Unmanned aerial vehicle base station deployment method and device and electronic equipment

[0071] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of the present invention.

[0072] In order to improve the deployment efficiency of UAV base stations, embodiments of the present invention provide a UAV base station deployment method, device and electronic equipment.

[0073] figure 1 This is a schematic flowchart of a method for deploying a drone base station provided by an embodiment of the present invention, including:

[0074] Step 101: Obtain user location information.

[0075] Step 102: Input the user location information into the neural network model obtained by pre-training to obtain deployment location information of the drone, so that the drone completes the deployment operation of the drone base station based on the deployment location information.

[0076] Among them, the neural network model is trained based on the position information of the sample users and the optimal UAV deployment position information corresponding to the position information of the sample users.

[0077] It can be seen from the above embodiment that in the embodiment of the present invention, the neural network model can be trained in advance. When the user position is changed, there is no need to re-train iteratively based on the changed user position, but only need to change the user position Then input into the neural network model obtained by the above training to obtain the deployment location information of the UAV, that is: in the actual deployment process, no iterative training is required, therefore, the efficiency of obtaining the deployment location information of the UAV is improved, and further Improve the efficiency of UAV base station deployment.

[0078] See figure 2 , figure 2 for figure 1 The flow diagram of neural network model training in the illustrated embodiment specifically includes:

[0079] Step 201: Obtain sample user location information and optimal drone deployment location information corresponding to the sample user location information.

[0080] Specifically, the process of obtaining the location information of the sample user may be: first obtain the area of ​​the target area where the sample user is located, and the user density in the target area to obtain the number of sample users in the target area, and then use the target area The location information of the sample users is obtained by randomly distributing the above-mentioned number of sample users in the method. There may be multiple sets of sample user location information, and each time the above-mentioned number of sample users are randomly distributed, one set of sample user location information is obtained.

[0081] After obtaining the location information of the sample users, traditional exhaustive algorithms or reinforcement learning methods, such as DQN, can be used to obtain the optimal drone deployment location information corresponding to the location information of the sample users. The optimal UAV deployment location information is: when the number of UAVs is determined, the UAV base station system can maximize the UAV deployment location information of the wireless network. Here, there is no limitation on the specific method for obtaining the optimal deployment position information of the drone.

[0082] When the number of users is determined, the ratio of the number of users to the maximum number of users that can be accessed by a single drone can be rounded up to get the number of drones corresponding to the number of users. For example, if the number of users is 95, and the maximum number of users that can be accessed by a single drone is 10, the number of drones is 10.

[0083] For each set of sample user location information, the process of using the DQN method to obtain the optimal drone deployment location information corresponding to the set of sample user location information can be: input the set of sample user location information and the initial location information of each drone In the Q network, according to the preset period time, the Q network periodically outputs an action information based on its internal parameters. After calculating the drone to perform the above action information, at the end of the period, the updated position information of the drone and When the UAV base station is deployed in the updated location, the wireless network capacity provided by the UAV base station system will determine the initial location information of the user and the UAV, the updated location information of the UAV, and the unmanned aircraft in each cycle. The wireless network capacity that the base station system can provide is a set of data. Analyze each group of data in each period to determine the target action information that maximizes the wireless network capacity. After that, adjust the Q network parameters so that the output action information is the above target action information, and the wireless network capacity reaches the preset threshold. At this time, The final position of the UAV in the period is determined as the optimal UAV deployment position information corresponding to the set of sample position information.

[0084] In the process of obtaining the information of the optimal deployment position of the drone, the wireless network capacity that the drone base station system can provide is equal to: the sum of the capacity of a single communication link between the drone and the individual users that it provides services. The communication link capacity can be calculated by the following formula:

[0085]

[0086] Among them, C is the capacity of a single communication link between the UAV and the single user it provides services; W is the bandwidth; P is the power of the useful signal received by a single user, I is the signal interference, and N is the noise.

[0087] It should be noted that the optimal UAV deployment location information in the embodiment of the present invention refers to the deployment location information of all UAVs obtained through traditional exhaustive algorithms or reinforcement learning methods. Specifically, when the number of UAVs is 1, the optimal UAV deployment location information is the UAV deployment location information obtained through traditional exhaustive methods or reinforcement learning methods; When the number is multiple, the optimal UAV deployment location information includes: the deployment location information of all UAVs obtained through traditional exhaustive methods or reinforcement learning methods.

[0088] Step 202: Input the position information of the sample user into the neural network model to obtain the first drone deployment position information corresponding to the position information of the sample user.

[0089] The neural network model in this step can be any neural network model, such as a fully connected neural network model, a feedforward neural network model, etc., which can obtain the first UAV deployment location information corresponding to the sample user location information. The specific structure of the network model is not limited.

[0090] Corresponding to the optimal UAV deployment location information, the first UAV deployment location information in the embodiment of the present invention refers to the deployment location information of all UAVs obtained through the neural network model. Specifically, when the number of UAVs is 1, the optimal UAV deployment location information is the UAV deployment location information obtained through the neural network model; when the number of UAVs is multiple, The optimal UAV deployment location information includes: the deployment location information of all UAVs obtained through the neural network model.

[0091] Step 203: Construct a loss function based on the first UAV deployment location information and the optimal UAV deployment location information.

[0092] Specifically, the mean square error between the first UAV deployment location information and the optimal UAV deployment location information may be used as the loss function.

[0093] Step 204: Determine whether the loss function is less than a preset threshold. If yes, the training is ended, and the trained neural network model is obtained; if not, step 205 is executed.

[0094] Step 205, adjust the network parameters of the neural network model, and return to step 201.

[0095] The network parameters in this step may include: the weights and biases between neurons in the neural network model that have a connection relationship between each level.

[0096] See image 3 , image 3 It is a schematic diagram of another process of the deployment method of a drone base station according to an embodiment of the present invention, which specifically includes:

[0097] Step 301: Obtain user location information.

[0098] Step 302: Calculate the number of users contained in each sub-region based on the user location information.

[0099] It should be noted that the target area based on the neural network model training phase is the same area as the area to be deployed in the actual application process, or the two areas have the same size and the wireless network environment in the area is also the same. At the same time, the number of sample users in the target area is the same or close to the number of users in the area to be deployed.

[0100] In the neural network model training stage, the target area where the user is located can be divided into a preset number of sub-areas. Correspondingly, in this step, the area to be deployed can also be divided into a preset number of sub-areas in the same way. To calculate the number of users included in each sub-area.

[0101] Step 303: Input the number of users contained in each sub-region into each input neuron in the input layer of the neural network model obtained by pre-training to obtain the deployment position information of the UAV.

[0102] In the input layer of the neural network model obtained by pre-training, the number of input neurons is the same as the number of sub-regions obtained after dividing the region to be deployed. In this step, the number of users contained in each sub-region can be input separately In each input neuron in the input layer of the neural network model obtained by pre-training, the deployment position information of the UAV can be obtained after processing by the neural network model obtained by the pre-training.

[0103] Step 304: Determine the number information of the sub-region where the drone is located based on the deployment location information of the drone.

[0104] In order to facilitate the understanding of this step, the following uses an example to illustrate:

[0105] Assuming: the area to be deployed is a square area of ​​1500m*1500m. The square area is divided into 50m*50m sub-areas in advance, and coding information is generated for each sub-areas in order from left to right and top to small. Specifically: the coding information of the first sub-area in the upper left corner is No. 1, followed by No. 2, No. 3... Assuming that the number of drones is 3, and the deployment position information of the drones obtained in step 303 are: (125,125), (325,1325) and (1225,325), the unit is m.

[0106] In this step, the number information of the sub-region where the drone is located is calculated by the following method: For the deployment location information of the drone (125, 125), the number information of the sub-region where the drone is located is: [125/50]+[125/50 ]×30=3+30×3=93, where [] means rounding up; for the UAV deployment location information (325, 1325), the number information of the sub-region where it is located is: [325/50]+[ 1325/50]×30=7+27×30=817; for the UAV deployment position information (1225,325), the number information of the sub-region where it is located is: [1225/50]+[325/50]*30 =25+7*30=235. In summary, the deployment location information (125,125), (325,1325) and (1225,325) of the UAV's sub-region number information are respectively: No. 93, No. 817, and No. 235.

[0107] Step 305, based on the size order of the number information of the sub-region where the drone is located, and the preset correspondence between each output neuron in the output layer of the neural network model obtained by pre-training and the size order of the number information, the nerve obtained through pre-training Each output neuron in the output layer of the network model outputs the deployment position information of the UAV, so that the UAV can complete the UAV base station deployment operation based on the deployment position information.

[0108] Taking the position information of the three drones deployed in step 304 as an example, the output layer of the neural network model obtained by pre-training also contains three output neurons. Assuming that each output neuron also has number information, the number 1 output neuron No. 2, output neuron No. 2 and No. 3 output neuron, and set the default corresponding relationship between each output neuron and the order of the number information: The deployment position of the drone with the smallest number in the subregion where the output neuron No. 1 output is located Information, the deployment location information of the drone with the middle number in the subregion where the output neuron No. 2 outputs, and the deployment location information of the drone with the largest number in the subregion where the output neuron output is located. Therefore, the output of the output neuron No. 1 is located The UAV deployment location information of the sub-region number 93, the UAV deployment location information of the sub-region number 235 where the output neuron No. 2 outputs, and the sub-region number 817 where the output neuron output No. 3 is located. Human-machine deployment location information. Then in step 305, the UAV deployment location information output by the output neuron No. 1 should be (125, 125), the UAV deployment location information output by the output neuron No. 2 should be (1225, 325), and the output neuron No. 3 The UAV deployment location information output by the meta should be (325,1325).

[0109] in image 3 In the illustrated embodiment, the neural network model can be trained in advance. When the user's position is changed, there is no need to re-train iteratively based on the changed user's position. Instead, it is only necessary to input the changed user's position into the training obtained above. In the neural network model, the deployment location information of the UAV can be obtained, that is, no iterative training is required in the actual deployment process. Therefore, the efficiency of obtaining the deployment location information of the UAV is improved, thereby improving the deployment of the UAV base station s efficiency.

[0110] See Figure 4 , Figure 4 for image 3 In the illustrated embodiment, a schematic diagram of the process of acquiring the deployment location information of the drone. First, obtain the location information of each user in the area to be deployed, and then, based on each sub-area obtained after the division of the area to be deployed, and the location information of each user, obtain the number of users contained in each sub-area, and compare each sub-area The number of users included is input to each input neuron in the input layer of the trained neural network model. After processing, the deployment location information of the drone is obtained, and then based on the deployment location information of the drone, the unmanned The number information of the sub-region where the drone is located, and finally, based on the size order of the number information of the sub-region where the drone is located, and the preset correspondence between each output neuron in the output layer of the neural network model obtained in advance and the size order of the number information , Through each output neuron in the output layer of the neural network model obtained by pre-training, output the deployment position information of the UAV.

[0111] image 3 In the illustrated embodiment, the neural network model can be trained in advance. When the user's position is changed, there is no need to re-train iteratively based on the changed user's position. Instead, it is only necessary to input the changed user's position into the training obtained above. In the neural network model, the deployment location information of the drone can be obtained, that is: in the actual deployment process, no iterative training is required, therefore, the efficiency of obtaining the deployment location information of the drone is improved, thereby improving the deployment of the drone base station s efficiency.

[0112] See Figure 5 , Figure 5 for image 3 The schematic diagram of the training process of the neural network model in the illustrated embodiment, the specific training process includes:

[0113] Step 501: Obtain sample user location information and optimal drone deployment location information corresponding to the sample user location information.

[0114] Step 502: Divide the area to be deployed where the sample user is located into a preset number of sub-areas.

[0115] Step 503: Calculate the number of sample users contained in each sub-region based on the position information of the sample users.

[0116] Step 504: Input the number of sample users contained in each sub-region into each input neuron in the input layer of the neural network model to obtain the first drone deployment location information corresponding to the sample user location information.

[0117] Step 505: Determine the number information of the sub-region where the drone is located based on the deployment location information of the first drone.

[0118] Step 506: Based on the size order of the number information of the sub-region where the drone is located, and the preset correspondence between each output neuron in the output layer of the neural network model and the size order of the number information, each output in the output layer of the neural network model The neuron outputs the first UAV deployment location information corresponding to the sample user location information.

[0119] Step 507: Construct a loss function based on the first UAV deployment location information and the optimal UAV deployment location information.

[0120] Step 508: Determine whether the loss function is less than a preset threshold. If yes, the training is ended, and the trained neural network model is obtained; if not, step 509 is executed.

[0121] Step 509: Adjust the network parameters of the neural network model. After that, return to step 501.

[0122] Since the number of drones can be multiple, when the number of drones is more than one, there will be multiple corresponding first drone deployment location information. Similarly, the optimal drone deployment location information is also There will be multiple, and in the neural network model training process, the loss function is calculated for each drone, that is to say: it is based on the first drone deployment location information corresponding to each drone and The optimal UAV deployment location information corresponding to the UAV can be used to construct the loss function. Specifically, the first UAV deployment location information corresponding to the same UAV can be judged with the optimal UAV corresponding to the UAV. Whether the mean square error between the human-machine deployment position information is less than the preset threshold, it is finally determined whether the training of the neural network model is completed.

[0123] If each output neuron in the output layer randomly outputs the first UAV deployment location information, because there are multiple first UAV deployment location information and multiple optimal UAV deployment location information, at this time, it cannot be determined The first UAV deployment location information output by each neuron corresponds to which of the multiple optimal UAV deployment location information corresponds to the optimal UAV deployment location information, which leads to the problem of inaccurate loss function calculation.

[0124] For example: if the number of drones is 3, then the first drone deployment location information and the optimal drone deployment location information each contain 3 specific location information, among which, each specific first drone There is theoretically a corresponding relationship between the deployment location information of the drone and the specific optimal deployment location information of the drone. If each output neuron randomly outputs 3 specific first drone deployment location information, that is to say , Without limiting "which output neuron corresponds to which specific first UAV deployment location information", it is impossible to determine the deployment location information of each UAV obtained through neural network model processing, which is different from the traditional exhaustive method or enhanced The corresponding relationship between the deployment location information of the UAVs obtained by the learning method leads to inaccurate calculation of the loss function, which in turn makes the neural network model inaccurate.

[0125] In the embodiment of the present invention, the preset corresponding relationship between the size and order of each output neuron and the number information is preset. Based on the above corresponding relationship, it can be determined that the first UAV deployment position information output by each output neuron and Correspondence between the optimal UAV deployment position information, therefore, the accuracy of the neural network model can be improved.

[0126] Based on the same inventive concept, according to the UAV base station deployment method provided in the above embodiment of the present invention, correspondingly, the embodiment of the present invention also provides a UAV base station deployment device. The structure diagram of the device is as Image 6 Shown, including:

[0127] The user location information obtaining module 601 is used to obtain user location information;

[0128] The deployment location information obtaining module 602 is used to input user location information into a pre-trained neural network model to obtain deployment location information of the drone, so that the drone can complete the deployment operation of the drone base station based on the deployment location information; The neural network model is trained based on the position information of the sample users and the optimal UAV deployment position information corresponding to the position information of the sample users.

[0129] Further, the device also includes: a model training module;

[0130] The model training module includes: a sample information acquisition submodule, a first UAV deployment location information acquisition submodule, a loss function construction submodule, a judgment submodule, and a parameter adjustment submodule;

[0131] The sample information acquisition sub-module is used to acquire the location information of the sample user and the optimal drone deployment location information corresponding to the location information of the sample user;

[0132] The first drone deployment location information obtaining sub-module is used to input the sample user location information into the neural network model to obtain the first drone deployment location information corresponding to the sample user location information;

[0133] The loss function construction sub-module is used to construct the loss function based on the first UAV deployment location information and the optimal UAV deployment location information;

[0134] The judgment sub-module is used to judge whether the loss function is less than the preset threshold; if it is, the training is ended, and the trained neural network model is obtained; if not, the parameter adjustment sub-module is triggered;

[0135] The parameter adjustment sub-module is used to adjust the network parameters of the neural network model and trigger the first UAV deployment position information to obtain the sub-module.

[0136] Further, the first UAV deployment location information obtains the sub-module, which is specifically used for:

[0137] Divide the area to be deployed where the sample users are located into a preset number of sub-areas;

[0138] Based on the location information of the sample users, calculate the number of sample users contained in each sub-region;

[0139] Input the number of sample users contained in each sub-area into each input neuron in the input layer of the neural network model to obtain the first drone deployment location information corresponding to the sample user location information;

[0140] Through each output neuron in the output layer of the neural network model, the first UAV deployment location information corresponding to the sample user location information is output; the number of input neurons in the input layer of the neural network model is equal to the number of sub-regions, The number of output neurons in the output layer of the neural network model is equal to the number of drones. The number of drones is determined based on the number of sample users and the maximum number of users that a single drone can serve;

[0141] The deployment location information obtaining module 602 is specifically used for:

[0142] Based on user location information, calculate the number of users included in each sub-area;

[0143] Input the number of users contained in each sub-area into each input neuron in the input layer of the neural network model obtained in advance to obtain the deployment location information of the drone;

[0144] Each output neuron in the output layer of the neural network model obtained by pre-training outputs the deployment position information of the UAV.

[0145] Further, each sub-area of ​​the area to be deployed has number information; the number of drones is multiple;

[0146] The first UAV deployment location information obtains the sub-module, which is specifically used when performing the step of outputting the first UAV deployment location information corresponding to the sample user location information through each output neuron in the output layer of the neural network model :

[0147] Based on the deployment location information of the first drone, determine the numbering information of the subregion where the drone is located; based on the size order of the numbering information of the subregion where the drone is located, and the output neurons and numbering information in the output layer of the neural network model The preset corresponding relationship in the order of magnitude, through each output neuron in the output layer of the neural network model, output the first UAV deployment location information corresponding to the sample user location information;

[0148] The deployment location information obtaining module 602 is specifically used to: when executing the steps of outputting the deployment location information of the drone by each output neuron in the neural network model output layer obtained through pre-training:

[0149] Based on the deployment location information of the drone, determine the number information of the sub-region where the drone is located; based on the size order of the number information of the sub-region where the drone is located, and each output neuron in the output layer of the neural network model obtained in advance. The preset corresponding relationship with the size and order of the number information, through each output neuron in the output layer of the neural network model obtained by pre-training, outputs the deployment position information of the UAV.

[0150] In the present invention Image 6 In the illustrated embodiment, the neural network model can be trained in advance. When the user's position is changed, there is no need to re-train iteratively based on the changed user's position. Instead, it is only necessary to input the changed user's position into the training obtained above. In the neural network model, the deployment location information of the drone can be obtained, that is, no iterative training is required in the actual deployment process, therefore, the efficiency of obtaining the deployment location information of the drone is improved, and the deployment of the drone base station is improved s efficiency.

[0151] The embodiment of the present invention also provides an electronic device, such as Figure 7 As shown, it includes a processor 701, a communication interface 702, a memory 703, and a communication bus 704. The processor 701, the communication interface 702, and the memory 703 communicate with each other through the communication bus 704.

[0152] The memory 703 is used to store computer programs;

[0153] The processor 701 is configured to implement the following steps when executing a program stored in the memory 703:

[0154] Obtain user location information;

[0155] Input user location information into the pre-trained neural network model to obtain the deployment location information of the drone, so that the drone can complete the deployment operation of the drone base station based on the deployment location information; the neural network model is based on the sample user location information , And the optimal UAV deployment location information training corresponding to the sample user location information.

[0156] Further, it may also include other processing procedures in the above-mentioned UAV base station deployment method provided by the embodiment of the present invention, which will not be described in detail here.

[0157] The aforementioned communication bus mentioned by the electronic device may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The communication bus can be divided into address bus, data bus, control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0158] The communication interface is used for communication between the aforementioned electronic device and other devices.

[0159] The memory may include random access memory (Random Access Memory, RAM for short), or non-volatile memory (Non-Volatile Memory, NVM for short), such as at least one disk storage. Optionally, the memory may also be at least one storage device located far away from the foregoing processor.

[0160] The aforementioned processor may be a general-purpose processor, including a central processing unit (Central Processing Unit, CPU for short), a network processor (Network Processor, NP), etc.; it may also be a digital signal processor (DSP) , Application Specific Integrated Circuit (ASIC), Field-Programmable Gate Array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components.

[0161] In yet another embodiment provided by the present invention, a computer-readable storage medium is also provided. The computer-readable storage medium stores instructions that, when run on a computer, cause the computer to execute any of the above-mentioned embodiments. The deployment method of the UAV base station.

[0162] In yet another embodiment provided by the present invention, a computer program product containing instructions is also provided, which when running on a computer, causes the computer to execute the drone base station deployment method described in any of the above embodiments.

[0163] In the above-mentioned embodiments, it may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented by software, it can be implemented in the form of a computer program product in whole or in part. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on the computer, the processes or functions described in the embodiments of the present invention are generated in whole or in part. The computer may be a general-purpose computer, a dedicated computer, a computer network, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from a website, computer, server, or data center. Transmission to another website, computer, server, or data center via wired (for example, coaxial cable, optical fiber, Digital Subscriber Line (DSL)) or wireless (for example, infrared, wireless, microwave, etc.). The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or data center integrated with one or more available media. The usable medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a high-density digital video disc (Digital Video Disc, DVD)), or a semiconductor medium (for example, a solid state disk (Solid State Disk, Referred to as SSD)) etc.

[0164] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply one of these entities or operations. There is any such actual relationship or order between. Moreover, the terms "include", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes those that are not explicitly listed Other elements of, or also include elements inherent to this process, method, article or equipment. If there are no more restrictions, the element defined by the sentence "including a..." does not exclude the existence of other same elements in the process, method, article, or equipment that includes the element.

[0165] The various embodiments in this specification are described in a related manner, and the same or similar parts between the various embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and for related parts, please refer to the part of the description of the method embodiments.

[0166] The foregoing descriptions are only preferred embodiments of the present invention, and are not used to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are all included in the protection scope of the present invention.