Air base station deployment method, system, device, electronic equipment and storage medium

Abstract

The present disclosure provides an air base station deployment method, system, device, electronic equipment and storage medium, and relates to the technical field of communication. The present disclosure obtains the communication state information of each air base station in a target three-dimensional space, wherein each air base station serves at least one user terminal, and the communication state information of each air base station includes the deployment position of each air base station in the target three-dimensional space and the spectrum efficiency of each user terminal served by each air base station; based on a pre-constructed deep reinforcement learning model with the target of maximizing system spectrum efficiency, the deployment position of each air base station in the target three-dimensional space when the system spectrum efficiency is maximum is determined. The present disclosure utilizes the deep reinforcement learning model to realize adaptive and efficient deployment of air base stations, meets real-time changing communication demands, improves the efficiency and reliability of the communication system, increases the spectrum efficiency of the communication system, and improves the network experience of users.

Description

Technical Field

[0001] This disclosure relates to the field of communication technology, and in particular to a method, system, apparatus, electronic device and storage medium for deploying an airborne base station. Background Technology

[0002] With the development of mobile internet, people have increasingly higher demands for communication experiences in various application scenarios. Users expect to obtain a smooth service experience in ultra-dense scenarios such as stadiums, open-air gatherings, and concerts, which put enormous load pressure on networks. Due to its advantages such as hovering capability, ease of deployment, flexibility, and low deployment cost, drone-based aerial base stations are considered an important supplement to terrestrial communication networks, effectively enhancing wireless capacity and coverage on the ground to meet the requirements of mobile communication.

[0003] The location of aerial base stations can affect users' network experience. How to deploy aerial base stations to provide users with better network quality services is a technical problem that urgently needs to be solved.

[0004] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] This disclosure provides a method, system, apparatus, electronic device, and storage medium for deploying an airborne base station, which at least to some extent overcomes the problem of poor user network experience caused by unsuitable deployment locations of airborne base stations in related technologies.

[0006] Other features and advantages of this disclosure will become apparent from the following detailed description, or may be learned in part from practice of this disclosure.

[0007] According to one aspect of this disclosure, a method for deploying airborne base stations is provided, comprising: acquiring communication status information of each airborne base station in a target three-dimensional space, wherein each airborne base station serves at least one user terminal, and the communication status information of each airborne base station includes: the deployment location of each airborne base station in the target three-dimensional space and the spectral efficiency of each user terminal served by each airborne base station; determining the deployment location of each airborne base station in the target three-dimensional space when the system spectral efficiency is maximized based on a pre-constructed deep reinforcement learning model with the goal of maximizing system spectral efficiency, wherein the system spectral efficiency is the sum of the average spectral efficiencies of all airborne base stations in the target three-dimensional space, and the average spectral efficiency of each airborne base station is the average of the spectral efficiencies of the multiple user terminals served by each airborne base station.

[0008] In some embodiments, based on a pre-built deep reinforcement learning model aimed at maximizing system spectral efficiency, the deployment position of each airborne base station in the target three-dimensional space when the system spectral efficiency is maximized is determined, including: deploying a deep reinforcement learning model aimed at maximizing system spectral efficiency on a high-altitude platform; deploying a deep reinforcement learning model aimed at maximizing average efficiency on each airborne base station; and constructing a federated learning model based on the high-altitude platform and each airborne base station to obtain the deployment position of each airborne base station in the target three-dimensional space when the system spectral efficiency is maximized.

[0009] In some embodiments, a federated learning model is constructed based on the high-altitude platform and each airborne base station to obtain the deployment position of each airborne base station in the target three-dimensional space when the system spectral efficiency is maximized. This includes: uploading the local model parameters of each airborne base station to the high-altitude platform so that the high-altitude platform can train a deep reinforcement learning model with the goal of maximizing the system spectral efficiency based on the local model parameters uploaded by each airborne base station; and distributing the global model parameters trained by the high-altitude platform to each airborne base station.

[0010] In some embodiments, after determining the deployment position of each airborne base station in the target three-dimensional space when the system spectral efficiency is maximized based on a pre-built deep reinforcement learning model aimed at maximizing system spectral efficiency, the method further includes: activating a motion system for each airborne base station according to the deployment position of each airborne base station in the target three-dimensional space when the system spectral efficiency is maximized, so as to change the deployment position of each airborne base station in the target three-dimensional space.

[0011] In some embodiments, obtaining communication status information of each airborne base station in the target three-dimensional space includes: obtaining deployment location data of each airborne base station in the target three-dimensional space, the deployment location data including: the height of each airborne base station and the horizontal distance from the airborne base station to the serving user terminal; and calculating the spectral efficiency of each user terminal based on the height of each airborne base station and the horizontal distance from the airborne base station to the serving user terminal.

[0012] According to another aspect of this disclosure, a communication system is also provided, comprising: a base station deployment server, a plurality of air base stations, and at least one user terminal served by each air base station; the base station deployment server is configured to acquire communication status information of each air base station in a target three-dimensional space, wherein each air base station serves at least one user terminal, and the communication status information of each air base station includes: the deployment location of each air base station in the target three-dimensional space and the spectral efficiency of each user terminal served by each air base station; based on a pre-constructed deep reinforcement learning model with the goal of maximizing system spectral efficiency, the deployment location of each air base station in the target three-dimensional space when the system spectral efficiency is maximized is determined, wherein the system spectral efficiency is the sum of the average spectral efficiencies of all air base stations in the target three-dimensional space, and the average spectral efficiency of each air base station is the average of the spectral efficiencies of the plurality of user terminals served by each air base station.

[0013] In some embodiments, the communication system further includes: a high-altitude platform; a base station deployment server, which is also used to deploy a deep reinforcement learning model on the high-altitude platform with the goal of maximizing system spectral efficiency; deploy a deep reinforcement learning model on each airborne base station with the goal of maximizing average efficiency; and construct a federated learning model based on the high-altitude platform and each airborne base station to obtain the deployment position of each airborne base station in the target three-dimensional space when the system spectral efficiency is maximized.

[0014] According to another aspect of this disclosure, an airborne base station deployment apparatus is also provided. The airborne base station deployment apparatus includes a data acquisition module for acquiring communication status information of each airborne base station within a target three-dimensional space, wherein each airborne base station serves at least one user terminal, and the communication status information of each airborne base station includes: the deployment location of each airborne base station within the target three-dimensional space and the spectral efficiency of each user terminal served by each airborne base station; and a deployment location determination module for determining the deployment location of each airborne base station within the target three-dimensional space when the system spectral efficiency is maximized, based on a pre-constructed deep reinforcement learning model aimed at maximizing system spectral efficiency, wherein the system spectral efficiency is the sum of the average spectral efficiencies of all airborne base stations within the target three-dimensional space, and the average spectral efficiency of each airborne base station is the average of the spectral efficiencies of the multiple user terminals served by each airborne base station.

[0015] According to another aspect of this disclosure, an electronic device is also provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to perform the airborne base station deployment method of any of the above via executing the executable instructions.

[0016] According to another aspect of this disclosure, a computer-readable storage medium is also provided, on which a computer program is stored, which, when executed by a processor, implements the airborne base station deployment method of any of the above.

[0017] According to another aspect of this disclosure, a computer program product is also provided, including a computer program that, when executed by a processor, implements the airborne base station deployment method of any of the above.

[0018] The airborne base station deployment method, system, apparatus, electronic device, and storage medium provided in the embodiments of this disclosure acquire communication status information of each airborne base station in a target three-dimensional space. Each airborne base station serves at least one user terminal. The communication status information of each airborne base station includes: the deployment location of each airborne base station in the target three-dimensional space and the spectral efficiency of each user terminal served by each airborne base station. Based on a pre-constructed deep reinforcement learning model aimed at maximizing system spectral efficiency, the deployment location of each airborne base station in the target three-dimensional space when the system spectral efficiency is maximized is determined. The system spectral efficiency is the sum of the average spectral efficiencies of all airborne base stations in the target three-dimensional space, and the average spectral efficiency of each airborne base station is the average of the spectral efficiencies of the multiple user terminals served by each airborne base station. This improves the efficiency of airborne base station deployment by monitoring the deployment location of airborne base stations in the target three-dimensional space and the spectral efficiency of the multiple user terminals they serve. It identifies changes in user needs and the communication environment, and uses a deep reinforcement learning model to achieve adaptive deployment of airborne base stations, meeting real-time changing communication needs, improving system spectral efficiency, enhancing communication quality, and improving user experience.

[0019] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0020] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure. It is obvious that the drawings described below are merely some embodiments of this disclosure, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0021] Figure 1 A schematic diagram of a communication system architecture according to an embodiment of this disclosure is shown;

[0022] Figure 2 A schematic diagram of yet another communication system architecture is shown in an embodiment of this disclosure;

[0023] Figure 3This diagram illustrates a method for deploying an airborne base station according to an embodiment of the present disclosure.

[0024] Figure 4 This diagram illustrates yet another method for deploying an airborne base station according to an embodiment of the present disclosure;

[0025] Figure 5 This diagram illustrates yet another method for deploying an airborne base station according to an embodiment of the present disclosure;

[0026] Figure 6 This diagram illustrates yet another method for deploying an airborne base station according to an embodiment of the present disclosure;

[0027] Figure 7 This diagram illustrates an airborne base station deployment device according to an embodiment of the present disclosure;

[0028] Figure 8 This diagram illustrates a structural block diagram of an electronic device according to an embodiment of the present disclosure.

[0029] Figure 9 A schematic diagram of a computer-readable storage medium is shown in an embodiment of the present disclosure. Detailed Implementation

[0030] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that this disclosure will be more comprehensive and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0031] Furthermore, the accompanying drawings are merely illustrative of this disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0032] As mentioned in the background section, with the development of mobile internet, people have increasingly higher demands for communication experiences across various application scenarios. Users expect a consistent service experience in ultra-dense environments such as stadiums, open-air gatherings, and concerts, which place enormous load pressure on networks. Due to its advantages such as hovering capability, ease of deployment, flexibility, and low deployment cost, using drone-based aerial base stations for temporary communication is considered an important supplement to terrestrial communication networks. It can effectively enhance wireless capacity and coverage on the ground, meeting the requirements of mobile communication.

[0033] In the three-dimensional deployment of aerial base stations, traditional heuristic algorithms require repetitive calculations. Due to the continuous updates and evolution of mobile communication technologies and network architectures, deploying aerial base stations still requires significant research and development work to build more intelligent, flexible, and reliable systems. Meanwhile, considering cost and economic efficiency, how to maximize the use of existing network infrastructure and resources without impacting user experience, and how to deploy aerial base stations to provide users with a high-quality network experience, are pressing technical problems that need to be solved.

[0034] The specific implementation methods of the embodiments of this disclosure will now be described in detail with reference to the accompanying drawings.

[0035] Figure 1 A schematic diagram of an exemplary application communication system architecture to which the airborne base station deployment method of the embodiments of this disclosure can be applied is shown. Figure 1 As shown, the communication system architecture includes a base station deployment server 101, an air base station 102, and a user terminal 103.

[0036] The base station deployment server 101 is used to acquire the communication status information of each air base station 102 in the target three-dimensional space. Each air base station 102 serves at least one user terminal 103. The communication status information of each air base station 102 includes: the deployment position of each air base station 102 in the target three-dimensional space and the spectral efficiency of each user terminal 103 served by each air base station 102. Based on a pre-constructed deep reinforcement learning model with the goal of maximizing system spectral efficiency, the deployment position of each air base station 102 in the target three-dimensional space when the system spectral efficiency is maximized is determined. The system spectral efficiency is the sum of the average spectral efficiency of all air base stations 102 in the target three-dimensional space, and the average spectral efficiency of each air base station 102 is the average of the spectral efficiency of the multiple user terminals 103 served by each air base station 102.

[0037] The airborne base station 102 can be a mobile base station, relay, or access point. The airborne base station can be a 5G or later version base station (e.g., 5G NR NB), or a base station in other communication systems (e.g., eNB base station). It should be noted that the specific type of airborne base station is not limited in this embodiment.

[0038] User terminal 103 can be various electronic devices, including but not limited to smartphones, laptops, desktop computers, smart speakers, smartwatches, wearable devices, augmented reality devices, virtual reality devices, etc. Optionally, the user terminal in this embodiment can also be referred to as UE (User Equipment). In specific implementations, the user terminal can be a mobile phone, tablet computer, laptop computer, personal digital assistant (PDA), mobile internet device (MID), wearable device, or in-vehicle device, etc. It should be noted that the specific type of user terminal is not limited in this embodiment of the invention.

[0039] Optionally, the application clients installed on different user terminals 103 may be the same, or clients of the same type of application based on different operating systems. Depending on the user terminal platform, the specific form of the application client may also differ; for example, the application client may be a mobile client, a PC client, etc.

[0040] There is also a network between the airborne base station and the user terminal. The network is used as a medium to provide a communication link between the user terminal and the airborne base station. It can be a wired network or a wireless network.

[0041] Optionally, the aforementioned wireless or wired networks use standard communication technologies and / or protocols. The network is typically the Internet, but can also be any network, including but not limited to Local Area Networks (LANs), Metropolitan Area Networks (MANs), Wide Area Networks (WANs), mobile, wired or wireless networks, private networks, or any combination of virtual private networks. In some embodiments, technologies and / or formats, including Hyper Text Markup Language (HTML), Extensible Markup Language (XML), etc., are used to represent data exchanged over the network. Furthermore, conventional encryption technologies such as Secure Socket Layer (SSL), Transport Layer Security (TLS), Virtual Private Networks (VPNs), and Internet Protocol Security (IPSec) can be used to encrypt all or some links. In other embodiments, custom and / or dedicated data communication technologies can be used to replace or supplement the aforementioned data communication technologies.

[0042] Those skilled in the art will know that Figure 1 The number of user terminals and air base stations shown is merely illustrative; any number of user terminals and air base stations can be used as needed. This disclosure does not limit this.

[0043] In some embodiments of this disclosure, such as Figure 2 As shown, the above-mentioned communication system also includes a high-altitude platform 201.

[0044] The base station deployment server 101 is also used to deploy a deep reinforcement learning model with the goal of maximizing system spectral efficiency on the high-altitude platform; to deploy a deep reinforcement learning model with the goal of maximizing average efficiency on each airborne base station; and to construct a federated learning model based on the high-altitude platform and each airborne base station to obtain the deployment position of each airborne base station in the target three-dimensional space when the system spectral efficiency is maximized.

[0045] Under the above-described communication system architecture, this disclosure provides a method for deploying an airborne base station, which can be executed by any electronic device with computing capabilities.

[0046] In some embodiments, the airborne base station deployment method provided in this disclosure can be executed by the base station deployment server in the above system architecture; in other embodiments, the airborne base station deployment method provided in this disclosure can be implemented by the user terminal and the base station deployment server in the above system architecture through interaction.

[0047] Figure 3 This diagram illustrates a flowchart of an airborne base station deployment method according to an embodiment of the present disclosure, such as... Figure 3 As shown, the method for deploying an airborne base station provided in this embodiment includes the following steps:

[0048] S302, acquire the communication status information of each air base station in the target three-dimensional space, wherein each air base station serves at least one user terminal, and the communication status information of each air base station includes: the deployment location of each air base station in the target three-dimensional space and the spectrum efficiency of each user terminal served by each air base station.

[0049] It should be noted that the target three-dimensional space refers to the area in a three-dimensional space where the airborne base station needs to be deployed. This area is considered the communication target area, containing all devices, users, and airborne base stations that need to conduct wireless communication. The target three-dimensional space is typically a specific area, such as an urban area, a large building, a stadium, or other locations that require wireless services. Within this target three-dimensional space, various wireless devices and users need to communicate; therefore, the airborne base station deployment method analyzes and optimizes the devices and users within this space to improve the efficiency of the entire system.

[0050] An airborne base station is a mobile communication base station device deployed in the air to provide wireless communication services. Compared to traditional terrestrial mobile communication base stations, airborne base stations can cover a wider area and provide better data transmission rates and reliability. Airborne base stations are typically deployed on aircraft, drones, and other airborne objects to achieve wider-range data transmission and communication services. For scenarios where communication needs cannot be met by traditional ground equipment (such as in power lines, mountainous areas, and at sea), airborne base stations provide an effective solution. Furthermore, in disaster relief and special application scenarios, airborne base stations can also serve as temporary communication infrastructure, providing critical communication assurance.

[0051] In some embodiments, the airborne base station is a millimeter-wave airborne base station, which is a wireless communication device using millimeter-wave technology to provide high-speed and high-capacity wireless data transmission services in the air. Millimeter waves refer to radio frequency bands between 30 GHz and 300 GHz, which have higher frequencies and shorter wavelengths, providing higher transmission rates and capacities. Compared with traditional 3G and 4G networks, millimeter-wave technology can provide greater bandwidth and transmission rates, enabling simultaneous high-speed communication for multiple users. Millimeter-wave airborne base stations are typically deployed on aircraft, balloons, and drones at altitudes ranging from hundreds to thousands of meters. Through wireless connections with ground equipment and user terminals, it constructs a wireless communication network with wider coverage and faster transmission speeds, applicable to communication and data transmission in fields such as high-speed trains, intelligent transportation, and mining factories. Millimeter-wave airborne base stations are one of the important directions for the development of future communication networks and are considered one of the key technologies of the "5G+" era, with broad application prospects.

[0052] It's important to explain that communication status information refers to various data and parameter information recorded and fed back to the system during wireless communication regarding the interaction between the airborne base station and the user terminal. For the airborne base station, its communication status information may include information such as the base station's deployment location, the number of users served, and frequency adjustments. For the user terminal, its communication status information may include information such as its location, communication rate, and signal strength. Collecting and utilizing this communication status information during communication helps the system better understand the interaction between the user and the device, thereby improving communication quality and enhancing the efficiency and performance of the entire system.

[0053] Spectral efficiency is a performance metric in communication systems or networks, typically used to describe the amount of data that can be transmitted within a given spectrum bandwidth. Specifically, it represents the amount of data transmitted per unit of spectrum (usually Hertz per second), typically expressed in bits per Hertz (bps / Hz) or minutes (e.g., Mbps, Gbps). Therefore, higher spectral efficiency means more information can be transmitted per unit of time; conversely, lower spectral efficiency means less information can be transmitted per unit of time. Spectrum utilization is also an important factor to consider in communication system optimization, requiring careful consideration of the allocation and utilization of various resources within the communication system to achieve higher spectral efficiency.

[0054] S304, based on a pre-built deep reinforcement learning model aimed at maximizing system spectral efficiency, determines the deployment location of each air base station in the target three-dimensional space when the system spectral efficiency is maximized. Here, the system spectral efficiency is the sum of the average spectral efficiencies of all air base stations in the target three-dimensional space, and the average spectral efficiency of each air base station is the average of the spectral efficiencies of the multiple user terminals served by each air base station.

[0055] It's important to note that the pre-built deep reinforcement learning model, aimed at maximizing system spectral efficiency, is an artificial intelligence algorithm used to optimize communication system performance. This model uses reinforcement learning to learn how to optimize the layout of airborne base stations within a target three-dimensional space to maximize the overall system's spectral efficiency. In this model, the system continuously trials and evaluates different base station layout schemes to determine the optimal location for each base station in a given target area. This evaluation process is based on the reward mechanism in reinforcement learning; the system collects and analyzes data from base stations and user terminals, evaluates the spectral efficiency of each scheme, and provides rewards or penalties based on these results to help the model optimize the overall system's spectral utilization efficiency. The depth of the model lies in its reliance on deep learning networks to process and analyze complex data and patterns in order to more accurately predict the optimal scheme.

[0056] Specifically, in some embodiments of this disclosure, the construction of a deep reinforcement learning model involves the following steps:

[0057] Reward Function Design: To maximize system spectral efficiency, we need to design a reward function for the agent in the deep reinforcement learning model, where the agent is the airborne base station. Considering system spectral efficiency as the objective, the reward function is designed as the average spectral efficiency of the airborne base station in the current state. That is, it calculates the average spectral efficiency of multiple user terminals served by the airborne base station based on the current state, and returns this average as the reward value to the deep reinforcement learning model.

[0058] State space and action space definition: In a deep reinforcement learning model, the state represents the current position of the airborne base station, and the action represents the movable direction of the airborne base station. For a drone-based airborne base station, there are seven actions: forward, backward, left, right, up, down, and hover.

[0059] Establishing an interactive environment for reinforcement learning agents: For the deployment environment of airborne base stations, we can create a simulation environment to simulate the movement of airborne base stations in three-dimensional space, and place agents in this environment to interact with them. At each time step, the agent selects an action based on the current state, and the environment returns a new state and reward value to the agent. The agent uses the current state, the selected action, and the obtained reward to update its policy network, enabling it to better maximize the system's spectral efficiency when taking actions in the future.

[0060] Deep Reinforcement Learning Model Training: For the task of deploying airborne base stations, we can use deep reinforcement learning models such as deep Q-networks for training. During training, the agent continuously updates its policy network by exploring and taking actions, enabling it to select the optimal action for the current state. Simultaneously, the agent's policy is evaluated and adjusted based on the reward function, thereby maximizing the system's spectral efficiency.

[0061] It's also important to clarify that system spectral efficiency refers to the sum of the average spectral efficiencies of all airborne base stations within the target three-dimensional space. The average spectral efficiency of each airborne base station, however, is the average of the spectral efficiencies of the multiple user terminals served by that base station. This value represents the efficiency of a base station in providing service within the target area, and it is influenced by many factors, including the base station's deployment location, channel conditions, signal interference, modulation scheme, bit error rate, and so on.

[0062] This embodiment of the disclosure obtains the communication status information of each air base station in the target three-dimensional space, including the deployment location of each air base station in the target three-dimensional space and the spectral efficiency of each user terminal served by each air base station. Based on a pre-built deep reinforcement learning model with the goal of maximizing system spectral efficiency, the deployment location of each air base station in the target three-dimensional space when the system spectral efficiency is maximized is determined based on the obtained communication status information. This allows for a faster and more efficient finding of the optimal deployment scheme, improving system spectral efficiency and optimizing the performance and reliability of the wireless communication system.

[0063] It should be noted that the acquisition, storage, use, and processing of data in this disclosed technical solution comply with the relevant provisions of national laws and regulations. The various types of data, such as personal identity data, operational data, and behavioral data related to individuals, customers, and groups, obtained in the embodiments of this disclosure have all been authorized.

[0064] In some embodiments of this disclosure, such as Figure 4 As shown, based on a pre-built deep reinforcement learning model that aims to maximize system spectral efficiency, the deployment location of each airborne base station in the target three-dimensional space is determined when the system spectral efficiency is maximized. This also includes the following steps:

[0065] S402 deploys a deep reinforcement learning model on a high-altitude platform with the goal of maximizing system spectral efficiency.

[0066] It should be noted that a High Altitude Platform (HAP) refers to a long-endurance UAV platform located at an altitude of 17 to 22 km in the stratosphere, capable of providing multi-purpose communication payloads over a large coverage area. HAPs play a crucial role in integrated air-to-ground network architectures. Compared to traditional terrestrial cellular networks, HAP channels offer advantages such as high channel capacity, wide coverage area, and low free-space fading. This is because HAPs have a line-of-sight channel with serving users and are less affected by building obstruction and rain attenuation.

[0067] The primary purpose of deploying deep reinforcement learning models aimed at maximizing system spectral efficiency on high-altitude platforms is to assist in determining the optimal deployment location for each airborne base station, thereby further improving the overall spectral efficiency of the communication system. Specifically, high-altitude platforms offer the following advantages:

[0068] High-altitude platforms can provide more accurate and comprehensive environmental perception information. Within the target's three-dimensional space, signal reflection and building obstruction can affect the quality of wireless communication, thus impacting spectrum efficiency. Because high-altitude platforms are deployed outside the atmosphere, they interact less with signals on the ground, making it easier to acquire more accurate and comprehensive environmental perception information. This allows them to accurately locate the optimal deployment position for each aerial base station, thereby improving spectrum efficiency.

[0069] High-altitude platforms can cover a wider area. Due to their higher altitude, they can more easily cover large areas within a certain range, including sparsely populated areas that are often difficult for ground base stations to cover. Therefore, deploying deep reinforcement learning models aimed at maximizing system spectral efficiency using high-altitude platforms in these areas can provide more comprehensive and extensive support and assurance for the entire communication system.

[0070] S404 deploys a deep reinforcement learning model on each airborne base station with the goal of maximizing average efficiency.

[0071] The main purpose of deploying a deep reinforcement learning model with the goal of maximizing average efficiency on each airborne base station is to enable each airborne base station, as an agent of the deep reinforcement learning model, to have autonomous adaptability and dynamically adjust its deployment position according to changes in the environment and the needs of the communication network, thereby further improving the efficiency of the entire communication system.

[0072] S406, based on the high-altitude platform and each airborne base station, constructs a federated learning model to obtain the deployment position of each airborne base station in the target three-dimensional space when the system spectral efficiency is maximized.

[0073] It's important to note that federated learning is an emerging machine learning model designed to address data protection and privacy concerns when training models on distributed devices. Federated learning allows data from different geographical locations and devices to be processed and trained locally, rather than centralizing training in one place, thus better protecting data privacy.

[0074] Federated learning utilizes distributed computing and deep learning technologies to distribute the model training process across different airborne base stations. Each device trains on a portion of the data and feeds the training results back to the high-altitude platform for aggregation. In this process, the data does not leave the airborne base stations; only the model parameters are exchanged between the airborne base stations and the high-altitude platform, thus protecting data privacy.

[0075] In summary, deploying a deep reinforcement learning model aimed at maximizing system spectral efficiency on a high-altitude platform can provide more accurate, comprehensive, and extensive environmental awareness information, helping to determine the optimal deployment location for each airborne base station, thereby improving the overall spectral efficiency of the communication system. Furthermore, a federated learning model is constructed between the high-altitude platform and the airborne base stations. Local nodes (airborne base stations) upload model parameters to the global node (high-altitude platform), which then aggregates the parameters and sends them back to each local node. During this process, different local nodes do not exchange information or transmit sensitive user information, thus protecting user information from different operators.

[0076] In some embodiments of this disclosure, such as Figure 5 As shown, a federated learning model is constructed based on the high-altitude platform and each airborne base station to obtain the deployment position of each airborne base station in the target three-dimensional space when the system's spectral efficiency is maximized. The model also includes the following steps:

[0077] S502 uploads the local model parameters of each air base station to the high-altitude platform, so that the high-altitude platform can train a deep reinforcement learning model with the goal of maximizing the system's spectral efficiency based on the local model parameters uploaded by each air base station.

[0078] It's important to note that in federated learning, each airborne base station possesses its own local training data and trains its deep reinforcement learning model locally. However, to utilize the training data from all airborne base stations, these devices need to collaboratively train a universal model in a distributed manner. Transmitting raw data is impractical and insecure to ensure user privacy and data security. Therefore, we need to perform training without leaking local device data.

[0079] A federated learning model is used to construct a deep reinforcement learning model to complete the deployment and training of all airborne base stations. The local model parameters for each airborne base station are uploaded to the high-altitude platform. The high-altitude platform uses these parameters to train a deep reinforcement learning model that aims to maximize system spectral efficiency. Because only local model parameters are uploaded, instead of raw data, the data and privacy of each airborne base station are protected. Finally, the high-altitude platform distributes the trained global model parameters to each airborne base station, where they are applied on the local machine to complete model training.

[0080] Therefore, a federated learning model is used to protect the local training data of the airborne base stations and to train the global deep reinforcement learning model. This avoids the risks of uploading raw data and fully utilizes the data from each airborne base station to improve the system's spectral efficiency.

[0081] S504 distributes the global model parameters trained by the high-altitude platform to each airborne base station.

[0082] In the federated learning model, each device trains locally to obtain local model parameters. These local model parameters are then uploaded to the high-altitude platform for global model training. Once the high-altitude platform has trained the global model parameters, these parameters are distributed to each airborne base station so that they can use these parameters locally to complete model training.

[0083] The high-altitude platform distributes the trained global model parameters to each airborne base station. These parameters include network weights, learning rates, and other information. This distribution is done periodically to ensure that the model parameters on each device are updated, while also protecting data privacy, as the original data is not exposed during transmission. This allows each airborne base station to autonomously adjust its deployment location to adapt to different communication environments, thereby achieving high-efficiency operation of the communication system.

[0084] This method uses a federated learning model to train data distributed across different airborne base stations and leverages a high-altitude platform to train a global model. By distributing the global model parameters trained on the high-altitude platform to each airborne base station, each base station can access the global model information and complete its local model training. This improves the efficiency and reliability of the communication system while protecting user data privacy.

[0085] Specifically, in some embodiments of this disclosure, the federated learning model is introduced into the deep reinforcement learning model. Each airborne base station agent periodically uploads network parameters and receives aggregated model parameters. The agents do not communicate with each other to protect user privacy. In the federated learning architecture, the high-altitude platform is selected as the global node for federated learning, and each airborne base station is a local node.

[0086] The parameter aggregation formula for federated learning is as follows:

[0087] (1)

[0088] in These are the Q network parameters of the high-altitude platform. These are the Q network parameters for each airborne base station, and D is the total number of users. It represents the number of users served by the m-th airborne base station.

[0089] Federated deep reinforcement learning training mainly consists of the following steps:

[0090] Each airborne base station has a locally deployed reinforcement learning model that makes independent decisions regarding deployment in 3D space. A high-altitude platform is deployed as a global node for federated learning, with the airborne base stations uploading their model parameters to the high-altitude platform.

[0091] The high-altitude platform performs federated aggregation and sends the aggregated global deployment model parameters back to each airborne base station.

[0092] Each airborne base station updates its locally deployed model and continues to train the model using local data.

[0093] Repeat the above steps to train the model until it converges.

[0094] The specific training process of the federated deep Q-network learning algorithm is as follows:

[0095] 1) First, initialize the experience replay pool D, with a capacity of N;

[0096] 2) Initialize the Q-network and its neural network parameters ω; initialize the target Q-network and its neural network parameters ω−; initialize the aggregation frequency F of the federated reinforcement learning;

[0097] 3) Loop through all rounds, episode = 1, 2, ..., M;

[0098] 3.1) Initialize the state set S;

[0099] 3.2) Iterate through the loop step=1,2,…,T:

[0100] 3.2.1) Using The greedy strategy adopts action strategy A;

[0101] 3.2.2) Execute action A, calculate the reward R for the system to take action A in state S, and the system will reach the new base station planning state S' in the next time step to obtain the reward and the new state S';

[0102] 3.2.3) Store the sample (S,A,R,S') into the experience replay pool D;

[0103] 3.2.4) Calculate the target Q value using uniformly random samples from the experience replay pool, and update the Q network parameters ω to reduce the loss function;

[0104] 3.2.5) Update the parameters of the target Q network for base station planning every C steps;

[0105] 3.3) Based on the aggregation frequency F of federated reinforcement learning, when the number of iterations (episodes) is divisible by F, the model of each agent is uploaded, aggregated, and transmitted back, and the Q network parameters are updated.

[0106] 4) The algorithm ends.

[0107] In real-world applications, different airborne base stations may come from different operators. These base stations do not want to share user information with other airborne base stations. Instead, they choose a high-altitude platform that is recognized by each operator to aggregate model parameters and then distribute the aggregated global model to each airborne base station. Federated learning accelerates training convergence and can solve the problem of insufficient data. For example, there may be situations that agent A has not encountered, but agent B has. Periodically uploading, aggregating, and downloading models can improve learning efficiency.

[0108] In some embodiments of this disclosure, after determining the deployment location of each airborne base station in the target three-dimensional space when the system spectral efficiency is maximized based on a pre-built deep reinforcement learning model aimed at maximizing system spectral efficiency, the method further includes:

[0109] Based on the deployment position of each air base station in the target three-dimensional space when the system spectral efficiency is maximized, the motion system of each air base station is activated to change the deployment position of each air base station in the target three-dimensional space.

[0110] Based on the determined optimal deployment position of each airborne base station in the target three-dimensional space when the system's spectral efficiency is maximized, a motion system is activated for each airborne base station to change its deployment position within the target three-dimensional space. This means that airborne base stations can achieve adaptive deployment to adapt to real-time changes in communication needs and the environment. For example, in densely populated areas, airborne base stations need to move to locations with higher system spectral efficiency to improve the quality of communication services.

[0111] Changes to the motion system should be subject to certain limitations to ensure safety and stability. For example, when the base station is in motion, it is necessary to stably maintain the continuity of signal transmission and communication, and avoid risks such as collisions with other aircraft or buildings.

[0112] This embodiment of the disclosure uses a deep reinforcement learning model to determine the deployment position of each airborne base station in the target three-dimensional space when the system's spectral efficiency is maximized. Next, based on these deployment positions, the airborne base stations initiate a motion system to change their deployment positions, enabling them to adapt to different communication needs and environments, thereby achieving high efficiency and stability of the communication system. This method is real-time and adaptive, allowing for timely adjustments when communication needs and environments change, thus better meeting user requirements.

[0113] In some embodiments of this disclosure, such as Figure 6 As shown, obtaining the communication status information of each airborne base station within the target's three-dimensional space also includes the following steps:

[0114] S602, acquire the deployment location data of each air base station in the target three-dimensional space. The deployment location data includes: the height of each air base station and the horizontal distance from the air base station to the user terminal it serves.

[0115] The deployment location data of each airborne base station within the target three-dimensional space is obtained, including the altitude of each airborne base station and the horizontal distance from the airborne base station to the served user terminal. This location data is the basis for determining the deployment location of each airborne base station when the system's spectral efficiency is maximized, and is the basis for frequency planning and airborne base station adjustments.

[0116] S604 calculates the spectral efficiency of each user terminal based on the altitude of each air base station and the horizontal distance from the air base station to the served user terminal.

[0117] After obtaining the altitude and horizontal distance of each airborne base station, the spectral efficiency of each user terminal is calculated. This helps us determine the communication quality of users within the target area and identify whether adjustments to the network deployment are needed to optimize data transmission and reception quality.

[0118] By acquiring the deployment location data of each airborne base station within the target three-dimensional space and calculating the spectral efficiency of each user terminal based on this data, communication status information of each airborne base station within the target area can be obtained. This information will be used for more accurate and efficient frequency planning, helping to adjust the deployment location of each airborne base station to meet ever-changing communication needs and achieve a highly efficient and reliable communication system.

[0119] Specifically, the altitude of the millimeter-wave aerial base station is , For usersi The horizontal distance from the vertical projection of the airborne base station on the ground is the distance for millimeter-wave users. i The elevation angle to the airborne base station can be expressed as:

[0120] (2)

[0121] The line-of-sight link probability between an aerial base station and a ground user is the user's i The elevation angle to the airborne base station and the environmental parameters of the airborne base station can be expressed as a function:

[0122] (3)

[0123] in, a and b It depends on environmental parameters. Similarly, non-line-of-sight (LOS) NLoS The probability can be derived from... LoS The probabilities are derived as follows:

[0124] (4)

[0125] Path loss can be calculated using the following formula:

[0126] (5)

[0127] (6)

[0128] in, It is by Given a fixed path loss, The frequency of millimeter waves used in airborne base stations. and Let be a log-normal random variable, and respectively represent... LoS and NLoS Shadow effects in the scene. and They are LoS and NLoS Path loss index in the scenario. For users i Distance to the airborne base station.

[0129] (7)

[0130] By airborne base station j Active users of network services i The received signal power is expressed as RSRP ,in, b i It is the total bandwidth allocated to users i That part. Besides, a useri Total received noise power N i It consists of two parts: thermal noise power and user equipment noise power, and can be expressed as:

[0131] (8)

[0132] Where ρ i It is the noise figure of the user equipment, determined by the over-the-air base station. j Users providing services i The signal-to-noise ratio (SINR) can be expressed as:

[0133] (9)

[0134] According to Shannon's theorem, from the airborne base station j Users providing services i The spectral efficiency (SE) can be expressed by the formula:

[0135] (10)

[0136] user i The spectral efficiency can be expressed as:

[0137] (11)

[0138] Based on the above formula, we establish the following optimization problem model for maximizing the average spectrum efficiency for all users, with the optimization variables being the three-dimensional deployment location of the airborne base station and the user association selection.

[0139]

[0140] Among them, constraints ( a )-( d The three-dimensional coordinates of the aerial and station deployment locations are constrained and cannot exceed the area boundary; constraints ( e )-( g The user association constraint is the minimum proportion of users that need to be served; constraint ( h )and( i ) represents the user's QoS constraints. RSRP and SINR They need to be higher than the threshold respectively and .

[0141] In some embodiments of this disclosure, the above formula can also be used to construct a spectrum efficiency model, construct a deep reinforcement learning model based on the spectrum efficiency model, add federated learning on the basis of the deep reinforcement learning model, and deploy an airborne base station through the deep reinforcement learning model.

[0142] Based on the same inventive concept, this disclosure also provides an airborne base station deployment device, as described in the following embodiments. Since the principle by which this device embodiment solves the problem is similar to that of the above-described method embodiments, the implementation of this device embodiment can refer to the implementation of the above-described method embodiments, and repeated details will not be elaborated further.

[0143] Figure 7 This diagram illustrates an airborne base station deployment device according to an embodiment of the present disclosure, such as... Figure 6 As shown, the airborne base station deployment device 70 includes:

[0144] The data acquisition module 701 is used to acquire the communication status information of each air base station in the target three-dimensional space. Each air base station serves at least one user terminal. The communication status information of each air base station includes: the deployment location of each air base station in the target three-dimensional space and the spectral efficiency of each user terminal served by each air base station.

[0145] The deployment location determination module 702 is used to determine the deployment location of each air base station in the target three-dimensional space when the system spectral efficiency is maximized, based on a pre-built deep reinforcement learning model with the goal of maximizing system spectral efficiency. Here, the system spectral efficiency is the sum of the average spectral efficiencies of all air base stations in the target three-dimensional space, and the average spectral efficiency of each air base station is the average of the spectral efficiencies of the multiple user terminals served by each air base station.

[0146] It should be noted that the examples and application scenarios implemented by the above modules and corresponding steps are the same, but are not limited to the content disclosed in the above method embodiments. It should also be noted that the above modules, as part of an apparatus, can be executed in a computer system such as a set of computer-executable instructions.

[0147] Those skilled in the art will understand that various aspects of this disclosure can be implemented as a system, method, or program product. Therefore, various aspects of this disclosure can be specifically implemented in the following forms: a completely hardware implementation, a completely software implementation (including firmware, microcode, etc.), or a combination of hardware and software aspects, collectively referred to herein as a "circuit," "module," or "system."

[0148] The following reference Figure 8 To describe an electronic device 800 according to such an embodiment of the present disclosure. Figure 8 The electronic device 800 shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments disclosed herein.

[0149] like Figure 8As shown, the electronic device 800 is manifested in the form of a general-purpose computing device. The components of the electronic device 800 may include, but are not limited to: at least one processing unit 810, at least one storage unit 820, and a bus 830 connecting different system components (including storage unit 820 and processing unit 810).

[0150] The storage unit stores program code that can be executed by the processing unit 810, causing the processing unit 810 to perform the steps described in the "Exemplary Methods" section of this specification according to various exemplary embodiments of this disclosure. For example, the processing unit 810 can perform the following steps of the above method embodiments: obtaining communication status information of each airborne base station in the target three-dimensional space, wherein each airborne base station serves at least one user terminal, and the communication status information of each airborne base station includes: the deployment location of each airborne base station in the target three-dimensional space and the spectral efficiency of each user terminal served by each airborne base station; determining the deployment location of each airborne base station in the target three-dimensional space when the system spectral efficiency is maximized based on a pre-constructed deep reinforcement learning model aimed at maximizing system spectral efficiency, wherein the system spectral efficiency is the sum of the average spectral efficiencies of all airborne base stations in the target three-dimensional space, and the average spectral efficiency of each airborne base station is the average of the spectral efficiencies of the multiple user terminals served by each airborne base station.

[0151] Storage unit 820 may include a readable medium in the form of a volatile storage unit, such as random access memory (RAM) 8201 and / or cache memory 8202, and may further include a read-only memory (ROM) 8203.

[0152] The storage unit 820 may also include a program / utility 8204 having a set (at least one) of program modules 8205, such program modules 8205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each or some combination of these examples may include an implementation of a network environment.

[0153] Bus 830 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the various bus structures.

[0154] Electronic device 800 can also communicate with one or more external devices 840 (e.g., keyboard, pointing device, Bluetooth device, etc.), and with one or more devices that enable a user to interact with electronic device 800, and / or with any device that enables electronic device 800 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 850. Furthermore, electronic device 800 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 860. As shown, network adapter 860 communicates with other modules of electronic device 800 via bus 830. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 800, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0155] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, terminal device, or network device, etc.) to execute the methods according to the embodiments of this disclosure.

[0156] In particular, according to embodiments of this disclosure, the process described above with reference to the flowchart can be implemented as a computer program product, which includes a computer program that, when executed by a processor, implements the above-described airborne base station deployment method.

[0157] In exemplary embodiments of this disclosure, a computer-readable storage medium is also provided, which may be a readable signal medium or a readable storage medium. Figure 9 This illustration shows a schematic diagram of a computer-readable storage medium according to an embodiment of the present disclosure, such as... Figure 9 As shown, the computer-readable storage medium 900 stores a program product capable of implementing the methods described above in this disclosure. In some possible embodiments, various aspects of this disclosure may also be implemented as a program product comprising program code that, when the program product is run on a user terminal, causes the user terminal to perform the steps described in the "Exemplary Methods" section of this specification according to various exemplary embodiments of this disclosure.

[0158] More specific examples of computer-readable storage media in this disclosure may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0159] In this disclosure, a computer-readable storage medium may include a data signal propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium, capable of transmitting, propagating, or transmitting a program for use by or in connection with an instruction execution system, apparatus, or device.

[0160] Optionally, the program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0161] In practical implementation, program code for performing the operations of this disclosure can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0162] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to embodiments of this disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0163] Furthermore, although the steps of the method in this disclosure are described in a specific order in the accompanying drawings, this does not require or imply that the steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps may be omitted, multiple steps may be combined into one step, and / or a step may be broken down into multiple steps.

[0164] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, mobile terminal, or network device, etc.) to execute the methods according to the embodiments of this disclosure.

[0165] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the appended claims.

Claims

1. A method for deploying an aerial base station, characterized in that, include: The communication status information of each air base station in the target three-dimensional space is obtained, wherein each air base station serves at least one user terminal, and the communication status information of each air base station includes: the deployment location of each air base station in the target three-dimensional space and the spectral efficiency of each user terminal served by each air base station. Deploy a deep reinforcement learning model on a high-altitude platform with the goal of maximizing system spectral efficiency; Deploy a deep reinforcement learning model on each airborne base station with the goal of maximizing average spectral efficiency; A federated learning model is constructed based on the high-altitude platform and each airborne base station to obtain the deployment position of each airborne base station in the target three-dimensional space when the system spectral efficiency is maximized; wherein, the system spectral efficiency is the sum of the average spectral efficiency of all airborne base stations in the target three-dimensional space, and the average spectral efficiency of each airborne base station is the average of the spectral efficiency of multiple user terminals served by each airborne base station; The construction process of the federated learning model includes: each air base station uses local training data to train a locally deployed deep reinforcement learning model with the goal of maximizing average spectral efficiency; the local model parameters of each air base station are uploaded to the high-altitude platform, so that the high-altitude platform can train the deep reinforcement learning model with the goal of maximizing system spectral efficiency based on the local model parameters uploaded by each air base station; and the global model parameters trained by the high-altitude platform are distributed to each air base station, so that each air base station can update the locally deployed deep reinforcement learning model with the goal of maximizing average spectral efficiency using the global model parameters.

2. The method for deploying an aerial base station according to claim 1, characterized in that, After constructing a federated learning model based on the high-altitude platform and each airborne base station to obtain the deployment position of each airborne base station in the target three-dimensional space when the system spectral efficiency is maximized, the method further includes: Based on the deployment position of each air base station in the target three-dimensional space when the system spectral efficiency is maximized, the motion system of each air base station is activated to change the deployment position of each air base station in the target three-dimensional space.

3. The method for deploying an aerial base station according to claim 1, characterized in that, Obtain communication status information for each airborne base station within the target's three-dimensional space, including: Acquire deployment location data for each airborne base station within the target three-dimensional space, the deployment location data including: the altitude of each airborne base station and the horizontal distance from the airborne base station to the user terminal it serves; The spectral efficiency of each user terminal is calculated based on the altitude of each air base station and the horizontal distance from the air base station to the served user terminal.

4. A communication system, characterized in that, include: The base station deployment includes a server, multiple airborne base stations, and at least one user terminal served by each airborne base station. The base station deployment server is used to acquire communication status information of each airborne base station in the target three-dimensional space, wherein each airborne base station serves at least one user terminal, and the communication status information of each airborne base station includes: the deployment location of each airborne base station in the target three-dimensional space and the spectral efficiency of each user terminal served by each airborne base station; deploying a deep reinforcement learning model on a high-altitude platform with the goal of maximizing system spectral efficiency; deploying a deep reinforcement learning model on each airborne base station with the goal of maximizing average spectral efficiency; constructing a federated learning model based on the high-altitude platform and each airborne base station to obtain the deployment location of each airborne base station in the target three-dimensional space when the system spectral efficiency is maximized, wherein the system spectral efficiency is the sum of the average spectral efficiency of all airborne base stations in the target three-dimensional space, and the average spectral efficiency of each airborne base station is the average of the spectral efficiencies of the multiple user terminals served by each airborne base station; The construction process of the federated learning model includes: each air base station uses local training data to train a locally deployed deep reinforcement learning model with the goal of maximizing average spectral efficiency; the local model parameters of each air base station are uploaded to the high-altitude platform, so that the high-altitude platform can train the deep reinforcement learning model with the goal of maximizing system spectral efficiency based on the local model parameters uploaded by each air base station; and the global model parameters trained by the high-altitude platform are distributed to each air base station, so that each air base station can update the locally deployed deep reinforcement learning model with the goal of maximizing average spectral efficiency using the global model parameters.

5. An aerial base station deployment device, characterized in that, include: The data acquisition module is used to acquire the communication status information of each air base station in the target three-dimensional space. Each air base station serves at least one user terminal. The communication status information of each air base station includes: the deployment location of each air base station in the target three-dimensional space and the spectral efficiency of each user terminal served by each air base station. The deployment location determination module is used to deploy a deep reinforcement learning model on a high-altitude platform with the goal of maximizing system spectral efficiency; deploy a deep reinforcement learning model on each airborne base station with the goal of maximizing average spectral efficiency; and construct a federated learning model based on the high-altitude platform and each airborne base station to obtain the deployment location of each airborne base station in the target three-dimensional space when the system spectral efficiency is maximized. The system spectral efficiency is the sum of the average spectral efficiencies of all airborne base stations in the target three-dimensional space, and the average spectral efficiency of each airborne base station is the average of the spectral efficiencies of multiple user terminals served by each airborne base station. The construction process of the federated learning model includes: each air base station uses local training data to train a locally deployed deep reinforcement learning model with the goal of maximizing average spectral efficiency; the local model parameters of each air base station are uploaded to the high-altitude platform, so that the high-altitude platform can train the deep reinforcement learning model with the goal of maximizing system spectral efficiency based on the local model parameters uploaded by each air base station; and the global model parameters trained by the high-altitude platform are distributed to each air base station, so that each air base station can update the locally deployed deep reinforcement learning model with the goal of maximizing average spectral efficiency using the global model parameters.

6. An electronic device, characterized in that, include: processor; as well as Memory for storing the executable instructions of the processor; The processor is configured to execute the airborne base station deployment method according to any one of claims 1 to 3 by executing the executable instructions.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the airborne base station deployment method according to any one of claims 1 to 3.

Images