Feedback-free transmission and dynamic resource allocation method for fully-decoupled space-air-ground integrated network

Abstract

Disclosed in the present invention is a feedback-free transmission and dynamic resource allocation method for a fully-decoupled space-air-ground integrated network, which provides a solution for resource management of space-air-ground multi-node cooperation. The method comprises: on the basis of a deep learning-based channel state information prediction model, replacing a traditional feedback signal by means of user geographic location information, so as to implement multiple-input multiple-output transmission; and designing a many-to-one matching model for enabling adaptive allocation of spectrum resource blocks between space-air-ground heterogeneous nodes, so as to ensure matching stability and have the characteristics of low complexity and fast convergence. For cooperative transmission of multiple nodes and resource allocation between the heterogeneous nodes, the spectrum efficiency of a system is improved by means of feedback-free channel state information prediction and a resource allocation mechanism. Compared with a single-connection mode and polling-based resource scheduling in a traditional network, the flexible resource allocation algorithm of the present invention significantly increases the network capacity and improves user communication quality.

Description

A method for feedback-free transmission and dynamic resource allocation in fully decoupled air-space-ground integrated networks Technical Field

[0001] This invention belongs to the field of fully decoupled network transmission mechanism and resource allocation for integrated air-space-ground networks, and relates to a method for feedback-free transmission and dynamic resource allocation in fully decoupled networks for integrated air-space-ground networks. Background Technology

[0002] In recent years, space-air-ground integrated networks (SAGIN) have gradually become a key paradigm for achieving ubiquitous connectivity, high capacity, and strong reliability in 6G. Satellite projects, such as Starlink, have launched products like the STARLINK WiFi router, providing users with uplink speeds of 8-25 Mbps and downlink speeds of 40-220 Mbps. This system receives signals from low-Earth orbit satellites and then converts the signals into WiFi for user terminals. Furthermore, unmanned aerial vehicles (UAVs) play a crucial role in the low-altitude economy, providing enhancement and support services for terrestrial communications. However, integrating these heterogeneous nodes into a unified network presents challenges in resource allocation and coordination between satellites, UAVs, and ground base stations. According to the 3rd Generation Partnership Project (3GPP) non-terrestrial networks (NTN) standard, satellites use frequency division duplex (FDD) mode, while terrestrial 5G networks use time division duplex (TDD) mode. The difference between the two exacerbates the complexity of network integration, highlighting the need to adapt terrestrial networks and develop new technologies to achieve heterogeneous integration of SAGIN.

[0003] From 4G to 5G, the core network architecture has achieved the separation of the control plane and user plane, improving network flexibility. However, in the 5G radio access network (RAN), each base station still needs to be configured with control and user plane functions. In 2019, Academician Yu Quan's team published a paper titled "A Fully-Decoupled RAN Architecture for 6G Inspired by Neurotransmission" in the *Journal of Communications and Information Networks*, which describes a fully-decoupled radio access network (FD-RAN) where base stations are physically decoupled into control base stations, uplink base stations, and downlink base stations. This decoupling design allows for flexible allocation of spectrum resources and promotes more efficient collaboration among heterogeneous nodes. Against this backdrop, this paper proposes an integrated air-space-ground FD-RAN architecture, enabling terrestrial network infrastructure to adopt FD-RAN and combining it with satellite and UAV nodes. Achieving flexible collaboration and resource management among these heterogeneous nodes remains a key challenge.

[0004] A review of existing literature revealed few studies proposing a base station feedback-free mechanism for fully decoupled air-space-ground integrated networks, or dynamic resource joint scheduling of base stations, satellite WiFi router nodes, and drones. Kai Yu et al., in their paper "Fully-Decoupled Radio Access Networks: A Flexible Downlink Multi-Connectivity and Dynamic Resource Cooperation Framework" published in *IEEE Transactions on Wireless Communications*, proposed a multi-base station joint transmission algorithm that requires user feedback of the complete channel. However, this paper did not consider the overhead and real-time requirements of feedback of the complete channel, especially since uplink and downlink transmissions occur through different base stations after base station decoupling, making it impossible to utilize channel heterogeneity to obtain the complete downlink transmission channel.

[0005] In summary, the existing technologies have the following problems: (1) The traditional downlink multiple-input multiple-output (MIMO) transmission method based on channel feedback is no longer applicable to FD-RAN, and a feedback-free MIMO downlink transmission method needs to be redesigned. (2) Unlike the user-access node single-connection mode and the spectrum resource block (RB) allocation method based on polling scheduling in existing mobile communication networks, the multi-user-multi-node-multi-resource block allocation in the integrated air-space-ground FD-RAN is a very large-scale non-convex integer programming problem, which requires the design of a low-complexity 0-1 integer programming algorithm. The significance of solving the above technical problems is that the integrated air-space-ground network is one of the typical application scenarios of 6G, and the proposed solution helps to lay out the 6G network technology reserves and provides theoretical guidance and reference for 6G architecture and algorithm design. Summary of the Invention

[0006] Purpose of the invention: In view of the problems existing in the prior art, the purpose of this invention is to provide a method for feedbackless transmission and dynamic resource allocation in a fully decoupled network for integrated air-space-ground systems, realize a feedbackless MIMO transmission mechanism for downlink base stations, and enable flexible collaboration and dynamic allocation of spectrum resource blocks among multiple heterogeneous air-space-ground nodes.

[0007] Technical solution: To achieve the above-mentioned objectives, the present invention adopts the following technical solution:

[0008] A method for feedback-free transmission and dynamic resource allocation in a fully decoupled air-space-ground integrated network includes the following steps:

[0009] Step 1: The edge cloud collects historical channel state information (CSI) data, including user location and corresponding spectrum resource block CSI information;

[0010] Step 2: For each spectrum resource block of each downlink base station, the edge cloud uses a neural network to train historical CSI data. The input is the user location, and the output is CSI information.

[0011] Step 3: Control the base station to collect user access requests and real-time geographical location, and feed them back to the edge cloud;

[0012] Step 4: The edge cloud dynamically allocates resources based on the user-heterogeneous node-spectrum resource block allocation algorithm of the many-to-one matching model, and sends the user location and resource block allocation decision to the downlink base station;

[0013] Step 5: The downlink base station predicts CSI information on the resource block to be transmitted;

[0014] Step 6: The downlink base station performs feedbackless MIMO transmission based on the closed-loop spatial multiplexing mode.

[0015] Furthermore, the user-heterogeneous node-spectrum resource block allocation algorithm based on the many-to-one matching model constructs a spectrum resource block allocation problem and realizes dynamic resource allocation based on the many-to-one matching model. The spectrum resource block allocation problem aims to maximize system throughput and ensures through constraints that each user is allocated at least one node's spectrum resource block, each spectrum resource block can only be allocated to one user, and the total transmission rate of WiFi nodes will not exceed the maximum downlink transmission rate of the satellite-to-satellite WiFi router.

[0016] Furthermore, the spectrum resource block allocation problem (RAP) is expressed as:

[0017] Where X, Y, and Z represent the matching matrices from base station (BS)-resource block (RB) pairs, WiFi-RB pairs, and UAV-RB pairs to users, respectively, and B, S, and U represent the number of BS-RB pairs, WiFi-RB pairs, and UAV-RB pairs, respectively. Let R represent the sets of BS-RB pairs, WiFi-RB pairs, and UAV-RB pairs, respectively. B (X) represents the throughput of the BS-RB pair, R S (Y) represents the throughput of the satellite WiFi-RB pair, R U (Z) represents the throughput of the UAV-RB pair, M represents the total number of users, and T S The maximum downlink throughput rate from satellite to WiFi is represented by: X(b,m), a two-dimensional 0-1 variable, indicating whether the BS-RB serves user m on b (1 if yes, 0 otherwise); Y(s,m), a two-dimensional 0-1 variable, indicating whether the WiFi-RB serves user m on s (1 if yes, 0 otherwise); and Z(u,m), a two-dimensional 0-1 variable, indicating whether the UAV-RB serves user m on u (1 if yes, 0 otherwise).

[0018] Furthermore, after the neural network on each spectrum resource block of each downlink base station is trained, the edge cloud sends the trained neural network to each downlink base station.

[0019] Furthermore, the dynamic resource allocation based on the many-to-one matching model involves assigning the most suitable user to each node-resource block pair during the matching initialization process, based on the user preference list. During the matching and exchange phase, the two node-resource block pairs are traversed sequentially, and if the matched users are different, an exchange attempt is initiated. In each round of matching, three matching and exchange strategies are tried for each user, and only the strategy that improves the overall system throughput is accepted.

[0020] Furthermore, the specific steps for achieving dynamic resource allocation based on a many-to-one matching model are as follows:

[0021] Step 4.1: In the algorithm initialization phase, first input the physical layer throughput estimate from the user to each node-resource block pair;

[0022] Step 4.2: Users generate a preference list based on their physical layer throughput estimates, from highest to lowest.

[0023] Step 4.3: Each user initiates a request in turn, requesting the best node-resource block pair;

[0024] Step 4.4: The node-resource block pair accepts the request if and only if no other match is received;

[0025] Step 4.5: Once each user has been assigned a node-resource block pair, the matching and exchange phase begins;

[0026] Step 4.6: In the matching and swapping phase, traverse the two node-resource block pairs in sequence. If the matching users are different, start the swapping attempt.

[0027] Step 4.7: Calculate the current total system throughput;

[0028] Step 4.8: Swap Attempt 1: Swap two nodes - resource block pairs that match the users;

[0029] Step 4.9: Check if the previous user has matched other resource blocks. If so, perform swap attempt 2: make both nodes and resource blocks match the previous user at the same time; otherwise, skip this step.

[0030] Step 4.10: Check if the next user has matched other resource blocks. If so, perform swap attempt 3: make both nodes and resource blocks match the next user at the same time; otherwise, skip this step.

[0031] Step 4.11: Compare swap attempts 1-3 with the initial throughput, and select the optimal strategy to update the match;

[0032] Step 4.12: Perform the next round of matching and swapping until all iterations are completed.

[0033] Furthermore, in the fully decoupled air-space-ground integrated network, users are served simultaneously by multiple heterogeneous air-space-ground nodes, and spectrum resource blocks are dynamically allocated among the heterogeneous nodes; the heterogeneous air-space-ground nodes include UAV nodes, satellite WiFi routers, and downlink base stations of the fully decoupled network.

[0034] Furthermore, downlink base station transmission relies on channel state information to determine the transmission parameters for each resource block, including: rank indicator (RI), channel quality indicator (CQI), and precoding matrix indicator (PMI). The base station determines the specific parameters for downlink MIMO transmission based on these parameters. Referring to the 3GPP standard, the code rate is calculated based on CQI, the number of transport streams is calculated based on RI, and the physical layer throughput estimate is calculated by combining the predicted or feedback CSI.

[0035] Based on the same inventive concept, this invention provides a fully decoupled network system for integrated air-space-ground communication with no feedback transmission and dynamic resource allocation, including an edge cloud, a control base station, and a downlink base station;

[0036] The edge cloud is used to collect historical channel state information (CSI) data, including user location and corresponding spectrum resource block CSI information; for each spectrum resource block of each downlink base station, a neural network is used to train historical CSI data, with user location as input and CSI information as output; and a user-heterogeneous node-spectrum resource block allocation algorithm based on a many-to-one matching model is used to dynamically allocate resources, and user location and resource block allocation decisions are sent to the downlink base station.

[0037] The control base station is used to collect user access requests and real-time geographical location, and feed them back to the edge cloud;

[0038] The downlink base station is used to predict CSI information on the resource block to be transmitted; and to perform feedbackless MIMO transmission based on closed-loop spatial multiplexing mode.

[0039] Beneficial Effects: Compared with existing technologies, the beneficial effects of this invention are as follows: First, the feedback-free transmission and dynamic resource allocation method for fully decoupled air-space-ground integrated networks can effectively improve the capacity and resource utilization of integrated air-space-ground network systems. Second, considering the specific decoupling mode of uplink and downlink networks, this invention proposes a deep learning-based CSI prediction method and a downlink base station feedback-free transmission mechanism. Finally, facing the problem of cooperation and resource allocation between users and multiple heterogeneous air-space-ground access nodes on multiple spectrum resource blocks, this invention proposes a multi-heterogeneous node cooperation and spectrum resource block allocation algorithm based on a many-to-one matching model, which can effectively improve system performance. Compared with the traditional spectrum resource block allocation based on round-robin scheduling and the power-equal distribution single-user-single-node connection method in networks, the flexible multi-node cooperation and efficient dynamic resource allocation algorithm can effectively improve network capacity and enhance the service quality of mobile user communication. Attached Figure Description

[0040] Figure 1 is a scenario diagram of the fully decoupled network feedback-free transmission and dynamic resource allocation method for air-space-ground integration according to an embodiment of the present invention.

[0041] Figure 2 is a flowchart of the CSI prediction and downlink base station feedback-free MIMO transmission mechanism based on deep learning proposed in an embodiment of the present invention.

[0042] Figure 3 is a flowchart of the user-heterogeneous node-spectrum resource block allocation algorithm based on a many-to-one matching model proposed in an embodiment of the present invention. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of this invention clearer, the embodiments of this invention are described in detail below with reference to the accompanying drawings. These embodiments are implemented based on the technical solutions of this invention, providing detailed implementation methods and specific operating procedures. It should be understood that the specific examples described herein are merely illustrative of this invention, but the scope of protection of this invention is not limited to the following embodiments.

[0044] This invention discloses a method for feedbackless transmission and dynamic resource allocation in a fully decoupled air-space-ground integrated network. It mainly involves a deep learning-based CSI prediction and a downlink base station feedbackless MIMO transmission mechanism, as well as a multi-heterogeneous node cooperation and spectrum resource block allocation algorithm based on a many-to-one matching model. A deep learning-based channel state information prediction model is used, replacing traditional feedback signals with user geographic location information to achieve feedbackless MIMO transmission at the downlink base station. In the fully decoupled air-space-ground integrated network, users can be served simultaneously by multiple heterogeneous air-space-ground nodes, and spectrum resource blocks can be dynamically allocated among these nodes to improve system performance.

[0045] The 6G fully decoupled network architecture considered in this embodiment, as shown in Figure 1, consists of multiple downlink base stations (DBS), satellite WiFi routers, drones, and users at multiple random locations. For downlink transmission in the integrated air-space-ground FD-RAN, we define matrices X, Y, and Z to represent the matching status of base station (BS)-resource block (RB) pairs, WiFi-RB pairs, and UAV (UAV)-RB pairs to users, respectively. A two-dimensional 0-1 variable X(b,m) indicates whether BS-RB pair b serves user m; if yes, X(b,m) is 1, otherwise X(b,m) is 0. Similarly, a two-dimensional 0-1 variable Y(s,m) is defined to indicate whether WiFi-RB pair s serves user m; if yes, Y(s,m) is 1, otherwise Y(s,m) is 0. A two-dimensional 0-1 variable Z(u,m) is defined to indicate whether UAV-RB pair u serves user m; if yes, Z(u,m) is 1, otherwise Z(u,m) is 0.

[0046] According to the 3GPP physical layer standard, for feedback-based MIMO transmission mode, i.e., closed-loop spatial multiplexing (CLSM), downlink base station transmission relies on Channel State Information (CSI) to determine the transmission parameters for each Resource Block (RB), including: 1. Rank Indicator (RI): The proposed number of spatially multiplexed data streams, indicating the number of parallel data streams supported by the current channel. 2. Channel Quality Indicator (CQI): The proposed channel coding rate and modulation scheme to adapt to the current channel quality. 3. Precoding Matrix Indicator (PMI): The proposed precoding matrix index, used to select the optimal precoding matrix from the codebook in the terminal design. Based on these parameters, the base station can determine the specific parameters for downlink MIMO transmission. Referring to the 3GPP standard, the code rate can be calculated based on CQI, and the number of transport streams can be calculated based on RI. Therefore, by combining the predicted or feedback CSI, a physical layer throughput estimate can be calculated.

[0047] In the integrated air-space-ground FD-RAN downlink transmission scenario, focusing on the downlink base station, since real-time CSI information feedback from users is unavailable, we propose a deep learning-based CSI prediction and feedback-free MIMO transmission mechanism for the downlink base station, as shown in Figure 2. First, the edge cloud collects historical CSI data, including user location and CSI information for the corresponding spectrum resource block. Next, for each spectrum resource block, the edge cloud trains the historical CSI data using a neural network (such as a fully connected layer neural network or other deep neural networks), with user location as input and CSI information as output. The control base station collects user access requests and real-time geographic location data and feeds it back to the edge cloud. The edge cloud dynamically allocates resources and sends user location and resource block allocation decisions to the downlink base station. Then, the downlink base station predicts CSI information on the resource block to be transmitted. Finally, the downlink base station performs feedback-free MIMO transmission based on a closed-loop spatial multiplexing mode.

[0048] Therefore, the multi-heterogeneous node cooperation and spectrum resource block allocation problem (RAP) of the integrated air-space-ground FD-RAN can be modeled as follows:

[0049] Where B, S, and U represent the number of BS-RB pairs, WiFi-RB pairs, and UAV-RB pairs, respectively. Let R represent the sets of BS-RB pairs, WiFi-RB pairs, and UAV-RB pairs, respectively. B (X) represents the throughput of the BS-RB pair, RS (Y) represents the throughput of the satellite WiFi-RB pair, R U (Z) represents the throughput of the UAV-RB pair, M represents the total number of users, and T S This represents the maximum downlink throughput rate from satellite to WiFi. In the objective function, R... B (X(b,m)) represents the estimated throughput of the physical layer based on CSI when the BS-RB serves user m to user b. The WiFi-RB throughput R is derived from the user's CSI feedback. S (Y(s,m)) and UAV-RB throughput R U (Z(u,m)). Constraint (C1) ensures that each user is assigned to at least one RB on a node to guarantee minimum quality of service. Constraint (C2) ensures that each RB can only be assigned to one user to avoid interference between nodes. Constraint (C3) represents the maximum downlink transmission rate of the satellite-to-satellite WiFi router.

[0050] Considering that the variables in this problem are all 0-1 variables, we adopt a many-to-one matching model algorithm for solving it. This algorithm has the characteristics of low complexity and fast convergence. During the matching initialization process, based on the user preference list, the most suitable user is assigned to each node-resource block pair for connection. In the matching and exchange phase, the two node-resource block pairs are traversed sequentially. If the matched users are different, an exchange attempt is initiated. In each round of matching, for each user, three matching and exchange strategies are tried. Only the strategy that improves the overall throughput of the system is accepted.

[0051] The corresponding algorithm execution flow is shown in Figure 3. The specific steps are as follows:

[0052] Step 1: In the algorithm initialization phase, first input the physical layer throughput estimate from the user to each node-resource block pair.

[0053] Step 2: Users generate a preference list based on their physical layer throughput estimates, from highest to lowest.

[0054] Step 3: Next, each user initiates a request in turn, requesting the best node-resource block pair.

[0055] Step 4: The node-resource block pair accepts the request if and only if no other match has been received.

[0056] Step 5: When resources are sufficient, it means that each user is assigned a node-resource block pair and enters the matching and exchange phase.

[0057] Step 6: In the matching and swapping phase, traverse the two node-resource block pairs in sequence. If the matching users are different, start the swapping attempt.

[0058] Step 7: First, calculate the current total system throughput.

[0059] Step 8: Swap Attempt 1: Swap two nodes - resource block pairs that match the user.

[0060] Step 9: Check if the previous user has matched other resource blocks. If so, perform swap attempt 2: make both nodes and resource blocks match the previous user at the same time. Otherwise, skip this step.

[0061] Step 10: Check if the next user has matched other resource blocks. If so, perform a swap attempt 3: make both nodes and resource blocks match the previous user at the same time. Otherwise, skip this step.

[0062] Step 11: Compare swap attempts 1-3 with the initial throughput, and select the optimal strategy to update the match.

[0063] Step 12: Then proceed to the next round of matching and swapping until all iterations are completed.

[0064] Based on the same inventive concept, this invention discloses a feedback-free transmission and dynamic resource allocation system for a fully decoupled air-space-ground integrated network, comprising an edge cloud, a control base station, and a downlink base station. The edge cloud collects historical Channel State Information (CSI) data, including user location and CSI information for corresponding spectrum resource blocks. For each spectrum resource block of each downlink base station, a neural network is used to train historical CSI data, with user location as input and CSI information as output. A user-heterogeneous node-spectrum resource block allocation algorithm based on a many-to-one matching model is used for dynamic resource allocation, and the user location and resource block allocation decision are sent to the downlink base station. The control base station collects user access requests and real-time geographic locations and feeds them back to the edge cloud. The downlink base station predicts CSI information on the resource blocks to be transmitted and performs feedback-free MIMO transmission based on a closed-loop spatial multiplexing mode.

[0065] The user-heterogeneous node-spectrum resource block allocation algorithm based on a many-to-one matching model constructs a spectrum resource block allocation problem and realizes dynamic resource allocation based on a many-to-one matching model. The spectrum resource block allocation problem aims to maximize system throughput. Through constraints, it ensures that each user is allocated at least one spectrum resource block of a node, each spectrum resource block can only be allocated to one user, and the total transmission rate of WiFi nodes will not exceed the maximum downlink transmission rate of the satellite-to-satellite WiFi router.

[0066] For details of specific implementation methods, please refer to the above method implementation methods, which will not be repeated here.

[0067] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention. Any aspects not detailed in the present invention are well-known techniques to those skilled in the art.

Claims

A method for fully decoupled network transmission without feedback and dynamic resource allocation in an integrated air-space-ground network, characterized in that: Includes the following steps: Step 1: The edge cloud collects historical Channel State Information (CSI) data, including user location and corresponding spectrum resource block CSI information; Step 2: For each spectrum resource block of each downlink base station, the edge cloud uses a neural network to train historical CSI data. The input is the user location, and the output is CSI information. Step 3: Control the base station to collect user access requests and real-time geographical location, and feed them back to the edge cloud; Step 4: The edge cloud dynamically allocates resources based on the user-heterogeneous node-spectrum resource block allocation algorithm of the many-to-one matching model, and sends the user location and resource block allocation decision to the downlink base station; Step 5: The downlink base station predicts CSI information on the resource block to be transmitted; Step 6: The downlink base station performs feedbackless MIMO transmission based on the closed-loop spatial multiplexing mode. A method for feedback-free transmission and dynamic resource allocation in a fully decoupled network for integrated air-space-ground communication, as described in claim 1, is characterized in that: The user-heterogeneous node-spectrum resource block allocation algorithm based on a many-to-one matching model constructs a spectrum resource block allocation problem and realizes dynamic resource allocation based on a many-to-one matching model. The spectrum resource block allocation problem aims to maximize system throughput. Through constraints, it ensures that each user is allocated at least one spectrum resource block of a node, each spectrum resource block can only be allocated to one user, and the total transmission rate of WiFi nodes will not exceed the maximum downlink transmission rate of the satellite-to-satellite WiFi router. A method for feedback-free transmission and dynamic resource allocation in a fully decoupled network for integrated air-space-ground communication, as described in claim 2, is characterized in that: The spectrum resource block allocation problem (RAP) is represented as follows: Where X, Y, and Z represent the matching matrices from base station (BS)-resource block (RB) pairs, WiFi-RB pairs, and UAV-RB pairs to users, respectively, and B, S, and U represent the number of BS-RB pairs, WiFi-RB pairs, and UAV-RB pairs, respectively. Let R represent the sets of BS-RB pairs, WiFi-RB pairs, and UAV-RB pairs, respectively. B (X) represents the throughput of the BS-RB pair, R S (Y) represents the throughput of the satellite WiFi-RB pair, R U (Z) represents the throughput of the UAV-RB pair, M represents the total number of users, and T S The maximum downlink throughput rate from satellite to WiFi is represented by: X(b,m), a two-dimensional 0-1 variable, indicating whether the BS-RB serves user m on b (1 if yes, 0 otherwise); Y(s,m), a two-dimensional 0-1 variable, indicating whether the WiFi-RB serves user m on s (1 if yes, 0 otherwise); and Z(u,m), a two-dimensional 0-1 variable, indicating whether the UAV-RB serves user m on u (1 if yes, 0 otherwise). A method for feedback-free transmission and dynamic resource allocation in a fully decoupled network for integrated air-space-ground communication, as described in claim 3, is characterized in that: After the neural network on each spectrum resource block of each downlink base station is trained, the edge cloud sends the trained neural network to each downlink base station. A method for feedback-free transmission and dynamic resource allocation in a fully decoupled network for integrated air-space-ground communication, as described in claim 2, is characterized in that: The resource dynamic allocation based on the many-to-one matching model involves assigning the most suitable user to each node-resource block pair during the matching initialization process, based on the user preference list; during the matching and exchange phase, the two node-resource block pairs are traversed sequentially, and if the matched users are different, an exchange attempt is initiated. In each round of matching, for each user, three matching and exchange strategies are tried, and only the strategy that improves the overall throughput of the system is accepted. A method for feedback-free transmission and dynamic resource allocation in a fully decoupled network for integrated air-space-ground communication, as described in claim 5, is characterized in that: The specific steps for achieving dynamic resource allocation based on a many-to-one matching model are as follows: Step 4.1: In the algorithm initialization phase, first input the physical layer throughput estimate from the user to each node-resource block pair; Step 4.2: Users generate a preference list based on their physical layer throughput estimates, from highest to lowest. Step 4.3: Each user initiates a request in turn, requesting the best node-resource block pair; Step 4.4: The node-resource block pair accepts the request if and only if no other match is received; Step 4.5: Once each user has been assigned a node-resource block pair, the matching and exchange phase begins; Step 4.6: In the matching and swapping phase, traverse the two node-resource block pairs in sequence. If the matching users are different, start the swapping attempt. Step 4.7: Calculate the current total system throughput; Step 4.8: Swap Attempt 1: Swap two nodes - resource block pairs that match the users; Step 4.9: Check if the previous user has matched other resource blocks. If so, perform swap attempt 2: make both nodes and resource blocks match the previous user at the same time; otherwise, skip this step. Step 4.10: Check if the next user has matched other resource blocks. If so, perform swap attempt 3: make both nodes and resource blocks match the next user at the same time; otherwise, skip this step. Step 4.11: Compare swap attempts 1-3 with the initial throughput, and select the optimal strategy to update the match; Step 4.12: Perform the next round of matching and swapping until all iterations are completed. A method for feedback-free transmission and dynamic resource allocation in a fully decoupled network for integrated air-space-ground communication, as described in claim 1, is characterized in that: In a fully decoupled air-space-ground integrated network, users are served simultaneously by multiple heterogeneous air-space-ground nodes, and spectrum resource blocks are dynamically allocated among the heterogeneous nodes; the heterogeneous air-space-ground nodes include drones, satellite WiFi routers, and downlink base stations of the fully decoupled network. A method for feedback-free transmission and dynamic resource allocation in a fully decoupled network for integrated air-space-ground communication, as described in claim 1, is characterized in that: Downlink base station transmission relies on channel state information to determine the transmission parameters for each resource block, including: rank indicator (RI), channel quality indicator (CQI), and precoding matrix indicator (PMI). The base station determines the specific parameters for downlink MIMO transmission based on these parameters. Referring to the 3GPP standard, the code rate is calculated based on CQI, the number of transport streams is calculated based on RI, and the physical layer throughput estimate is calculated by combining the predicted or feedback CSI. A fully decoupled network system for integrated air-space-ground communication with feedback-free transmission and dynamic resource allocation, characterized in that: This includes edge cloud, control base stations, and downlink base stations; The edge cloud is used to collect historical channel state information (CSI) data, including user location and corresponding spectrum resource block CSI information; for each spectrum resource block of each downlink base station, a neural network is used to train historical CSI data, with user location as input and CSI information as output; and a user-heterogeneous node-spectrum resource block allocation algorithm based on a many-to-one matching model is used to dynamically allocate resources, and user location and resource block allocation decisions are sent to the downlink base station. The control base station is used to collect user access requests and real-time geographical location, and feed them back to the edge cloud; The downlink base station is used to predict CSI information on the resource block to be transmitted; Feedbackless MIMO transmission based on closed-loop spatial multiplexing mode. A fully decoupled network system for integrated air-space-ground communication with feedback-free transmission and dynamic resource allocation as described in claim 9, characterized in that: The user-heterogeneous node-spectrum resource block allocation algorithm based on a many-to-one matching model constructs a spectrum resource block allocation problem and realizes dynamic resource allocation based on a many-to-one matching model. The spectrum resource block allocation problem aims to maximize system throughput. Through constraints, it ensures that each user is allocated at least one spectrum resource block of a node, each spectrum resource block can only be allocated to one user, and the total transmission rate of WiFi nodes will not exceed the maximum downlink transmission rate of the satellite-to-satellite WiFi router.

Images