Unmanned aerial vehicle low-altitude network communication management method based on 5G and Internet of Things

Abstract

An unmanned aerial vehicle low-altitude network communication management method based on 5G and Internet of Things relates to the technical field of communication control, and comprises the following steps: an unmanned aerial vehicle selects a 5G network slice to construct a virtual network, and sets a network slice bandwidth according to an allocated task, meanwhile, a plurality of available network slices are selected, bandwidth resources of the network slices are preset according to tasks, an unmanned aerial vehicle carries out air testing and evaluation, a communication environment is monitored and calculated, and the network slices capable of smoothly completing the tasks are screened by taking transmitting power as a judgment standard. The unmanned aerial vehicle is connected with the network slices and executes the flight task, the communication parameters of the unmanned aerial vehicle are dynamically adjusted in the network slices, so that the unmanned aerial vehicle is in the optimal parameter configuration to provide better communication quality, and the unmanned aerial vehicle automatically switches the network slices according to the communication parameter adjustment result. Through a mode of firstly screening network slices and then adjusting communication parameters in real time, the performance of the network slices is brought into full play, and the connection stability is ensured.

Description

Technical Field

[0001] The present invention relates to the field of communication control technology, and specifically to a method for managing low-altitude network communications of unmanned aerial vehicles based on 5G and the Internet of Things. Background Art

[0002] With the increasing application of drones in various fields, real-time data transmission and communication quality have become key issues. Traditional communication networks often cannot meet the drones' requirements for low latency and high bandwidth, especially in complex environments, they are easily interfered with, resulting in unstable data transmission. In addition, fixed communication parameter settings often cannot adapt to dynamically changing environments, further affecting communication quality. In response to these challenges, existing technologies mainly rely on improving network hardware performance or increasing transmission power, but these methods cannot fundamentally solve the problem and increase energy consumption and costs.

[0003] According to the method, device and medium for improving the communication quality of low-altitude communication of unmanned aerial vehicle terminals disclosed in publication number CN202210960796.X, it includes assigning a dedicated fifth-generation mobile communication technology service quality identifier 5QI value to the unmanned aerial vehicle terminal; adjusting the switching parameters of the unmanned aerial vehicle terminal assigned the 5QI value in the 5G public network, wherein the adjusted switching parameters of the unmanned aerial vehicle terminal are greater than the switching parameters corresponding to the user terminal in the 5G public network; sending the 5QI value and the adjusted switching parameters to the unmanned aerial vehicle terminal, so that the unmanned aerial vehicle terminal switches according to the 5QI value and the adjusted switching parameters.

[0004] Although the above patent realizes 5G-based network switching for drones, the method of judging switching is too single and does not fully utilize the performance of network slicing. Using a single evaluation criterion can easily lead to frequent switching due to waste of network performance, thus affecting connection stability. Summary of the invention

[0005] 1. Technical issues to be resolved

[0006] In view of the shortcomings of the existing technology, the present invention provides a low-altitude network communication management method for UAVs based on 5G and the Internet of Things. During the test flight of the UAV, multiple data such as signal strength and loss are detected and introduced, and multiple communication parameters are adjusted and optimized in real time through reinforcement learning during the actual mission execution, fully ensuring the stability of the connection side.

[0007] (II) Technical solution

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0009] A method for managing low-altitude network communication of unmanned aerial vehicles based on 5G and the Internet of Things, comprising:

[0010] Step 1: The drone selects a 5G network slice to build a virtual network and sets the network slice bandwidth according to the assigned task;

[0011] Step 2: The drone conducts aerial testing and evaluation, and selects network slices that can successfully complete the mission based on the communication environment;

[0012] Step 3: The drone connects to the network slice and performs the flight mission, and dynamically adjusts its communication parameters within the network slice;

[0013] Step 4: The drone automatically switches network slices based on the communication parameter adjustment results.

[0014] Furthermore, the drone selects 5G network slices to build a virtual network, and sets the logic of network slice bandwidth according to the assigned tasks. The drone initially selects a network slice based on the bandwidth, latency, and QoS required to perform the task, sets the number of tasks that the drone needs to perform and the weight of each task in the overall task, and allocates bandwidth resources of the network slice according to the weights of these tasks. The formula is as follows:

[0015]

[0016] R i is the bandwidth allocated to task i;

[0017] W i is the weight of task i, reflecting its priority;

[0018] is the sum of all task weights;

[0019] n is the total number of tasks;

[0020] R is the total bandwidth of the network slice;

[0021] The number of network slices selected in step 1 is more than one, and bandwidth allocation is preset in the same manner for all selected network slices.

[0022] Furthermore, step 2 is to conduct aerial testing and evaluation of the UAV, select network slices that can successfully complete the task according to the communication environment, test-fly the UAV and monitor the communication environment data. The communication environment data includes the signal strength received by the UAV, the distance between the UAV and the base station, and the signal frequency. The reflection, refraction and diffraction loss, phase noise power and background noise power between the UAV and the base station are obtained through the spectrum analyzer. The multipath signal analyzer detects the loss caused by the multipath effect. The signal strength detection instrument obtains the penetration loss by detecting the signal strength change of the UAV behind different obstacles. The signal strength, signal-to-noise ratio and delay are calculated respectively to judge the communication status of the current UAV communication in the current environment. The signal strength RSSI is based on the following formula:

[0023] RSSI = P received -P loss

[0024] FSPL = 20log 10 (d)+20log 10 (f)+20log 10 (4π / c)

[0025] P loss =FSPL+FZ+DS+CS

[0026] Among them, P received is the signal strength received by the drone, P loss It is the power loss suffered by the signal during propagation, FSPL is the free space path loss, d is the communication distance, f is the signal frequency, c is the speed of light, d, f, c are dimensionless, FZ is the reflection, refraction and diffraction loss, DS is the multipath effect loss, CS is the penetration loss, when the RSSI value is higher than -70dBm, it is determined that the communication quality is good and the signal strength is high, which can meet the current communication needs, otherwise the signal strength is determined to not meet the communication needs, and the signal transmission power is increased to improve the signal strength;

[0027] The phase noise and environmental background noise are measured by a spectrum analyzer, and the signal-to-noise ratio (SNR) is based on the following formula:

[0028]

[0029] N=kTB

[0030] P signal is the signal power, N is the thermal noise power, k is the Boltzmann constant, T is the absolute temperature of the environment, B is the bandwidth, XZ phase noise power, BZ is the background noise power,

[0031] When SNR>25dB, it is determined that the signal-to-noise ratio required for communication is met, otherwise it is determined that the signal-to-noise ratio required for communication is not met, and the transmission power is increased to enhance the signal reception effect;

[0032] The formula for delay is as follows:

[0033] Latency = T received -T sent

[0034] Among them, T received is the packet arrival time, T sent It is the time it takes to send a data packet. When Latency < 10ms, it is considered that the delay required for communication is met. Otherwise, it is considered that the delay is too high and does not meet the communication requirements. The transmission power is increased to reduce the delay.

[0035] Increase the transmission power so that the signal strength, signal-to-noise ratio and delay meet the communication requirements. fs Make a judgment, when P fs >2 / 3P za When P fs ≤2 / 3P za When P za is the maximum transmit power of the base station;

[0036] Step 2 detects all network slices selected in step 1, retains the slices that enable the drone to successfully complete the mission, and removes the network slices that cannot enable the drone to successfully complete the mission.

[0037] Furthermore, in step 3, the drone connects to the network slice and performs a flight mission, and dynamically adjusts its communication parameters within the network slice. The drone selects a network slice from step 2, and the remaining network slices are used as backup. The drone performs the flight mission and adjusts the communication parameters within the network slice in real time. The communication parameters recorded by the drone are signal strength, broadband utilization, and interference strength. The signal strength and interference strength are dimensionlessly processed, and the dynamic learning method is used to adjust the communication parameters. The reward function is set as follows:

[0038] R=α·RSSI+β·B-γ·P loss

[0039] R is the instant reward, RSSI is the signal strength, B is the bandwidth utilization, P is the bandwidth utilization, loss is the interference intensity, α, β, γ are weight coefficients, α>0, β>0, γ>0;

[0040] The actions of setting parameter adjustment include increasing the transmission power a1, switching the modulation mode a2, switching the antenna mode a3, and increasing the bandwidth a4. The Q(s1, a4) of each action is set in the initial state s1. y ) values ​​are all 0, and then Q-learning reinforcement learning is performed. In the current state of the communication parameter s i Next, select an action a y , the state of the communication system enters the next state s i+1 , and generates an immediate reward R based on the new state, in the next state s i+1 Select the action with the largest Q value, and update the current state through weighted operation to perform action a y The Q value of each action in each state is continuously updated and recorded into a Q table. The formula is as follows:

[0041]

[0042] α is the learning rate, α>0, R is the immediate reward, γ is the discount factor, γ>0, s i+1 is the new state after executing the action, Q(s i ,a y ) means in state s i Next, perform action a y Q value, i represents the retrieval variable of the state sequence, i∈N, y represents the retrieval variable of the action, y∈N and 1≤y≤3, Indicates that in the next state s i+1 The maximum Q value generated by all actions in , before taking the first step, the next state s i+1 The maximum Q value is set to 0, Q new (s i ,a y ) is the updated state s i Conduct a y Q value, Q old (s i ,a y ) is the state before update s i Conduct a y Q value;

[0043] When the Q value update amplitude is less than 0.01, the communication parameters at that time are considered to be in the optimal state, and the transmission power, modulation mode, antenna mode, and bandwidth are set according to this state. Otherwise, training and learning will continue to cope with the real-time changing environment. When the transmission power is greater than 100% or the total bandwidth of all tasks exceeds the maximum allowable bandwidth of this network slice and the Q value update is still greater than 0.01, the network slice is switched.

[0044] Furthermore, in the Q-learning reinforcement learning process, the ε-greedy strategy is used to select actions to continuously reduce the Q value update amplitude. The ε value is artificially set in the (0,1) interval, and an action is randomly selected with probability ε to update the Q value. At the same time, the action with the highest Q value in the current state is selected with probability 1-ε to update the Q value, and each time an action is selected, ε is decreased. The formula is:

[0045] ∈ t =∈0e -λt

[0046] λ is the control deceleration speed parameter, λ>0, ∈0 is the initial setting value, ∈ t is the value after t actions are selected.

[0047] Furthermore, the step 4 is that the drone automatically switches the network slice according to the communication parameter adjustment result. During the step 3, after the demand for switching the network slice is obtained, the drone does not disconnect the connection of the current network slice. While maintaining the current network slice, the drone connects to another network slice, so that the drone is connected to two network slices at the same time. The current network slice is calibrated as network slice I, and the other network slice is calibrated as network slice II. The drone collects the state data of network slice I that has undergone reinforcement learning and moves it to network slice II. The state data includes the current transmission power, modulation mode, antenna mode, bandwidth and Q table. Step 3 is continued in network slice II to determine whether network slice II can obtain the optimal state of communication parameters. If network slice II obtains the optimal state of communication parameters, the transmission power, modulation mode, antenna mode and bandwidth of network slice II are set according to the optimal state, the connection of network slice I is cut off, and all traffic is redirected to network slice II that meets the conditions;

[0048] When network slice II also concludes that it is necessary to switch network slices, another backup network slice is used as the new network slice II to execute state data transfer and step 3 again, traversing all backup network slices until the optimal state of the communication parameters is obtained. When the optimal state of the communication parameters cannot be obtained after traversing all backup networks, the network slice with the smallest Q value update is selected as the optimal network slice and switched to this network slice.

[0049] (III) Beneficial effects

[0050] The present invention provides a method for managing low-altitude network communications for UAVs based on 5G and the Internet of Things. By screening network slices for missions and test flights, preparations can be made for the execution of UAV missions. At the same time, during the execution of UAV missions, the connection performance between the UAV and the base station is continuously improved through reinforcement learning methods, thereby giving full play to the performance of network slices and ensuring the stability of the connection. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0052] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0053] A method for managing low-altitude network communication of unmanned aerial vehicles based on 5G and the Internet of Things, comprising:

[0054] Step 1: The drone selects a 5G network slice to build a virtual network and sets the network slice bandwidth according to the assigned task;

[0055] As a preferred embodiment, the drone selects 5G network slices to build a virtual network, and the logic of setting the network slice bandwidth according to the assigned tasks is that the drone preliminarily selects the network slice according to the bandwidth, delay and QoS required to perform the task, sets the number of tasks that the drone needs to perform and the weight of each task in the overall task, and allocates the bandwidth resources of the network slice according to the weights of these tasks, based on the formula:

[0056]

[0057] R i is the bandwidth allocated to task i;

[0058] W i is the weight of task i, reflecting its priority;

[0059] is the sum of all task weights;

[0060] n is the total number of tasks;

[0061] R is the total bandwidth of the network slice;

[0062] The number of network slices selected in step 1 is more than one, and bandwidth allocation is preset in the same manner for all selected network slices.

[0063] By preliminarily selecting multiple network slices, the optimal network slice can be screened out and preparations can be made for the UAV to switch network slices during mission execution. At the same time, the slices preliminarily set are all available network slices that can meet all UAV tasks, which provides a guarantee for the automatic switching of network slices in step 4, avoids fluctuations in network slices during the switching process and possible verification failures, and ensures the stability of connection performance.

[0064] Step 2: The drone conducts aerial testing and evaluation, and selects network slices that can successfully complete the mission based on the communication environment;

[0065] As a preferred embodiment, the UAV is tested and the communication environment data is monitored. The communication environment data includes the signal strength received by the UAV, the distance between the UAV and the base station and the signal frequency. The reflection, refraction and diffraction loss, phase noise power and background noise power between the UAV and the base station are obtained by the spectrum analyzer. The multipath signal analyzer detects the loss caused by the multipath effect. The signal strength detection instrument obtains the penetration loss by detecting the signal strength change of the UAV behind different obstacles. The signal strength, signal-to-noise ratio and delay are calculated respectively to judge the communication status of the current UAV communication in the current environment. The signal strength RSSI is based on the following formula:

[0066] RSSI = P received -P loss

[0067] FSPL = 20log 10 (d)+20log 10 (f)+20log 10 (4π / c)

[0068] P loss =FSPL+FZ+DS+CS

[0069] Among them, P received is the signal strength received by the drone, P loss It is the power loss suffered by the signal during propagation, FSPL is the free space path loss, d is the communication distance, f is the signal frequency, c is the speed of light, d, f, c are dimensionless, FZ is the reflection, refraction and diffraction loss, DS is the multipath effect loss, CS is the penetration loss, when the RSSI value is higher than -70dBm, it is determined that the communication quality is good and the signal strength is high, which can meet the current communication needs, otherwise the signal strength is determined to not meet the communication needs, and the signal transmission power is increased to improve the signal strength;

[0070] The phase noise and environmental background noise are measured by a spectrum analyzer, and the signal-to-noise ratio (SNR) is based on the following formula:

[0071]

[0072] N=kTB

[0073] P signal is the signal power, N is the thermal noise power, k is the Boltzmann constant, T is the absolute temperature of the environment, B is the bandwidth, XZ phase noise power, BZ is the background noise power,

[0074] When SNR>25dB, it is determined that the signal-to-noise ratio required for communication is met, otherwise it is determined that the signal-to-noise ratio required for communication is not met, and the transmission power is increased to enhance the signal reception effect;

[0075] The formula for delay is as follows:

[0076] Latency = T received -T sent

[0077] Among them, T received is the packet arrival time, T sent It is the time it takes to send a data packet. When Latency < 10ms, it is considered that the delay required for communication is met. Otherwise, it is considered that the delay is too high and does not meet the communication requirements. The transmission power is increased to reduce the delay.

[0078] Increase the transmission power so that the signal strength, signal-to-noise ratio and delay meet the communication requirements. fs Make a judgment, when P fs >2 / 3P za When P fs ≤2 / 3P za When P za is the maximum transmit power of the base station;

[0079] Step 2 detects all network slices selected in step 1, retains the slices that enable the drone to successfully complete the mission, and removes the network slices that cannot enable the drone to successfully complete the mission.

[0080] During the test flight, by real-time collection and analysis of signal strength, signal-to-noise ratio, and delay parameters, each network slice selected from step 1 will be judged to obtain a network slice with excellent connection performance. At the same time, preparations will be made for the automatic switching of the drone network slice to ensure a smooth connection.

[0081] Step 3: The drone connects to the network slice and performs the flight mission, and dynamically adjusts its communication parameters within the network slice;

[0082] As a preferred embodiment, the drone selects a network slice from step 2, and the remaining network slices are used as backup. The drone performs a flight mission and adjusts the communication parameters in the network slice in real time. The communication parameters recorded by the drone are signal strength, broadband utilization, and interference strength. The signal strength and interference strength are dimensionlessly processed, and the dynamic learning method is used to adjust the communication parameters. The reward function is set as follows:

[0083] R=α·RSSI+β·B-γ·Ploss

[0084] R is the instant reward, RSSI is the signal strength, B is the bandwidth utilization, P is the bandwidth utilization, loss is the interference intensity, α, β, γ are weight coefficients, α>0, β>0, γ>0;

[0085] The actions of setting parameter adjustment include increasing the transmission power a1, switching the modulation mode a2, switching the antenna mode a3, and increasing the bandwidth a4. The Q(s1, a4) of each action is set in the initial state s1. y ) values ​​are all 0, and then Q-learning reinforcement learning is performed. In the current state of the communication parameter s i Next, select an action a y , the state of the communication system enters the next state s i+1 , and generates an immediate reward R based on the new state, in the next state s i+1 Select the action with the largest Q value, and update the current state through weighted operation to perform action a y The Q value of each action in each state is continuously updated and recorded into a Q table. The formula is as follows:

[0086]

[0087] α is the learning rate, α>0, R is the immediate reward, γ is the discount factor, γ>0, s i+1 is the new state after executing the action, Q(s i ,a y ) means in state s i Next, perform action a y Q value, i represents the retrieval variable of the state sequence, i∈N, y represents the retrieval variable of the action, y∈N and 1≤y≤3, Indicates that in the next state s i+1 The maximum Q value generated by all actions in , before taking the first step, the next state s i+1 The maximum Q value is set to 0, Q new (s i ,a y ) is the updated state s i Conduct a y Q value, Q old (s i ,a y ) is the state before update s i Conduct a y Q value;

[0088] When the Q value update amplitude is less than 0.01, the communication parameters at that time are considered to be in the optimal state, and the transmission power, modulation mode, antenna mode, and bandwidth are set according to this state. Otherwise, training and learning will continue. When the transmission power is greater than 100% or the total bandwidth of all tasks exceeds the maximum allowed bandwidth of this network slice and the Q value update is still greater than 0.01, the network slice will be switched.

[0089] As a preferred embodiment, in the Q-learning reinforcement learning process, the ε-greedy strategy is used to select actions to continuously reduce the Q value update amplitude. The ε value is artificially set in the (0,1) interval, and an action is randomly selected with probability ε to update the Q value. At the same time, the action with the highest Q value in the current state is selected with probability 1-ε to update the Q value, and each time an action is selected, ε is decreased. The formula is:

[0090] ∈ t =∈0e -λt

[0091] λ is the control deceleration speed parameter, λ>0, ∈0 is the initial setting value, ∈ t is the value after t actions are selected.

[0092] Through the dynamic adaptive adjustment of reinforcement learning, drones can automatically optimize status data in different environments. During the execution of tasks, they continuously optimize the communication connection performance within the network slice to ensure stable and efficient data transmission. By optimizing within the network slice that already has the connection conditions, the quality of the communication connection is further improved.

[0093] Step 4: The drone automatically switches network slices based on the communication parameter adjustment results.

[0094] As a preferred embodiment, during step 3, after the demand for switching network slices is obtained, the drone does not disconnect the connection of the current network slice. While maintaining the current network slice, the drone connects to another network slice, so that the drone is connected to two network slices at the same time. The current network slice is marked as network slice I, and the other network slice is marked as network slice II. The drone collects the state data of network slice I that has undergone reinforcement learning and moves it to network slice II. The state data includes the current transmission power, modulation mode, antenna mode, bandwidth and Q table. Step 3 is continued in network slice II to determine whether network slice II can obtain the optimal state of communication parameters. If network slice II obtains the optimal state of communication parameters, the transmission power, modulation mode, antenna mode and bandwidth of network slice II are set according to the optimal state, the connection of network slice I is cut off, and all traffic is redirected to network slice II that meets the conditions;

[0095] When network slice II also concludes that it is necessary to switch network slices, another backup network slice is used as the new network slice II to execute state data transfer and step 3 again, traversing all backup network slices until the optimal state of the communication parameters is obtained. When the optimal state of the communication parameters cannot be obtained after traversing all backup networks, the network slice with the smallest Q value update is selected as the optimal network slice and switched to this network slice.

[0096] By maintaining the connection of one network slice while calculating the internal parameters of the second network slice connection, the uninterrupted connection of the drone is guaranteed, while the short dual network slice connection also saves energy consumption.

[0097] The above formulas are all dimensionless and numerical calculations. The formula is a formula that is closest to the actual situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0098] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0099] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for managing low-altitude network communication of unmanned aerial vehicles based on 5G and the Internet of Things, characterized in that the steps include: Step 1: The drone selects a 5G network slice to build a virtual network and sets the network slice bandwidth according to the assigned task; Step 2: The drone conducts aerial testing and evaluation, and selects network slices that can successfully complete the mission based on the communication environment; Step 3: The drone connects to the network slice and performs the flight mission, and dynamically adjusts its communication parameters within the network slice; Step 4: The drone automatically switches network slices based on the communication parameter adjustment results.

2. The method for managing low-altitude network communication of unmanned aerial vehicles based on 5G and the Internet of Things according to claim 1 is characterized in that: The logic of the drone selecting 5G network slices to build a virtual network and setting the network slice bandwidth according to the assigned tasks is that the drone initially selects the network slice according to the bandwidth, latency and QoS required to perform the task, sets the number of tasks that the drone needs to perform and the weight of each task in the overall task, and allocates the bandwidth resources of the network slice according to the weights of these tasks. The formula is: R i is the bandwidth allocated to task i; W i is the weight of task i, reflecting its priority; is the sum of all task weights; n is the total number of tasks; R is the total bandwidth of the network slice; The number of network slices selected in step 1 is more than one, and bandwidth allocation is preset in the same manner for all selected network slices.

3. The method for managing low-altitude network communication of unmanned aerial vehicles based on 5G and the Internet according to claim 2 is characterized in that: Step 2 is to conduct aerial testing and evaluation of the drone. According to the communication environment, the network slices that can successfully complete the task are selected. The drone is tested and the communication environment data is monitored. The communication environment data includes the signal strength received by the drone, the distance between the drone and the base station, and the signal frequency. The reflection, refraction and diffraction loss, phase noise power and background noise power between the drone and the base station are obtained through the spectrum analyzer. The multipath signal analyzer detects the loss caused by the multipath effect. The signal strength detection instrument obtains the penetration loss by detecting the signal strength change of the drone behind different obstacles. The signal strength, signal-to-noise ratio and delay are calculated respectively to judge the communication status of the current drone in the current environment. The signal strength RSSI is based on the following formula: RSSI=P received -P loss FSPL=20log 10 (d)+20log 10 (f)+20log 10 (4π / c) P loss =FSPL+FZ+DS+CS Among them, P received is the signal strength received by the drone, P loss It is the power loss suffered by the signal during propagation, FSPL is the free space path loss, d is the communication distance, f is the signal frequency, c is the speed of light, d, f, c are dimensionless, FZ is the reflection, refraction and diffraction loss, DS is the multipath effect loss, CS is the penetration loss, when the RSSI value is higher than -70dBm, it is determined that the communication quality is good and the signal strength is high, which can meet the current communication needs, otherwise the signal strength is determined to not meet the communication needs, and the signal transmission power is increased to improve the signal strength; The phase noise and environmental background noise are measured by a spectrum analyzer, and the signal-to-noise ratio (SNR) is based on the following formula: N=kTB P signal is the signal power, N is the thermal noise power, k is the Boltzmann constant, T is the absolute temperature of the environment, B is the bandwidth, XZ phase noise power, BZ is the background noise power, When SNR>25dB, it is determined that the signal-to-noise ratio required for communication is met, otherwise it is determined that the signal-to-noise ratio required for communication is not met, and the transmission power is increased to enhance the signal reception effect; The formula for delay is as follows: Latency=T received -T sent Among them, T received is the packet arrival time, T sent It is the time it takes to send a data packet. When Latency < 10ms, it is considered that the delay required for communication is met. Otherwise, it is considered that the delay is too high and does not meet the communication requirements. The transmission power is increased to reduce the delay. Increase the transmission power so that the signal strength, signal-to-noise ratio and delay meet the communication requirements. fs Make a judgment, when P fs >2 / 3P za When P fs ≤2 / 3P za When P za is the maximum transmit power of the base station; Step 2 detects all network slices selected in step 1, retains the slices that enable the drone to successfully complete the mission, and removes the network slices that cannot enable the drone to successfully complete the mission.

4. The method for managing low-altitude network communication of unmanned aerial vehicles based on 5G and the Internet according to claim 3 is characterized in that: In step 3, the drone connects to the network slice and performs the flight mission, and dynamically adjusts its communication parameters within the network slice. The drone selects a network slice from step 2, and the remaining network slices are used as backup. The drone performs the flight mission and adjusts the communication parameters within the network slice in real time. The communication parameters recorded by the drone are signal strength, broadband utilization, and interference strength. The signal strength and interference strength are dimensionlessly processed, and the dynamic learning method is used to adjust the communication parameters. The reward function is set as follows: R=α·RSSI+β·B-γ·P loss R is the instant reward, RSSI is the signal strength, B is the bandwidth utilization, P is the bandwidth utilization, loss is the interference intensity, α, β, γ are weight coefficients, α>0, β>0, γ>0; The actions of setting parameter adjustment include increasing the transmission power a1, switching the modulation mode a2, switching the antenna mode a3, and increasing the bandwidth a4. The Q(s1, a4) of each action is set in the initial state s1. y ) values ​​are all 0, and then Q-learning reinforcement learning is performed. In the current state of the communication parameter s i Next, select an action a y , the state of the communication system enters the next state s i+1 , and generates an immediate reward R based on the new state, in the next state s i+1 Select the action with the largest Q value, and update the current state through weighted operation to perform action a y The Q value of each action in each state is continuously updated and recorded into a Q table. The formula is as follows: α is the learning rate, α>0, R is the immediate reward, γ is the discount factor, γ>0, s i+1 is the new state after executing the action, Q(s i ,a y ) means in state s i Next, perform action a y Q value, i represents the retrieval variable of the state sequence, i∈N, y represents the retrieval variable of the action, y∈N and 1≤y≤3, Indicates that in the next state s i+1 The maximum Q value generated by all actions in , before taking the first step, the next state s i+1 The maximum Q value is set to 0, Q new (s i ,a y ) is the updated state s i Conduct a y Q value, Q old (s i ,a y ) is the state s before updating i Conduct a y Q value; When the Q value update amplitude is less than 0.01, the communication parameters at that time are considered to be in the optimal state, and the transmission power, modulation mode, antenna mode, and bandwidth are set according to this state. Otherwise, training and learning will continue. When the transmission power is greater than 100% or the total bandwidth of all tasks exceeds the maximum allowed bandwidth of this network slice and the Q value update is still greater than 0.01, the network slice will be switched.

5. The method for managing low-altitude network communication of unmanned aerial vehicles based on 5G and the Internet according to claim 4 is characterized in that: In the Q-learning reinforcement learning process, the ε-greedy strategy is used to select actions to continuously reduce the Q value update amplitude. The ε value is artificially set in the (0,1) interval, and an action is randomly selected with probability ε to update the Q value. At the same time, the action with the highest Q value in the current state is selected with probability 1-ε to update the Q value. Each time an action is selected, ε decreases. The formula is: ∈ t ＝∈0e -λt λ is the control deceleration speed parameter, λ>0, ∈0 is the initial setting value, and ∈t is the value after t action selections.

6. The method for managing low-altitude network communication of unmanned aerial vehicles based on 5G and the Internet according to claim 5 is characterized in that: The step 4 is that the drone automatically switches the network slice according to the communication parameter adjustment result. In the process of step 3, after the demand for switching the network slice is obtained, the drone does not disconnect the connection of the current network slice. While maintaining the current network slice, the drone connects to another network slice, so that the drone is connected to two network slices at the same time. The current network slice is marked as network slice I, and the other network slice is marked as network slice II. The drone collects the state data of network slice I that has undergone reinforcement learning and moves it to network slice II. The state data includes the current transmission power, modulation mode, antenna mode, bandwidth and Q table. Step 3 is continued in network slice II to determine whether network slice II can obtain the optimal state of communication parameters. If network slice II obtains the optimal state of communication parameters, the transmission power, modulation mode, antenna mode and bandwidth of network slice II are set according to the optimal state, the connection of network slice I is cut off, and all traffic is redirected to network slice II that meets the conditions; When network slice II also concludes that it is necessary to switch network slices, another backup network slice is used as the new network slice II to execute state data transfer and step 3 again, traversing all backup network slices until the optimal state of the communication parameters is obtained. When the optimal state of the communication parameters cannot be obtained after traversing all backup networks, the network slice with the smallest Q value update is selected as the optimal network slice and switched to this network slice.

Images