A communication method combining network AI and drone air-ground integration

Abstract

The present invention discloses a communication method for network AI combined with unmanned aerial vehicle (UAV) air-ground integration, including: loading an SRv6 network chip and a Beidou signal receiver on a UAV, establishing an aerial wireless network for UAVs through the SRv6 protocol, assigning an independent IPv6 address to each UAV, and automatically assigning an initial path identifier to the network routing; obtaining the aerial coordinates of the UAV through the Beidou signal receiver, combining possible obstacles between the UAVs, building a collision model, and calculating the probability of collision between the UAV and the obstacle, and storing it in the Argument of the SRH; using a Markov chain to build an early warning model, obtaining the probability of an alarm associated with the initial path identifier of the current UAV identifier, and storing it in the Argument of the SRH; the network routing updates the initial path identifier based on the probability of collision and the alarm probability. The communication method of the present invention reduces the resource consumption of message forwarding and avoids the occurrence of aerial failures.

Description

Technical Field

[0001] The present invention relates to the field of unmanned aerial vehicle (UAV) air-ground integrated communication technology, and in particular to a UAV air-ground integrated communication method combining network AI with UAV. Background Art

[0002] As an emerging aerial technology, drones offer low-altitude flight, wide coverage, flexible viewing angles, grid-based deployment, short preparation times, and ready deployment. They are also capable of performing high-risk missions in hazardous weather or polluted environments. Consequently, they are widely used in industries such as environmental monitoring, traffic management, fire emergency response, pollution monitoring, urban management, and commodity transportation. Using drone base stations as carriers, drone management platforms as interfaces, and geospatial intelligence as a core, they integrate multifunctional sensors, powerful mapping, and AI computing modules to build an air-ground collaborative networking model, enabling high-precision, high-frequency, fully automated, and three-dimensional monitoring of surface resources, creating a drone aerial intelligent remote sensing network.

[0003] Segment Routing (SR) is a source routing technology, and SRv6 is the application of SR technology to IPv6 networks. The emergence of SRv6 is a major innovation. Combined with SDN technology, it enables programmable networks, providing innovative opportunities for network infrastructure and value-added services in drone-based intelligent remote sensing networks.

[0004] In existing technologies, the main function of the aerial SRv6 network system is to comprehensively transmit information and track and locate drones, enabling drones to collect and transmit information in all directions through relevant network media. If the SRv6 network data connection medium is disconnected, the drone will be unable to be effectively operated and controlled, ultimately causing the drone to make a hard landing. Summary of the Invention

[0005] In response to the problems existing in the prior art, the present invention provides a communication method that combines network AI with UAV air-ground integration.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a network AI combined with UAV air-ground integrated communication method, specifically comprising the following steps:

[0007] Step S1: SRv6 network chips and BeiDou signal receivers are installed on each UAV. The UAV aerial wireless network is established through the SRv6 protocol, and an independent IPv6 address is assigned to each UAV. The network routing automatically allocates the nearest path table to the target UAV and generates an initial path identifier based on the IPv6 address.

[0008] Step S2: The aerial coordinates of each drone are obtained in real time through the Beidou signal receiver. A collision model is constructed based on the aerial coordinates of all drones and possible aerial obstacles between drones. The probability of collision between all drones and aerial obstacles is calculated, and the collision probability of each drone is stored in the Argument of the corresponding IPv6 extension header SRH.

[0009] Step S3: Use the Markov chain to build an early warning model. According to the health status of the drone, the probability of the drone associated with the initial path identifier of the current drone identifier generating an alarm is obtained, and the probability of the current drone generating an alarm is stored in the Argument of the corresponding IPv6 extension header SRH;

[0010] Step S4: Network routing updates the initial path identifier based on the probability of collision and the warning probability of the drone stored in Argument to achieve integrated air-ground communication of the drone.

[0011] Furthermore, the process of building a collision model is as follows:

[0012] The aerial obstacle is taken as the center of the coverage area, the center coordinate is (x0, y0), the area radius is r, and the standard discrete coefficient value β is obtained by derivative in the time region Δt; at the same time, based on the aerial coordinate set G of the UAV t And the time point set T of the corresponding coordinate point, combined with the optimized wireless positioning algorithm, obtain the average dispersion coefficient value Z of the corresponding drone;

[0013] Within the current area radius and movement time range, the standard dispersion coefficient value β and the average dispersion coefficient value Z are compared. If Z is greater than β, it means that the current drone has no intersection with the coverage area, and there will be no collision between the current drone and the aerial obstacle. Conversely, if Z is less than or equal to β, the larger the difference between the two, K = β - Z, the more frequently the current drone intersects with the coverage area, and the greater the probability of collision between the current drone and the aerial obstacle.

[0014] Furthermore, the calculation process of the average dispersion coefficient value Z is as follows:

[0015] The aerial coordinate set G of the UAV t The aerial coordinates of a certain UAV are (x t ,y t ), the coordinates correspond to the time point t, and the discrete average value s is calculated by combining the optimized wireless positioning algorithm:

[0016]

[0017] The aerial coordinate set G of the UAV t Perform matrix operations on the time point set T corresponding to the coordinate point to obtain the vector parameter value

[0018]

[0019] The discrete mean value s and the vector parameter value Multiply to get the average dispersion coefficient value Z:

[0020]

[0021] Furthermore, the calculation process of the standard dispersion coefficient value β is as follows:

[0022]

[0023] Among them, K represents the difference between the standard dispersion coefficient value β and the average dispersion coefficient value Z. K≤0 means that there is no intersection between the current drone and the coverage area, and K>0 means that there is an intersection between the current drone and the coverage area.

[0024] Furthermore, the construction process of the early warning model is as follows:

[0025] X(t+1)=X(t)×P

[0026] Among them, X(t) represents the state vector of trend analysis and drone health status at time t, P represents the one-step transition probability matrix, and X(t+1) represents the state vector of trend analysis and drone health status at time t+1.

[0027] Furthermore, the specific implementation process of step S4 is as follows: sort the collision probabilities and alarm probabilities of all drones from small to large, the smaller the collision probability or alarm probability, the higher the priority, obtain the IPv6 address of the drone with high priority in the identifier from the Argument of the IPv6 extension header SRH of the current drone, update the initial path identifier, and set the priority parameter value in the 802.1Q-Tag cos field through the IPv6 message header information. The value range is 8-15, and the priority parameter value is set to 8 as the highest priority and 15 as the lowest priority. The SRv6 message forwarding priority setting is completed to realize the integrated air-ground communication of drones.

[0028] Compared with the existing technology, the present invention has the following beneficial effects: the network AI combined with the drone air-ground integrated communication method of the present invention adds AI fault prediction and path intelligent optimization compared to the existing SRv6 message forwarding mechanism, provides a new solution for the forwarding of complex business messages in aerial wireless networking, and reduces the resource consumption of message forwarding; when the drone network encounters collisions and failures, the risk can be predicted in advance. While taking avoidance procedures, the drone will quickly switch the network information that should belong to it to transmit signals to other drones, thereby ensuring signal quality and avoiding the occurrence of aerial failures; the network AI combined with the drone air-ground integrated communication method of the present invention breaks through the unscientific nature of the current SRv6 protocol routing algorithm for automatically allocating the IPv6 address closest to the destination drone, and optimizes the initial path selection by combining artificial intelligence analysis and prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 This is a framework diagram of the network AI combined with the UAV air-ground integrated communication method of the present invention;

[0030] Figure 2 This is the SRv6 networking diagram of the aerial drone of the present invention;

[0031] Figure 3 This is the SRv6 protocol structure diagram. DETAILED DESCRIPTION

[0032] The technical solution of the present invention will be further explained below with reference to the accompanying drawings.

[0033] like Figure 1 This is a framework diagram of the network AI combined with the UAV air-ground integrated communication method of the present invention, which specifically includes the following steps:

[0034] Step S1: Figure 2, SRv6 network chips and BeiDou signal receivers are installed on each UAV, and the SRv6 protocol communication format is adopted to improve communication efficiency and enhance confidentiality; BeiDou signal receivers are used to improve the positioning accuracy of UAVs; UAV aerial wireless networking is established through the SRv6 protocol, and the SRv6 networking method is adopted to improve inter-machine communication efficiency and reduce the probability of packet loss; and an independent IPv6 address is assigned to each UAV, and the network routing automatically assigns the nearest path table Segment List to reach the target UAV, which is stored in the SRv6 message protocol Segment List. By extracting the IPv6 addresses of all SIDs in the Segment List, they are concatenated from back to front, separated by ###, to generate an initial path identifier based on the IPv6 address. This method is used for all UAVs to obtain the initial path identifier of the wireless networking of aerial UAVs. The format of the initial path identifier in the present invention is: IPv6 address of the current UAV ###IPv6 address 2###IPv6 address N###IPv6 address of the target UAV.

[0035] like Figure 3 The IPv6 extension header SRH is mainly composed of Locator, Function, and Argument. Among them, Locator is a location identifier, which is an identifier assigned to a network node in the network and can be used to route and forward data packets. Locator has two important attributes: routing and aggregation. In SRv6 SID, Locator is a variable-length part used to adapt to networks of different sizes; Function is an ID value assigned by the device to the local forwarding instruction. This value can be used to express the forwarding action that the device needs to perform, which is equivalent to the operation code of the computer instruction. In SRv6 network programming, different forwarding behaviors are expressed by different function IDs; Argument is an optional field, which is a supplement to Function. It is the parameters required when the forwarding instruction is executed. These parameters may contain flows, services or any other related variable information.

[0036] Step S2: The aerial coordinates of each drone are obtained in real time through the Beidou signal receiver. A collision model is constructed based on the aerial coordinates of all drones and possible aerial obstacles between drones. The probability of collision between all drones and aerial obstacles is calculated, and the probability of collision between each drone is stored in the Argument of the corresponding IPv6 extension header SRH. The probability of collision between each drone is used as an identification string and as a carrier for transmission between drone nodes.

[0037] The specific process of constructing the collision model in the present invention is as follows: taking the air obstacle as the center of the coverage area, the center coordinates are (x0, y0), the area radius is r, and taking the derivative in the time area Δt to obtain the standard discrete coefficient value β; at the same time, based on the aerial coordinate set G of the UAV t And the time point set T of the corresponding coordinate point, combined with the optimized wireless positioning algorithm, obtain the average dispersion coefficient value Z of the corresponding drone;

[0038] Within the current area radius and movement time range, the standard dispersion coefficient value β and the average dispersion coefficient value Z are compared. If Z is greater than β, it means that the current drone has no intersection with the coverage area, and there will be no collision between the current drone and the aerial obstacle. Conversely, if Z is less than or equal to β, the larger the difference between the two, K = β - Z, the more frequently the current drone intersects with the coverage area, and the greater the probability of collision between the current drone and the aerial obstacle.

[0039] The calculation process of the average dispersion coefficient value Z is as follows:

[0040] The aerial coordinate set G of the UAV t The aerial coordinates of a certain UAV are (x t ,y t ), the coordinates correspond to the time point t, and the discrete average value s is calculated by combining the optimized wireless positioning algorithm:

[0041]

[0042] The aerial coordinate set G of the UAV t Perform matrix operations on the time point set T corresponding to the coordinate point to obtain the vector parameter value

[0043]

[0044] The discrete mean value s and the vector parameter value Multiply to get the average dispersion coefficient value Z:

[0045]

[0046] The calculation process of the standard dispersion coefficient value β is as follows:

[0047]

[0048] Among them, K represents the difference between the standard dispersion coefficient value β and the average dispersion coefficient value Z. K≤0 means that there is no intersection between the current drone and the coverage area, and K>0 means that there is an intersection between the current drone and the coverage area.

[0049] Step S3, using the Markov chain to build an early warning model, according to the health status of the drone, obtain the probability of the drone associated with the initial path identifier of the current drone identifier to issue an alarm. The advantage of the Markov chain early warning is to predict the next probability through the current value. Except for the initial data which requires historical data, all the data are analyzed based on the previous data, which is very suitable for the scenario with limited data storage and computing resources in this application. The probability of the current drone issuing an alarm is stored in the Argument of the corresponding IPv6 extension header SRH, which can be passed to the next drone node along with the network data, and by parsing the identifier in the Argument, the preset alarm probability of the current drone node can be immediately obtained, avoiding the need to calculate the alarm probability at the next drone node. At the same time, as the IPv6 extension header SRH transmits data, the result avoids a large amount of analyzed data and reduces the resource consumption of data transmission.

[0050] The construction process of the early warning model in the present invention is:

[0051] X(t+1)=X(t)×P

[0052] Among them, X(t) represents the state vector of trend analysis and drone health status at time t, P represents the one-step transition probability matrix, and X(t+1) represents the state vector of trend analysis and drone health status at time t+1.

[0053] In step S4, the network router updates the initial path identifier based on the collision probability and warning probability of the drone stored in the Argument, thus achieving integrated air-to-ground communication between drones. Since the drone node has just entered the air network, the shortest path to the destination drone node automatically assigned by the network router is the shortest path, but not necessarily the optimal path for integrated air-to-ground communication between drones. Therefore, it is necessary to confirm the health of the passing drone nodes using the warning probability and the safety factor of the passing drone nodes using the collision probability. Specifically, all drones are sorted from small to large by their collision probability and warning probability. The lower the collision probability or warning probability, the higher the priority. The IPv6 address of the drone with the highest priority is obtained from the Argument of the current drone's IPv6 extension header (SRH). The initial path identifier is updated. The priority parameter value is set in the 802.1Q-Tag cos field using the IPv6 packet header information. The value range is 8-15, with a priority parameter value of 8 being the highest priority and 15 being the lowest priority. This completes the SRv6 packet forwarding priority setting, thus achieving integrated air-to-ground communication between drones.

[0054] The network AI of the present invention is combined with the integrated air-to-ground communication method of drones to add AI fault prediction and intelligent path optimization, providing a new solution for the forwarding of complex business messages in aerial wireless networks and reducing the resource consumption of message forwarding. When the drone network encounters collisions and failures, the risks can be predicted in advance. While taking avoidance procedures, the drone will quickly switch the network information that should belong to it to transmit signals to other drones, thereby ensuring signal quality and avoiding the occurrence of aerial failures.

[0055] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions based on the principles of the present invention are within the scope of protection of the present invention. It should be noted that for those skilled in the art, various improvements and modifications that do not depart from the principles of the present invention should be considered within the scope of protection of the present invention.

Claims

1. A network AI combined with UAV air-ground integrated communication method, characterized in that: The specific steps include: Step S1: SRv6 network chips and BeiDou signal receivers are installed on each UAV. The UAV aerial wireless network is established through the SRv6 protocol, and an independent IPv6 address is assigned to each UAV. The network routing automatically allocates the nearest path table to the target UAV and generates an initial path identifier based on the IPv6 address. Step S2: The aerial coordinates of each drone are obtained in real time through the Beidou signal receiver. A collision model is constructed based on the aerial coordinates of all drones and possible aerial obstacles between drones. The probability of collision between all drones and aerial obstacles is calculated, and the collision probability of each drone is stored in the Argument of the corresponding IPv6 extension header SRH. The process of building a collision model is as follows: The aerial obstacle is taken as the center of the coverage area, and the center coordinates are , the area radius is , in the time zone Derivative to obtain the standard dispersion coefficient value At the same time, based on the aerial coordinate set of the UAV and the set of time points corresponding to the coordinate points , combined with the optimized wireless positioning algorithm, the average dispersion coefficient value of the corresponding drone is obtained ; Within the current area radius and movement time range, the standard dispersion coefficient value and the average coefficient of variation For comparison, if Greater than , indicating that the current drone has no intersection with the coverage area, and there will be no collision between the current drone and the air obstacles; on the contrary, if Less than or equal to , then the difference between the two The larger the value, the more frequently the current drones intersect in the coverage area, and the greater the probability of collision between the current drone and aerial obstacles. Step S3: Use the Markov chain to build an early warning model. According to the health status of the drone, the probability of the drone associated with the initial path identifier of the current drone identifier generating an alarm is obtained, and the probability of the current drone generating an alarm is stored in the Argument of the corresponding IPv6 extension header SRH; Step S4: Network routing updates the initial path identifier based on the probability of collision and the warning probability of the drone stored in Argument to achieve integrated air-ground communication of the drone.

2. The method for integrating network AI with drone air-ground communication according to claim 1, characterized in that: The average dispersion coefficient value The calculation process is as follows: Aerial coordinates set of drones The aerial coordinates of a certain drone are , the coordinates corresponding to the time point are , combined with the optimized wireless positioning algorithm to calculate the discrete average value for: The aerial coordinates of the drone are set and the set of time points corresponding to the coordinate points Perform matrix operations to obtain vector parameter values : By discrete mean With vector parameter values Multiply to get the average dispersion coefficient value : 。 3. The method for integrating network AI with drone air-ground communication according to claim 1, characterized in that: The standard coefficient of variation value The calculation process is as follows: in, K Indicates the standard coefficient of variation value and the average coefficient of variation The difference between K ≤0, indicating that there is no intersection between the current drone and the coverage area. K >0, indicating that there is an intersection between the current drone and the coverage area.

4. The network AI combined with UAV air-ground integrated communication method according to claim 1 is characterized in that: The construction process of the early warning model is as follows: X(t+1)=X(t)×P Among them, X(t) represents the state vector of trend analysis and drone health status at time t, P represents the one-step transition probability matrix, and X(t+1) represents the state vector of trend analysis and drone health status at time t+1.

5. The network AI combined with UAV air-ground integrated communication method according to claim 1 is characterized in that: The specific implementation process of step S4 is as follows: sort the collision probability and alarm probability of all drones from small to large. The smaller the collision probability or alarm probability, the higher the priority. Obtain the IPv6 address of the drone with high priority in the identifier from the Argument of the IPv6 extension header SRH of the current drone, update the initial path identifier, and set the priority parameter value in the 802.1Q-Tag cos field through the IPv6 message header information. The value range is 8-15. Set the priority parameter value 8 as the highest priority and 15 as the lowest priority. Complete the SRv6 message forwarding priority setting to realize the integrated air-ground communication of drones.

Images