An emergency communication resource optimization method with dynamic spectrum allocation

Abstract

This application provides a dynamic spectrum allocation method for optimizing emergency communication resources, applied to a self-organizing emergency communication network consisting of several UAV nodes and at least one edge aggregation node. The method includes: an offline pre-training step; a model initialization step; and an online optimization step: UAV nodes perform online optimization, sensing the current environment's spectrum information in real time; based on the spectrum information, the initial local model is fine-tuned online to update the initial local model; using the updated initial local model, the channel quality of candidate communication frequency bands is predicted, and the target communication frequency band is determined based on the channel quality; after determining the target communication frequency band, a consensus mechanism is initiated, allowing other nodes in the communication network to verify the selection of the target communication frequency band, and after successful verification, the switching to the target communication frequency band is executed. This application can solve the problems of slow convergence, low decision accuracy, and untimely response caused by a lack of training samples in emergency communication tasks.

Description

Technical Field

[0001] This invention relates to the field of wireless communication technology, and in particular to an emergency communication resource optimization method with dynamic spectrum allocation. Background Technology

[0002] Following natural disasters such as earthquakes and floods, ground-based public communication networks (such as 4G / 5G base stations) often suffer widespread paralysis due to physical damage or power outages, turning rescue sites into "information islands." Utilizing drone swarms to rapidly deploy aerial emergency communication networks to restore command and dispatch, data transmission, and public communication capabilities has become a recognized key technology in the industry.

[0003] However, the electromagnetic environment in emergency scenarios is extremely complex: Unknown spectrum: surviving commercial, broadcast, or dedicated communication signals in the disaster area are unknown, posing numerous potential sources of interference; Spectrum congestion: multiple rescue teams (such as firefighters, medical personnel, and police) operating simultaneously with radios of different standards can easily cause severe congestion and conflicts in specific frequency bands; Dynamic changes: the movement of rescue teams, the addition of new equipment, and environmental changes cause the spectrum occupancy status to change rapidly. Existing technical solutions face a key technical bottleneck when applied to emergency scenarios: in the early stages of the mission, due to the lack of effective channel samples on-site, each UAV node's local model needs to be trained from random weights. Federated learning requires multiple rounds (usually dozens) of communication and aggregation before the model accuracy converges to a usable level. In the critical timeframe of emergency rescue, such minute-level model initialization delays are unacceptable, potentially leading to unstable initial communication links or even missing the optimal rescue opportunity. Summary of the Invention

[0004] The technical problem to be solved by this application is to provide a dynamic spectrum allocation method for optimizing emergency communication resources, so as to solve the problem of "cold start" in emergency situations, that is, in emergency communication tasks, the model converges slowly, has low decision accuracy, and responds untimely due to a lack of training samples.

[0005] To achieve the aforementioned technical effects, this application provides a dynamic spectrum allocation method for optimizing emergency communication resources, applied to a self-organizing emergency communication network consisting of several UAV nodes and at least one edge aggregation node, comprising the following steps: Offline pre-training step: Based on a preset spectrum dataset, a general channel feature pre-trained model is trained offline and generated. The general channel feature pre-trained model is used to predict the corresponding channel quality based on the input spectrum information. Model initialization step: At the start of the emergency communication mission, the general channel feature pre-trained model is loaded into at least one UAV node in the communication network as the initial local model of the UAV node. Online optimization step: The UAV node performs online optimization, which includes: real-time sensing of the spectrum information of the current environment; online fine-tuning of the initial local model based on the real-time sensed spectrum information to update the initial local model; using the updated initial local model to predict the channel quality of candidate communication frequency bands, and determining the target communication frequency band based on the predicted channel quality of the candidate communication frequency bands. Consensus verification step: After determining the target communication frequency band, a consensus mechanism is initiated, and other nodes in the communication network verify the selection of the target communication frequency band. After successful verification, the switching to the target communication frequency band is executed.

[0006] The solution provided in this application is based on a pre-set spectrum dataset. Before mission execution, a general channel feature pre-trained model with strong generalization ability is pre-trained offline. In emergency communication missions, this general channel feature pre-trained model is deployed as the initial local model on at least one UAV node. The UAV uses its onboard scanning equipment to perceive the spectrum information of the current environment in real time, generating a three-dimensional spectrum heatmap consistent with the input format of the pre-set spectrum dataset. Based on the real-time perceived spectrum information, the initial local model is fine-tuned and updated online to predict the channel quality of candidate communication frequency bands. This achieves the effect of the UAV having high-precision channel quality prediction capability at the moment of takeoff. The target communication frequency band is determined based on the channel quality of the candidate communication frequency bands. After the target communication frequency band is determined, a consensus mechanism is initiated, and other nodes in the communication network verify the selection of the target communication frequency band. After the verification is successful, the switching to the target communication frequency band is executed. This solves the problems of slow convergence, low decision accuracy, and untimely response caused by the lack of training samples in emergency communication missions, thus shortening the rescue time for emergency missions.

[0007] As an improvement to the technical solution of this application, the online fine-tuning method is federated learning; the online fine-tuning of the initial local model based on the real-time perceived spectrum information to update the initial local model includes: the UAV node using the real-time perceived spectrum information to perform local training on the initial local model to generate a fine-tuned local model; the UAV node interacting with edge aggregation nodes in the network to perform federated updates on the local model based on the fine-tuned local model and the model information of other UAV nodes in the network.

[0008] The solution provided in this application employs a federated learning method, which uses the spectrum information perceived in real time by UAV nodes to train an initial local model locally, thereby generating a fine-tuned local model. This achieves the effect of rapid adaptation to existing models and reduced computational overhead. Furthermore, by interacting with edge aggregation nodes in the network, the UAV nodes perform federated updates to the local model based on the fine-tuned local model and the model information of other UAV nodes in the network, enabling it to quickly adapt to the specific environment in the current task scenario.

[0009] As an improvement to the technical solution of this application, the federated update steps include: the drone node calculating and encrypting its model update information; uploading the encrypted model update information to the edge aggregation node; the edge aggregation node aggregating the received model update information to generate global model update parameters; and the drone node downloading the global model update parameters to update its local model.

[0010] Building upon the previous improvement, the model update information is further defined as the model gradient.

[0011] Building upon the previous improvement, the encryption further employs a homomorphic encryption algorithm.

[0012] As an improvement to the technical solution of this application, the preset spectrum dataset further includes at least one of the following: a radio signal spectrum dataset covering multiple modulation signals and signal-to-noise ratios; and a simulated spectrum dataset generated based on channel modeling technology, wherein the simulated spectrum dataset is used to simulate the electromagnetic propagation and interference characteristics under typical disaster area terrain.

[0013] As an improvement to the technical solution of this application, the spectrum information is further defined as a three-dimensional spectrum heatmap, which includes time, frequency, and signal energy dimensions.

[0014] As an improvement to the technical solution of this application, the general channel feature pre-training model is further described as a deep neural network containing convolutional neural network (CNN) layers and long short-term memory network (LSTM) layers.

[0015] As an improvement to the technical solution of this application, determining the target communication frequency band includes: using a reinforcement learning algorithm to select the target communication frequency band based on the channel quality of the predicted candidate communication frequency band and at least one other preset decision factor.

[0016] Furthermore, the reinforcement learning algorithm is the Q-Learning algorithm; the at least one other preset decision factor includes the channel capacity of the candidate communication frequency band and / or the task importance level of the UAV node.

[0017] The beneficial effects of this application are as follows: 1. The solution provided in this application is based on a pre-set spectrum dataset. Before the mission is executed, a general channel feature pre-trained model with strong generalization ability is pre-trained offline. In the emergency communication mission, this general channel feature pre-trained model is deployed as the initial local model on at least one UAV node. The UAV uses its onboard scanning equipment to perceive the spectrum information of the current environment in real time, generate a three-dimensional spectrum heatmap consistent with the input format of the pre-set spectrum dataset, and fine-tune and update the initial local model online based on the real-time perceived spectrum information to predict the channel quality of candidate communication frequency bands. This achieves the effect of the UAV having high-precision channel quality prediction capability at the moment of takeoff. The target communication frequency band is determined based on the channel quality of the candidate communication frequency bands. After the target communication frequency band is determined, a consensus mechanism is initiated, and other nodes in the communication network verify the selection of the target communication frequency band. After the verification is passed, the switching to the target communication frequency band is executed. This solves the problems of slow convergence, low decision accuracy, and untimely response caused by the lack of training samples in the model in emergency communication missions, thus shortening the rescue time for emergency missions.

[0018] 2. The solution provided in this application adopts a federated learning method, which uses the spectrum information perceived in real time by UAV nodes to train the initial local model locally, so as to generate a fine-tuned local model. This achieves the effect of rapid adaptation to existing models and reducing computational overhead. Furthermore, by interacting with edge aggregation nodes in the network, the UAV nodes perform federated updates on the local model based on the fine-tuned local model and the model information of other UAV nodes in the network, enabling it to quickly adapt to the specific environment in the current task scenario. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating an emergency communication resource optimization method for dynamic spectrum allocation, as described in an embodiment of this application. Detailed Implementation

[0021] The embodiments of the technical solution of this application will be described in detail below. The following embodiments are only used to more clearly illustrate the technical solution of this application, and are therefore merely examples and should not be used to limit the scope of protection of this application. Furthermore, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0022] The following is in conjunction with the appendix Figure 1 This application provides a detailed description of the technical solutions provided in each embodiment.

[0023] like Figure 1 The diagram shown illustrates a flowchart of an emergency communication resource optimization method for dynamic spectrum allocation, provided in one embodiment of this application. This method is applied to a self-organizing emergency communication network consisting of several UAV nodes and at least one edge aggregation node, and specifically includes the following steps: S1: Offline pre-training step: Based on a preset spectrum dataset, an offline training and generation of a general channel feature pre-training model is performed. The general channel feature pre-training model is used to predict the corresponding channel quality based on the input spectrum information. In this embodiment of the application, before executing the task, a general channel feature pre-trained model with strong generalization ability is trained offline and generated based on a preset spectrum dataset. In one implementation, in order to achieve the high generalization ability of the model, the preset spectrum dataset includes at least one of the following: a radio signal spectrum dataset covering a variety of modulation signals and signal-to-noise ratios, such as the RML2016.10a dataset, and a simulation spectrum dataset generated based on channel modeling technology, wherein the simulation spectrum dataset is used to simulate the electromagnetic propagation and interference characteristics under typical disaster area terrain.

[0024] S2: Model initialization steps: At the start of the emergency communication mission, the general channel feature pre-trained model is loaded into at least one UAV node in the communication network as the initial local model of the UAV node. In this embodiment of the application, at the start of an emergency communication mission, the general channel feature pre-trained model is loaded into the computing unit of at least one UAV node in the communication network, so as to serve as the initial local weight model of the UAV node. In one implementation, in order to construct a deep neural network as a pre-trained model, the general channel feature pre-trained model is a deep neural network containing convolutional neural network (CNN) layers and long short-term memory network (LSTM) layers.

[0025] The CNN layer is used to extract local correlation features of spectral data in the frequency and time dimensions, thereby effectively identifying the "shape" of interference signals. The LSTM layer is used to learn the time-series dependencies of the relevant features extracted by the CNN layer, thereby capturing the dynamic changes in spectrum occupancy.

[0026] For example, during the training process of the general channel feature pre-trained model, mean squared error (MSE) can be used as the loss function, and the Adam optimizer can be used for backpropagation training until the model converges on the validation set, thereby obtaining the general channel feature pre-trained model.

[0027] S3: Online optimization step: The drone node performs online optimization; In one implementation, to significantly improve emergency response speed and enable the UAV to have high-precision channel quality prediction capabilities and rapid understanding of the specific electromagnetic environment of the current mission scenario at the moment of takeoff, this step includes the following sub-steps: S31: Real-time sensing of the spectrum information of the current environment; In this embodiment, the UAV utilizes its onboard scanning device to perform real-time and continuous scanning of a preset frequency range. The data acquired during this scanning process can be converted into a three-dimensional spectrum heatmap. The format of this three-dimensional spectrum heatmap is consistent with the pre-trained input format, thereby enabling direct data feeding and analysis. In one implementation, the three-dimensional spectrum heatmap includes time, frequency, and signal energy dimensions. The time dimension records the time-series information of the scanned data; the frequency dimension represents the frequency components of the scanned signal; and the signal energy dimension quantifies the signal strength or power at specific time and frequency points.

[0028] S32: Based on the real-time perceived spectrum information, the initial local model is fine-tuned online to update the initial local model; In one implementation, to reduce model computational power and avoid waiting for multiple rounds of convergence in federated learning, this step includes the following sub-steps: S321: The UAV node uses the real-time sensed spectrum information to train the initial local model locally, so as to generate a fine-tuned local model. In this embodiment, the fine-tuning method is federated learning. In one implementation, the current environmental spectrum data collected in real time in the computing unit of the UAV node is used to quickly train the loaded general channel feature pre-trained model, that is, to quickly adapt the existing spectrum information data, thereby realizing online fine-tuning and updating of the initial local model, ensuring the generalization and specificity of the model in unknown environments.

[0029] S322: The drone node interacts with the edge aggregation node in the network to perform federated updates on the local model based on the fine-tuned local model and the model information of other drone nodes in the network.

[0030] In this embodiment, the local model update information is the model gradient, that is, when the preset global update time node is reached, the UAV calculates the gradient increment of the locally fine-tuned model relative to the global model. The federated update steps include the following sub-steps: S3221: The UAV node calculates and encrypts its model update information; In one implementation, to encrypt the model gradients and ensure privacy and security, a homomorphic encryption algorithm, such as the Paillier algorithm, is specifically used.

[0031] S3222: Upload the encrypted model update information to the edge aggregation node; For example, the encrypted model gradient is uploaded to an edge aggregation server, such as a ground command vehicle or a high-altitude drone.

[0032] S3223: The edge aggregation node aggregates the received model update information to generate global model update parameters; In one implementation, the server decrypts and aggregates all model gradients (e.g., using the FedAvg algorithm) to further update the global model.

[0033] S3224: The UAV node downloads the global model update parameters to update its local model.

[0034] In one implementation, the updated global model parameters are distributed to the drone nodes, and each drone node updates its local model accordingly.

[0035] S33: Using the updated initial local model, predict the channel quality of candidate communication frequency bands, and determine the target communication frequency band based on the predicted channel quality of candidate communication frequency bands; In this embodiment of the application, the UAV uses the updated high-precision initial local model and the real-time input spectrum information to predict the channel quality of each selected communication frequency band and determine the target communication frequency band.

[0036] In one implementation, determining the target communication frequency band includes: employing a reinforcement learning algorithm to select the target communication frequency band based on the channel quality of the predicted candidate communication frequency band and at least one other preset decision factor, wherein the reinforcement learning algorithm is a Q-Learning algorithm; the at least one other preset decision factor includes the channel capacity of the candidate communication frequency band and / or the mission importance level of the UAV node.

[0037] For example, construct the state space s={channel capacity C of each frequency band, UAV mission importance level P, predicted channel quality I of each frequency band}; Construct an action space a = {Keep the current frequency band, switch to candidate frequency band 1, switch to candidate frequency band 2, ...}; Using the Q-Learning algorithm, the optimal action (i.e., the optimal communication frequency band) is solved by iterating through the reward function R(s,a). R(s,a)= α ⋅ C actual + β ⋅(1− I normalized )− c ⋅Δ f cost + d ⋅ P task_bonus The reward function R(s,a) is used to quantify the benefits obtained by the UAV after performing action a in state s. The reward function R(s,a) aims to guide the algorithm to minimize unnecessary resource consumption while ensuring communication quality.

[0038] C actual : Refers to the current frequency band channel capacity that is actually measured or estimated after the UAV performs action a.

[0039] I normalized : Refers to the normalized actual interference level after performing action a (for example, the signal-to-noise ratio is inversely mapped to between 0 and 1, and the higher the value, the greater the interference).

[0040] Δ f cost This refers to the overhead incurred by the frequency band switching operation, including but not limited to switching latency, instantaneous packet loss rate, and energy consumption increment. If the instruction in the task is "Keep current frequency band", this item will be zero.

[0041] P task_bonus Based on the current task importance level P task Additional rewards or penalties can be set. For example, when communication quality reaches a preset threshold under important tasks, an additional reward can be given; conversely, when communication is impaired under important tasks, an additional penalty can be given.

[0042] α , β , c , d These are all preset weighting coefficients used to adjust the relative importance of each factor in the reward function. These weights can be optimized and adjusted through offline simulation, expert experience, or heuristic algorithms to meet the needs of different application scenarios.

[0043] S4: Consensus Verification Step: After determining the target communication frequency band, a consensus mechanism is initiated, and other nodes in the communication network verify the selection of the target communication frequency band. After successful verification, the switching to the target communication frequency band is executed.

[0044] In one implementation, after a drone node selects the optimal frequency band, a consensus mechanism is triggered to prevent malicious or faulty nodes from making incorrect decisions. Other drone nodes in the communication network verify the selection of the target communication frequency band. For example, "2f+1" nodes (where f is the number of tolerable malicious nodes) are randomly selected from nearby drone nodes as verification nodes. Only after more than 2 / 3 of the verification nodes confirm the decision will the frequency band switch be officially executed and broadcast to nearby drone nodes. Optionally, this consensus mechanism can be a Byzantine fault-tolerant consensus mechanism.

[0045] As can be seen from the above embodiments, the solution provided in this application is based on a preset spectrum dataset. Before the task is executed, a general channel feature pre-trained model with strong generalization ability is pre-trained offline. In the emergency communication task, the general channel feature pre-trained model is deployed as the initial local model on at least one UAV node. The UAV uses its onboard scanning equipment to perceive the spectrum information of the current environment in real time, generate a three-dimensional spectrum heatmap consistent with the input format of the preset spectrum dataset, and fine-tune and update the initial local model online based on the real-time perceived spectrum information, so that it can quickly adapt to the specific environment in the current task scenario and achieve the prediction of the channel quality of candidate communication frequency bands. This achieves the effect that the UAV has high-precision channel quality prediction capability at the moment of takeoff. The target communication frequency band is determined according to the channel quality of the candidate communication frequency bands. After the target communication frequency band is determined, a consensus mechanism is initiated, and other nodes in the communication network verify the selection of the target communication frequency band. After the verification is passed, the switching to the target communication frequency band is executed. This solves the problems of slow convergence, low decision accuracy, and untimely response caused by the lack of training samples in the model in the emergency communication task, and shortens the rescue time for the emergency task.

[0046] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing emergency communication resources through dynamic spectrum allocation, applied to a self-organizing emergency communication network consisting of several UAV nodes and at least one edge aggregation node, characterized in that, Includes the following steps: Offline pre-training step: Based on a preset spectrum dataset, an offline training and generation of a general channel feature pre-training model is performed. The general channel feature pre-training model is used to predict the corresponding channel quality based on the input spectrum information. Model initialization steps: At the start of the emergency communication mission, the general channel feature pre-trained model is loaded into at least one UAV node in the communication network as the initial local model of the UAV node; Online optimization steps: The drone node performs online optimization, which includes: Real-time perception of the spectrum information of the current environment; Based on the real-time perceived spectrum information, the initial local model is fine-tuned online to update the initial local model; Using the updated initial local model, the channel quality of candidate communication frequency bands is predicted, and the target communication frequency band is determined based on the predicted channel quality of the candidate communication frequency bands. Consensus verification steps: After determining the target communication frequency band, a consensus mechanism is initiated, and other nodes in the communication network verify the selection of the target communication frequency band. After successful verification, the switching to the target communication frequency band is executed.

2. The method according to claim 1, characterized in that, The online fine-tuning method is federated learning; The online fine-tuning of the initial local model based on the real-time sensed spectrum information to update the initial local model includes: The drone node uses the real-time perceived spectrum information to train the initial local model locally, so as to generate a fine-tuned local model. The drone node interacts with edge aggregation nodes in the network to perform federated updates to the local model based on the fine-tuned local model and model information from other drone nodes in the network.

3. The method according to claim 2, characterized in that, The steps of the federal update include: The drone node calculates and encrypts its model update information; The encrypted model update information is uploaded to the edge aggregation node; The edge aggregation node aggregates the received model update information to generate global model update parameters; The drone node downloads the global model update parameters to update its local model.

4. The method according to claim 3, characterized in that, The model update information is the model gradient.

5. The method according to claim 4, characterized in that, The encryption uses a homomorphic encryption algorithm.

6. The method according to claim 1, characterized in that, The preset spectrum dataset includes at least one of the following: A radio signal spectrum dataset covering various modulation signals and signal-to-noise ratios; The simulated spectrum dataset is generated based on channel modeling technology and is used to simulate the electromagnetic propagation and interference characteristics under typical disaster area terrain.

7. The method according to claim 1, characterized in that, The spectrum information is a three-dimensional spectrum heatmap, which includes time, frequency, and signal energy dimensions.

8. The method according to claim 1, characterized in that, The general channel feature pre-training model is a deep neural network containing convolutional neural network (CNN) layers and long short-term memory network (LSTM) layers.

9. The method according to claim 1, characterized in that, Determining the target communication frequency band includes: The target communication frequency band is selected by using a reinforcement learning algorithm based on the channel quality of the predicted candidate communication frequency band and at least one other preset decision factor.

10. The method according to claim 9, characterized in that, The reinforcement learning algorithm is the Q-Learning algorithm; The at least one other preset decision factor includes the channel capacity of the candidate communication frequency band and / or the mission importance level of the UAV node.

Images