Optimization method and device for predicting resource allocation, equipment and storage medium

Abstract

The invention discloses an optimization method and device for predicting resource allocation, equipment and a storage medium. The method comprises the following steps: a network device obtains first moving track information of a first terminal; and processing the first moving track information by using a first model to obtain a first resource allocation strategy of the first terminal in the future time.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of communication technology, in particular to an optimization method and device for predicting resource allocation, equipment and storage medium. BACKGROUND

[0002] In order to further improve the capacity of 5G and future 6G networks and reduce system energy consumption, some research has begun to use predicted user future movement trajectory or large-scale information for predictive resource allocation. The current scheme uses a machine learning model to predict the user's future movement trajectory or large-scale channel information according to the user's historical trajectory, and then optimizes the user's data rate or power allocation strategy in the future according to the predicted movement trajectory or large-scale channel information. For details, refer to Figure 1 The base station can transmit more data when the user's future channel condition is good, and transmit less data when the future channel condition is poor, which can improve network throughput or reduce system energy consumption.

[0003] However, this approach has the following problems: the information prediction process and the resource allocation process are independent of each other, the former usually obtains the prediction result by minimizing the mean square error (MSE, Mean Square Error), and the latter optimizes the allocation strategy by maximizing the data rate or minimizing the energy consumption. Because the goals of the two processes are inconsistent, the "predict first and then optimize" method performs poorly in reducing energy consumption. In addition, because the scheme includes information prediction and resource optimization processes, and the resource optimization stage often uses a numerical algorithm with high complexity, the online decision-making complexity of the scheme is high. Whenever the environment changes, information prediction and resource optimization need to be performed again, which has high computation and storage costs. SUMMARY

[0004] To solve the existing technical problems, the embodiments of the present application provide an optimization method, device, equipment and storage medium for predicting resource allocation.

[0005] To achieve the above-mentioned purpose, the technical scheme of the embodiments of the present application is as follows:

[0006] The embodiments of the present application provide an optimization method for predicting resource allocation, which is applied to a network device, and the method comprises:

[0007] Obtain the first movement trajectory information of the first terminal;

[0008] Process the first movement trajectory information using a first model to obtain a first resource allocation strategy of the first terminal at a future time.

[0009] In the above scheme, the method further comprises: obtaining the playing time of the video requested by the first terminal in the future;

[0010] Correspondingly, the future time corresponding to the resource allocation strategy at least includes the playing time.

[0011] In the above solution, the method further includes: executing the first resource allocation strategy according to time, and transmitting data of the first terminal based on a first performance corresponding to the first resource allocation strategy.

[0012] In the above solution, the method further includes: obtaining a first performance parameter during execution of the first resource allocation strategy.

[0013] When the performance indicated by the first performance parameter does not meet a set performance requirement, re-training the first model.

[0014] In the above solution, the first performance parameter includes a data rate and / or energy consumption; the method further includes: comparing the data rate with a first threshold value, and determining that the performance indicated by the first performance parameter does not meet the set performance requirement when the data rate is less than the first threshold value; and / or,

[0015] Comparing the energy consumption with a second threshold value, and determining that the performance indicated by the first performance parameter does not meet the set performance requirement when the energy consumption is greater than or equal to the second threshold value.

[0016] In the above solution, the re-training of the first model includes: obtaining second mobile trajectory information of a second terminal within a first time range, processing the second mobile trajectory information by using the first model, and obtaining a second resource allocation strategy of the second terminal within a second time range;

[0017] Obtaining third mobile trajectory information of the second terminal within the second time range, and determining large-scale channel information based on the third mobile trajectory information;

[0018] Calculating a second performance parameter based on the large-scale channel information and the second resource allocation strategy according to a set performance function, and training the first model according to the second performance parameter.

[0019] In the above solution, the training of the first model according to the second performance parameter includes:

[0020] When the second performance parameter does not meet a set threshold value, determining a loss function based on the performance function; wherein the loss function is a gradient of the performance function with respect to the second resource allocation strategy;

[0021] Adjusting model parameters of the first model based on the loss function.

[0022] The embodiment of the present application also provides an optimization device for predicting resource allocation, which is applied to a network device and comprises an information collecting unit and a decision unit.

[0023] The information collecting unit is used for obtaining first mobile trajectory information of a first terminal.

[0024] The decision unit is used for processing the first mobile trajectory information by using a pre-trained first model to obtain a first resource allocation strategy of the first terminal in future time.

[0025] In the above scheme, the information collecting unit is also used for obtaining a playing time of a video requested by the first terminal in future time.

[0026] Correspondingly, the future time corresponding to the resource allocation strategy at least comprises the playing time.

[0027] In the above scheme, the device further comprises an execution unit, which is used for executing the first resource allocation strategy according to time and transmitting data of the first terminal based on a first performance corresponding to the first resource allocation strategy.

[0028] In the above scheme, the device further comprises a monitoring unit.

[0029] The information collecting unit is also used for obtaining a first performance parameter in the process of executing the first resource allocation strategy by the execution unit and transmitting the first performance parameter to the monitoring unit.

[0030] The monitoring unit is used for transmitting a first instruction to the decision unit when a performance represented by the first performance parameter does not meet a set performance requirement, and the first instruction is used for instructing to retrain the first model.

[0031] In the above scheme, the decision unit is also used for transmitting a first time range and a second time range to the information collecting unit.

[0032] The information collecting unit is also used for obtaining second mobile trajectory information of a second terminal in the first time range and third mobile trajectory information of the second terminal in the second time range, transmitting the second mobile trajectory information to the decision unit and transmitting the third mobile trajectory information to the monitoring unit.

[0033] The decision unit is also used for processing the second mobile trajectory information by using the first model to obtain a second resource allocation strategy of the second terminal in the second time range and transmitting the second resource allocation strategy to the monitoring unit.

[0034] The monitoring unit is further configured to determine large-scale channel information based on the third mobile trajectory information, calculate a second performance parameter according to a set performance function based on the large-scale channel information and the second resource allocation strategy, and train the first model according to the second performance parameter.

[0035] In the above scheme, the monitoring unit is configured to determine a loss function based on the performance function when the second performance parameter does not satisfy a set threshold value, and send the loss function to the decision unit; wherein the loss function is a gradient of the performance function with respect to the second resource allocation strategy.

[0036] The decision unit is configured to adjust model parameters of the first model based on the loss function.

[0037] The embodiment of the present application further provides a computer readable storage medium, which has a computer program stored thereon, and the program is executed by a processor to realize the steps of the optimization method for predicting resource allocation.

[0038] The embodiment of the present application further provides a network device, which comprises a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor realizes the steps of the optimization method for predicting resource allocation when executing the program.

[0039] The embodiment of the present application further provides a computer program product, which comprises computer program instructions, and the computer program instructions enable a computer to execute the steps of the optimization method for predicting resource allocation.

[0040] The optimization method, device, equipment and storage medium for predicting resource allocation provided by the embodiment of the present application are as follows: a network device obtains first mobile trajectory information of a first terminal; a first model is used to process the first mobile trajectory information, and a first resource allocation strategy of the first terminal at a future time is obtained. In this way, the end-to-end directly obtains the future resource allocation strategy from the historical trajectory information of the user, saves the information prediction and large-scale channel information conversion process in the prior art, simplifies the process of predicting resource allocation, further improves the system performance, and reduces the complexity of online decision. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 It is a schematic diagram of the prior art scheme of predicting mobile trajectory and then optimizing allocation strategy;

[0042] Figure 2 It is a schematic diagram of the optimization scheme for predicting resource allocation of the embodiment of the present application;

[0043] Figure 3A flowchart of the optimization method for the predicted resource allocation of the embodiment of the present application;

[0044] Figure 4 A schematic diagram of the observation window and the prediction window in the optimization method for the predicted resource allocation of the embodiment of the present application;

[0045] Figure 5 A schematic diagram of the composition structure of the optimization device for the predicted resource allocation of the embodiment of the present application;

[0046] Figure 6 An interaction flowchart of the optimization method for the predicted resource allocation of the embodiment of the present application Figure 1 ;

[0047] Figure 7 An interaction flowchart of the optimization method for the predicted resource allocation of the embodiment of the present application Figure 2 ;

[0048] Figure 8 A schematic diagram of the hardware composition structure of the network device of the embodiment of the present application. DETAILED DESCRIPTION

[0049] The present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0050] The technical solution of the embodiment of the present application can be applied to various communication systems, such as: Global System of Mobile communication (GSM) system, Long Term Evolution (LTE) system or 5G system, etc. Optionally, the 5G system or 5G network can also be referred to as New Radio (NR) system or NR network.

[0051] For example, the communication system to which the embodiment of the present application is applied can include a network device and a terminal device (also referred to as a terminal, a communication terminal, etc.); the network device can be a device that communicates with the terminal device. Among them, the network device can provide communication coverage in a certain area range, and can communicate with terminals located in the area. Optionally, the network device can be a base station in each communication system, such as an Evolutional Node B (eNB) in the LTE system, and for example, a base station (gNB) in the 5G system or the NR system.

[0052] It should be understood that the device with communication function in the network / system in the embodiments of the present application can be referred to as a communication device. The communication device can include network devices and terminals with communication functions, and the network devices and terminal devices can be the specific devices described above, which will not be described here again; the communication device can also include other devices in the communication system, such as network controllers, mobile management entities and other network entities, which are not limited in the embodiments of the present application.

[0053] It should be understood that the terms "system" and "network" are often used interchangeably in this paper. The term "and / or" in this paper is only a description of the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B, which can represent the three cases of A alone, A and B together, and B alone. In addition, the character " / " in this paper generally represents an "or" relationship between the front and rear associated objects.

[0054] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not necessarily limit to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0055] Figure 2 The flowchart of the optimization scheme of the prediction resource allocation of the embodiments of the present application is shown in Figure 1. Figure 2 As shown in Figure 1, the active optimization module simplifies the user's prediction resource allocation, and the future resource allocation strategy is obtained directly from the user's historical trajectory information through the end-to-end active optimization module, which saves the information prediction and large-scale channel information conversion process in the prior art scheme, and simplifies the process of prediction resource allocation.

[0056] The embodiments of the present application provide an optimization method for prediction resource allocation. Figure 3 The flowchart of the optimization method for prediction resource allocation of the embodiments of the present application is shown in Figure 1. Figure 3 As shown in Figure 1, the method comprises:

[0057] Step 101: obtaining first mobile trajectory information of a first terminal;

[0058] Step 102: processing the first mobile trajectory information by using a first model to obtain a first resource allocation strategy of the first terminal at a future time. Step 102: processing the first mobile trajectory information by using a first model to obtain a first resource allocation strategy of the first terminal at a future time.

[0059] The optimization method of predicting resource allocation in the embodiment is applied in a network device, which can be an access network or an access network device, such as a base station. In some optional embodiments, Figure 2 The active optimization module shown in FIG. 1 can be implemented by a logical function in the base station.

[0060] In the embodiment, for predicting the resource allocation strategy of the first terminal, the network device first obtains the first mobile trajectory information of the first terminal. In some optional embodiments, the first terminal can determine the mobile trajectory information according to its own position information in a time range, and then report the mobile trajectory information to the network device. In other optional embodiments, the first terminal can also obtain the position information reported by the first terminal in a time range, and then determine the mobile trajectory information according to the reported position information.

[0061] It should be noted that, in order to distinguish from the mobile trajectory information of the terminal used in the subsequent retraining process of the first model, the terminal involved in the resource allocation strategy inference process is referred to as the first terminal, and the mobile trajectory information involved is referred to as the first mobile trajectory information; the terminal involved in the subsequent retraining process of the first model is referred to as the second terminal, and the mobile trajectory information involved is referred to as the second mobile trajectory information, the third mobile trajectory information, and so on.

[0062] In the embodiment, the network device takes the first mobile trajectory information as the input data of the first model, and obtains the first resource allocation strategy of the first terminal at a future time through the processing of the first model on the first mobile trajectory information. The future time refers to a future time or time range compared with the current time.

[0063] In the embodiment, the first model can adopt a deep neural network (DNN) model. The DNN model is a multi-layer, unsupervised neural network, which is composed of multiple neuron layers, each layer containing multiple neurons; wherein the first layer is the input layer, the last layer is the output layer, and the middle layers are hidden layers. The DNN transmits and processes data through a forward propagation algorithm, the output of each layer is taken as the input of the next layer, and finally the result of the output layer is obtained.

[0064] Due to the universal approximation of the DNN model, that is, given a sufficient number of deep hidden layers and a sufficient number of wide hidden layer nodes, the DNN has the ability to fit any function. Considering that the future resource allocation strategy of the user in the problem of predicting the resource allocation strategy can be directly regarded as a function of the mobile trajectory information of the user, in this embodiment, the first model includes a first function, the input element of the first function is the mobile trajectory information, and the output is the resource allocation strategy at the future time. The corresponding rule of the function is learned by the DNN model to obtain the first model.

[0065] In the embodiment of the application, the first model directly obtains the future resource allocation strategy from the historical trajectory information of the user in an end-to-end manner, thereby omitting the information prediction and large-scale channel information conversion process in the prior art solution, simplifying the process of predicting the resource allocation, further improving the system performance, and reducing the complexity of online decision-making.

[0066] In the embodiments of the application, the resource allocation strategy (including the first resource allocation strategy and the subsequent second resource allocation strategy) can specifically include a data rate (or transmission rate) allocation strategy and / or a power allocation strategy; wherein the data rate (or transmission rate) allocation strategy specifically refers to the data rate (or transmission rate) allocated to the terminal in the process of transmitting data between the network device and the terminal; and the power allocation strategy specifically refers to the transmission power of the network device in the process of transmitting data between the network device and the terminal. That is, the resource allocation strategy represents the resources that can be allocated to the terminal in the process of transmitting data between the network device and the terminal.

[0067] In some optional embodiments, the method further includes: obtaining the playing time of the video requested by the first terminal in the future; and correspondingly, the future time corresponding to the resource allocation strategy at least includes the playing time.

[0068] In this embodiment, the first terminal obtains the playing time of the video in the future in the process of requesting the video. For example, the user requests to play the video by using the first terminal, and after the request reaches the application server, the application server feeds back the video content to the first terminal. In this process, the first terminal obtains the playing time of the video in the future, and then reports the playing time of the video in the future to the network device. The playing time of the video in the future can specifically refer to the playing duration. Correspondingly, the future time corresponding to the resource allocation strategy at least includes the playing time. That is, the first resource allocation strategy of the first terminal in the future obtained by the network device by using the first model at least includes the resource allocation strategy in the video playing duration, for example, if the wireless environment is good, a larger data rate (or transmission rate) can be allocated in the video playing duration of the first terminal, so as to guarantee the video watching experience of the user.

[0069] In some optional embodiments of the present application, the method further comprises: executing the first resource allocation strategy according to time, and transmitting data of the first terminal based on the first performance corresponding to the first resource allocation strategy.

[0070] In the embodiment, after obtaining the first resource allocation strategy of the first terminal, the active optimization module in the network device feeds back the first resource allocation strategy to the network device, and the network device executes the first resource allocation strategy according to time sequence. For example, the first resource allocation strategy is a resource allocation strategy including a time range, that is, the network device can monitor the current time to reach the starting time of the first resource allocation strategy, execute the first resource allocation strategy, and perform data transmission with the first terminal according to the resource allocated by the first resource allocation strategy.

[0071] In some optional embodiments of the present application, the method further comprises: obtaining a first performance parameter during execution of the first resource allocation strategy; and retraining the first model when the performance indicated by the first performance parameter does not meet the set performance requirement.

[0072] In the embodiment, during data transmission between the network device and the first terminal according to the resource allocated by the first resource allocation strategy, that is, during data transmission between the network device and the first terminal according to the data rate and / or power allocated by the first resource allocation strategy, due to the possibility of changes in the actual wireless environment, the actual performance of data transmission between the network device and the first terminal may be different from the resource allocated by the first resource allocation strategy. Therefore, the network device needs to monitor the actual performance parameter (i.e., the first performance parameter) during data transmission with the first terminal, and the first performance parameter can be data rate (or transmission rate) and / or energy consumption. Then, the performance is determined according to the first performance parameter. If the performance indicated by the first performance parameter is better, the first model does not need to be retrained. If the performance indicated by the first performance parameter is worse, the first model needs to be retrained.

[0073] In the embodiment, the network device can compare the first performance parameter with a preset performance threshold value, so as to determine the performance indicated by the first performance parameter.

[0074] In some optional embodiments, the first performance parameter includes data rate and / or energy consumption. The method further comprises: comparing the data rate with a first threshold value, and determining that the performance indicated by the first performance parameter does not meet the set performance requirement when the data rate is less than the first threshold value; and / or comparing the energy consumption with a second threshold value, and determining that the performance indicated by the first performance parameter does not meet the set performance requirement when the energy consumption is greater than or equal to the second threshold value.

[0075] In the embodiment, when the first performance parameter is data rate (or transmission rate), if the data rate is greater than or equal to a first threshold, it can be indicated that the performance represented by the first performance parameter is superior to or meets the set performance requirement; if the data rate is less than the first threshold, it can be indicated that the performance represented by the first performance parameter is inferior to or does not meet the set performance requirement. When the first performance parameter is energy consumption, if the energy consumption is less than a second threshold, it can be indicated that the performance represented by the first performance parameter is superior to or meets the set performance requirement; if the energy consumption is greater than or equal to the second threshold, it can be indicated that the performance represented by the first performance parameter is inferior to or does not meet the set performance requirement. The energy consumption can be the energy consumption of the network device for the first terminal, or can be the total energy consumption of the network device.

[0076] In some optional embodiments, the retraining of the first model comprises: obtaining second movement trajectory information of the second terminal in a first time range, processing the second movement trajectory information by using the first model to obtain a second resource allocation strategy of the second terminal in a second time range; obtaining third movement trajectory information of the second terminal in the second time range, determining large-scale channel information based on the third movement trajectory information; calculating a second performance parameter based on the large-scale channel information and the second resource allocation strategy according to a set performance function, and training the first model according to the second performance parameter.

[0077] In the embodiment, the first model has the capability of online training and updating; the network device obtains the second movement trajectory information and the third movement trajectory information of the second terminal in the first time range and the second time range respectively, processes the second movement trajectory information by using the first model to be trained to obtain a second resource allocation strategy of the second terminal in the second time range, obtains corresponding large-scale channel information from the third movement trajectory information, and calculates a second performance parameter based on the second resource allocation strategy and the large-scale channel information according to a performance function, and trains the first model by using the second performance parameter, thereby realizing online training and updating of the first model.

[0078] In some optional embodiments, during the retraining of the first model, the prediction of the resource allocation strategy by using the first model is stopped or paused.

[0079] In the embodiment, the first time range can be referred to as an observation window, and the length of the first time range, i.e., the length of the observation window; the second time range can be referred to as a prediction window, and the length of the second time range can also be referred to as the length of the prediction window. The first time range is earlier than the second time range, and the specific representation of time can refer to the coordinate axes in FIG. 1. Figure 4 Figure 4

[0080] ​​In this embodiment, the large-scale channel information can include path loss (Pass Loss) and / or shadowing. In this embodiment, the network device can determine the large-scale channel information corresponding to the third mobile trajectory information according to a large-scale channel feature model. The large-scale channel feature model can be any model that can obtain path loss (Pass Loss) and / or shadowing. For example, the large-scale channel feature model can include a path loss model, a ray tracing model, etc. The large-scale channel feature model used in this embodiment is not limited.

[0081] In this embodiment, the network device calculates a second performance parameter according to the large-scale channel information in the second time range (prediction window) and the predicted second resource allocation strategy according to a set performance function. Taking the data rate as an example, the performance function can be represented as Data rate (a p ,d p ); where a p represents the large-scale channel information, d p represents the second resource allocation strategy, that is, the performance function is a function related to the large-scale channel information and the second resource allocation strategy, and the output result of the function is the second performance parameter of the data rate and / or energy consumption. Then the second performance parameter is obtained. The second performance parameter can include the data rate and / or energy consumption.

[0082] In some optional embodiments, the training of the first model according to the second performance parameter includes: determining a loss function based on the performance function when the second performance parameter does not satisfy a set threshold value; wherein the loss function is the gradient of the performance function with respect to the second resource allocation strategy; and adjusting the model parameters of the first model based on the loss function.

[0083] In this embodiment, the calculated second performance parameter is compared with a set threshold value. If the second performance parameter satisfies the set threshold value, it indicates that the performance is good or meets the performance standard, and the online training of the first model is stopped. If the second performance parameter does not satisfy the set threshold value, it indicates that the performance is poor or does not meet the standard, and it is necessary to further determine a loss function according to the performance function and adjust the model parameters of the first model according to the loss function.

[0084] For example, if the data rate is greater than or equal to a set threshold, it indicates that the second performance parameter meets the set threshold; if the data rate is less than the set threshold, it indicates that the second performance parameter does not meet the set threshold. For example, if the energy consumption is less than the set threshold, it indicates that the second performance parameter meets the set threshold; if the energy consumption is greater than or equal to the set threshold, it indicates that the second performance parameter does not meet the set threshold.

[0085] In this embodiment, the loss function is the gradient of the performance function with respect to the second resource allocation strategy. For example, if the second performance parameter is the data rate, the corresponding loss function can be specifically represented as:

[0086]

[0087] The network device can calculate the loss function according to the large-scale channel information and the second resource allocation strategy, and according to the above expression for calculating the gradient of the performance function with respect to the second resource allocation strategy, and is used for updating the model parameters of the first model by back propagation; further, the second mobile trajectory information of the first time range and the third mobile trajectory information of the second time range are obtained in the above manner, and the second resource allocation strategy in the second time range is obtained by processing the second mobile trajectory information using the first model; and the corresponding large-scale channel information is obtained by using the third mobile trajectory information, and then the second performance parameter is calculated according to the performance function based on the second resource allocation strategy and the large-scale information, and the first model is trained using the second performance parameter, until the performance meets the set threshold or the performance meets the standard, or the iteration reaches the termination iteration number, and the online training of the first model is terminated.

[0088] The model online optimization mechanism introduced in the embodiments of the present application is beneficial to implementation in a dynamically changing wireless environment.

[0089] Based on the above embodiments, the embodiments of the present application also provide an optimization device for predicting resource allocation, which is applied to a network device. As shown in Figure 5 The device includes an information acquisition unit 21 and a decision unit 22; wherein,

[0090] The information acquisition unit 21 is configured to obtain the first mobile trajectory information of a first terminal.

[0091] The decision unit 22 is configured to process the first mobile trajectory information using a pre-trained first model to obtain a first resource allocation strategy of the first terminal in the future time.

[0092] In some optional embodiments of the present application, the information acquisition unit 21 is further configured to obtain the playing time of the video requested by the first terminal in the future.

[0093] Correspondingly, the future time corresponding to the resource allocation strategy at least includes the playing time.

[0094] In some optional embodiments of the present application, the apparatus further comprises an execution unit 24 for executing the first resource allocation strategy according to time, and transmitting data of the first terminal based on the first performance of the first resource allocation strategy.

[0095] In some optional embodiments of the present application, the apparatus further comprises a monitoring unit 23;

[0096] The information acquisition unit 21 is further configured to obtain a first performance parameter during execution of the first resource allocation strategy by the execution unit 24, and transmit the first performance parameter to the monitoring unit 23.

[0097] The monitoring unit 23 is configured to transmit a first instruction to the decision unit 22 when the performance indicated by the first performance parameter does not meet the set performance requirement, the first instruction being used to instruct to re-train the first model.

[0098] In some optional embodiments of the present application, the first performance parameter comprises a data rate and / or energy consumption; the monitoring unit 23 is configured to compare the data rate with a first threshold value, and determine that the performance indicated by the first performance parameter does not meet the set performance requirement when the data rate is less than the first threshold value; and / or compare the energy consumption with a second threshold value, and determine that the performance indicated by the first performance parameter does not meet the set performance requirement when the energy consumption is greater than or equal to the second threshold value.

[0099] In some optional embodiments of the present application, the decision unit 22 is further configured to transmit a first time range and a second time range to the information acquisition unit 21.

[0100] The information acquisition unit 21 is further configured to obtain second mobile trajectory information of a second terminal within the first time range and third mobile trajectory information of the second terminal within the second time range, transmit the second mobile trajectory information to the decision unit 22, and transmit the third mobile trajectory information to the monitoring unit 23.

[0101] The decision unit 22 is further configured to process the second mobile trajectory information by using the first model, obtain a second resource allocation strategy of the second terminal within the second time range, and transmit the second resource allocation strategy to the monitoring unit 23.

[0102] The monitoring unit 23 is further configured to determine large-scale channel information based on the third movement trajectory information, calculate second performance parameters based on the large-scale channel information and the second resource allocation strategy according to a set performance function, and train the first model according to the second performance parameters.

[0103] In some optional embodiments of the present invention, the monitoring unit 23 is configured to determine a loss function based on the performance function when the second performance parameter does not meet a set threshold value, and send the loss function to the decision unit 22; wherein, the loss function is the gradient of the performance function with respect to the second resource allocation strategy;

[0104] The decision unit 22 is used to adjust the model parameters of the first model based on the loss function.

[0105] In this embodiment of the invention, the decision-making unit 22 and the monitoring unit 23 in the device can be considered as... Figure 2 The active optimization module shown includes the following functional modules. In practical applications, the decision-making unit 22, execution unit 24, and monitoring unit 23 in the device can all be implemented by the central processing unit (CPU), digital signal processor (DSP), microcontroller unit (MCU), or field-programmable gate array (FPGA) in the terminal. The information acquisition unit 21 in the device can be implemented in practical applications by a communication module (including: basic communication kit, operating system, communication module, standardized interface and protocol, etc.) and transceiver antennas.

[0106] It should be noted that the resource allocation optimization device provided in the above embodiments is only illustrated by the division of the above-described program modules when performing the optimization process of resource allocation. In practical applications, the above processing can be assigned to different program modules as needed, that is, the internal structure of the device can be divided into different program modules to complete all or part of the processing described above. In addition, the resource allocation optimization device and the resource allocation optimization method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process can be found in the method embodiments, which will not be repeated here.

[0107] The optimization method for predictive resource allocation according to embodiments of the present invention will be described below with reference to specific examples. In the following embodiments, a network device is used as a base station as an example, and in conjunction with... Figure 5 The functional modules shown will be described in detail in this embodiment.

[0108] Figure 6 An interaction flow diagram of the optimization method for predicting resource allocation of an embodiment of the present application Figure 1 As shown in Figure 6 The method comprises the following steps:

[0109] Step 301: After the terminal accesses the base station, the terminal reports the mobile trajectory information to the information collection unit of the base station, or reports the real-time position information, and the information collection unit generates the mobile trajectory information according to the obtained real-time position information.

[0110] Here, the terminal also sends the requested video playing time in the future to the information collection unit; wherein, the playing time can be reported simultaneously with the mobile trajectory information, or can be reported separately from the mobile trajectory information.

[0111] Here, the information collection unit receives the information reported by the terminal, and sends a retransmission instruction to the terminal when the interference is large, and saves and records after confirming that the information is correct.

[0112] Step 302: The information collection unit transmits the mobile trajectory information of the user to the decision unit of the active optimization module, the decision unit inputs the received mobile trajectory information into the DNN for inference after vectorization processing, and obtains the resource allocation strategy for the terminal in the future time. Wherein, the resource allocation strategy can include data rate and / or power allocation strategy.

[0113] Here, the DNN can be equivalent to the first model in the above embodiment.

[0114] Step 303: The decision unit feeds back the resource allocation strategy in the future time to the execution unit of the base station in advance, and the execution unit executes in time sequence.

[0115] Here, the future time corresponding to the resource allocation strategy at least includes the playing time.

[0116] Step 304: After the execution unit of the base station transmits the data of the terminal, the information collection unit reports the performance data in the execution process to the monitoring unit. Wherein, the performance data at least includes the actual data rate and / or the actual total energy consumption.

[0117] Here, the execution unit of the base station executes the resource allocation strategy to perform data transmission, which at least includes performing data transmission of the video requested by the terminal according to the resource allocation strategy.

[0118] Step 305: The monitoring unit determines whether to retrain the model according to the performance data. Specifically, the monitoring unit compares the actual performance data with the required performance threshold value, and determines to retrain the model when the actual performance does not meet the performance requirement according to the comparison result.

[0119] Here, when the actual performance meets the performance requirement, the monitoring unit does not issue any instruction, and the active optimization module continues to perform the inference process in the next period; when the actual performance does not meet the performance requirement, the monitoring unit issues a retraining instruction to the decision unit.

[0120] For example, if the actual data rate is greater than or equal to the first threshold value, it indicates that the actual performance meets the performance requirement; if the actual data rate is less than the first threshold value, it indicates that the actual performance does not meet the performance requirement. If the actual total energy consumption is less than the second threshold value, it indicates that the actual performance meets the performance requirement; if the actual total energy consumption is greater than or equal to the second threshold value, it indicates that the actual performance does not meet the performance requirement.

[0121] In actual application, due to the dynamic change of the wireless environment, the model performance may need to be optimized frequently, and therefore the first model in the embodiment has the capability of online updating.

[0122] Figure 7 An interaction flow diagram of the optimization method for the prediction resource allocation of the embodiment of the application Figure 2 ; as shown in Figure 7 , the method comprises:

[0123] Step 401: When the monitoring unit of the active optimization module finds that the actual performance of the model does not meet the performance requirement, the monitoring unit sends a first instruction to the decision unit, and the first instruction is used to indicate that the model inference process is suspended and the model training is re-performed.

[0124] Here, the inference process can specifically be the process in which the decision unit obtains the data rate and / or power allocation strategy in steps 301 to 302. Figure 6

[0125] Step 402: The decision unit informs the information collection unit of the observation window T o and the prediction window T p needed for one online training.

[0126] Here, the observation window T o and the prediction window T p may respectively correspond to the first time range and the second time range in the above embodiment.

[0127] Step 403: The information collection unit sends the observation window T o and the prediction window T p to the terminal, and the terminal reports the second mobile trajectory information to the information collection unit within the observation window T o .

[0128] Step 404: The information collection unit transmits the second mobile trajectory information to the decision unit.

[0129] ​Step 405: the DNN of the decision unit performs forward propagation to obtain the predicted window T p resource allocation strategy d p and sends the resource allocation strategy d p to the monitoring unit.

[0130] Step 406: within the predicted window T p , the terminal sends the third mobile trajectory information to the monitoring unit.

[0131] Here, the terminal can also report the third mobile trajectory information within the predicted window T p to the information collection unit, and the information collection unit sends the third mobile trajectory information to the monitoring unit.

[0132] Step 407: the monitoring unit determines the large-scale channel information a p from the third mobile trajectory information within the predicted window T p , calculates a performance parameter according to the large-scale channel information a p and the resource allocation strategy d p according to a set performance function, and stops the online training when the performance parameter meets the performance requirement, or calculates a loss function according to the performance function and transmits the loss function to the decision unit when the performance parameter does not meet the performance requirement, and the decision unit updates the model parameters of the DNN according to the loss function.

[0133] Here, taking the data rate as an example, the performance function can be expressed as: Data rate (a p ,d p ); the loss function calculated based on the performance function is the gradient of the resource allocation strategy, which can be specifically expressed as follows:

[0134]

[0135] The monitoring unit can calculate the loss function according to the large-scale channel information a p and the resource allocation strategy d p , according to the above expression of the gradient of the resource allocation strategy d p calculated based on the performance function, and transmit the loss function to the decision unit, and the decision unit updates the model parameters of the DNN through back propagation using the loss function, thereby realizing one-time model training.

[0136] Step 408: if the online training has not stopped, continue to execute step 402-step 407 to start the next round of model online training.

[0137] This example can be combined with Figure 4As shown, the corresponding steps 401-402 are performed before online training; the steps 403-405 are performed within an observation window; and the steps 406-407 are performed within a prediction window.

[0138] The embodiment of the present application also provides a network device. Figure 8 A schematic diagram of a hardware component structure of the network device of the embodiment of the present application is shown in Fig. 2. Figure 8 As shown, the network device comprises a memory 52, a processor 51, and a computer program stored in the memory 52 and executable on the processor 51, and the processor 51 implements the steps of the prediction resource allocation optimization method of the embodiment of the present application when executing the program.

[0139] Optionally, the network device can further comprise at least one network interface 53. In the network device, various components are coupled together through a bus system 54. It can be understood that the bus system 54 is used to realize the connection communication between the components. The bus system 54 comprises not only a data bus, but also a power supply bus, a control bus, and a status signal bus. However, for the purpose of clear illustration, all kinds of buses are marked as the bus system 54 in the figure. Figure 8

[0140] ​It can be understood that the memory 52 can be a volatile memory or a non-volatile memory, and can also include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a ferromagnetic random access memory (FRAM), a flash memory, a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM). The magnetic surface memory can be a disk memory or a tape memory. The volatile memory can be a random access memory (RAM) used as an external cache. By way of example but not limitation, many forms of RAM can be used, such as static random access memory (SRAM), synchronous static random access memory (SSRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), sync link dynamic random access memory (SLDRAM), and direct rambus random access memory (DRRAM).The memory 52 described in the embodiments of the present application is intended to include, but not limited to, these and any other suitable type of memory.

[0141] The method disclosed in the embodiments of the present application can be applied in or implemented by the processor 51. The processor 51 can be an integrated circuit chip having a processing capability of signals. In the implementation process, each step of the above method can be completed by the integrated logic circuit of hardware in the processor 51 or the instruction in the form of software. The processor 51 described above can be a general processor, a DSP, or other programmable logic device, discrete gate or transistor logic device, discrete hardware component, etc. The processor 51 can implement or execute the disclosed methods, steps and logic block diagrams in the embodiments of the present application. The general processor can be a microprocessor or any conventional processor, etc. In combination with the steps of the method disclosed in the embodiments of the present application, the hardware decoding processor can be directly embodied to be executed by the decoding processor or executed by the combination of hardware and software modules in the decoding processor. The software module can be located in the storage medium, which is located in the memory 52. The processor 51 reads the information in the memory 52 and combines the hardware to complete the steps of the above method.

[0142] In the exemplary embodiments, the network device can be implemented by one or more ASICs (Application Specific Integrated Circuits), DSPs, PLDs (Programmable Logic Devices), CPLDs (Complex Programmable Logic Devices), FPGAs, general-purpose processors, controllers, MCUs, microprocessors, or other electronic elements, for executing the above method.

[0143] In the exemplary embodiments, the embodiments of the present application also provide a computer readable storage medium, such as the memory 52 including a computer program, which can be executed by the processor 51 of the network device to complete the steps of the above method. The computer readable storage medium can be a FRAM, a ROM, a PROM, an EPROM, an EEPROM, a Flash Memory, a magnetic surface memory, an optical disc, or a CD-ROM, etc. The computer readable storage medium can also be various devices including one or any combination of the above memories.

[0144] The computer readable storage medium provided by the embodiments of the present application has a computer program stored thereon, which is executed by the processor to implement the steps of the prediction resource allocation optimization method of the embodiments of the present application.

[0145] The embodiment of the present application further provides a computer program product comprising a computer program, which can be executed by a network device (such as the processor 51 of the network device) to complete the steps of any of the preceding prediction resource allocation optimization methods.

[0146] The methods disclosed in the several method embodiments provided by the present application can be combined arbitrarily without conflict to obtain new method embodiments.

[0147] The features disclosed in the several product embodiments provided by the present application can be combined arbitrarily without conflict to obtain new product embodiments.

[0148] The features disclosed in the several method or device embodiments provided by the present application can be combined arbitrarily without conflict to obtain new method embodiments or device embodiments.

[0149] In the several embodiments provided by the present application, it should be understood that the disclosed devices and methods can be implemented in other manners. The embodiments described above are merely schematic, and the division of units is merely logical function division, and there can be another division manner in actual implementation, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed coupling, or direct coupling or communication connection between the components can be indirect coupling or communication connection through some interfaces, devices or units, and can be electrical, mechanical or other forms.

[0150] The units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, that is, can be located in one place, or can be distributed on a plurality of network units; some or all of the units can be selected according to actual needs to achieve the purpose of the embodiments.

[0151] In addition, each functional unit in the embodiments of the present application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in the form of hardware, or in the form of hardware plus software function units.

[0152] Those skilled in the art can understand that all or part of the steps of the above-mentioned method embodiments can be completed by program instruction related hardware, and the foregoing program can be stored in a computer readable storage medium, and the program is executed to perform the steps of the above-mentioned method embodiments; and the foregoing storage medium includes mobile storage devices, ROM, RAM, magnetic disks or optical disks and various storage media that can store program codes.

[0153] Alternatively, the above-mentioned integrated unit of the present application, if realized in the form of a software function module and sold or used as an independent product, can also be stored in a computer-readable storage medium. Based on such an understanding, the technical solutions of the embodiments of the present application can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: mobile storage devices, ROM, RAM, magnetic disks or optical disks, and various media that can store program codes.

[0154] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for predicting optimization of resource allocation, characterized in that, The method is applied to a network device, and the method comprises: obtaining first mobile trajectory information of a first terminal; processing the first mobile trajectory information by using a first model to obtain a first resource allocation strategy of the first terminal at a future time.

2. The method of claim 1, wherein, The method further comprises: obtaining a playing time of a video requested by the first terminal in the future; Correspondingly, the future time corresponding to the resource allocation strategy at least comprises the playing time.

3. The method of claim 1, wherein, The method further comprises: executing the first resource allocation strategy according to time, and transmitting data of the first terminal based on a first performance corresponding to the first resource allocation strategy.

4. The method of claim 3, wherein, The method further comprises: obtaining a first performance parameter during execution of the first resource allocation strategy; re-training the first model when a performance indicated by the first performance parameter does not meet a set performance requirement.

5. The method of claim 4, wherein, The first performance parameter comprises a data rate and / or energy consumption; the method further comprises: comparing the data rate with a first threshold value, and determining that the performance indicated by the first performance parameter does not meet the set performance requirement when the data rate is less than the first threshold value; and / or comparing the energy consumption with a second threshold value, and determining that the performance indicated by the first performance parameter does not meet the set performance requirement when the energy consumption is greater than or equal to the second threshold value.

6. The method of claim 4, wherein, The re-training of the first model comprises: obtaining second mobile trajectory information of a second terminal within a first time range, processing the second mobile trajectory information by using the first model to obtain a second resource allocation strategy of the second terminal within a second time range; obtaining third mobile trajectory information of the second terminal within the second time range, and determining large-scale channel information based on the third mobile trajectory information; calculating a second performance parameter based on the large-scale channel information and the second resource allocation strategy according to a set performance function, and training the first model according to the second performance parameter.

7. The method of claim 6, wherein, The training of the first model according to the second performance parameter comprises: determining a loss function based on the performance function when the second performance parameter does not meet a set threshold value; wherein the loss function is a gradient of the performance function with respect to the second resource allocation strategy; adjusting model parameters of the first model based on the loss function.

8. An optimization device for predicting resource allocation, characterized in that, The device is applied to a network device, and the device comprises an information collection unit and a decision unit; wherein The information collection unit is configured to obtain first mobile trajectory information of a first terminal. The decision unit is configured to process the first mobile trajectory information by using a pre-trained first model to obtain a first resource allocation strategy of the first terminal at a future time.

9. The apparatus of claim 8, wherein, The information collection unit is further configured to obtain a playing time of a video requested by the first terminal in the future. Correspondingly, the future time corresponding to the resource allocation strategy at least comprises the playing time.

10. The apparatus of claim 8, wherein, The device further comprises an execution unit configured to execute the first resource allocation strategy according to time, and transmit data of the first terminal based on a first performance corresponding to the first resource allocation strategy.

11. The apparatus of claim 10, wherein, The device further comprises a monitoring unit. The information collection unit is further configured to obtain a first performance parameter during execution of the first resource allocation strategy by the execution unit, and send the first performance parameter to the monitoring unit. The monitoring unit is configured to send a first instruction to the decision unit when the performance indicated by the first performance parameter does not meet a set performance requirement, where the first instruction is used to instruct to retrain the first model.

12. The apparatus of claim 11, wherein, The decision unit is further configured to send a first time range and a second time range to the information collection unit. The information collection unit is further configured to obtain second movement trajectory information of a second terminal within the first time range and third movement trajectory information of the second terminal within the second time range, send the second movement trajectory information to the decision unit, and send the third movement trajectory information to the monitoring unit. The decision unit is further configured to process the second movement trajectory information by using the first model, obtain a second resource allocation strategy of the second terminal within the second time range, and send the second resource allocation strategy to the monitoring unit. The monitoring unit is further configured to determine large-scale channel information based on the third movement trajectory information, calculate a second performance parameter based on the large-scale channel information and the second resource allocation strategy according to a set performance function, and train the first model according to the second performance parameter.

13. The apparatus of claim 12, wherein, The monitoring unit is configured to determine a loss function based on the performance function when the second performance parameter does not meet a set threshold value, and send the loss function to the decision unit, where the loss function is a gradient of the performance function with respect to the second resource allocation strategy. The decision unit is configured to adjust model parameters of the first model based on the loss function.

14. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the steps of the method of any one of claims 1 to 7.

15. A network device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the computer program comprises instructions for causing the processor to perform the method of any one of claims 1 to 14. The processor executes the program to implement the steps of the method of any one of claims 1 to 7.

16. A computer program product, characterised in that, The computer program instructions cause the computer to implement the steps of the method of any one of claims 1 to 7.

Images