Satellite auxiliary pilot frequency distribution method based on quantum convolutional neural network

Abstract

The invention relates to the technical field of communication, in particular to a satellite auxiliary pilot frequency distribution method based on a quantum convolutional neural network, and the method comprises the steps: determining all ground large-scale fading coefficients between each ground access point and each single-antenna user, all satellite large-scale fading coefficients between each satellite access point and each single-antenna user are obtained; determining a coefficient maximum value in all the ground large-scale fading coefficients, and determining a target communication network architecture according to the coefficient maximum value; and using the target communication network architecture to serve all the single-antenna users, and inputting all the large-scale fading coefficients corresponding to the target communication network architecture into a preset quantum convolutional neural network model to obtain a pilot frequency index of each single-antenna user, so as to perform pilot frequency distribution according to the pilot frequency index. Therefore, the problem of pilot frequency pollution caused by pilot frequency reuse is solved, pilot frequency distribution can be optimized, and pilot frequency pollution is reduced.

Description

Technical Field

[0001] This application relates to the field of communication technologies, and particularly to a satellite-assisted pilot allocation method based on a quantum convolutional neural network. Background Art

[0002] The sixth-generation mobile communication system is envisioned to meet the stringent requirements of Internet of Things applications such as intelligent transportation and industrial automation, aiming to achieve an air interface latency of less than 0.1 ms, a reliability of higher than 99.9999%, a connection density of more than 107 / km2, and global connectivity. To meet the strict requirements of 6G for low latency, high reliability, high energy efficiency, large-scale connection, and scalability, the cell-free massive multiple-input and multiple-output (CF mMIMO) technology has become a research hotspot.

[0003] With the continuous growth of global connectivity requirements, the low Earth orbit (LEO) satellite network has become a key technology for global connectivity due to its low latency and high data transmission rate. In the case where the ground communication link conditions are poor and cannot support effective high-reliability and low-latency communication, it provides necessary communication link connections for devices. At the same time, with the intensification of the competition for space resources, satellite deployments are becoming more and more dense, and it is becoming a reality that more than a dozen satellites appear above a user at the same time. Therefore, machine-to-machine communication can utilize multiple satellites to provide services for users simultaneously, improving communication quality.

[0004] However, machine-to-machine communication usually uses short-packet communication. With the access of a large number of devices, the number of orthogonal pilots that can be allocated under a finite block length is very limited. When the pilot length is less than the number of users, pilot reuse will bring serious pilot contamination.

[0005] Related technologies formulate the pilot allocation problem as a convex optimization problem and try different objective functions, or formulate the pilot allocation problem as a graph partitioning problem and introduce different heuristic algorithms, such as graph coloring algorithms and Hungarian algorithms. However, as the number of users increases and the channel dimension becomes higher, the complexity of traditional algorithms will increase sharply. In addition, these traditional algorithms usually rely on explicit mathematical models or rules and have low generalization ability, making it difficult to cope with the ever-changing requirements in practical applications. In addition, related technologies also propose an artificial intelligence-based pilot allocation algorithm, such as an algorithm based on K-means, an algorithm based on deep reinforcement learning, and an algorithm based on a convolutional neural network. However, when the number of access UEs is large, these algorithms are still limited by their explosive complexity. Summary of the Invention

[0006] This application provides a satellite-assisted pilot allocation method based on a quantum convolutional neural network to solve the pilot contamination problem caused by pilot reuse under harsh conditions, which can optimize pilot allocation, reduce pilot contamination, and improve the reliability of the system under ultra-low latency conditions.

[0007] In the first aspect of the embodiments of this application, a satellite-assisted pilot allocation method based on a quantum convolutional neural network is provided. The satellite-assisted pilot allocation method based on a quantum convolutional neural network is applied to a preset communication network architecture. Among them, the preset communication network architecture includes a preset terrestrial communication network architecture and a preset satellite communication network architecture. The preset terrestrial communication network architecture includes at least one terrestrial access point and at least one single-antenna user. The preset satellite communication network architecture includes at least one satellite access point. Among them, the method includes the following steps: Determine all the terrestrial large-scale fading coefficients between each terrestrial access point and each single-antenna user, and all the satellite large-scale fading coefficients between each satellite access point and each single-antenna user; Determine the maximum value of the coefficients among all the terrestrial large-scale fading coefficients, and determine the target communication network architecture from the preset terrestrial communication network architecture and the preset satellite communication network architecture according to the maximum value of the coefficients; Use the target communication network architecture to serve all single-antenna users, and input all the large-scale fading coefficients corresponding to the target communication network architecture into a preset quantum convolutional neural network model to obtain the pilot index of each single-antenna user, so as to perform pilot allocation for each single-antenna user according to the pilot index of each single-antenna user.

[0008] Optionally, in some embodiments, the determining the target communication network architecture from the preset terrestrial communication network architecture and the preset satellite communication network architecture according to the maximum value of the coefficients includes: Judge whether the maximum value of the coefficients is greater than a preset threshold; If the maximum value of the coefficients is greater than the preset threshold, then use the preset terrestrial communication network architecture as the target communication network architecture; otherwise, use the preset satellite communication network architecture as the target communication network architecture.

[0009] Optionally, in some embodiments, before inputting all the large-scale fading coefficients corresponding to the target communication network architecture into the preset quantum convolutional neural network model, it includes: Obtain training set data; Input the training set data into a preset quantum circuit. In the preset quantum circuit, use the RX gate to encode the training set data to obtain encoded data; Perform quantum operations on the encoded data to obtain a quantum state, and use a CNOT gate to perform quantum entanglement on the quantum state to obtain a quantum state after quantum entanglement; Use Pauli-Z to observe the quantum state after quantum entanglement to obtain a quantum computing result, input the quantum computing result into an initial neural network model for training to obtain a trained neural network model, calculate the cross-entropy loss function of the trained neural network model to obtain a loss value, and use the loss value to optimize the trained neural network model to obtain a new trained neural network model; Use the new trained neural network model as the initial neural network model, and re-execute the step of inputting the training set data into a preset quantum circuit until the cumulative number of training rounds reaches a preset number of training rounds, and output the finally trained neural network model as the preset quantum convolutional neural network model.

[0010] Optionally, in some embodiments, the encoding the training set data using an RX gate to obtain encoded data includes: Encoding the training set data using an RX gate according to a preset encoding formula to obtain encoded data, where the preset encoding formula is: ; where is a rotation gate around the X-axis, is the angle by which the quantum state rotates around the X-axis, is the imaginary unit.

[0011] Optionally, in some embodiments, the calculation formula for all the large-scale fading coefficients on the ground between each ground access point and each single-antenna user is: ; The calculation formula for all the large-scale fading coefficients of the satellite between each satellite access point and each single-antenna user is: ; where is the large-scale fading coefficient on the ground between the th access point and the th single-antenna user, is the path loss between the th access point and the th single-antenna user, is the shadow fading with a standard deviation of between the th access point and the th single-antenna user, is the The standard deviation between the th access point and the th single-antenna user, The standard normal distribution random variable between the th access point and the th single-antenna user, The satellite large-scale fading coefficient between the th access point and the th single-antenna user, The path loss between the th access point and the th single-antenna user, The shadow fading with a standard deviation of between the th access point and the th single-antenna user, The standard deviation between the th access point and the th single-antenna user, The standard normal distribution random variable between the th access point and the

[0012] In the second aspect of the present application, an embodiment provides a satellite-assisted pilot allocation device based on a quantum convolutional neural network. The satellite-assisted pilot allocation device based on a quantum convolutional neural network is applied to a preset communication network architecture. Among them, the preset communication network architecture includes a preset terrestrial communication network architecture and a preset satellite communication network architecture. The preset terrestrial communication network architecture includes at least one terrestrial access point and at least one single-antenna user. The preset satellite communication network architecture includes at least one satellite access point. Among them, the device includes: A first determination module, configured to determine all terrestrial large-scale fading coefficients between each terrestrial access point and each single-antenna user, and all satellite large-scale fading coefficients between each satellite access point and each single-antenna user; A second determination module, configured to determine the maximum value of the coefficients among all the terrestrial large-scale fading coefficients, and determine a target communication network architecture from the preset terrestrial communication network architecture and the preset satellite communication network architecture according to the maximum value of the coefficients; A pilot allocation module, configured to serve all single-antenna users by using the target communication network architecture, and input all the large-scale fading coefficients corresponding to the target communication network architecture into a preset quantum convolutional neural network model to obtain the pilot index of each single-antenna user, so as to perform pilot allocation for each single-antenna user according to the pilot index of each single-antenna user.

[0013] Optionally, in some embodiments, the second determination module includes: A judgment unit, configured to judge whether the maximum value of the coefficient is greater than a preset threshold; A generation unit, configured to use the preset terrestrial communication network architecture as the target communication network architecture when the maximum value of the coefficient is greater than the preset threshold, otherwise, use the preset satellite communication network architecture as the target communication network architecture.

[0014] Optionally, in some embodiments, before inputting all the large-scale fading coefficients corresponding to the target communication network architecture into the preset quantum convolutional neural network model, the pilot allocation module includes: An acquisition unit, configured to acquire training set data; An encoding unit, configured to input the training set data into a preset quantum circuit, and in the preset quantum circuit, use an RX gate to encode the training set data to obtain encoded data; An operation unit, configured to perform quantum operations on the encoded data to obtain a quantum state, and use a CNOT gate to perform quantum entanglement on the quantum state to obtain a quantum state after quantum entanglement; A training unit, configured to use Pauli-Z to observe the quantum state after quantum entanglement to obtain a quantum calculation result, input the quantum calculation result into an initial neural network model for training to obtain a trained neural network model, calculate the cross-entropy loss function of the trained neural network model to obtain a loss value, and use the loss value to optimize the trained neural network model to obtain a new trained neural network model; A generation unit, configured to use the new trained neural network model as the initial neural network model, and re-execute the step of inputting the training set data into the preset quantum circuit until the cumulative number of training rounds reaches a preset number of training rounds, and output the finally trained neural network model as the preset quantum convolutional neural network model.

[0015] Optionally, in some embodiments, the encoding unit includes: An encoding subunit, configured to encode the training set data using an RX gate according to a preset encoding formula to obtain encoded data, where the preset encoding formula is: ; Where is a rotation gate around the X-axis, is the angle of rotation of the quantum state around the X-axis, is the imaginary unit.

[0016] Optionally, in some embodiments, the calculation formula for all the terrestrial large-scale fading coefficients between each terrestrial access point and each single-antenna user is: ; The calculation formula for all the satellite large-scale fading coefficients between each satellite access point and each single-antenna user is: ; Wherein, is the terrestrial large-scale fading coefficient between the th access point and the th single-antenna user, is the path loss between the th access point and the th single-antenna user, is the shadow fading with a standard deviation of between the th access point and the th single-antenna user, is the standard deviation between the th access point and the th single-antenna user, is the standard normal distribution random variable between the th access point and the th single-antenna user, is the satellite large-scale fading coefficient between the th access point and the th single-antenna user, is the path loss between the th access point and the th single-antenna user, is the shadow fading with a standard deviation of between the th access point and the th single-antenna user, is the standard deviation between the th access point and the th single-antenna user, is the standard normal distribution random variable between the th access point and the th single-antenna user.

[0017] An embodiment of the third aspect of the present application provides an electronic device, including: a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor executes the program to implement the satellite-assisted pilot allocation method based on a quantum convolutional neural network as described in the above embodiments.

[0018] The fourth aspect of the present application provides a computer-readable storage medium, on which a computer program is stored, and the program is executed by a processor to implement the satellite-assisted pilot allocation method based on a quantum convolutional neural network as described in the above embodiments.

[0019] Thus, by determining all the terrestrial large-scale fading coefficients between each terrestrial access point and each single-antenna user, and all the satellite large-scale fading coefficients between each satellite access point and each single-antenna user, and determining the maximum value among all the terrestrial large-scale fading coefficients, and determining the target communication network architecture from the preset terrestrial communication network architecture and the preset satellite communication network architecture according to the maximum value; using the target communication network architecture to serve all single-antenna users, and inputting all the large-scale fading coefficients corresponding to the target communication network architecture into the preset quantum convolutional neural network model to obtain the pilot index of each single-antenna user, so as to perform pilot allocation for each single-antenna user according to the pilot index of each single-antenna user. Thus, the problem of pilot contamination caused by pilot reuse under harsh conditions can be solved, pilot allocation can be optimized, pilot contamination can be reduced, and the reliability of the system under ultra-low latency conditions can be improved.

[0020] Additional aspects and advantages of the present application will be given in part in the following description, will become apparent in part from the following description, or will be understood through the practice of the present application. Description of the Drawings

[0021] The above and / or additional aspects and advantages of the present application will become apparent and be readily understood from the following description of the embodiments in conjunction with the drawings, where: Figure 1 It is a schematic diagram of a preset communication network architecture according to an embodiment of the present application; Figure 2 It is a flowchart of a satellite-assisted pilot allocation method based on a quantum convolutional neural network according to an embodiment of the present application; Figure 3 It is a flowchart of a satellite-assisted pilot allocation method based on a quantum convolutional neural network according to an embodiment of the present application; Figure 4 It is a flowchart of constructing a preset quantum convolutional neural network model according to an embodiment of the present application; Figure 5 It is a block schematic diagram of a satellite-assisted pilot allocation device based on a quantum convolutional neural network according to an embodiment of the present application; Figure 6 It is a schematic diagram of the structure of an electronic device according to an embodiment of the present application. Detailed Embodiments

[0022] Embodiments of the present application will be described in detail below. Examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to explain the present application, but should not be construed as limiting the present application.

[0023] A satellite-assisted pilot allocation method based on a quantum convolutional neural network according to an embodiment of the present application will be described below with reference to the accompanying drawings. In view of the pilot contamination problem caused by pilot reuse under harsh conditions mentioned in the above background art, the present application provides a satellite-assisted pilot allocation method based on a quantum convolutional neural network. In this method, by determining all the large-scale fading coefficients between each ground access point and each single-antenna user, and all the satellite large-scale fading coefficients between each satellite access point and each single-antenna user, and determining the maximum value among all the ground large-scale fading coefficients, and determining the target communication network architecture from a preset ground communication network architecture and a preset satellite communication network architecture according to the maximum value; using the target communication network architecture to serve all single-antenna users, and inputting all the large-scale fading coefficients corresponding to the target communication network architecture into a preset quantum convolutional neural network model to obtain the pilot index of each single-antenna user, so as to perform pilot allocation for each single-antenna user according to the pilot index of each single-antenna user. Thus, the pilot contamination problem caused by pilot reuse under harsh conditions can be solved, the pilot allocation can be optimized, the pilot contamination can be reduced, and the reliability of the system under ultra-low latency conditions can be improved.

[0024] Before introducing the satellite-assisted pilot allocation method based on a quantum convolutional neural network according to an embodiment of the present application, the preset communication network architecture will be introduced first, as Figure 1 shown, the preset communication network architecture includes a preset ground communication network architecture and a preset satellite communication network architecture, where the preset ground communication network architecture includes at least one ground access point and at least one single-antenna user, and the preset satellite communication network architecture includes at least one satellite access point.

[0025] Specifically, the preset ground communication network architecture includes ground access points (Ground Access Point, GAP) each having antennas, and single-antenna users (User Equipment, UE), GAPs serve UEs at the same time, represents the large-scale fading coefficient between the th GAP and the th UE; the preset satellite communication network architecture includes satellite access points each having The satellite access point (SAP) of the root antenna, if necessary simultaneously serves UEs, denotes the large-scale fading coefficient between the th SAP and the th UE.

[0026] Specifically, Figure 2 is a schematic flowchart of a satellite-assisted pilot allocation method based on a quantum convolutional neural network provided by an embodiment of this application.

[0027] As Figure 2 shown, the satellite-assisted pilot allocation method based on a quantum convolutional neural network includes the following steps: In step S101, determine all the terrestrial large-scale fading coefficients between each terrestrial access point and each single-antenna user, and all the satellite large-scale fading coefficients between each satellite access point and each single-antenna user.

[0028] It can be understood that an embodiment of this application defines and analyzes the pilot contamination term of the kth user. Specifically, when terrestrial GAP is used to provide services, the pilot contamination term can be expressed as: ; When the celestial SAP is used to provide services, the pilot contamination term can be expressed as: ; Among them, is the pilot contamination term of the th single-antenna user, is the number of terrestrial access points, is a terrestrial access point, is a single-antenna user different from , is a single-antenna user different from and , is the set of users that reuse a pilot different from that of user k , is the set of users that reuse the same pilot as user , is the terrestrial large-scale fading coefficient between the th access point and the th single-antenna user, is the terrestrial large-scale fading coefficient between the th access point and the th single-antenna user, is the The large-scale fading coefficient between the -th access point and the -th single-antenna user on the ground, is the large-scale fading coefficient between the -th access point and the -th access point and the -th single-antenna user via satellite, is the large-scale fading coefficient between the -th access point and the -th single-antenna user via satellite, is the number of antennas of each ground access point, is the number of antennas of each satellite access point.

[0029] It can be seen from the above formula that represents the set of users that reuse pilots with the UE . It can be seen from this formula that when the UE is fixed, and are smaller, the pilot contamination is smaller. In addition, if the difference between , and (or , and ) is large enough, then is also smaller, which means fewer users reuse pilots and the pilot contamination is smaller. Moreover, compared with each other, UEs that are far enough apart geographically reusing pilots can effectively alleviate pilot contamination and obtain a higher SINR. That is to say, the pilot contamination term is closely related to the large-scale fading coefficient.

[0030] Specifically, the calculation formula for all the large-scale fading coefficients between each ground access point and each single-antenna user is: ; The calculation formula for all the large-scale fading coefficients between each satellite access point and each single-antenna user is: ; where is the large-scale fading coefficient between the -th access point and the -th single-antenna user on the ground, is the path loss between the -th access point and the -th single-antenna user, is the -th access point to the The standard deviation among shadow fading for single-antenna users is the standard deviation between the th access point and the th single-antenna user, is the standard normal distribution random variable between the th access point and the th single-antenna user, i.e., the satellite large-scale fading coefficient between the th access point and the th single-antenna user, is the path loss between the th access point and the shadow fading with a standard deviation of for the th access point and the th single-antenna user, is the standard deviation between the th access point and the th

[0031] All ground large-scale fading coefficients are obtained through the above calculation formula and all satellite large-scale fading coefficients .

[0032] In step S102, determine the maximum coefficient among all ground large-scale fading coefficients, and determine the target communication network architecture from the preset ground communication network architecture and the preset satellite communication network architecture according to the maximum coefficient.

[0033] Furthermore, in some embodiments, determining the target communication network architecture from the preset ground communication network architecture and the preset satellite communication network architecture according to the maximum coefficient includes: determining whether the maximum coefficient is greater than a preset threshold; if the maximum coefficient is greater than the preset threshold, using the preset ground communication network architecture as the target communication network architecture, otherwise, using the preset satellite communication network architecture as the target communication network architecture.

[0034] Wherein, the preset threshold can be preset by the user, can be obtained through a finite number of experiments, or can be obtained through a finite number of computer simulations, and is not specifically limited herein.

[0035] Specifically, determine the maximum value among all the large-scale fading coefficients on the ground. When the maximum value of the coefficients is greater than a preset threshold then use the preset ground communication network architecture as the target communication network architecture. If the maximum value of the coefficients is less than the preset threshold then use the preset satellite communication network architecture as the target communication network architecture.

[0036] In step S103, use the target communication network architecture to serve all single-antenna users, and input all the large-scale fading coefficients corresponding to the target communication network architecture into a preset quantum convolutional neural network model to obtain the pilot index of each single-antenna user, so as to perform pilot allocation for each single-antenna user according to the pilot index of each single-antenna user.

[0037] Specifically, when the maximum value of the coefficients is greater than the preset threshold, the embodiments of the present application use the preset ground communication network architecture to serve all single-antenna users, and input all the large-scale fading coefficients on the ground into the preset quantum convolutional neural network model. The preset quantum convolutional neural network model outputs the first pilot index of each single-antenna user, so as to perform pilot allocation for each single-antenna user according to the first pilot index of each single-antenna user; when the maximum value of the coefficients is greater than the preset threshold, use the preset satellite communication network architecture to serve all single-antenna users, and input all the large-scale fading coefficients of the satellite into the preset quantum convolutional neural network model. The preset quantum convolutional neural network model outputs the second pilot index of each single-antenna user, so as to perform pilot allocation for each single-antenna user according to the second pilot index of each single-antenna user.

[0038] That is to say, when in the GAP that serves users, the maximum value of the large-scale fading coefficients is greater than the threshold then use the ground-based CF mMIMO (Cell-Free Massive Multiple-Input Multiple-Output) architecture to serve these users. If the maximum value of the large-scale fading coefficients in the GAP that serves users is less than the threshold then use the space-based CF mMIMO architecture to serve users.

[0039] Optionally, in some embodiments, before inputting all the large-scale fading coefficients corresponding to the target communication network architecture into a preset quantum convolutional neural network model, it includes: obtaining training set data; inputting the training set data into a preset quantum circuit, in which the training set data is encoded by an RX (rotation around the X-axis) gate to obtain encoded data; performing quantum operations on the encoded data to obtain a quantum state, and using a CNOT (controlled NOT gate or controlled anti-gate) gate to perform quantum entanglement on the quantum state to obtain a quantum state after quantum entanglement; using a Pauli-Z (Pauli Z gate) to observe the quantum state after quantum entanglement to obtain a quantum calculation result, and inputting the quantum calculation result into an initial neural network model for training to obtain a trained neural network model, and calculating the cross-entropy loss function of the trained neural network model to obtain a loss value, using the loss value to optimize the trained neural network model to obtain a new trained neural network model; taking the new trained neural network model as the initial neural network model, and re-executing the step of inputting the training set data into the preset quantum circuit until the cumulative number of training rounds reaches a preset number of training rounds, and outputting the finally trained neural network model as the preset quantum convolutional neural network model.

[0040] Embodiments of the present application can construct a preset quantum convolutional neural network model, expecting to utilize the parallelism of quantum computing to improve the computing speed and reliability of the system through quantum superposition and quantum interference, and construct an HQCNN (hybrid quantum-classical neural network) model. The model includes quantum circuit design (including data encoding, quantum entanglement, and quantum observation design), quantum convolutional layer, and fully connected layer. In the quantum circuit, the RX gate is used to encode the large-scale fading coefficient information of the channel into a quantum state, and the CNOT gate is used to establish quantum correlations between qubits. The Pauli-Z measurement converts quantum information into classical data. Through training and optimization, the preset quantum convolutional neural network model outputs the pilot index of each user to achieve pilot allocation.

[0041] To enable those skilled in the relevant art to further understand the satellite-assisted pilot allocation method based on a quantum convolutional neural network in embodiments of the present application, the following will be elaborated in detail with specific embodiments.

[0042] As Figure 3 shown, Figure 3 is a flowchart of a satellite-assisted pilot allocation method based on a quantum convolutional neural network according to an embodiment of the present application.

[0043] Step 201: Initialize the number of users K, the number of APs M, the number of satellites N, the pilot length , the large-scale fading coefficient on the ground ; the preset communication network architecture includes APs, LEO satellites and users. Assume that all APs are connected through an ideal backhaul without transmission errors and bandwidth limitations, providing infinite capacity for the CPU.

[0044] Step 202: Determine whether the maximum value of the large-scale fading coefficient of the ground for the AP serving each user is greater than the threshold . If it is greater than , use GAP to serve the user and execute Step 203; if it is less than , use SAP to serve the user and execute Step 204. Step 203: Set the preset ground communication network architecture GAP as the target communication network architecture.

[0045] Step 204: Set the preset satellite communication network architecture SAP as the target communication network architecture.

[0046] Step 205: Input all the large-scale fading coefficients corresponding to the target communication network architecture into the preset quantum neural network model.

[0047] Step 206: Output the pilot sequence number corresponding to each user.

[0048] Before using the satellite-assisted pilot allocation method based on quantum convolutional neural network in the embodiments of the present application, a preset quantum convolutional neural network model needs to be constructed. Among them, the specific steps for constructing the preset quantum convolutional neural network model are as Figure 4 shown: Step 301: Input the large-scale fading coefficient matrix B of the training data set into the quantum circuit. Step 302: Encode the input data using the RX gate: ; Among them, is a rotation gate around the X-axis, is the angle of rotation of the quantum state around the X-axis, is the imaginary unit.

[0049] After passing through four qubits, calculate the quantum state: ; Among them, is a four-qubit quantum state, is the quantum state of the first qubit, is the quantum state of the second qubit, is the quantum state of the third qubit, is the quantum state of the fourth qubit.

[0050] Step 303: Use the CNOT gate for quantum entanglement to fully learn the features: Among them,

[0051] Step 304: Use Pauli-Z measurement to extract the qubit state, reduce noise, and reduce the variance of the output data, and input it into a traditional fully connected layer and a ReLU activation function for feature learning.

[0052] Step 305: Calculate the loss function.

[0053] Step 306: Update the weights of the quantum convolutional layer and the fully connected layer.

[0054] Step 307: Determine whether the number of training epochs has reached the maximum number of training epochs R. If so, execute Step 308; otherwise, jump back to execute Step 301.

[0055] Step 308: Output the preset quantum convolutional neural network model.

[0056] In summary, the embodiments of the present application make full use of the coverage advantage of LEO satellites, provide a reliable option when the ground de-cellular architecture cannot provide a reliable communication link, and enhance the continuity of coverage. In addition, compared with traditional neural networks, processing high-dimensional data usually requires a long time and higher computational complexity. The quantum convolutional neural network can utilize the quantum superposition state to process multiple signals in parallel, accelerate the data processing process, and improve the learning efficiency of the model. And compared with traditional neural networks that rely on the gradient descent method to find local optimal solutions, the quantum convolutional neural network can effectively avoid falling into local optima through the superposition and interference effects of quantum states.

[0057] According to the satellite-assisted pilot allocation method based on a quantum convolutional neural network proposed by the embodiments of the present application, by determining all the ground large-scale fading coefficients between each ground access point and each single-antenna user, and all the satellite large-scale fading coefficients between each satellite access point and each single-antenna user, and determining the maximum value among all the ground large-scale fading coefficients, and determining the target communication network architecture from the preset ground communication network architecture and the preset satellite communication network architecture according to the maximum value; using the target communication network architecture to serve all single-antenna users, and inputting all the large-scale fading coefficients corresponding to the target communication network architecture into the preset quantum convolutional neural network model to obtain the pilot index of each single-antenna user, so as to perform pilot allocation for each single-antenna user according to the pilot index of each single-antenna user. Thereby, the problem of pilot contamination caused by pilot reuse under harsh conditions can be solved, the pilot allocation can be optimized, the pilot contamination can be reduced, and the reliability of the system under ultra-low latency conditions can be improved.

[0058] Next, a satellite-assisted pilot allocation device based on a quantum convolutional neural network proposed by the embodiments of the present application is described with reference to the accompanying drawings.

[0059] It should be noted that the satellite-assisted pilot allocation based on the quantum convolutional neural network is applied to a preset communication network architecture, where the preset communication network architecture includes a preset terrestrial communication network architecture and a preset satellite communication network architecture. The preset terrestrial communication network architecture includes at least one terrestrial access point and at least one single-antenna user, and the preset satellite communication network architecture includes at least one satellite access point.

[0060] Figure 5 It is a block diagram of the satellite-assisted pilot allocation device based on the quantum convolutional neural network according to an embodiment of the present application.

[0061] As Figure 5 shown, the satellite-assisted pilot allocation device 10 based on the quantum convolutional neural network includes: a first determination module 100, a second determination module 200, and a pilot allocation module 300.

[0062] Among them, the first determination module 100 is used to determine all the terrestrial large-scale fading coefficients between each terrestrial access point and each single-antenna user, and all the satellite large-scale fading coefficients between each satellite access point and each single-antenna user.

[0063] The second determination module 200 is used to determine the maximum value of the coefficients among all the terrestrial large-scale fading coefficients, and determine the target communication network architecture from the preset terrestrial communication network architecture and the preset satellite communication network architecture according to the maximum value of the coefficients.

[0064] The pilot allocation module 300 is used to serve all single-antenna users by using the target communication network architecture, and input all the large-scale fading coefficients corresponding to the target communication network architecture into the preset quantum convolutional neural network model to obtain the pilot index of each single-antenna user, so as to perform pilot allocation for each single-antenna user according to the pilot index of each single-antenna user.

[0065] Optionally, in some embodiments, the second determination module includes: a judgment unit and a generation unit.

[0066] Among them, the judgment unit is used to judge whether the maximum value of the coefficients is greater than a preset threshold.

[0067] The generation unit is used to use the preset terrestrial communication network architecture as the target communication network architecture when the maximum value of the coefficients is greater than the preset threshold, otherwise, use the preset satellite communication network architecture as the target communication network architecture.

[0068] Optionally, in some embodiments, before inputting all the large-scale fading coefficients corresponding to the target communication network architecture into the preset quantum convolutional neural network model, the pilot allocation module 300 includes: an acquisition unit, an encoding unit, an operation unit, a training unit, and a generation unit.

[0069] Among them, an acquisition unit is used to acquire training set data.

[0070] An encoding unit is used to input the training set data into a preset quantum circuit. In the preset quantum circuit, the RX gate is used to encode the training set data to obtain encoded data.

[0071] An operation unit is used to perform quantum operations on the encoded data to obtain a quantum state, and the CNOT gate is used to perform quantum entanglement on the quantum state to obtain a quantum state after quantum entanglement.

[0072] A training unit is used to obtain a quantum computing result by using the Pauli-Z observable on the quantum state after quantum entanglement, input the quantum computing result into an initial neural network model for training to obtain a trained neural network model, calculate the cross-entropy loss function of the trained neural network model to obtain a loss value, and use the loss value to optimize the trained neural network model to obtain a new trained neural network model.

[0073] A generation unit is used to use the new trained neural network model as the initial neural network model, and re-execute the step of inputting the training set data into the preset quantum circuit until the cumulative number of training rounds reaches the preset number of training rounds, and output the finally trained neural network model as the preset quantum convolutional neural network model.

[0074] Optionally, in some embodiments, the encoding unit includes: an encoding subunit.

[0075] Among them, the encoding subunit is used to encode the training set data by using the RX gate according to a preset encoding formula to obtain encoded data, where the preset encoding formula is: ; Among them, is a rotation gate around the X-axis, is the angle by which the quantum state rotates around the X-axis, is the imaginary unit.

[0076] Optionally, in some embodiments, the second determination unit includes: a determination subunit.

[0077] Among them, the determination subunit is used to determine all the ground large-scale fading coefficients between each ground access point and each single-antenna user, and all the satellite large-scale fading coefficients between each satellite access point and each single-antenna user by using a preset large-scale fading coefficient calculation formula, where the preset large-scale fading coefficient calculation formula: ; Among them, is the The large-scale fading coefficient between the th access point and the th single-antenna user, The path loss between the th access point and the th single-antenna user, The shadow fading with a standard deviation of between the th access point and the th single-antenna user, The standard deviation between the th access point and the th single-antenna user, The standard normal distribution random variable between the th access point and the th single-antenna user, i.e.,

[0078] It should be noted that the foregoing explanation of the embodiments of the satellite-assisted pilot allocation method based on the quantum convolutional neural network also applies to the satellite-assisted pilot allocation device based on the quantum convolutional neural network of this embodiment, and will not be elaborated here.

[0079] According to the satellite-assisted pilot allocation device based on the quantum convolutional neural network proposed in the embodiments of the present application, by determining all the ground large-scale fading coefficients between each ground access point and each single-antenna user, and all the satellite large-scale fading coefficients between each satellite access point and each single-antenna user, and determining the maximum value of the coefficients among all the ground large-scale fading coefficients, and determining the target communication network architecture from the preset ground communication network architecture and the preset satellite communication network architecture according to the maximum value of the coefficients; using the target communication network architecture to serve all single-antenna users, and inputting all the large-scale fading coefficients corresponding to the target communication network architecture into the preset quantum convolutional neural network model to obtain the pilot index of each single-antenna user, so as to perform pilot allocation for each single-antenna user according to the pilot index of each single-antenna user. Thus, the problem of pilot contamination caused by pilot reuse under harsh conditions can be solved, the pilot allocation can be optimized, the pilot contamination can be reduced, and the reliability of the system under ultra-low latency conditions can be improved.

[0080] Figure 6 It is a schematic structural diagram of the electronic device provided in the embodiments of the present application. The electronic device may include: A memory 601, a processor 602, and a computer program stored on the memory 601 and executable on the processor 602.

[0081] When the processor 602 executes the program, it implements the satellite-assisted pilot allocation method based on the quantum convolutional neural network provided in the above embodiments.

[0082] Further, the electronic device further includes: A communication interface 603 for communication between the memory 601 and the processor 602.

[0083] The memory 601 for storing a computer program that can run on the processor 602.

[0084] The memory 601 may include a high-speed RAM (Random Access Memory) memory, and may also include a non-volatile memory, such as at least one disk memory.

[0085] If the memory 601, the processor 602, and the communication interface 603 are independently implemented, the communication interface 603, the memory 601, and the processor 602 can be interconnected through a bus and communicate with each other. The bus can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 6 only a thick line is used to represent it in the figure, but it does not mean that there is only one bus or one type of bus.

[0086] Optionally, in a specific implementation, if the memory 601, the processor 602, and the communication interface 603 are integrated on a single chip, the memory 601, the processor 602, and the communication interface 603 can communicate with each other through an internal interface.

[0087] The processor 602 may be a CPU (Central Processing Unit), or an ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of the present application.

[0088] The embodiments of the present application further provide a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the satellite-assisted pilot allocation method based on a quantum convolutional neural network as described above is implemented.

[0089] In the description of this specification, the descriptions referring to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of this application. In this specification, the schematic expressions of the above terms are not necessarily directed to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any one or N embodiments or examples in a suitable manner. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0090] In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" can explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "N" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0091] Any process or method description shown in a flowchart or described in other ways herein can be understood to represent a module, segment, or part of code including one or more N executable instructions for implementing a customized logic function or process, and the scope of the preferred embodiments of this application includes additional implementations, where the functions can be executed in a manner that is not in the order shown or discussed, including in a substantially simultaneous manner according to the involved functions or in the reverse order, which should be understood by those skilled in the art to which the embodiments of this application belong.

[0092] It should be understood that each part of this application can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware as in another embodiment, any one or a combination of the following technologies well-known in the art can be used: discrete logic circuits with logic gate circuits for implementing logic functions on data signals, application-specific integrated circuits with appropriate combinational logic gate circuits, programmable gate arrays, field-programmable gate arrays, etc.

[0093] Those of ordinary skill in the art of this technology can understand that all or part of the steps carried by the methods of the above embodiments can be completed by instructing relevant hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiments.

[0094] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present application. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present application.

Claims

1. A satellite-assisted pilot allocation method based on a quantum convolutional neural network, characterized in that The satellite-assisted pilot allocation method based on a quantum convolutional neural network is applied to a preset communication network architecture. Among them, the preset communication network architecture includes a preset terrestrial communication network architecture and a preset satellite communication network architecture. The preset terrestrial communication network architecture includes at least one terrestrial access point and at least one single-antenna user. The preset satellite communication network architecture includes at least one satellite access point. Among them, the method includes the following steps: Determine all terrestrial large-scale fading coefficients between each terrestrial access point and each single-antenna user, and all satellite large-scale fading coefficients between each satellite access point and each single-antenna user; Determine the maximum value of the coefficients among all the terrestrial large-scale fading coefficients, and determine the target communication network architecture from the preset terrestrial communication network architecture and the preset satellite communication network architecture according to the maximum value of the coefficients; Use the target communication network architecture to serve all single-antenna users, and input all the large-scale fading coefficients corresponding to the target communication network architecture into a preset quantum convolutional neural network model to obtain the pilot index of each single-antenna user, so as to perform pilot allocation for each single-antenna user according to the pilot index of each single-antenna user.

2. The method according to claim 1, characterized in that, The determining the target communication network architecture from the preset terrestrial communication network architecture and the preset satellite communication network architecture according to the maximum value of the coefficients includes: Judge whether the maximum value of the coefficients is greater than a preset threshold; If the maximum value of the coefficients is greater than the preset threshold, use the preset terrestrial communication network architecture as the target communication network architecture; otherwise, use the preset satellite communication network architecture as the target communication network architecture.

3. The method according to claim 1, wherein Before inputting all the large-scale fading coefficients corresponding to the target communication network architecture into the preset quantum convolutional neural network model, it includes: Obtain training set data; Input the training set data into a preset quantum circuit. In the preset quantum circuit, use the RX gate to encode the training set data to obtain encoded data; Perform quantum operations on the encoded data to obtain a quantum state, and use the CNOT gate to perform quantum entanglement on the quantum state to obtain a quantum state after quantum entanglement; Use Pauli-Z to observe the quantum state after quantum entanglement to obtain a quantum calculation result, input the quantum calculation result into an initial neural network model for training to obtain a trained neural network model, calculate the cross-entropy loss function of the trained neural network model to obtain a loss value, and use the loss value to optimize the trained neural network model to obtain a new trained neural network model; Use the new trained neural network model as the initial neural network model, and re-execute the step of inputting the training set data into the preset quantum circuit until the cumulative number of training rounds reaches the preset number of training rounds, and output the finally trained neural network model as the preset quantum convolutional neural network model.

4. The method according to claim 3, characterized in that, The using the RX gate to encode the training set data to obtain encoded data includes: Encode the training set data using RX gates according to a preset encoding formula, where the preset encoding formula is: ； Among them, is a revolving door around the X-axis, is the angle of rotation of the quantum state around the X-axis, is the imaginary unit.

5. The method according to claim 1, characterized in that The calculation formula for all the large-scale fading coefficients between each ground access point and each single-antenna user is: ； The calculation formula for all the large-scale fading coefficients between each satellite access point and each single-antenna user is: ； Among them, is the ground large-scale fading coefficient between the th access point and the th single-antenna user, is the path loss between the th access point and the th single-antenna user, is the shadow fading with a standard deviation of between the th access point and the th single-antenna user, is the standard deviation between the th access point and the th single-antenna user, is the standard normal distribution random variable between the th access point and the th single-antenna user, is the satellite large-scale fading coefficient between the th access point and the th single-antenna user, is the path loss between the th access point and the th single-antenna user, is the shadow fading with a standard deviation of between the th access point and the th single-antenna user, is the standard deviation between the th access point and the th single-antenna user, is the standard normal distribution random variable between the th access point and the th single-antenna user.

6. A satellite-assisted pilot allocation device based on a quantum convolutional neural network, characterized in that, The satellite-assisted pilot allocation device based on a quantum convolutional neural network is applied to a preset communication network architecture, where the preset communication network architecture includes a preset terrestrial communication network architecture and a preset satellite communication network architecture. The preset terrestrial communication network architecture includes at least one ground access point and at least one single-antenna user, and the preset satellite communication network architecture includes at least one satellite access point. The device includes: A first determination module for determining all the large-scale fading coefficients between each ground access point and each single-antenna user, and all the large-scale fading coefficients between each satellite access point and each single-antenna user; A second determination module for determining the maximum value among all the large-scale fading coefficients, and determining a target communication network architecture from the preset terrestrial communication network architecture and the preset satellite communication network architecture according to the maximum value; A pilot allocation module for serving all single-antenna users using the target communication network architecture, and inputting all the large-scale fading coefficients corresponding to the target communication network architecture into a preset quantum convolutional neural network model to obtain the pilot index of each single-antenna user, so as to perform pilot allocation for each single-antenna user according to the pilot index of each single-antenna user.

7. The satellite-assisted pilot allocation device based on a quantum convolutional neural network according to claim 6, wherein The second determination module includes: A judgment unit for judging whether the maximum value is greater than a preset threshold; A generation unit for, when the maximum value is greater than the preset threshold, taking the preset terrestrial communication network architecture as the target communication network architecture, otherwise, taking the preset satellite communication network architecture as the target communication network architecture.

8. The satellite-assisted pilot allocation device based on a quantum convolutional neural network according to claim 6, characterized in that Before inputting all the large-scale fading coefficients corresponding to the target communication network architecture into the preset quantum convolutional neural network model, the pilot allocation module includes: An acquisition unit for acquiring training set data; An encoding unit for inputting the training set data into a preset quantum circuit, and in the preset quantum circuit, encoding the training set data using RX gates to obtain encoded data; An operation unit for performing quantum operations on the encoded data to obtain a quantum state, and using CNOT gates to perform quantum entanglement on the quantum state to obtain a quantum state after quantum entanglement; A training unit, configured to use Pauli-Z to observe the quantum state after quantum entanglement to obtain a quantum computing result, input the quantum computing result into an initial neural network model for training to obtain a trained neural network model, calculate the cross-entropy loss function of the trained neural network model to obtain a loss value, and use the loss value to optimize the trained neural network model to obtain a new trained neural network model; A generating unit, configured to use the new trained neural network model as the initial neural network model, and re-execute the step of inputting the training set data into a preset quantum circuit until the cumulative number of training rounds reaches a preset number of training rounds, and output the finally trained neural network model as the preset quantum convolutional neural network model.

9. An electronic device, characterized in that, Comprising: A memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor executes the program to implement the satellite-aided pilot allocation method based on a quantum convolutional neural network according to any one of claims 1-5.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to be used to implement the satellite-aided pilot allocation method based on a quantum convolutional neural network according to any one of claims 1-5.

Images