A method for power control of a multi-beam antenna

Abstract

This invention discloses a multi-beam antenna power control method, belonging to the field of antenna power control technology. This invention utilizes the ARIMA model to predict user demand over time, decomposing the user demand sequence into autoregressive and moving average components, and combining them with long-term averages to recursively predict future demand trends across multiple time slots, improving sensitivity to demand fluctuations and solving the problem of insufficient response to instantaneous demand changes in traditional methods. Secondly, based on the prediction results, a joint model for power allocation and latency optimization is constructed, incorporating service quality, power consumption constraints, and user fairness into a unified framework. Simultaneously, dynamic weight adjustment achieves a balance between latency and power consumption, improving the rationality of resource allocation and avoiding the limitations of single-beam optimization. Furthermore, the optimization model is iteratively solved using the gradient descent method to dynamically generate resource allocation strategies, prioritizing resource allocation to high-demand areas.

Description

Technical Field

[0001] This invention relates to the field of antenna power control technology, and in particular to a method for controlling the power of a multi-beam antenna. Background Technology

[0002] In low-Earth orbit satellite communication systems, multi-beam antenna technology can provide high-quality coverage in multiple directions simultaneously, and has become an important means to meet global communication needs.

[0003] In practical applications, user demands are often dynamic, with communication loads in some areas surging while resources in others remain idle. This uneven distribution is one of the factors affecting the effectiveness of beam resource scheduling and power allocation. Traditional power control methods rely on preset power allocation strategies and lack the ability to respond quickly to instantaneous changes in demand. In highly dynamic environments, it is difficult to balance system efficiency and user experience. At the same time, low-Earth orbit satellites have shorter channel availability windows due to their lower orbital altitude. This means that when user traffic surges, if power resources cannot be allocated in time, it will not only lead to a decline in service quality for users in hotspot areas but also affect the energy efficiency of the entire system due to excessive power consumption. Traditional power control methods mainly focus on individual beams and lack a global perspective on power resource adjustment, making it unable to effectively address resource bottlenecks in highly dynamic communication scenarios.

[0004] To alleviate these problems, some traditional solutions optimize power allocation through resource reallocation strategies, such as prioritizing the power of high-demand beams. However, this approach lacks control over the balance between latency and energy efficiency, resulting in limited performance improvement. As demand continues to grow and become more complex, exploring power control methods that can dynamically adapt to changes in user needs while balancing latency and energy consumption is the core issue in the current optimization of multi-beam antenna technology. Summary of the Invention

[0005] In view of the aforementioned existing problems, the present invention is proposed.

[0006] This invention provides a multi-beam antenna power control method to address the problems in low-Earth orbit satellite communication where dynamic changes in user demand lead to uneven load distribution, traditional power allocation strategies lack rapid response capabilities, and it is difficult to balance latency and energy efficiency; at the same time, the communication window is short and resource bottlenecks are significant, and a single beam optimization method cannot meet the global dynamic adjustment requirements.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0008] This invention provides a method for controlling the power of a multi-beam antenna, comprising:

[0009] Step S1: The satellite collects user information, including location information, channel status information, and queue length data;

[0010] Step S2: Based on user information, the user demand in multiple future scheduling slots is predicted using an autoregressive moving average (ARIMA) model to obtain the prediction results. The prediction results include the total demand within the beam coverage area and the specific resource requirements of different users.

[0011] Step S3: Based on the prediction results, construct a joint optimization model for power allocation and latency optimization. The joint optimization model combines service quality requirements, power consumption constraints and user fairness constraints to solve the optimization model and generate a resource allocation strategy.

[0012] Step S4: Adjust beam direction and power allocation according to resource allocation strategy;

[0013] After the adjustments in steps S4 and S5 are completed, monitor network performance metrics, including latency, throughput, and power consumption. Combine these performance metrics to evaluate the effectiveness of the resource allocation strategy in step S3. Use the evaluation results as feedback to input into the ARIMA model and the optimization model in step S3 to correct the resource allocation strategy.

[0014] As a preferred embodiment of the multi-beam antenna power control method described in this invention, the step of using an autoregressive moving average (ARIMA) model to predict user demand within multiple future scheduling time slots is as follows:

[0015] Define the user requirement sequence as : ,in, For the current demand sequence, For the past The demand for each time slot This is the current timeslot number. The number of historical time slots used;

[0016] Autoregressive modeling is performed, and the formula for calculating the autoregressive part is as follows:

[0017] ,

[0018] in, These are the predicted values ​​from the autoregressive component. Let the order be the autoregressive order. These are the autoregressive coefficients. For a moment The actual demand value, For the current autoregressive order, from arrive ,

[0019] The moving average component is modeled, and the formula for calculating the moving average component is as follows:

[0020] ,

[0021] in, This is the predicted value of the moving average portion. The moving average order is... The moving average coefficient is... For a moment The residual, For the current moving average order, from arrive .

[0022] As a preferred embodiment of the multi-beam antenna power control method described in this invention, the step of performing timing prediction further includes...

[0023] A comprehensive forecast is conducted, and the comprehensive forecast formula is as follows:

[0024] ,

[0025] in, For a moment Forecast demand, This is the long-run mean of the demand series.

[0026] For the future Recursive prediction is performed for each time slot, and the prediction formula is:

[0027] ,

[0028] in, For the future Predicted demand sequence for each time slot, For the future The predicted value for each time slot, For the predicted number of time slots, The currently predicted time slot number, from arrive ;

[0029] The formula for calculating the total demand within the beam coverage area is as follows:

[0030] ,

[0031] in, For total demand, The total number of users covered by the beam. For the first Predicted demand for individual users.

[0032] As a preferred embodiment of the multi-beam antenna power control method described in this invention, the step of constructing a joint optimization model for power allocation and time delay optimization is as follows:

[0033] Construct the objective function, which is:

[0034] ,

[0035] in, The objective function value, For dynamic weight parameters, For the first Power allocation for individual users For the first The needs of individual users For the first latency per user The total number of users covered by the beam. For the current user ID, from arrive ,

[0036] Define constraints that the optimization model must satisfy:

[0037] ,

[0038] in, Due to power limitations, To limit the maximum latency, As an indicator of power allocation fairness, This is the fairness threshold.

[0039] As a preferred embodiment of the multi-beam antenna power control method described in this invention, the step of solving the optimization model to generate a resource allocation strategy is as follows:

[0040] Iterative optimization is performed using the gradient descent method, and the calculation formula is as follows:

[0041] ,

[0042] in, For power allocation in the current and next iterations, For the current iteration round, For learning rate,

[0043] The gradient of the objective function with respect to power allocation;

[0044] Solve for the output resource allocation strategy. The optimal power allocation strategy is:

[0045] ,

[0046] in, For the optimal power allocation set, For the first Optimal power allocation for each user.

[0047] As a preferred embodiment of the multi-beam antenna power control method described in this invention, the joint optimization model dynamically adjusts the weights, and the weight adjustment formula is:

[0048] ,

[0049] in,

[0050] This is the current average latency. For target latency, This is for adjusting the coefficient.

[0051] In a preferred embodiment of the multi-beam antenna power control method of the present invention, in step S3, a weighting parameter is introduced to dynamically adjust the balance between power efficiency and time delay.

[0052] As a preferred embodiment of the multi-beam antenna power control method of the present invention, the step of evaluating the effectiveness of the resource allocation strategy in step S3 by combining performance indicators, and using the evaluation result as feedback to input into the ARIMA model and the optimization model of step S3 to correct the resource allocation strategy is as follows:

[0053] Record performance metrics; the set of network performance metrics is as follows:

[0054] ,

[0055] in, For the observed time delay, For the observed throughput, This represents the observed power consumption.

[0056] As a preferred embodiment of the multi-beam antenna power control method of the present invention, the step of evaluating the effectiveness of the resource allocation strategy in step S3 by combining performance indicators, and using the evaluation results as feedback to input into the ARIMA model and the optimization model of step S3 to correct the resource allocation strategy further includes...

[0057] The evaluation effect is calculated using the following formula:

[0058] ,

[0059] in, For evaluation purposes, For dynamic weight parameters, For network performance metrics,

[0060] Calculate the feedback error, which is:

[0061] ,

[0062] in,

[0063] For feedback error, The target evaluation value, This is the current evaluation value.

[0064] As a preferred embodiment of the multi-beam antenna power control method of the present invention, in step S5, the final prediction based on the error correction requirement is as follows:

[0065] ,

[0066] in, For the revised demand forecast, For the original forecast demand, For correction factor,

[0067] The model has been updated and optimized; the revised optimization update formula is as follows:

[0068] ,

[0069] in, To provide feedback on the updated learning rate,

[0070] This is the feedback error.

[0071] The beneficial effects of this invention are as follows: First, this invention utilizes the ARIMA model to predict user demand over time, decomposing the user demand sequence into autoregressive and moving average components. Combined with long-term averages, it recursively predicts demand trends for multiple time slots in the future, improving sensitivity to demand fluctuations and solving the problem of insufficient response to instantaneous demand changes in traditional methods. Second, based on the prediction results, a joint model for power allocation and latency optimization is constructed, incorporating service quality, power consumption constraints, and user fairness into a unified framework. Simultaneously, dynamic weight adjustment achieves a balance between latency and power consumption, improving the rationality of resource allocation and avoiding the limitations of single-beam optimization. Furthermore, the optimization model is iteratively solved using gradient descent, dynamically generating resource allocation strategies. Resources are prioritized for high-demand areas, and dynamic weight adjustment balances different performance objectives, taking into account both service quality in hotspot areas and overall system energy efficiency. Simultaneously, a real-time monitoring and feedback mechanism is introduced, using network performance indicators as feedback information to dynamically correct demand prediction and optimization model parameters, thereby improving resource utilization, service quality, and overall system efficiency. Attached Figure Description

[0072] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. 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.

[0073] Figure 1 This is a flowchart of the multi-beam antenna power control method of the present invention. Detailed Implementation

[0074] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0075] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0076] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0077] Example 1, referring to Figure 1 This embodiment provides a multi-beam antenna power control method, including the following steps:

[0078] Step S1: The satellite collects user information, including location information, channel status information, and queue length data;

[0079] Step S2: Based on user information, the user demand in multiple future scheduling slots is predicted using an autoregressive moving average (ARIMA) model to obtain the prediction results. The prediction results include the total demand within the beam coverage area and the specific resource requirements of different users.

[0080] The steps for time-series forecasting of user demand over multiple future scheduling slots using an autoregressive moving average (ARIMA) model are as follows:

[0081] Define the user requirement sequence as : ,in, For the current demand sequence, For the past The demand for each time slot This is the current timeslot number. The number of historical time slots used;

[0082] Autoregressive modeling is performed, and the formula for calculating the autoregressive part is as follows:

[0083] ,

[0084] in, These are the predicted values ​​from the autoregressive component. Let the order be the autoregressive order. These are the autoregressive coefficients. For a moment The actual demand value, For the current autoregressive order, from arrive ,

[0085] The moving average component is modeled, and the formula for calculating the moving average component is as follows:

[0086] ,

[0087] in, This is the predicted value of the moving average portion. The moving average order is... The moving average coefficient is... For a moment The residual, For the current moving average order, from arrive ;

[0088] The steps for performing time series forecasting also include,

[0089] A comprehensive forecast is conducted, and the comprehensive forecast formula is as follows:

[0090] ,

[0091] in, For a moment Forecast demand, This is the long-run mean of the demand series.

[0092] For the future Recursive prediction is performed for each time slot, and the prediction formula is:

[0093] ,

[0094] in, For the future Predicted demand sequence for each time slot, For the future The predicted value for each time slot, For the predicted number of time slots, The currently predicted time slot number, from arrive ;

[0095] The formula for calculating the total demand within the beam coverage area is as follows:

[0096] ,

[0097] in, For total demand, The total number of users covered by the beam. For the first Predicted demand for individual users;

[0098] Specifically, the ARIMA model constructs a user demand forecasting model by combining autoregressive and moving average components with long-term mean, and obtains demand trends over multiple time slots in the future through recursive forecasting.

[0099] Step S3: Based on the prediction results, construct a joint optimization model for power allocation and latency optimization. The joint optimization model combines service quality requirements, power consumption constraints and user fairness constraints to solve the optimization model and generate a resource allocation strategy.

[0100] The steps to construct a joint optimization model for power allocation and time delay optimization are as follows:

[0101] Construct the objective function, which is:

[0102] ,

[0103] in, The objective function value, For dynamic weight parameters, For the first Power allocation for individual users For the first The needs of individual users For the first latency per user The total number of users covered by the beam. For the current user ID, from arrive ,

[0104] Define constraints that the optimization model must satisfy:

[0105] ,

[0106] in, Due to power limitations, To limit the maximum latency, As an indicator of power allocation fairness, This is the fairness threshold;

[0107] The steps to solve the optimization model to generate the resource allocation strategy are as follows:

[0108] Iterative optimization is performed using the gradient descent method, and the calculation formula is as follows:

[0109] ,

[0110] in, For power allocation in the current and next iterations, For the current iteration round, For learning rate,

[0111] The gradient of the objective function with respect to power allocation;

[0112] Solve for the output resource allocation strategy. The optimal power allocation strategy is:

[0113] ,

[0114] in, For the optimal power allocation set, For the first Optimal power allocation for each user;

[0115] In the joint optimization model, the weights are dynamically adjusted, and the weight adjustment formula is:

[0116] ,

[0117] in,

[0118] This is the current average latency. For target latency, To adjust the coefficient,

[0119] Specifically, by using dynamic weights and gradient optimization, a balance is struck between power consumption and latency, and combined with constraints, a resource allocation strategy that meets the quality of service requirements is generated.

[0120] In step S3, a weighting parameter is introduced to dynamically adjust the balance between power efficiency and time delay;

[0121] Step S4: Adjust beam direction and power allocation according to resource allocation strategy;

[0122] After the adjustments in steps S5 and S4 are completed, monitor network performance indicators, including latency, throughput and power consumption. Combine the performance indicators to evaluate the effectiveness of the resource allocation strategy in step S3. Use the evaluation results as feedback to input into the ARIMA model and the optimization model in step S3 to correct the resource allocation strategy.

[0123] The effectiveness of the resource allocation strategy in step S3 is evaluated using performance metrics. The evaluation results are then used as feedback and input into the ARIMA model and the optimization model in step S3. The steps to correct the resource allocation strategy are as follows:

[0124] Record performance metrics; the set of network performance metrics is as follows:

[0125] ,

[0126] in, For the observed time delay, For the observed throughput, The observed power consumption,

[0127] The effectiveness of the resource allocation strategy in step S3 is evaluated using performance metrics. The evaluation results are then used as feedback and input into the ARIMA model and the optimization model in step S3. The steps to correct the resource allocation strategy also include...

[0128] The evaluation effect is calculated using the following formula:

[0129] ,

[0130] in, For evaluation purposes, For dynamic weight parameters, For network performance metrics,

[0131] Calculate the feedback error, which is:

[0132] ,

[0133] in,

[0134] For feedback error, The target evaluation value, This is the current assessment value.

[0135] In step S5, the final prediction is based on the error correction requirements, and the correction formula is as follows:

[0136] ,

[0137] in, For the revised demand forecast, For the original forecast demand, For correction factor,

[0138] The model has been updated and optimized; the revised optimization update formula is as follows:

[0139] ,

[0140] in, To provide feedback on the updated learning rate,

[0141] For feedback error,

[0142] Specifically, based on real-time monitored performance metrics, the demand forecasting and resource allocation strategies are dynamically adjusted in conjunction with a feedback mechanism to enhance the adaptability and robustness of the resource allocation strategies; the learning rate differentiation effectively ensures the independence of model optimization and feedback updates, avoiding mutual conflicts.

[0143] In summary, this invention utilizes the ARIMA model for time-series prediction of user demand, decomposing the user demand sequence into autoregressive and moving average components. By combining long-term averages and recursively predicting future demand trends across multiple time slots, it enhances sensitivity to demand fluctuations and addresses the insufficient response to instantaneous demand changes in traditional methods. Secondly, based on the prediction results, a joint model for power allocation and latency optimization is constructed, incorporating service quality, power consumption constraints, and user fairness into a unified framework. Simultaneously, dynamic weight adjustment achieves a balance between latency and power consumption, improving the rationality of resource allocation and avoiding the limitations of single-beam optimization. Furthermore, the optimization model is iteratively solved using gradient descent to dynamically generate resource allocation strategies, prioritizing resource allocation to high-demand areas. Dynamic weight adjustment balances different performance objectives, considering both service quality in hotspot areas and overall system energy efficiency. Real-time monitoring and feedback mechanisms are introduced, using network performance indicators as feedback information to dynamically correct demand prediction and optimization model parameters. This improves resource utilization, service quality, and overall system efficiency.

[0144] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method of power control for a multi-beam antenna, characterized by: Comprising, Step S1, the satellite collects user information, including location information, channel state information and queue length data; Step S2, based on user information, through autoregressive moving average ARIMA model, the user demand in future multiple scheduling time slots is time series predicted, the prediction result is obtained, the prediction result contains the total demand in the beam coverage area and the specific demand of different users for resources; Step S3, based on the prediction result, a joint optimization model of power allocation and delay optimization is constructed, the joint optimization model combines the quality of service demand, power consumption limit and user fairness constraint condition, the optimization model is solved to generate resource allocation strategy; Step S4, according to the resource allocation strategy, the beam direction and power allocation are adjusted; Step S5, after the adjustment of step S4 is completed, the network performance index is monitored, including delay, throughput and power consumption, the effect of resource allocation strategy of step S3 is evaluated combined with performance index, the evaluation result is taken as feedback, input to ARIMA model and optimization model of step S3, and the resource allocation strategy is corrected; The step of predicting the user demand in future multiple scheduling time slots through autoregressive moving average ARIMA model is: The user demand sequence is defined as : wherein, is the demand sequence of the current time slot, is the demand quantity of the past time slots, is the current time slot number, is the number of historical time slots used; The autoregressive part is modeled, and the autoregressive part calculation formula is: ， wherein, is an autoregressive partial prediction value, is an autoregressive order, is an autoregressive coefficient, is a time slot of an actual demand value, is a current autoregressive order from to , The moving average part is modeled, and the moving average part calculation formula is: ， wherein, is a moving average partial prediction value, is a moving average order, is a moving average coefficient, is a time slot of a residual, is a current moving average order from to ; Comprehensive prediction is carried out, and the comprehensive prediction formula is: ， wherein is the predicted demand for a time slot of the prediction horizon, is the long-term mean of the demand sequence, For future Recursive prediction is performed for future slots, with the prediction formula being: ， wherein, is the predicted demand sequence for the future slots, is the predicted value for the future slot, is the number of predicted slots, is the current predicted slot number, from to ; The total demand is calculated, and the calculation formula of the total demand in the beam coverage area is: ， wherein, is the total demand, is the total number of users covered by the beam, is the predicted demand of the th user; The step of solving the optimization model to generate resource allocation strategy is: Iterative optimization is carried out, and the gradient descent method is used for iterative solution, and the calculation formula is: ， wherein, is the power allocation for the current and next iteration, is the current iteration round, is the learning rate, a gradient of the objective function with respect to the power allocation; The output resource allocation strategy is solved, and the optimal power allocation strategy is: ， wherein, is the optimal power allocation set, is the optimal power allocation for the th user; The step of combining the performance index to evaluate the effect of resource allocation strategy of step S3, taking the evaluation result as feedback, inputting to ARIMA model and optimization model of step S3, and correcting the resource allocation strategy is: The performance index is recorded, and the network performance index set is: ， wherein, is the observed latency, is the observed throughput, is the observed power consumption; The evaluation effect is calculated, and the evaluation formula is: ， wherein, is an evaluation value, is a dynamic weight parameter, is a network performance indicator, The feedback error is calculated, and the feedback error is: ， Wherein, is a feedback error, is a target evaluation value, is a current evaluation value; In step S5, the demand prediction is finally corrected according to the error, and the correction formula is: ， wherein, is the revised demand forecast, is the original demand forecast, is the revision coefficient, The optimization model is updated, and the updated optimization update formula is: ， wherein, is the learning rate for the feedback update, is the feedback error.

2. A method for controlling the power of a multi-beam antenna as claimed in claim 1, characterized in that: The step of constructing the joint optimization model of power allocation and delay optimization is: The objective function is constructed, and the objective function is: ， wherein, is the objective function value, is the dynamic weight parameter, is the power allocation for the th user, is the demand for the th user, is the latency for the th user, is the total number of users covered by the beam, is the current user number, from to , The constraint condition is defined, and the optimization model needs to meet: ， wherein, is a power limit, is a maximum latency limit, is a power allocation fairness index, is a fairness threshold.

3. A method of controlling the power of a multi-beam antenna as claimed in claim 2, characterized in that: In the joint optimization model, the weight is dynamically adjusted, and the weight adjustment formula is: ， Wherein, is a current average delay, is a target delay, is an adjustment coefficient.

4. A method of power control for a multiple beam antenna as claimed in claim 3, characterized in that: In step S3, the weight parameter is introduced to dynamically adjust the balance between power efficiency and delay.

Images