0052] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments are 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 these drawings without creative effort.

[0053] Figure 1 This is a schematic diagram of the hybrid path loss model according to an embodiment of the present invention;

[0054] Figure 2 This is a structural diagram of the hierarchical alternating minimization algorithm according to an embodiment of the present invention;

[0055] Figure 3 The following are the effect diagrams of the HPL model in the embodiment of the present invention, wherein (a) is a dual-ray path loss model and (b) is a hybrid path loss model.

[0056] Figure 4 This is a rendering of the LAM model according to an embodiment of the present invention;

[0057] Figure 5 This is a schematic diagram of a near-shore communication network optimization method according to an embodiment of the present invention. Detailed Implementation

[0058] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0059] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0060] Example 1

[0061] In nearshore MEC communication networks assisted by multiple UAVs, distributed collaborative optimization faces numerous technical challenges. First, the network architecture is complex and dynamically changing. Communication links between UAVs and nearshore equipment and user terminals are susceptible to environmental factors such as waves, wind speed, and signal interference, leading to latency fluctuations and increased energy consumption. Second, the collaborative optimization of data transmission and computing tasks requires consideration of multi-level factors, including task offloading and MEC server selection decisions, uplink and downlink power allocation, UAV trajectory control, and CPU frequency allocation. These factors are interconnected, increasing the optimization difficulty. Traditional centralized optimization methods cannot decouple these multi-level factors to reduce optimization difficulty; furthermore, the variability and dynamism of the nearshore environment require algorithms to adjust optimization strategies in real time, which traditional centralized optimization methods cannot meet.

[0062] like Figure 5 As shown in the figure, this embodiment provides a method for optimizing near-shore communication networks, the method comprising:

[0063] Based on the distance between mobile users and MEC servers, a hybrid path loss model, namely the HPL model, is constructed to adapt to near-shore communication scenarios.

[0064] Based on the HPL model, a near-shore communication network optimization model is constructed to simulate the actual communication scenarios of near-shore users.

[0065] By dynamically adjusting the nearshore communication strategy through a nearshore communication network optimization model, low-latency and low-energy nearshore communication can be achieved.

[0066] like Figure 1 As shown, for near-sea line-of-sight (LoS) transmission, the signal propagation environment becomes more complex and multipath effects become more pronounced with increasing transmit-receive distance. This invention considers... A dual-ray path loss model is used within the range. The three-ray path loss model is used within the range to simulate the situation of a moving ship receiving signals near the sea surface.

[0067] If in Moving ships at all times With index The straight-line distance between the only BS-MEC servers is , The two-ray path loss model is then used.

[0068] (1);

[0069] In formula (1) and They are respectively and Antenna height, Where is the carrier wavelength, and .

[0070] When the distance is greater than At that time, there exists a third ray from the captured signal in the evaporation pipe layer, therefore the path distance of the three rays is... The loss model is defined as:

[0071] (2);

[0072] in, , This is the effective height of the evaporation pipe. We also need to consider the small-scale Rayleigh fading caused by the multipath effect between the moving vessel and the BS-MEC.

[0073] The core idea of ​​the LAM algorithm proposed in this invention is to decompose a complex optimization problem into multiple subproblems, each subproblem corresponding to a level in the optimization process, and then gradually approach the global optimum by alternately minimizing the variables at each level. For example... Figure 2As shown, the decision-making problem of mobile user task offloading and MEC server selection is first solved based on game theory, and the Nash equilibrium solution is obtained as the decision matrix. The optimal solution. Then, the decision matrix. As known quantities, the uplink and downlink power optimization problem for task execution is solved using the geometric water injection algorithm. Then, the continuous convex approximation SCA algorithm is used to solve the UAV trajectory control problem piecewise. Finally, with the known decision matrix... Uplink and downlink power allocation for task execution and and drone flight trajectory In this case, CPU frequency allocation, which has become a standard convex optimization problem, is solved using a standard convex optimization algorithm. The aim of this distributed algorithm is to minimize the cost function of the entire near-shore communication system. :

[0074] (3);

[0075] s. (3a);

[0076] (3b);

[0077] (3c);

[0078] (3d);

[0079] (3e);

[0080] (3f);

[0081] (3g);

[0082] (3h);

[0083] (3i);

[0084] (3j);

[0085] (3k);

[0086] (31);

[0087] (3m);

[0088] (3n);

[0089] (3o);

[0090] (3p);

[0091] Constraints (3a) and (3b) state that, in any time slot, a user can only execute a task locally or on one of the MEC servers. Furthermore, (3c), (3d), and (3e) represent the total available transmission bandwidth and the minimum required transmission rate, respectively. Constraints (3f), (3g), and (3h) give the minimum and maximum transmission power for unloading computational tasks and downloading computation results. Constraints (3i), (3j), and (3k) define the upper and lower limits of the computational resources allocated to the task. This is the maximum computing resource allocated to mobile user k for local computation. Similarly, Indicates MEC server The maximum computing power. Furthermore, constraint (31) also stipulates that when a time span... When finished, the UAV-MEC will return to its initial position. Reach the end position Furthermore, (3m) indicates that the drone's flight trajectory is also affected by its initial velocity. and ending speed The influence of (3n). The maximum speed of the drone was specified. This ensures that the UAV can maintain flight, and (3o) indicates that the UAV is limited by the maximum acceleration within a certain time slot, and that the UAV's flight trajectory is a uniformly accelerating process. The constraint in (3p)... This gives the minimum distance between drones to prevent collisions. This is achieved by selecting a sufficiently small time slot. Assuming the drone's location is in each time slot The internal remains unchanged. Problem (3) is a typical mixed-integer non-linear programming (MINLP) problem. This invention proposes a solution—the Layered Alternating Minimization (LAM) algorithm—to find the optimal solution to this NP-hard problem.

[0092] (1) Decision game of mobile user task unloading and MEC server selection

[0093] Considering uplink and downlink power allocation, the locations of base stations and drones, and CPU frequency allocation, the task offloading decision and MEC server selection decision problem can be formulated as follows:

[0094] (4);

[0095] For offshore mobile user terminals An index function was introduced. ,Right now:

[0096] (5);

[0097] in, Mobile users The selection of variables, each variable will be in the time slot. The interior remains unchanged. If This indicates a mobile user. The calculations are performed locally, meaning no uninstallation is performed. , Indicates user The computational task in server When executing this, special attention should be paid to the following: This indicates that the uninstallation is performed on the BS-MEC server. This indicator function is used to monitor near-shore mobile user terminals. Total cost Simplified to:

[0098] (6);

[0099] in, Indicates mobile user and server The cost of communication between them.

[0100] Therefore, considering uplink and downlink power allocation, the location of base stations and drones, and CPU frequency allocation, and taking into account each time slot... Minimizing will lead to a longer time span Minimizing this subproblem allows us to reformulate the first subproblem as follows:

[0101] (7);

[0102] For the first subproblem, in order to avoid collecting a large number of user parameters, the first layer of the LAM algorithm proposes a game theory-based task offloading and MEC server selection decision algorithm. This algorithm is a low-complexity solution and is also conducive to quickly obtaining a distributed solution to the optimization problem proposed in this part.

[0103] This invention assumes that near-shore mobile user terminals are engaged in strategic game theory. ,in It refers to the number of mobile users in the system. It is the strategy space. User The function that minimizes the total cost. The possible values ​​are as follows:

[0104] (8);

[0105] In the above game, each near-shore user terminal attempts to minimize its own cost (6), that is, to find the optimal value of the choice variable for itself:

[0106] (9);

[0107] in, Indicates excluding user terminals The set of arbitrary choices made by the remaining users. This invention first needs to prove the game theory... Does a Nash equilibrium exist where no user can further reduce costs by changing their choice variables?

[0108] Game Theory The Nash equilibrium is the process of finding a set of choice variables and strategies. , so that:

[0109] (10);

[0110] According to the potential function game theory, if the cost function It can be represented as a potential function So, game theory It is a power game, which can be represented as:

[0111] (11);

[0112] in, .

[0113] Similarly, we can obtain:

[0114] (12);

[0115] therefore, User In selecting a decision set The best possible choice under given conditions. It is also defined as the potential function game of optimal choice, i.e.:

[0116] (13);

[0117] Since equation (13) satisfies the definition of the potential function, therefore It is a potential function game. It provides an optimal solution. As a Nash equilibrium solution, it ensures an optimal trade-off between local costs and total system costs.

[0118] Therefore, the dimension is The task unloading decision and MEC server selection decision set s can be given by the following elements:

[0119] (14);

[0120] in, This represents the optimal decision matrix for a near-shore communication system.

[0121] (2) Geometric water injection algorithm

[0122] Taking into account offloading and MEC server selection decisions, UAV location, and CPU frequency allocation, the second layer optimizes the transmit power for uplink offloading and downlink downloading respectively, by minimizing the transmit power of each time slot. The cost can extend the time span. The goal of this layer is to minimize the total cost within the layer. The simplified subproblem is described below:

[0123] (15);

[0124] Problem (15) will be solved in two parts: first, the power allocated to offloading the computational task to the MEC server; and second, the power allocated to downloading the computational results back to the user. The second layer uses the GWF algorithm to avoid the necessity of solving the nonlinear model from the Karush-Kuhn-Tucker (KKT) conditions of the target problem, thereby determining the power level. Furthermore, compared to traditional techniques, the GWF method requires less processing time and has similar memory costs.

[0125] The optimization of uplink offload power is allocated based on user terminal k, used for uplink offload calculation, to provide the optimal power allocation for efficient utilization of user equipment. The problem can be described as follows:

[0126] (16);

[0127] In this part of the algorithm, the path loss for communication between BS-MEC, UAV-MEC, and near-shore users in a noise-free environment must first be defined, namely:

[0128] (17);

[0129] in, Indicates UAV and mobile users Communication between them follows the free path loss model. When L(k)=0, it indicates that there is no communication link loss in the local calculation. Equation (17) determines the value of the function solely by the mobile user k because the task execution mode (local offload or offload to MEC) is selected first. Therefore, the uplink rate of near-shore communication in a noise-free environment is redefined as:

[0130] (18);

[0131] The GWF algorithm was used to dynamically solve the uplink and downlink power allocation. Assuming four unit width steps (K=4) are applied in the water tank, the number of near-shore users K is chosen to represent the number of steps in the GWF algorithm. This paper uses... This represents the "step depth" of the k-th step, i.e., the... The formula for calculating the height from the step to the bottom of the tank is:

[0132] (19);

[0133] Due to the channel gain sequence The sorting order is monotonically decreasing, therefore... The step depth for indexing is monotonically increasing.

[0134] let Give the use of steps below (in, The data rate achieved by the power of ) It is a sequence It is a set The cardinality, therefore It can be represented as .So, It can be represented as:

[0135] (20);

[0136] in, Therefore, according to the above formula, the exponential rate can be written as:

[0137] The cost function in problem (16) Cost minimization can be achieved by minimizing the offloading power. Therefore, the explicit power solution can be given by the following formula:

[0138] (twenty one);

[0139] in .

[0140] The power level for this step is:

[0141] (twenty two);

[0142] in, Indicates the first Uplink unloading power of the step; Indicates the first The "step depth" of a step; Indicates user The amount of data that needs to be uninstalled; Indicates the use of steps below the specified threshold. The data rate achieved by the power; This represents the optimal number of steps for unloading uplink power.

[0143] In the next time phase, a new state will emerge, generating an optimal power allocation procedure with state feedback. With a finite set The set continues to shrink until it reaches its final size. When exhausted, the framework will go through The optimal solution for power allocation is determined by using a series of cycles.

[0144] Power is allocated to each drone / base station transmitter for calculating download results. We minimize the connection to the MEC server m (m∈ The cost function is optimized by considering the download power of all users. The simplified subproblem for each transmitter m can be expressed as:

[0145] (twenty three);

[0146] Assumption It is a partition of the index set: For simplicity, we can... The elements in the array are monotonically increasing, that is... To solve problem (23), the following method is used:

[0147] (twenty four);

[0148] in, And the power level of this step It can be obtained using expressions (25) and (26) respectively;

[0149] (25);

[0150] (26);

[0151] on the other hand, The value can be obtained from the total transmit power. Deduction steps The water volume was calculated using the following formula:

[0152] (27);

[0153] Since power cannot be negative, therefore The symbol means: when hour, (n); when hour, ; Indicates the first The downlink power allocated to each user is obtained by subtracting the cumulative difference in power allocated to each user from the maximum available power, and then taking the non-negative value. This allocation method ensures efficient power utilization and fairness among users.

[0154] Indicates server Maximum total transmit power; This indicates the number of steps in the GWF algorithm; Indicates the number of mobile users in the near-shore area; This represents the maximum number of mobile users in the near-shore area. This indicates that the downlink power has been calculated up to the [number]th [number]. The "step depth" of a step; This indicates that the downlink power has been calculated up to the [number]th [number]. The "step depth" of the steps.

[0155] (3) Continuous convex approximation algorithm

[0156] Taking into account unloading decisions, uplink / downlink power allocation, and CPU frequency allocation, the UAV's flight trajectory will be optimized at the third layer. To minimize computation time, a segment-by-segment strategy is adopted to ignore variables. The influence of this divides the entire drone's flight path into smaller time periods. The simplified sub-problems are as follows:

[0157] (28);

[0158] The non-convex constraint in (3p) implies that problem (28) is neither concave nor quasi-concave. Therefore, we use Theorem 2 as defined below to transform the non-convexity and adopt an iterative SCA method, in which the main function is estimated at a given local point using a more manageable function in each iteration.

[0159] In the In the next iteration, we use This indicates the drone's trajectory. Because... It is about For convex functions, based on the local properties of convex functions, the constraint condition (3p) can be transformed into the following inequality:

[0160] (29);

[0161] in, This represents the transpose of a matrix. Using the first-order Taylor expansion, at any given local point... In the case of the lower bound in (29), the problem can be approximately represented as follows:

[0162] (30);

[0163] Where (3n) is a convex quadratic constraint, and (31), (3m), (3o), and (29) are all linear constraints. As a convex optimization problem, (30) can be solved efficiently using a traditional convex optimizer (such as CVX). Therefore, the optimal objective value obtained from (30) can be used as a general upper bound for (28). This represents the choice between near-shore users and MEC servers; Indicates the process The near-shore communication cost matrix of the next iteration; Indicates the number of iterations.

[0164] (4) Standard convex optimization algorithm

[0165] Considering the offloading decision s and power allocation scheme and and the location deployment of base stations and drones The situation, and taking into account the situation for each time slot. Minimization alone can lead to a longer time span. The goal is to minimize the total cost within the system. Therefore, this invention will ultimately allocate CPU frequency for task computation. The simplified subproblem is expressed as:

[0166] (31);

[0167] in, Represents the cost matrix for near-shore communications; This represents the CPU frequency allocation matrix.

[0168] The objective function (31) is a convex function, and the constraints are also convex sets. Therefore, problem (31) is a convex optimization problem, which can be solved using standard convex optimization algorithms.

[0169] like Figure 3 As shown, the HPL model takes into account the distance between the mobile vessel and the server, and can more accurately describe the path loss variation of communication between the mobile vessel and the fixed base station. Rayleigh fading is added on this basis to make the HPL model more consistent with the complex and ever-changing near-shore communication situation.

[0170] like Figure 4 As shown, the technical advantages of the LAM algorithm are as follows: First, the algorithm effectively distributes computational and storage loads, utilizing the resources of multiple multi-access edge computing (MEC) servers to process multiple tasks in parallel, significantly improving the overall processing and storage capabilities of the communication system, thereby achieving better scalability. Second, the algorithm has higher fault tolerance and reliability. Since there is no single point of failure, even if some MEC nodes fail, the near-shore communication system can continue to operate; this characteristic is particularly important in the complex and ever-changing near-shore communication environment. Third, the LAM algorithm allows MEC nodes to perform data processing and decision-making locally, reducing data transmission latency and improving the system's real-time response capability. Fourth, through intelligent task allocation and resource management decisions, the LAM algorithm effectively achieves load balancing and efficient resource utilization, improving the overall performance of the system and meeting the low-latency and low-energy consumption requirements of near-shore communication.

[0171] Example 2

[0172] The present invention also provides a near-shore communication network optimization system, the system being used to implement any of the methods described above, the system comprising: a first construction module, a second construction module, and an adjustment module;

[0173] The first building module is used to build a hybrid path loss model, i.e., the HPL model, adapted to near-shore communication scenarios based on the distance between the mobile user and the MEC server.

[0174] The second building module is used to simulate the actual communication scenarios of near-shore users based on the HPL model and to build an optimization model for the near-shore communication network.

[0175] The adjustment module is used to dynamically adjust the nearshore communication strategy through the nearshore communication network optimization model to achieve low-latency and low-energy nearshore communication.

[0176] Example 3

[0177] This embodiment is designed as a network architecture for near-shore MEC communication, which includes multiple UAV-MECs and one BS-MEC acting as transmitters to provide network access services to K mobile vessels (users), and the terminals used by the transmitters and mobile vessels each have only one antenna.

[0178] This invention proposes a near-shore communication network optimization method—the Layered Alternating Minimization (LAM) algorithm—based on the near-shore communication environment simulated by the HPL model. First, the LAM algorithm solves the decision problem of mobile user task offloading and MEC server selection using game theory, obtaining its Nash equilibrium solution as the near-shore communication strategy. Then, the Geometric Water-Fifilling (GWF) algorithm is used to solve the dynamic allocation problem of uplink and downlink power during task execution. Next, the Sequential Convex Approximation (SCA) algorithm is used to solve the UAV trajectory control problem piecewise. Finally, a rigorous convex optimization algorithm is applied to solve the CPU frequency allocation problem. The LAM algorithm proposed in this invention can effectively reduce the weighted sum of latency and energy consumption in near-shore communication under the HPL model environment.

[0179] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for optimizing a near-shore communication network, characterized in that, The method includes: Based on the distance between mobile users and the BS-MEC server, a hybrid path loss model, namely the HPL model, is constructed to adapt to near-shore communication scenarios; A dual-ray path loss model is used within the range. A three-ray path loss model is used within the range; communication between UAV-MEC and mobile user k follows a free path loss model. Based on the HPL model, a near-shore communication network optimization model is constructed to simulate the actual communication scenarios of near-shore users. By dynamically adjusting nearshore communication strategies through a nearshore communication network optimization model, low-latency and low-energy nearshore communication can be achieved, including: The hierarchical alternating minimization algorithm is used to solve the decision problem of mobile user task offloading and MEC server selection, and the Nash equilibrium solution is obtained as the optimal solution of the decision matrix; in any time slot, a user can only execute tasks locally or on one of the MEC servers. Using the decision matrix as a known quantity, the uplink and downlink power dynamic allocation problem for task execution is solved using the geometric water-filling algorithm. Then, a continuous convex approximation algorithm is employed to solve the UAV trajectory control problem piecewise. By minimizing the cost of each time slot t, the time span can be minimized. The goal is to minimize the total cost within the network; the optimization objective is to find the uplink and downlink transmit power matrix that minimizes the total cost. This is solved in two parts: first, by allocating the power for offloading computational tasks to the MEC server; and second, by allocating the power for downloading the computation results back to the user. The optimization of uplink offload power is performed on user terminal k, used for uplink offload calculations, to provide the optimal power allocation for efficient use of user equipment. This is achieved by minimizing the connection to MEC server m, where m ∈ [missing information]. The cost function is optimized based on the download power of all users; over a time span When finished, the UAV-MEC will return to its initial position. Reach the end position By selecting a sufficiently small time slot Assuming the drone's position remains constant within each time slot t, a segmented strategy is adopted to ignore the influence of variable t, dividing the entire drone's flight trajectory into smaller time segments; Given the decision matrix, uplink and downlink power allocation for task execution, and the UAV flight trajectory, a standard convex optimization algorithm is used to solve the CPU frequency allocation problem of the MEC, taking into account the offloading decision. The power allocation scheme and the location deployment of base stations and drones, taking into account the power allocation scheme for each time slot. t Minimization alone can lead to a significant impact over a long period of time. To minimize the total cost, frequency allocation will be performed on the CPU for task computation.

2. The method according to claim 1, characterized in that, Specifically: like t Moving ships at all times k With index M The straight-line distance between the only BS-MEC servers The two-ray path loss model is used: ； In the formula, h T and h R They are respectively and Antenna height, Where is the carrier wavelength, and ; When the distance is greater than At that time, there exists a third ray from the captured signal in the evaporation pipe layer, therefore the path distance of the three rays is... The loss model is defined as: ； in, , h E This is the effective height of the evaporator pipe.

3. The method according to claim 1, characterized in that, The hierarchical alternating minimization algorithm is used to solve the decision problem of mobile user task offloading and MEC server selection. The Nash equilibrium solution is obtained as the optimal solution of the decision matrix, including: ； in, For mobile users k The choice of variables, m For MEC servers, k For mobile users, This represents the optimal decision matrix for a near-shore communication system.

4. The method according to claim 1, characterized in that, Solving the dynamic allocation problem of uplink and downlink power during task execution using the geometric water injection algorithm includes: ； ； in, E For a finite set, For when hour, ;when hour, , Indicates the first Uplink unloading power of the step; Indicates the first The "step depth" of a step; Indicates user k The amount of data that needs to be uninstalled; Indicates the use of steps below the specified level. The data rate achieved by the power; This represents the optimal number of steps for uplink power offloading. K Indicates the number of mobile users in the near-shore area. Indicates the first Downlink power allocated to each user; Indicates server Maximum total transmit power; This indicates that the downlink power has been calculated up to the [number]th [number]. The "step depth" of a step; This indicates that the downlink power has been calculated up to the [number]th [number]. The "step depth" of the step.

5. The method according to claim 1, characterized in that, Solving the UAV trajectory control problem piecewise using a continuous convex approximation algorithm includes: ； in, For the drone's flight path, This represents the choice between near-shore users and MEC servers; Indicates the process Iterative near-shore communication cost matrix; Indicates the number of iterations.

6. The method according to claim 1, characterized in that, Solving the CPU frequency allocation problem using standard convex optimization algorithms includes: ； in, Represents the cost matrix for near-shore communications; This represents the CPU frequency allocation matrix.

7. A near-shore communication network optimization system, said system being used to implement the method according to any one of claims 1-6, characterized in that, The system includes: a first construction module, a second construction module, and an adjustment module; The first building module is used to build a hybrid path loss model, namely the HPL model, adapted to near-shore communication scenarios based on the distance between the mobile user and the BS-MEC server. The second building module is used to simulate the actual communication scenarios of near-shore users based on the HPL model and to build an optimization model for the near-shore communication network. The adjustment module is used to dynamically adjust the nearshore communication strategy through the nearshore communication network optimization model to achieve low-latency and low-energy nearshore communication.

Images