A resource allocation method for ultra-dense low earth orbit satellite network

Abstract

The application provides a resource allocation method for an ultra-dense low-orbit satellite network, comprising the following steps: constructing a joint optimization model with the target of minimizing the total weighted energy consumption of the system, the joint optimization model being limited by co-channel interference and time delay constraints; generating an initial feasible solution of the joint optimization model under the condition of meeting the constraints; under the premise of fixing the continuous resource variables, updating the calculation offloading decision variables of the joint optimization model by iteratively optimizing the matching relationship between the tasks and the satellite-channel units; under the premise of fixing the calculation offloading decision variables, jointly optimizing the user terminal transmission power and the satellite calculation resource allocation amount; alternately performing a discrete optimization stage and a continuous optimization stage until a preset convergence condition is met, and obtaining an optimal offloading decision and a resource allocation scheme; and controlling the user terminal and the satellite to perform task offloading and resource allocation according to the optimal offloading decision and the resource allocation scheme.

Description

Technical Field

[0001] This invention belongs to the field of satellite communication and edge computing technology, specifically relating to a resource allocation method for ultra-dense low-Earth orbit satellite networks. Background Technology

[0002] With the rapid development of computationally intensive and latency-sensitive applications such as autonomous driving, intelligent remote sensing, and immersive virtual reality, edge computing has become a key technology for alleviating terminal computing bottlenecks and reducing service latency. However, traditional terrestrial edge computing networks struggle to cover remote areas such as oceans, deserts, and polar regions. Ultra-dense low-Earth orbit (LEO) satellite networks (UDLSNs), with their advantages of low latency, high bandwidth, and seamless global coverage, have become an important solution for building next-generation communication networks. Industry and academia are working to integrate edge computing capabilities into UDLSNs to achieve ubiquitous global communication and computing services. In UDLSNs, Joint Compute Offloading and Resource Allocation (JCORA) is a core issue for ensuring computing service quality and improving resource utilization efficiency, but achieving efficient compute offloading faces many challenges: the size, weight, and power limitations of satellite platforms restrict their computing power; co-channel interference introduced by frequency reuse technology affects transmission quality; and the tight coupling of communication and computing resources and numerous offloading options complicate joint optimization.

[0003] Currently, research on the JCORA problem in satellite networks has made some progress, with the main technical solutions including the discrete variable relaxation-convex optimization scheme and the staged alternating optimization scheme. The discrete variable relaxation-convex optimization scheme simplifies the complex mixed integer programming problem and reduces the difficulty of solving it by relaxing discrete variables into continuous variables and using traditional convex optimization techniques. The staged alternating optimization scheme adopts a framework of discrete offloading correlation optimization + continuous resource allocation. It first determines the discrete offloading correlation through metaheuristic algorithms or deep reinforcement learning (DRL) methods, and then uses a staged optimization strategy to handle the allocation of transmission power and computing resources.

[0004] While existing technologies have addressed the JCORA problem to some extent, significant shortcomings remain. In UDLSNs, the discrete variable relaxation-convex optimization scheme suffers from significant deviations between the solution and the actual optimal solution due to the massive discrete decision space. Furthermore, most schemes neglect co-channel interference and the heterogeneous requirements of user tasks. In the phased alternating optimization scheme, metaheuristic algorithms struggle to generate high-quality discrete solutions due to the complex coupling of discrete and continuous variables, stringent interference, and resource constraints in UDLSNs. The DRL method, on the other hand, suffers from high training costs and unstable policy learning due to the large variable space and complex interference environment of UDLSNs. Moreover, phased optimization strategies tend to converge to local optima under stringent service delay constraints, failing to achieve a globally optimal match between communication and computing resources. The ultra-dense deployment of UDLSNs leads to complex interference environments and stringent resource constraints, while the JCORA problem itself exhibits strong coupling between discrete and continuous variables, making it difficult to simplify the solution while simultaneously ensuring model adaptability, optimal solution results, and algorithm efficiency and stability. In summary, existing solutions are mostly migrated from traditional satellite networks and do not fully consider the unique technical characteristics of UDLSNs, such as large-scale node deployment, strong interference, and heterogeneous user needs, making it difficult for the technical solutions to adapt to the application scenarios of UDLSNs. Summary of the Invention

[0005] This invention proposes a resource allocation method for ultra-dense low-orbit satellite networks, which solves the problem that existing methods are difficult to balance scenario adaptability, solution efficiency and solution stability.

[0006] To address the aforementioned technical problems, this invention provides a resource allocation method for ultra-dense low-Earth orbit satellite networks, comprising the following steps: Step S1: Construct a joint optimization model with the objective of minimizing the total weighted energy consumption of the system. The joint optimization model is constrained by co-channel interference and time delay constraints. Step S2: Generate an initial feasible solution for the joint optimization model under the constraints. Step S3: Under the premise of fixed continuous resource variables, update the computational offloading decision variables of the joint optimization model by iteratively optimizing the matching relationship between the task and the satellite-channel unit; Step S4: Under the premise of fixing the computational offloading decision variables, jointly optimize the user terminal transmit power and the satellite computational resource allocation; Step S5: Alternately execute steps S3 and S4 until the preset convergence condition is met to obtain the optimal unloading decision and resource allocation scheme; Step S6: Based on the optimal unloading decision and resource allocation scheme, control the user terminal and satellite to perform task unloading and resource allocation.

[0007] Preferably, the total weighted energy consumption of the system includes the uplink transmission energy consumption of the user terminal and the computing energy consumption of the satellite, and the objective function of the joint optimization model is expressed as: ; In the formula, , , These represent the unloading decision variable matrix, the transmission power variable vector, and the computing resource allocation variable matrix, respectively. , , These represent the task set, satellite set, and sub-channel set, respectively. Unload decision variables for binary when the task via sub-channel Unload to satellite The value is 1 if it is true, and 0 otherwise. For the task The uplink transmit power of the corresponding user terminal; Indicates when the task via sub-channel Unload to satellite At that time, satellite Assigned to task Computing resources; For the task The amount of input data; For the task Required number of CPU cycles; For the task The corresponding user's energy weighting coefficient; For satellite Energy consumption coefficient; For the task via sub-channel To satellite Uplink transmission rate during transmission.

[0008] Preferably, the constraints of the joint optimization model include at least: Offload uniqueness constraint: Each task must be offloaded to one and only one subchannel of a satellite; Channel exclusivity constraint: Each sub-channel of each satellite can serve at most one task at any given time; Co-channel interference constraint: The uplink transmission rate is calculated according to Shannon's formula. The denominator of the signal-to-interference-plus-noise ratio (SINR) of the uplink transmission rate includes co-channel interference from other tasks on the same sub-channel. The expression for the uplink transmission rate is: ; In the formula, Sub-channel bandwidth; For satellite The noise power; For the task With satellite Sub-channel Channel gain between; Indicates from user terminal via channel Send to satellite Interference signals at user terminals and satellite Channel gain between; Latency constraint: The sum of the uplink transmission delay and computation delay of each task shall not exceed the maximum tolerable delay of the current task. ; Resource constraints: The total computing resources allocated to any satellite shall not exceed the maximum computing capacity of the current satellite. And the transmit power of any user terminal Within the permissible power range Inside.

[0009] Preferably, step S2 includes the following steps: Step S21: Initialize the matching weights between the mission and the satellite-sub-channel element. : ; In the formula, For the task With satellite Sub-channel Channel gain between; It is a preset, sufficiently large constant; Step S22: Construct the matching problem between the mission and the satellite-sub-channel unit into a maximum weight bidirectional matching problem, and use the Hungarian algorithm to solve the maximum weight bidirectional matching problem to obtain the initial unloading decision variables; Step S23: Initialize the transmit power for each task. If the transmission of a task will not cause co-channel interference to other tasks, set it to the maximum transmit power; otherwise, set it to the minimum transmit power. Step S24: Perform inner iterations to check and eliminate all task latency violations and satellite computing resource violations until an initial feasible solution that satisfies all constraints is obtained.

[0010] Preferably, step S24 includes the following steps: Step S241: Calculate the maximum feasible computation delay for each task under the current resource allocation. : ; In the formula, For the task Maximum tolerable delay; This represents the transmission delay at the current transmission power. The computation latency under the current computing resource allocation; Step S242: If a task exists This was determined to be a time delay violation, and the following was selected. The minimum mission requires increasing transmission power; Step S243: If all tasks have no latency violations, calculate the minimum feasible computational resources for each task. : ; In the formula, To complete the task Required number of CPU cycles; If the total allocation of a satellite exceeds its maximum computing capacity, select the current satellite. The largest mission requires increased transmission power; Step S244: Repeat steps S241 to S243 until there are no violations or the preset number of iterations is reached; if no feasible solution is found after the preset number of iterations, reduce the corresponding matching weight. Then return to step S22 to regenerate the uninstallation decision.

[0011] Preferably, the iterative optimization of the matching relationship between the task and the satellite-channel element in step S3 includes the following steps: Step S31: Model the association problem between the task set and the satellite-sub-channel element set as a two-sided matching model; Step S32: Define the mission's preference for satellite-subchannel elements based on channel quality and current interference level; Step S33: Determine the set of interchangeable matching pairs. Only when two tasks are served by the same satellite or the coverage areas of the satellites serving the two tasks overlap, determine that the two tasks and their corresponding satellite-sub-channel elements constitute an interchangeable matching pair. Step S34: Traverse the set of interchangeable matching pairs according to preference degree, perform feasibility verification and optimization on candidate exchanges, and update the matching relationship if the total weighted energy consumption of the system is reduced and all constraints are satisfied after the exchange.

[0012] Preferably, step S34, which involves feasibility verification and optimization of candidate swaps, includes the following steps: when evaluating candidate swaps, calculating the constraint satisfaction of the two direct tasks involved in the swap, as well as the constraint satisfaction of other tasks affected by the change in the interference environment caused by the swap; if any related task is found to have a delay violation or a computational resource violation, increasing the transmit power of the violating direct task or reducing the transmit power of the interference source task that caused the violation; if all related tasks satisfy the constraints after power coordination adjustment, and the total weighted energy consumption of the system is reduced after continuous resource optimization, then confirming the execution of the current swap.

[0013] Preferably, the joint optimization of user terminal transmit power and satellite computing resource allocation in step S4 includes the following steps: Step S41: Based on the current unloading decision variables, construct a continuous resource allocation subproblem; Step S42: Introduce auxiliary variables The quadratic transformation technique is used to transform the fractional terms in the objective function of the joint optimization model. Convert to polynomial form; Step S43: With other variables fixed, solve using closed-form methods. Update auxiliary variables The expression for the closed-form solution is: ; Step S44: Transform the original non-convex continuous resource allocation problem into a series of convex optimization subproblems and solve them iteratively.

[0014] Preferably, step S44 includes the following steps: Step S441: Introduce slack variables and Handling non-convex signal-to-interference-to-noise ratio constraints, where This is the lower bound of the signal-to-interference-plus-noise ratio. satisfy , For the task With satellite Sub-channel Channel gain between; Step S442: Using the first-order Taylor expansion technique, perform a first-order Taylor expansion on the concave function terms in the objective function. Convex approximation is performed on the non-convex terms in the constraints, in the th... Constructing a convex approximation subproblem in the next iteration: ; In the formula, , These are the unloading decision variable matrix and the auxiliary decision variables, respectively; To unload decision variables, namely the matching relationship between the mission and the satellite-channel; Step S443: Solve the convex approximation subproblem using a standard convex optimization solver to obtain the optimal transmit power and computational resource allocation for the current iteration.

[0015] Preferably, the ultra-dense low-Earth orbit satellite network is divided into multiple satellite control domains, each managed by a master satellite; step S6 specifically includes the following steps: Step S61: The main satellite generates flow table entries and control commands based on the calculated optimal offloading decision and resource allocation scheme; Step S62: Send the flow table entries and control commands to each serving satellite and user terminal in the domain via the inter-satellite control channel; Step S63: The user terminal adjusts the transmission power according to the control command and sends the mission data to the designated service satellite via the designated sub-channel; Step S64: The service satellite allocates a corresponding computing frequency to the received task for processing according to the control command.

[0016] The beneficial effects of the present invention include at least the following: 1. Co-channel interference and time delay constraints were clearly considered during the model building phase. The complex interference environment and strict time delay requirements of UDLSNs are one of the main challenges in practical applications. By introducing these constraints into the optimization model, it can be ensured that the optimization results are operable and effective in real-world scenarios. 2. By adopting a stepwise optimization strategy, different variables are alternately fixed and optimized to gradually approach the optimal solution. This decomposes the complex joint optimization problem into several relatively simple sub-problems, reducing the difficulty of solving the problem and the computational complexity, making it more suitable for large-scale, dynamic UDLSNs scenarios. Through iterative optimization, the algorithm gradually approaches the optimal solution, which can effectively improve the convergence speed and solution efficiency of the algorithm, and reduce the computation time and resource consumption. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention; Figure 2 This is a system architecture diagram of an ultra-dense low-Earth orbit satellite network in an embodiment of the present invention; Figure 3 This is a convergence curve diagram of the algorithm in an embodiment of the present invention; Figure 4(a) shows an embodiment of the present invention. Performance box plot of the time-limited algorithm; Figure 4(b) shows an embodiment of the present invention. Box plot of the algorithm's performance; Figure 4(c) shows an embodiment of the present invention. Box plot of the algorithm's performance; Figure 4(d) shows an embodiment of the present invention. Box plot of the algorithm's performance; Figure 5 This is a graph showing how the algorithm's performance changes with the number of tasks in an embodiment of the present invention. Detailed Implementation

[0018] 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 protection scope of the present invention.

[0019] like Figure 1 As shown, this embodiment of the invention provides a resource allocation method for ultra-dense low-Earth orbit satellite networks, including the following steps: Step S1: Construct a joint optimization model with the objective of minimizing the total weighted energy consumption of the system. The joint optimization model is constrained by co-channel interference and time delay.

[0020] Specifically, this step addresses the joint computation offloading and resource allocation problem in ultra-dense low-Earth orbit satellite networks by establishing a mixed-integer nonlinear programming model to minimize weighted energy consumption. The model aims to minimize the total weighted energy consumption of the system, which includes uplink transmission energy consumption of user terminals and satellite computational energy consumption. The model simultaneously considers uplink co-channel interference, differentiated latency constraints for user tasks, limited onboard energy and computational resources of satellites, and differentiated energy states of user terminals.

[0021] Step S11: Determine the system network architecture and basic parameters.

[0022] To address the highly dynamic, resource-constrained, and large-scale characteristics of ultra-dense low-Earth orbit (LEO) satellite networks, a hierarchical management architecture based on software-defined networking (SDN) is adopted. This architecture logically divides the ultra-dense LEO satellite network into multiple relatively independent satellite control domains. Each control domain contains several adjacent LEO satellites and is managed by a master satellite with high computing power. The master satellite maintains a global network view through an intra-satellite secure OpenFlow channel, enabling centralized real-time control of the forwarding, computing, and storage resources of all satellites within its domain. This model focuses on a single satellite control domain scenario, employing a discrete-time system model. The dynamic topology of the ultra-dense LEO satellite network is discretized into a series of time slots, assuming a quasi-static network state within each time slot. This model covers multiple LEO satellites, multiple computing task users, and multiple sets of orthogonal sub-channels within the satellite control domain. The system architecture is as follows: Figure 2 As shown.

[0023] For a given satellite control domain, the following parameters are defined: the set of low-Earth orbit satellites within the domain is... and determine each satellite Maximum computing power ; The set of computationally intensive tasks to be processed is and for each task Determine its parameters, including the amount of input data. Number of CPU cycles required to complete the calculation Maximum tolerable service latency and the selectable range of user terminal transmit power ; The system employs orthogonal frequency division multiple access (OFDMA) channel resources, determining the total set of orthogonal sub-channels as follows: and each satellite Available sub-channel subset .

[0024] Step S12: Define the decision variables for the joint optimization problem.

[0025] For the mission set, satellite set, and channel resources, the following three types of decision variables are defined to fully describe an offloading and resource allocation scheme. Offloading decision variables It is a binary variable. Time indicates task via sub-channel Unloaded to satellite Using sub-channels Perform an uplink transmission; otherwise, return 0.

[0026] Transmission power variable As a continuous variable, representing the task The uplink transmit power of the corresponding user terminal is within a preset power range.

[0027] Calculate resource variables As a continuous variable, it represents the satellite as its associated mission. The allocated computing resources are limited by the satellite's maximum computing power.

[0028] Step S13: Establish an uplink transmission model that includes co-channel interference.

[0029] Based on the channel gain between the satellite and the user terminal Computational tasks via sub-channel To satellite Actual uplink rate during transmission Channel gain The calculation incorporates factors such as antenna gain, path loss, atmospheric attenuation, and rain attenuation. The uplink transmission rate is calculated using Shannon's formula. The denominator of the signal-to-interference-plus-noise ratio (SINR) includes co-channel interference from other tasks on the same sub-channel, specifically expressed as: ; in, Sub-channel bandwidth; For satellite The noise power; For the task With satellite Sub-channel Channel gain between; Indicates from user terminal via channel Send to satellite Interference signals at user terminals and satellite Channel gain between.

[0030] The summation term in the denominator of the above formula explicitly represents the co-channel interference caused by other tasks on the same sub-channel, where... Indicates tasks Other tasks, Including satellites Other satellites besides [the main satellite].

[0031] Step S14: Construct an optimization problem with the objective of minimizing the weighted total energy consumption.

[0032] The mathematical model for the joint computational unloading and resource allocation problem is defined as the following mixed-integer nonlinear programming problem, with the objective function being: ; in, , , These represent the unloading decision variable matrix, the transmission power variable vector, and the computing resource allocation variable matrix, respectively. , , These represent the task set, satellite set, and sub-channel set, respectively. For the task The corresponding user's energy weighting coefficient is used to characterize the differentiated energy shortage level of different terminals; For satellite The energy consumption coefficient.

[0033] The objective function consists of two terms: the first term represents the weighted transmission energy consumption of the user terminal, and the second term represents the weighted computation energy consumption of the satellite. By minimizing the weighted sum of energy consumption by the user and the satellite, this model aims to improve the overall energy efficiency of the network.

[0034] Step S15: Establish differentiated service quality and resource constraints.

[0035] Based on the above variables and model, construct the constraints to ensure the feasibility of the solution: Offloading Uniqueness Constraint: Each task must and can only be offloaded to one subchannel of a satellite, ensuring that each computational task has one and only one unique offloading path, i.e.: .

[0036] Channel exclusivity constraint: Each satellite's sub-channel serves at most one task at a time to avoid intra-domain interference and ensure exclusive use of channel resources, i.e.: .

[0037] Co-channel interference constraint: The uplink transmission rate is calculated according to the Shannon formula. The denominator of the signal-to-interference-plus-noise ratio of the uplink transmission rate includes co-channel interference from other tasks on the same sub-channel. This constraint reflects the interference characteristics in ultra-dense deployment scenarios.

[0038] Latency constraint: The sum of the uplink transmission delay and computation delay of each task shall not exceed the maximum tolerable delay of the current task, that is: ; Transmission delay is determined by the amount of task input data and the sub-channel transmission rate, which is calculated by combining channel gain, transmission power, and co-channel interference; computation delay is determined by the number of CPU cycles required for the task and the allocated computation frequency. Time, i.e., task i via sub-channel k Unload to satellite j At that time, its uplink transmission delay and computation delay are respectively: ; ; In the above formula, This refers to the uplink transmission delay; Indicates task Corresponding user terminal to satellite The distance between them; The speed of light; To indicate the task via sub-channel Unloaded to satellite The computational delay during calculation; To complete the task The number of CPU cycles required.

[0039] Resource constraints: The total computing resources allocated to any satellite shall not exceed the maximum computing capacity of the current satellite, that is: ; And the transmit power of any user terminal is within the allowable power range, that is: .

[0040] Step S2: Generate an initial feasible solution for the joint optimization model under the condition of satisfying the constraints.

[0041] Specifically, this step designs a heuristic initialization strategy based on the Hungarian algorithm to generate a high-quality initial solution that satisfies the basic constraints of the system, ensuring stable execution and convergence speed of subsequent iterations.

[0042] Step S21: Initialize the matching weights between the task and the satellite-sub-channel element.

[0043] Set both the outer and inner iteration counters to 0. Initialize the matching weights between the task and the satellite-subchannel element. The weight calculation formula is: ; In the formula, For the task With satellite Sub-channel Channel gain between; The preset constant is large enough to avoid excessive matching bias towards high-gain channels, which could prevent some tasks from being offloaded. This ensures offloading fairness and prevents satellite overload, ensuring that as many tasks as possible can be offloaded.

[0044] Step S22: Solve the initial unloading association using the Hungarian algorithm.

[0045] The task-satellite-subchannel cell matching problem is constructed as a maximum-weight bidirectional matching problem, and the Hungarian algorithm is used to solve it to obtain the initial offloading decision variables. Determine whether a feasible solution has been obtained or the number of outer iterations. Reaching the preset outer iteration limit If not satisfied, then And continue execution. Unload decision variables. A value of 1 indicates that the task is offloaded to the corresponding satellite through the corresponding sub-channel, while a value of 0 indicates that the offloading relationship does not exist.

[0046] Step S23: Initialize transmit power.

[0047] Initialize transmit power for each mission If the transmission of a task will not cause co-channel interference to other tasks, then set it to the maximum transmission power. Otherwise, set to minimum transmission power. This initialization strategy ensures transmission quality while minimizing interference with other tasks.

[0048] Step S24: Perform inner iteration to eliminate constraint violations.

[0049] Perform inner-layer iterations to check and eliminate latency violations and satellite computing resource violations for all tasks until an initial feasible solution satisfying all constraints is obtained. Determine whether all tasks satisfy the transmission power constraint. Or the number of inner iterations reaches the preset upper limit for inner iterations. If the condition is not met, the inner iteration counter is incremented and execution continues.

[0050] First, calculate the maximum feasible computation delay for each task under the current resource allocation: ; in For the task Maximum tolerable delay; This represents the transmission delay at the current transmission power. This represents the computation latency under the current computing resource allocation.

[0051] If the maximum feasible computation latency of a task is less than or equal to zero, it is considered a latency violation, and the task with the smallest such value is selected. , Increase the transmission power and its transmission power Increase preset step size .

[0052] If all tasks have no latency violations, then calculate the minimum feasible computational resources for each task: ; in To complete the task The number of CPU cycles required.

[0053] If the total allocation of a satellite exceeds its maximum computing capacity, select the task with the largest minimum feasible computing resources on the current satellite and increase its launch power.

[0054] Repeat the above steps until no violations are found or the preset number of iterations is reached. If no feasible solution is found after the preset number of iterations, reduce the corresponding matching weight and return to step S22 to regenerate the unloading decision. If no delay violation or computational resource violation is found, output the initial feasible solution, including the initial unloading decision variables. Initial transmission power Initial computational resource allocation .

[0055] Step S3: Under the premise of fixed continuous resource variables, update the computational offloading decision variables of the joint optimization model by iteratively optimizing the matching relationship between the task and the satellite-channel unit.

[0056] Specifically, this step models the optimization problem of discrete unloading decision variables as a two-sided matching problem under the premise of fixed continuous resource variables, designs a matching optimization algorithm based on cooperative exchange, and optimizes the matching relationship through local exchange operations with cooperative mechanisms.

[0057] Step S31: Construct a bilateral matching model Combine the task set with the satellite-subchannel unit set The correlation problem is modeled as a bilateral matching model. A bilateral matching relationship is defined to match the task set with the satellite-sub-channel element set. The matching relationship must satisfy three constraints: Each task is matched with only one satellite-sub-channel element, that is: and , ; In the formula, This indicates a matching mapping relationship, that is, mapping from a task to a matching satellite-subchannel unit; Each satellite-subchannel element is assigned only one task, namely: and , .

[0058] The matching requires bidirectional uniqueness, meaning a task is matched to a satellite-sub-channel element if and only if that satellite-sub-channel element is assigned to the task. .

[0059] Step S32: Define the mission's preference for satellite-sub-channel elements.

[0060] Defining tasks based on channel quality and current interference levels. For satellite-subchannel units Preference : ; In the formula, For satellite Noise power.

[0061] Preference is quantified as the ratio of channel gain to the sum of interference power and noise power. A higher preference indicates lower potential transmission power consumption and lower latency. This indicator comprehensively reflects the impact of channel quality and interference environment on transmission performance.

[0062] Step S33: Determine the set of interchangeable matching pairs.

[0063] Define an exchange match as two existing matching pairs. and The association relationships are interchangeable. This only applies when two missions are served by the same satellite. Or the coverage areas of satellites serving two different missions may overlap. When determining that the two tasks and their corresponding satellite-sub-channel elements constitute an exchangeable matching pair, the specific expression is as follows: ; Initialize the preference list for tasks and satellite-subchannel units, sort all current matching pairs in descending order of energy consumption per unit of data, and prioritize high-energy-consuming matching pairs to improve optimization efficiency.

[0064] Step S34: Perform cooperation exchange and feasibility verification.

[0065] Traverse the set of commutable pairs by preference, perform feasibility checks and optimizations on candidate swaps, and update the matching relationships if the total weighted energy consumption of the system decreases and all constraints are satisfied after the swap. For each sorted pair, traverse its set of commutable pairs in descending order of preference. For each candidate swap pair, perform the following cooperative swap operation: When evaluating candidate swaps, computation involves two direct tasks. , The constraint satisfaction status, and other tasks affected by changes in the interference environment due to the exchange. , The constraint satisfaction status; If any related task is found to have a latency violation or a computational resource violation, the transmit power of the directly violating task will be increased. Or reduce the transmit power for missions that are causing interference that could lead to violations. ; If all relevant tasks meet the constraints after power coordination adjustment, and the total weighted energy consumption of the system decreases after continuous resource optimization, then the current exchange is confirmed to be executed.

[0066] After the traversal is complete, output the optimized discrete unloading decision variables. .

[0067] Step S4: Under the premise of fixing the computational offloading decision variables, jointly optimize the user terminal transmit power and the satellite computational resource allocation.

[0068] Specifically, this step, under the premise of fixing the decision variables for computational unloading, designs a fractional programming-assisted successive convex approximation algorithm to solve the continuous resource allocation problem.

[0069] Step S41: Based on the current unloading decision variables, construct a continuous resource allocation subproblem.

[0070] Based on the optimized unloading decision variables, the optimization objective of the continuous resource allocation subproblem is to minimize the total weighted energy consumption, with constraints including upper limits on satellite computing resources and mission latency. ; ; ; ; ; In the formula, This refers to the currently optimized unloading decision variables; This represents the signal-to-interference-to-noise ratio.

[0071] Because the objective function contains fractional terms and the constraint variables are coupled, this problem is non-convex and requires appropriate transformations to solve.

[0072] Step S42: Introduce auxiliary variables to perform a second transformation of fractional programming.

[0073] Introducing auxiliary variables The quadratic transformation technique is used to transform the fractional terms in the objective function of the joint optimization model. Transform it into an integer form to construct an equivalent parametric optimization problem: ; This transformation preserves the optimality of the original problem and lays the foundation for subsequent convex optimization solutions.

[0074] Step S43: Update auxiliary variables With other variables fixed, a closed-form solution can be used. Update the auxiliary variable; the expression for the closed-form solution is: ; For fixed transmission power and computational resource variables, the objective function is concave with respect to auxiliary variables, allowing us to obtain a closed-form optimal solution for the auxiliary variables. Substituting the optimal auxiliary variables into the equation yields the transformed minimization problem: .

[0075] Step S44: Successive convex approximation optimization solution.

[0076] The original non-convex continuous resource allocation problem is transformed into a series of convex optimization subproblems for iterative solution.

[0077] Introducing slack variables and Handling non-convex signal-to-noise ratio constraints, This is the lower bound of the signal-to-interference-plus-noise ratio (SIR), and satisfies... Transform the non-convex objective function into an equivalent problem: ; Using the first-order Taylor expansion technique, a convex approximation is performed on the concave terms in the objective function and the non-convex terms in the constraints. n Constructing a convex approximation subproblem in the next iteration: ; The objective function of this convex approximation subproblem includes a computational energy consumption term, a logarithmic transmission rate term, and a Taylor approximation term for the square root of power. The constraints include upper limits on satellite computational resources, upper limits on mission latency, and related constraints on slack variables.

[0078] The optimal transmit power for the current iteration is obtained by solving the convex approximation subproblem using a standard convex optimization solver. Computing resource allocation After updating the approximate point, repeat the above steps until the difference between the objective function values ​​of two iterations is less than a preset threshold or the maximum number of iterations is reached, then output the optimal continuous resource variable. , .

[0079] Step S5: Alternately execute steps S3 and S4 until the preset convergence condition is met, and obtain the optimal unloading decision and resource allocation scheme.

[0080] Specifically, steps S3 and S4 are executed alternately: after updating the discrete unloading decision, the continuous resource allocation is optimized, and then the unloading decision is improved based on the new resource allocation state. This process is iterated until the convergence condition is met. The convergence condition is that the difference in the total weighted energy consumption between two iterations is less than a preset threshold or the preset maximum number of iterations is reached, ensuring the feasibility and optimization performance of the solution. When the convergence condition is met, the final optimal unloading decision and resource allocation scheme are obtained.

[0081] Step S6: Based on the optimal offloading decision and resource allocation scheme, control the user terminal and satellite to perform task offloading and resource allocation.

[0082] The ultra-dense low-Earth orbit satellite network is divided into multiple satellite control domains, each managed by a master satellite. This step specifically includes the following sub-steps: Step S61: The master satellite generates flow table entries and control commands based on the calculated optimal offloading decision and resource allocation scheme. The flow table entries contain the mapping relationship between tasks and satellite-channel pairs, and the control commands contain the transmit power settings of each user terminal and the amount of computing resources allocated to each serving satellite.

[0083] Step S62: Flow table entries and control commands are sent to each serving satellite and user terminal within the domain via the inter-satellite control channel. The inter-satellite control channel employs a secure communication protocol to ensure reliable transmission of control information.

[0084] Step S63: The user terminal adjusts its transmit power according to the control command and transmits the mission data to the designated service satellite via the designated sub-channel. The user terminal transmits according to the allocated power value and establishes an uplink with the service satellite through the designated sub-channel.

[0085] Step S64: The service satellite allocates corresponding computing frequencies to the received tasks for processing according to the control commands. The service satellite provides computing services to each task according to the allocated computing resources, and returns the results to the user terminal after completing the calculation.

[0086] To verify the effectiveness of the method of this invention, a super-dense low-Earth orbit satellite network based on a multi-layer Walker-δ constellation was constructed for simulation experiments. This network consists of three independent constellation layers, with specific orbital parameters shown in Table 1. The entire network is logically divided into 100 software-defined network domains, each managed by a master satellite to simulate a distributed control architecture.

[0087] Table 1 Orbital parameters of ultra-dense low-Earth orbit satellite networks Each satellite has a computing power of 10 Gcycles / s and a computing energy consumption coefficient of 10^{-26}. The ultra-dense low-Earth orbit satellite network operates in the C-band (6 GHz) with a noise power spectral density of -203 dBm / Hz. Each satellite is equipped with 10 sub-channels, each with a bandwidth of 10 MHz. The data size for each computational offloading task is uniformly distributed between 1000 and 3000 Kb, and the required computation is uniformly distributed between 100 and 900 Mcycles. The maximum tolerable service latency for each task is between 0.5 and 1 second. User energy weighting coefficient. The values ​​are randomly selected from the set {0, 0.1, 1, 10, 100} to simulate the varying sensitivities of users to energy consumption. The maximum transmit power of the user terminal is uniformly distributed between 0.2W and 3.0W.

[0088] To evaluate the algorithm's performance under different network loads, six simulation scenarios were set up. The total number of offloading tasks was varied from 15,000 to 30,000 to adjust the average number of service tasks per satellite. The load is varied from 5 to 10 to simulate network service loads ranging from moderate to high intensity. To comprehensively evaluate the performance of the hierarchical iterative matching optimization and successive convex approximation algorithms proposed in this invention, they are compared with the following four representative benchmark algorithms: a random selection algorithm, where the user terminal randomly selects a feasible satellite and sub-channel for offloading, serving as a performance lower bound to measure the gain brought by intelligent association decision-making; a joint optimization algorithm based on stable matching, employing the classic Gale-Shapley stable matching algorithm to determine the association relationship between the user and the satellite channel, used to compare and verify the superiority of the cooperative switching matching optimization algorithm of this invention over the traditional stable matching algorithm; and an algorithm based on cooperative switching and decoupled iteration, employing a decoupled joint iterative algorithm in the continuous resource allocation phase, used to compare and verify the advantages of the joint optimization algorithm of this invention over the traditional decoupled iterative optimization method.

[0089] like Figure 3 As shown, the algorithm of this invention exhibits fast and stable convergence characteristics during the optimization process. Under different task scales, the algorithm can monotonically decrease the objective function and converge to a stable value within approximately 10 iterations. This fast convergence characteristic is crucial for ultra-dense low-Earth orbit satellite networks with rapidly changing topology and channel conditions. It ensures that optimization decisions can keep pace with changes in network status, thereby providing continuous and stable high-performance services.

[0090] As shown in Figures 4(a) to 4(d), box plot analysis shows that the solution obtained by the algorithm of the present invention has a tighter energy distribution and lower volatility, and the performance differences with each benchmark scheme pass the significance test, with p-values ​​less than 0.02. This statistically proves the high stability and strong robustness of the solution obtained by the algorithm of the present invention, which can reliably cope with the randomness in the network.

[0091] like Figure 5 As shown, the algorithm proposed in this invention achieves lower average weighted system energy than all benchmark schemes under different network loads. Compared to the best-performing benchmark scheme, the algorithm of this invention can reduce system energy consumption by an average of 10.24%. This is mainly attributed to: in the discrete association stage, the proposed cooperative exchange matching optimization algorithm can more effectively avoid co-channel interference through cooperative exchange operations between users, thus laying a better initial association state for subsequent resource optimization; in the continuous resource allocation stage, the proposed fractional programming and continuous convex approximation algorithm can jointly and collaboratively optimize user transmit power and satellite computing resources, avoiding suboptimal solutions that traditional alternating iterative methods may fall into, thereby achieving a globally better energy efficiency configuration.

[0092] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described; only preferred embodiments of the present invention are illustrated. The descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. As long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0093] It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of this invention. Therefore, the scope of protection of this invention should be determined by the appended claims.

Claims

1. A resource allocation method for ultra-dense low-Earth orbit satellite networks, characterized in that, Includes the following steps: Step S1: Construct a joint optimization model with the objective of minimizing the total weighted energy consumption of the system. The joint optimization model is constrained by co-channel interference and time delay constraints. Step S2: Generate an initial feasible solution for the joint optimization model under the constraints. Step S3: Under the premise of fixed continuous resource variables, update the computational offloading decision variables of the joint optimization model by iteratively optimizing the matching relationship between the task and the satellite-channel unit; Step S4: Under the premise of fixing the computational offloading decision variables, jointly optimize the user terminal transmit power and the satellite computational resource allocation; Step S5: Alternately execute steps S3 and S4 until the preset convergence condition is met to obtain the optimal unloading decision and resource allocation scheme; Step S6: Based on the optimal unloading decision and resource allocation scheme, control the user terminal and satellite to perform task unloading and resource allocation.

2. The resource allocation method for ultra-dense low-Earth orbit satellite networks according to claim 1, characterized in that: The total weighted energy consumption of the system includes the uplink transmission energy consumption of the user terminal and the computing energy consumption of the satellite. The objective function of the joint optimization model is expressed as follows: ； In the formula, , , These represent the unloading decision variable matrix, the transmission power variable vector, and the computing resource allocation variable matrix, respectively. , , These represent the task set, satellite set, and sub-channel set, respectively. Unload decision variables for binary when the task via sub-channel Unload to satellite The value is 1 if it is true, and 0 otherwise. For the task The uplink transmit power of the corresponding user terminal; Indicates when the task via sub-channel Unload to satellite At that time, satellite Assigned to task Computing resources; For the task The amount of input data; For the task Required number of CPU cycles; For the task The corresponding user's energy weighting coefficient; For satellite Energy consumption coefficient; For the task via sub-channel To satellite Uplink transmission rate during transmission.

3. The resource allocation method for ultra-dense low-Earth orbit satellite networks according to claim 2, characterized in that: The constraints of the joint optimization model include at least the following: Offload uniqueness constraint: Each task must be offloaded to one and only one subchannel of a satellite; Channel exclusivity constraint: Each sub-channel of each satellite can serve at most one task at any given time; Co-channel interference constraint: The uplink transmission rate is calculated according to Shannon's formula. The denominator of the signal-to-interference-plus-noise ratio (SINR) of the uplink transmission rate includes co-channel interference from other tasks on the same sub-channel. The expression for the uplink transmission rate is: ； In the formula, Sub-channel bandwidth; For satellite The noise power; For the task With satellite Sub-channel Channel gain between; Indicates from user terminal via channel Send to satellite Interference signals at user terminals and satellite Channel gain between; Latency constraint: The sum of the uplink transmission delay and computation delay of each task shall not exceed the maximum tolerable delay of the current task. ; Resource constraints: The total computing resources allocated to any satellite shall not exceed the maximum computing capacity of the current satellite. And the transmit power of any user terminal Within the permissible power range Inside.

4. The resource allocation method for ultra-dense low-Earth orbit satellite networks according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: Initialize the matching weights between the mission and the satellite-sub-channel element. : ； In the formula, For the task With satellite Sub-channel Channel gain between; It is a preset, sufficiently large constant; Step S22: Construct the matching problem between the mission and the satellite-sub-channel unit into a maximum weight bidirectional matching problem, and use the Hungarian algorithm to solve the maximum weight bidirectional matching problem to obtain the initial unloading decision variables; Step S23: Initialize the transmit power for each task. If the transmission of a task will not cause co-channel interference to other tasks, set it to the maximum transmit power; otherwise, set it to the minimum transmit power. Step S24: Perform inner iterations to check and eliminate all task latency violations and satellite computing resource violations until an initial feasible solution that satisfies all constraints is obtained.

5. A resource allocation method for ultra-dense low-Earth orbit satellite networks according to claim 4, characterized in that: Step S24 includes the following steps: Step S241: Calculate the maximum feasible computation delay for each task under the current resource allocation. : ； In the formula, For the task Maximum tolerable delay; This represents the transmission delay at the current transmission power. The computation latency under the current computing resource allocation; Step S242: If a task exists This was determined to be a time delay violation, and the following was selected. The minimum mission requires increasing transmission power; Step S243: If all tasks have no latency violations, calculate the minimum feasible computational resources for each task. : ； In the formula, To complete the task Required number of CPU cycles; If the total allocation of a satellite exceeds its maximum computing capacity, select the current satellite. The largest mission requires increased transmission power; Step S244: Repeat steps S241 to S243 until there are no violations or the preset number of iterations is reached; if no feasible solution is found after the preset number of iterations, reduce the corresponding matching weight. Then return to step S22 to regenerate the uninstallation decision.

6. A resource allocation method for ultra-dense low-Earth orbit satellite networks according to claim 1, characterized in that: Step S3, which iteratively optimizes the matching relationship between the task and the satellite-channel element, includes the following steps: Step S31: Model the association problem between the task set and the satellite-sub-channel element set as a two-sided matching model; Step S32: Define the mission's preference for satellite-subchannel elements based on channel quality and current interference level; Step S33: Determine the set of interchangeable matching pairs. Only when two tasks are served by the same satellite or the coverage areas of the satellites serving the two tasks overlap, determine that the two tasks and their corresponding satellite-sub-channel elements constitute an interchangeable matching pair. Step S34: Traverse the set of interchangeable matching pairs according to preference degree, perform feasibility verification and optimization on candidate exchanges, and update the matching relationship if the total weighted energy consumption of the system is reduced and all constraints are satisfied after the exchange.

7. A resource allocation method for ultra-dense low-Earth orbit satellite networks according to claim 6, characterized in that: Step S34 involves feasibility verification and optimization of candidate swaps, including the following steps: When evaluating candidate swaps, calculate the constraint satisfaction of the two direct tasks involved in the swap, as well as the constraint satisfaction of other tasks affected by the change in the interference environment caused by the swap; if any related task is found to have a delay violation or a computational resource violation, increase the transmit power of the violating direct task, or decrease the transmit power of the interference source task that caused the violation; if all related tasks meet the constraints after power coordination adjustment, and the total weighted energy consumption of the system is reduced after continuous resource optimization, then confirm the execution of the current swap.

8. A resource allocation method for ultra-dense low-Earth orbit satellite networks according to claim 2, characterized in that: The joint optimization of user terminal transmit power and satellite computing resource allocation in step S4 includes the following steps: Step S41: Based on the current unloading decision variables, construct a continuous resource allocation subproblem; Step S42: Introduce auxiliary variables The quadratic transformation technique is used to transform the fractional terms in the objective function of the joint optimization model. Convert to polynomial form; Step S43: With other variables fixed, solve using closed-form methods. Update auxiliary variables The expression for the closed-form solution is: ； Step S44: Transform the original non-convex continuous resource allocation problem into a series of convex optimization subproblems and solve them iteratively.

9. A resource allocation method for ultra-dense low-Earth orbit satellite networks according to claim 8, characterized in that: Step S44 includes the following steps: Step S441: Introduce slack variables and Handling non-convex signal-to-interference-to-noise ratio constraints, where This is the lower bound of the signal-to-interference-plus-noise ratio. satisfy , For the task With satellite Sub-channel Channel gain between; Step S442: Using the first-order Taylor expansion technique, perform a first-order Taylor expansion on the concave function terms in the objective function. Convex approximation is performed on the non-convex terms in the constraints, in the th... Constructing a convex approximation subproblem in the next iteration: ； In the formula, , These are the unloading decision variable matrix and the auxiliary decision variables, respectively; To unload decision variables, namely the matching relationship between the mission and the satellite-channel; Step S443: Solve the convex approximation subproblem using a standard convex optimization solver to obtain the optimal transmit power and computational resource allocation for the current iteration.

10. A resource allocation method for ultra-dense low-Earth orbit satellite networks according to claim 1, characterized in that: The ultra-dense low-Earth orbit satellite network is divided into multiple satellite control domains, each managed by a master satellite; step S6 specifically includes the following steps: Step S61: The main satellite generates flow table entries and control commands based on the calculated optimal offloading decision and resource allocation scheme; Step S62: Send the flow table entries and control commands to each serving satellite and user terminal in the domain via the inter-satellite control channel; Step S63: The user terminal adjusts the transmission power according to the control command and sends the mission data to the designated service satellite via the designated sub-channel; Step S64: The service satellite allocates a corresponding computing frequency to the received task for processing according to the control command.

Images