Multi-unmanned aerial vehicle formation communication link access optimization method

Abstract

The invention belongs to the technical field of multi-unmanned aerial vehicle formation control, and particularly relates to a multi-unmanned aerial vehicle formation communication link access optimization method, which comprises the following steps of S1, constructing an unmanned aerial vehicle formation control model and a communication network model; s2, dividing the data packets in the unmanned aerial vehicle data buffer area into data packet queues according to priorities; s3, judging the difference between the state of the unmanned aerial vehicle and an ideal state based on a Mamdani fuzzy rule; and S4, formulating the priority of the data sent by the unmanned aerial vehicle through the state difference of the unmanned aerial vehicle. According to the invention, in the existing unmanned aerial vehicle formation control, the influence of the actual communication time delay on the formation control is considered. A double-layer priority division mechanism is provided by analyzing formation network characteristics and communication requirements. A data link layer communication protocol which has the largest influence on time delay is optimized, network congestion is reduced, and the accuracy and safety of formation can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of multi-UAV formation control, and in particular to a method for optimizing communication link access of a multi-UAV formation. Background Art

[0002] Multi-UAV self-organizing networks have become a research hotspot in recent years. In order to perform tasks, drones are usually formed in a coordinated manner to control the flight attitude of each drone and adjust the relative position of each drone. Currently, the more common formation control methods include consistency-based methods, leader-follower methods, virtual structure methods, and behavior-based methods. Research on drone formations still focuses on the control level, and less attention is paid to the impact of the network level on the formation.

[0003] In fact, the formation of drones is achieved by transmitting a large amount of control information such as position and speed between drones. This type of data is called delay-sensitive traffic. Delay-sensitive traffic has very high requirements for transmission delay. Although there have been studies focusing on the impact of this delay, such as event-triggered control and model predictive control. However, they are all through control methods and have not fundamentally solved the delay in data transmission. In addition, when drone formations are performing tasks, they may generate a large amount of delay-tolerant traffic such as audio and video data and sensor data. In a formation, different drones also have different delay requirements. How to reduce the transmission delay of delay-sensitive traffic has become a problem that needs to be solved.

[0004] The protocols used in drone ad hoc networks are generally divided into two types: TDMA and IEEE802.11 DCF. TDMA can transmit without collision under high load due to its pre-allocated time slots, but when the topology changes during drone formation flight, such as drones joining or leaving the formation, TDMA cannot adapt to the changing topology; IEEE 802.11's distributed coordination function DCF uses CSMA / CA carrier sense multiple access / collision avoidance. CSMA / CA reduces data packet collisions due to the handshake form. In long-distance transmission, the increase in RTS / CTS control packet propagation delay will lead to an increase in communication delay. However, in the formation network, the distance between drones is not too far and they generally only communicate with neighboring drones, so long-distance communication is rare. However, when the network load is too large, the collision probability of CSMA / CA will increase. How to design a reasonable link access protocol based on the communication needs of multi-drone formations has become a problem that needs attention. Summary of the invention

[0005] 1. Technical issues to be resolved

[0006] In view of the deficiencies in the prior art, the present invention provides a method for optimizing communication link access of a multi-UAV formation, which solves the problems raised in the above-mentioned background technology.

[0007] (II) Technical solution

[0008] In order to achieve the above-mentioned purpose, the present invention specifically adopts the following technical solutions:

[0009] A method for optimizing communication link access of a multi-UAV formation comprises the following steps:

[0010] S1: Construct UAV formation control model and communication network model;

[0011] S2: divide the data packets in the drone data buffer into data packet queues according to their priorities;

[0012] S3: Determine the difference between the drone state and the ideal state based on Mamdani fuzzy rules;

[0013] S4: Prioritize the data sent by drones based on their status differences;

[0014] S5: The drone adjusts the backoff parameters of the CSMA / CA mechanism through the given double-layer priority rules, coordinates the sending and receiving of data packets, and completes the formation control by continuously adjusting the state through the control rate.

[0015] Furthermore, in S1, a UAV kinematic model is established, given the following second-order integrator system of the UAV and the virtual leader:

[0016]

[0017] Where x i ∈R n , v i ∈R n ,u i ∈R n are the position, velocity and control input of UAV i, respectively, x1∈R n , v1∈R n is the position and velocity of the virtual leader, f:R n →R n Represents the nonlinear function existing in the UAV motion process;

[0018] The formation error of the system is defined as:

[0019]

[0020] where ζ xi is the position difference of UAV i relative to the virtual navigator, ζ xi Determines where the drones fly in the formation;

[0021] The formation uses the following consistency control rates:

[0022]

[0023] Where k>0 is a constant control gain, d i Represents the connection status between drone i and the virtual drone, d i =1 means the drone can obtain the information of the virtual leader, otherwise d i =0, l ij is the Laplace matrix element, the Laplace matrix L=(l ij )∈R N×N , N is the number of drones, l ij =-a ij (i≠j), a ij represents the communication connection between drones. Considering that all drones are equipped with the same communication equipment and have the same communication distance R, the graph formed by the drone network is an undirected graph. If there is a connection between drones, then a ij =a ji =1, otherwise a ij =a ji =0, adjacency matrix A = (a ij )∈R N×N ;

[0024] The sampling period of the drone is defined as T, that is, every T time, the drone encapsulates its own status information into a data packet, and then the drone node performs channel access according to the distributed coordination function in the IEEE 802.11 protocol, specifically, it contends for the channel through the CSMA / CA mechanism. When the node successfully contends for the channel, it sends its own status information to the neighboring drone. The neighboring drone receives the status information and updates its speed and position according to the above consistency control rate;

[0025] Assume that the drones are equipped with the same type of communication equipment, the network is distributed, and there is no central node to send a unified beacon frame to determine the cycle. Each drone uses its own timer to time the cycle. Assume that the timers in the network have completed time synchronization and have a synchronized sampling cycle start time. Each drone has a communication data transmission buffer of the same size. After the data packet is generated, it will enter the buffer and wait for transmission.

[0026] Since the drones that each drone in the network communicates with are determined by the adjacency matrix A, the drone does not need to communicate with all drones within the communication range, but only needs to communicate with a few designated drones within the one-hop communication range. This will greatly reduce the channel load and does not need to consider the multi-hop routing of data. The data transmission delay caused by routing can be reduced. In the formation network, due to the high node density and the large number of nodes within the one-hop communication range, data packet collisions are more likely to occur. Since the nodes in the broadcast network cannot receive feedback from the ACK data frame, there is no way to determine whether the data transmission is successful. In order to ensure the reliability of the transmission of this critical information, multiple unicasts are used to ensure data transmission.

[0027] Furthermore, in S2, the first layer of the double-layer priority division is performed, that is, the priority of the data packets is divided, and the same communication buffer is given to each drone, and each buffer is divided into three priority queues, which are used to store data packets of different priorities;

[0028] Data packets are divided into three levels: A, B and C based on the service type, where:

[0029] Grade A data is the highest grade data, which includes flight control data such as the position and speed of the drone. This type of data has very high requirements for latency, and it directly determines the accuracy and safety of system control, and is called delay-sensitive data. Grade B data is the second highest grade data, which includes data that needs to be transmitted in real time, such as mission information. This type of data has high requirements for latency and needs to be transmitted in real time to ensure the execution of the mission. Grade C data is the lowest grade data, which includes delay-tolerant data such as sensor information, and this type of data has low requirements for latency.

[0030] When a data packet enters the buffer, its type is determined and it enters the corresponding data queue. For different queues, the following working mechanisms are used:

[0031] Class A queue: When a class A data packet enters the queue, it will wait for the class A data packet in front of the queue to be sent. When it reaches the front of the queue, it will determine the current sending status. If the sending is idle, it will directly enter the sending state. If there is A or B class data transmission, it will wait for the current transmission to end and start the transmission directly. If the data currently being transmitted is C class data, it will directly suspend the transmission of the C class data packet and start the transmission directly.

[0032] Class B queue: When a Class B data packet enters the queue, it also needs to wait for the Class B data packet in front of the queue to be sent. When it reaches the front of the queue, it determines the sending status. If the sending is idle, it starts sending data directly. If there is Class A data to send and there are data packets waiting in the Class A queue, it needs to wait for all Class A data to be sent. If the current data packet is a Class C data packet, it will wait for it to be sent. At this time, regardless of whether there is data waiting in the Class C queue, it will start sending directly;

[0033] Class C queue: Similar to class A and B data packets, when it reaches the front of the queue, it needs to wait for all the A and B class data packets currently being sent and waiting to be sent to be transmitted before it can be sent. When its transmission is preempted by class A data, it will wait for the next idle time to retransmit.

[0034] Furthermore, in S3, the second layer of the two-layer priority division is performed, and in this step, the priority is divided based on the number of drones with large differences among the neighboring drones of the drone;

[0035] Based on the formation error system, the state difference of the UAVs is further defined as:

[0036]

[0037] Where N i (t) represents the set of neighbors of drone i, Represents N i (t), θ i (t) represents the heading angle of UAV i, Δθ i Represents the average value of the absolute value of the difference in heading angle between UAV i and all its neighbors, Δv i represents the average value of the 2-norm of the speed difference between drone i and all its neighbors, Δx i Represents the average value of the 2-norm of the expected position difference between UAV i and all its neighbors. It should be noted that, unlike the formation error system, in the state difference, the speed difference and heading angle are considered separately, and the state difference between the UAV and its neighbors is calculated in the state difference, not the state difference with the navigator. This is because in a distributed system, not every UAV can know the state of the virtual navigator, so it can only be compared with its neighbors during calculation;

[0038] However, it can be seen that the state difference is composed of three values ​​with different dimensions. Next, Mamdani fuzzy reasoning is used to adjust the comprehensive state difference;

[0039] Fuzzy rules use if-then rules:

[0040] ifΔθ i is A1 AndΔvi is A2 AndΔx i is A3 Then Δαis Z

[0041] In the formula, A1, A2, A3 are the three input fuzzy subsets, Z is the output fuzzy subset, and the fuzzy output is the comprehensive difference Δα*. For rule R j The membership function of its output It can be expressed as:

[0042]

[0043] After aggregating the outputs of all rules, the total output membership function μ output (Δα) is:

[0044]

[0045] The defuzzification uses the centroid method to output the value Δα*:

[0046]

[0047] The UAVs whose comprehensive difference Δα* is greater than a certain threshold α* are called difference UAVs.

[0048] Furthermore, in S4, the second-level priority division is performed. The difference of nodes has been obtained through S3. For drone node i, its node priority ρ is:

[0049]

[0050] Among them, ρ1<ρ2<ρ3 are constants of different priorities to be designed, ΔN i (t) is the number of difference drone nodes in the neighbors of drone i. By setting the node with more difference drones in the neighboring drones to a higher priority, its data packets can be sent to the difference drones with a higher success rate. The difference drones will converge to the expected formation more accurately through high-real-time data. It should be pointed out that at the beginning of each sampling cycle, the drone will update the number of difference drones and update the priority ρ according to the neighboring drone status information obtained in the previous round or most recently.

[0051] Furthermore, in S5, the parameters in the CSMA / CA protocol are adaptively adjusted according to the priority of the drone node. In the CSMA / CA protocol, the nodes randomly compete for the channel. When there is data to be sent and the channel is detected to be idle, a binary exponential backoff is used to wait for a random time slot. The backoff contention window CW is as follows:

[0052] CW=Random(0,2 BE -1) SlotTime

[0053] Where Random represents a random integer in the range, BE is the backoff index, and its initial value is BEmin. When a channel contention conflict occurs each time, BE=BE+1, and the contention window is recalculated until the set BEmax is reached. When the data is successfully sent or the data packet is discarded, BE is initialized to BEmin, and SlotTime is the length of time occupied by each time slot;

[0054] According to the priority of the node, the priority ρ is substituted into the calculation of the contention window, that is, BE min =ρ, by setting the BEmin of higher priority nodes to a smaller value, the high priority nodes can retreat smaller time slots in the initial state, so that they have a higher chance to access the channel and send data. At the beginning of each sampling cycle, the node will update the BEmin value according to the updated priority ρ, and only when the data packets of the B and C queues are ready to be sent, the priority ρ is not used for updating, but a value ρ4 greater than ρ1, ρ2, ρ3 is used for updating;

[0055] The UAV formation network completes the communication interaction of status information through the data packet queue mechanism given by the double-layer priority and the node priority contention window, and continuously updates its own status through a given control rate until the formation control is finally completed.

[0056] (III) Beneficial effects

[0057] Compared with the prior art, the present invention provides a method for optimizing communication link access of a multi-UAV formation, which has the following beneficial effects:

[0058] In the existing UAV formation control, the present invention considers the impact of actual communication delay on formation control. By analyzing the formation network characteristics and communication requirements, a two-layer priority division mechanism is proposed. The data link layer communication protocol, which has the greatest impact on delay, is optimized to reduce network congestion, which can effectively improve the accuracy and safety of the formation.

[0059] The present invention proposes a dual-layer priority division mechanism combining data packet priority and node priority. Data packet priority division reduces the waiting time of status information data in the queue. Node priority division improves the transmission success rate of nodes with higher requirements for data transmission success rate. It effectively reduces the end-to-end delay and collision probability of control information in the network; and increases the throughput of control information.

[0060] The dual-layer priority mechanism proposed in this invention is a fully distributed mechanism. Unlike many communication scenarios, when a UAV formation is performing many tasks, there is no way to receive control from the ground control station, so the formation control design is fully distributed. Therefore, the proposed link access optimization mechanism is applicable to more UAV formation scenarios without being restricted by centralized computing. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 is a flow chart of the method of the present invention;

[0062] Figure 2 A schematic diagram of a data packet priority queue of the present invention;

[0063] Figure 3 It is a priority queue workflow diagram of the present invention;

[0064] Figure 4 This is a flowchart of node priority division of the present invention. DETAILED DESCRIPTION

[0065] 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.

[0066] Example

[0067] like Figure 1-4 As shown, a method for optimizing communication link access of a multi-UAV formation proposed in one embodiment of the present invention includes the following steps:

[0068] S1: Construct UAV formation control model and communication network model;

[0069] In S1, the UAV kinematic model is established, given the following second-order integrator system of the UAV and the virtual leader:

[0070]

[0071] Where x i ∈R n , v i ∈R n ,u i ∈R n are the position, velocity and control input of UAV i, respectively, x1∈R n , v1∈R n is the position and velocity of the virtual leader, f:R n →Rn Represents the nonlinear function existing in the UAV motion process;

[0072] The formation error of the system is defined as:

[0073]

[0074] where ζ xi is the position difference of UAV i relative to the virtual navigator, ζ xi Determines where the drones fly in the formation;

[0075] The formation uses the following consistency control rates:

[0076]

[0077] Where k>0 is a constant control gain, d i Represents the connection status between drone i and the virtual drone, d i =1 means the drone can obtain the information of the virtual leader, otherwise d i =0, l ij is the Laplace matrix element, the Laplace matrix L=(l ij )∈R N×N , N is the number of drones, l ij =-a ij (i≠j), a ij represents the communication connection between drones. Considering that all drones are equipped with the same communication equipment and have the same communication distance R, the graph formed by the drone network is an undirected graph. If there is a connection between drones, then a ij =a ji =1, otherwise a ij =a ji =0, adjacency matrix A = (a ij )∈R N×N ;

[0078] The sampling period of the drone is defined as T, that is, every T time, the drone encapsulates its own status information into a data packet, and then the drone node performs channel access according to the distributed coordination function in the IEEE 802.11 protocol, specifically, it contends for the channel through the CSMA / CA mechanism. When the node successfully contends for the channel, it sends its own status information to the neighboring drone. The neighboring drone receives the status information and updates its speed and position according to the above consistency control rate;

[0079] Assume that the drones are equipped with the same type of communication equipment, the network is distributed, and there is no central node to send a unified beacon frame to determine the cycle. Each drone uses its own timer to time the cycle. Assume that the timers in the network have completed time synchronization and have a synchronized sampling cycle start time. Each drone has a communication data transmission buffer of the same size. After the data packet is generated, it will enter the buffer and wait for transmission.

[0080] Since the drones that each drone in the network communicates with are determined by the adjacency matrix A, the drone does not need to communicate with all drones within the communication range, but only needs to communicate with a few designated drones within the one-hop communication range. This will greatly reduce the channel load and does not need to consider the multi-hop routing of data. The data transmission delay caused by routing can be reduced. In the formation network, due to the high node density and the large number of nodes within the one-hop communication range, data packet collisions are more likely to occur. Since the nodes in the broadcast network cannot receive feedback from the ACK data frame, there is no way to determine whether the data transmission is successful. In order to ensure the reliability of the transmission of this critical information, multiple unicasts are used to ensure data transmission.

[0081] S2: divide the data packets in the drone data buffer into data packet queues according to their priorities;

[0082] In S2, the first layer of the two-layer priority division is performed, that is, the priority of the data packets is divided, and the same communication buffer is given to each drone. Each buffer is divided into three priority queues, which are used to store data packets of different priorities;

[0083] Data packets are divided into three levels: A, B and C based on the service type, where:

[0084] Grade A data is the highest grade data, which includes flight control data such as the position and speed of the drone. This type of data has very high requirements for latency, and it directly determines the accuracy and safety of system control, and is called delay-sensitive data. Grade B data is the second highest grade data, which includes data that needs to be transmitted in real time, such as mission information. This type of data has high requirements for latency and needs to be transmitted in real time to ensure the execution of the mission. Grade C data is the lowest grade data, which includes delay-tolerant data such as sensor information, and this type of data has low requirements for latency.

[0085] When a data packet enters the buffer, its type is determined and it enters the corresponding data queue. For different queues, the following working mechanisms are used:

[0086] Class A queue: When a class A data packet enters the queue, it will wait for the class A data packet in front of the queue to be sent. When it reaches the front of the queue, it will determine the current sending status. If the sending is idle, it will directly enter the sending state. If there is A or B class data transmission, it will wait for the current transmission to end and start the transmission directly. If the data currently being transmitted is C class data, it will directly suspend the transmission of the C class data packet and start the transmission directly.

[0087] Class B queue: When a Class B data packet enters the queue, it also needs to wait for the Class B data packet in front of the queue to be sent. When it reaches the front of the queue, it determines the sending status. If the sending is idle, it starts sending data directly. If there is Class A data to send and there are data packets waiting in the Class A queue, it needs to wait for all Class A data to be sent. If the current data packet is a Class C data packet, it will wait for it to be sent. At this time, regardless of whether there is data waiting in the Class C queue, it will start sending directly;

[0088] Class C queue: Similar to class A and B data packets, when it reaches the front of the queue, it needs to wait for all the A and B class data packets currently being sent and waiting to be sent to be transmitted before it can be sent. When its transmission is preempted by class A data, it will wait for the next idle time to retransmit.

[0089] S3: Determine the difference between the drone state and the ideal state based on Mamdani fuzzy rules;

[0090] In S3, the second level of the two-level priority division is performed, in which the priority is divided based on the number of drones with large differences among the drone's neighbors;

[0091] Based on the formation error system, the state difference of the UAVs is further defined as:

[0092]

[0093] Where N i (t) represents the set of neighbors of drone i, Represents N i (t), θ i (t) represents the heading angle of UAV i, Δθ i Represents the average value of the absolute value of the difference in heading angle between UAV i and all its neighbors, Δv i represents the average value of the 2-norm of the speed difference between drone i and all its neighbors, Δx iRepresents the average value of the 2-norm of the expected position difference between UAV i and all its neighbors. It should be noted that, unlike the formation error system, in the state difference, the speed difference and heading angle are considered separately, and the state difference between the UAV and its neighbors is calculated in the state difference, not the state difference with the navigator. This is because in a distributed system, not every UAV can know the state of the virtual navigator, so it can only be compared with its neighbors during calculation;

[0094] However, it can be seen that the state difference is composed of three values ​​with different dimensions. Next, Mamdani fuzzy reasoning is used to adjust the comprehensive state difference;

[0095] Fuzzy rules use if-then rules:

[0096] ifΔθ i is A1 AndΔv i is A2 AndΔx i is A3 Then Δαis Z

[0097] In the formula, A1, A2, A3 are the three input fuzzy subsets, Z is the output fuzzy subset, and the fuzzy output is the comprehensive difference Δα*. For rule R j The membership function of its output It can be expressed as:

[0098]

[0099] After aggregating the outputs of all rules, the total output membership function μ output (Δα) is:

[0100]

[0101] The defuzzification uses the centroid method to output the value Δα*:

[0102]

[0103] The UAVs whose comprehensive difference Δα* is greater than a certain threshold α* are called difference UAVs.

[0104] S4: Prioritize the data sent by drones based on their status differences;

[0105] In S4, the second level of priority division is performed. The difference of nodes has been obtained through S3. For drone node i, its node priority ρ is:

[0106]

[0107] Among them, ρ1<ρ2<ρ3 are constants of different priorities to be designed, ΔNi (t) is the number of difference drone nodes in the neighbors of drone i. By setting the node with more difference drones in the neighboring drones to a higher priority, its data packets can be sent to the difference drones with a higher success rate. The difference drones will converge to the expected formation more accurately through high-real-time data. It should be pointed out that at the beginning of each sampling cycle, the drone will update the number of difference drones and update the priority ρ according to the neighboring drone status information obtained in the previous round or most recently.

[0108] S5: The drone adjusts the backoff parameters of the CSMA / CA mechanism through the given double-layer priority rules, coordinates the sending and receiving of data packets, and completes the formation control by continuously adjusting the state through the control rate.

[0109] In S5, the parameters in the CSMA / CA protocol are adaptively adjusted according to the priority of the drone node. In the CSMA / CA protocol, nodes randomly compete for channels. When there is data to be sent and the channel is detected to be idle, a binary exponential backoff is used to wait for a random time slot. The backoff contention window CW is as follows:

[0110] CW=Random(0,2 BE -1) SlotTime

[0111] Where Random represents a random integer in the range, BE is the backoff index, and its initial value is BEmin. When a channel contention conflict occurs each time, BE=BE+1, and the contention window is recalculated until the set BEmax is reached. When the data is successfully sent or the data packet is discarded, BE is initialized to BEmin, and SlotTime is the length of time occupied by each time slot;

[0112] According to the priority of the node, the priority ρ is substituted into the calculation of the contention window, that is, BE min =ρ, by setting the BEmin of higher priority nodes to a smaller value, the high priority nodes can retreat smaller time slots in the initial state, so that they have a higher chance to access the channel and send data. At the beginning of each sampling cycle, the node will update the BEmin value according to the updated priority ρ, and only when the data packets of the B and C queues are ready to be sent, the priority ρ is not used for updating, but a value ρ4 greater than ρ1, ρ2, ρ3 is used for updating;

[0113] The UAV formation network completes the communication interaction of status information through the data packet queue mechanism given by the double-layer priority and the node priority contention window, and continuously updates its own status through a given control rate until the formation control is finally completed.

[0114] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for optimizing communication link access for a multi-UAV formation. Features: The following steps are involved: S1: Construct UAV formation control model and communication network model; S2: divide the data packets in the drone data buffer into data packet queues according to their priorities; S3: Determine the difference between the drone state and the ideal state based on Mamdani fuzzy rules; S4: Prioritize the data sent by drones based on their status differences; S5: The drone adjusts the backoff parameters of the CSMA / CA mechanism through the given double-layer priority rules, coordinates the sending and receiving of data packets, and completes the formation control by continuously adjusting the state through the control rate.

2. According to claim 1, a method for optimizing communication link access of a multi-UAV formation, Features: In S1, the UAV kinematic model is established, given the following second-order integrator system of the UAV and the virtual leader: Where x i ∈R n , v i ∈R n ,u i ∈R n are the position, velocity and control input of UAV i, respectively, 1 ∈R n , v 1 ∈R n is the position and velocity of the virtual leader, f:R n →R n Represents the nonlinear function existing in the UAV motion process; The formation error of the system is defined as: where ζ xi is the position difference of UAV i relative to the virtual navigator, ζ xi Determines where the drones fly in the formation; The formation uses the following consistency control rates: Where k>0 is a constant control gain, d i Represents the connection status between drone i and the virtual drone, d i =1 means the drone can obtain the information of the virtual leader, otherwise d i =0, l ij is the Laplace matrix element, the Laplace matrix L=(l ij )∈R N×N , N is the number of drones, a ij represents the communication connection between drones. Considering that all drones are equipped with the same communication equipment and have the same communication distance R, the graph formed by the drone network is an undirected graph. If there is a connection between drones, then a ij =a ji =1, otherwise a ij =a ji =0, adjacency matrix A = (a ij )∈R N×N ; The sampling period of the drone is defined as T, that is, every T time, the drone encapsulates its own status information into a data packet, and then the drone node performs channel access according to the distributed coordination function in the IEEE 802.11 protocol, specifically, it contends for the channel through the CSMA / CA mechanism. When the node successfully contends for the channel, it sends its own status information to the neighboring drone. The neighboring drone receives the status information and updates its speed and position according to the above consistency control rate; Assume that the drones are equipped with the same type of communication equipment, the network is distributed, and there is no central node to send a unified beacon frame to determine the cycle. Each drone uses its own timer to time the cycle. Assume that the timers in the network have completed time synchronization and have a synchronized sampling cycle start time. Each drone has a communication data transmission buffer of the same size. After the data packet is generated, it will enter the buffer and wait for transmission. Since the drones that each drone in the network communicates with are determined by the adjacency matrix A, the drone does not need to communicate with all drones within the communication range, but only needs to communicate with a few designated drones within the one-hop communication range. This will greatly reduce the channel load and does not need to consider the multi-hop routing of data. The data transmission delay caused by routing can be reduced. In the formation network, due to the high node density and the large number of nodes within the one-hop communication range, data packet collisions are more likely to occur. Since the nodes in the broadcast network cannot receive feedback from the ACK data frame, there is no way to determine whether the data transmission is successful. In order to ensure the reliability of the transmission of this critical information, multiple unicasts are used to ensure data transmission.

3. According to claim 2, a method for optimizing communication link access of a multi-UAV formation, Features: In S2, the first layer of the double-layer priority division is performed, that is, the priority of the data packets is divided, and the same communication buffer is given to each drone. Each buffer is divided into three priority queues, which are used to store data packets of different priorities; Data packets are divided into three levels: A, B and C based on the service type, where: Grade A data is the highest grade data, which includes flight control data such as the position and speed of the drone. This type of data has very high requirements for latency, and it directly determines the accuracy and safety of system control, and is called delay-sensitive data. Grade B data is the second highest grade data, which includes data that needs to be transmitted in real time, such as mission information. This type of data has high requirements for latency and needs to be transmitted in real time to ensure the execution of the mission. Grade C data is the lowest grade data, which includes delay-tolerant data such as sensor information, and this type of data has low requirements for latency. When a data packet enters the buffer, its type is determined and it enters the corresponding data queue. For different queues, the following working mechanisms are used: Class A queue: When a class A data packet enters the queue, it will wait for the class A data packet in front of the queue to be sent. When it reaches the front of the queue, it will determine the current sending status. If the sending is idle, it will directly enter the sending state. If there is A or B class data transmission, it will wait for the current transmission to end and start the transmission directly. If the data currently being transmitted is C class data, it will directly suspend the transmission of the C class data packet and start the transmission directly. Class B queue: When a class B data packet enters the queue, it also needs to wait for the class B data packet in front of the queue to be sent. When it reaches the front of the queue, it determines the sending status. If the sending is idle, it starts sending data directly. If there is class A data to send and there are data packets waiting in the class A queue, it needs to wait for all the class A data to be sent. If the current data packet is a class C data packet, it needs to wait for it to be sent. At this time, regardless of whether there is data waiting in the class C queue, it will start sending directly. Class C queue: Similar to class A and B data packets, when it reaches the front of the queue, it needs to wait for all the A and B class data packets currently being sent and waiting to be sent to be transmitted before it can be sent. When its transmission is preempted by class A data, it will wait for the next idle time to retransmit.

4. According to claim 3, a method for optimizing communication link access of a multi-UAV formation, Features: In S3, the second layer of the two-layer priority division is performed, in which the priority is divided based on the number of drones with large differences among the drone's neighbors; Based on the formation error system, the state difference of the UAVs is further defined as: Where N i (t) represents the set of neighbors of drone i, Represents N i (t), θ i (t) represents the heading angle of UAV i, Δθ i Represents the average value of the absolute value of the difference in heading angle between UAV i and all its neighbors, Δv i represents the average value of the 2-norm of the speed difference between drone i and all its neighbors, Δx i Represents the average value of the 2-norm of the expected position difference between UAV i and all its neighbors. It should be noted that, unlike the formation error system, in the state difference, the speed difference and heading angle are considered separately, and the state difference between the UAV and its neighbors is calculated in the state difference, not the state difference with the navigator. This is because in a distributed system, not every UAV can know the state of the virtual navigator, so it can only be compared with its neighbors during calculation; However, it can be seen that the state difference is composed of three values ​​with different dimensions. Next, Mamdani fuzzy reasoning is used to adjust the comprehensive state difference; Fuzzy rules use if-then rules: ifΔθ i is A 1 AndΔv i is A 2 AndΔx i is A 3 ThenΔαis Z Where A 1 ,A 2 ,A 3 The fuzzy subsets Z of the three inputs are the fuzzy subsets of the outputs. The fuzzy output is the comprehensive difference Δα*. For the rule R j The membership function of its output It can be expressed as: After aggregating the outputs of all rules, the total output membership function μ output (Δα) is: The defuzzification uses the centroid method to output the value Δα*: The UAVs whose comprehensive difference Δα* is greater than a certain threshold α* are called difference UAVs.

5. According to claim 4, a method for optimizing communication link access of a multi-UAV formation, Features: In S4, the second-level priority division is performed. The difference of nodes has been obtained through S3. For drone node i, its node priority ρ is: where ρ 1 <ρ 2 <ρ 3 are the different priority constants to be designed, ΔN i (t) is the number of difference drone nodes in the neighbors of drone i. By setting the node with more difference drones in the neighboring drones to a higher priority, its data packets can be sent to the difference drones with a higher success rate. The difference drones will converge to the expected formation more accurately through high-real-time data. It should be pointed out that at the beginning of each sampling cycle, the drone will update the number of difference drones and update the priority ρ according to the neighboring drone status information obtained in the previous round or most recently.

6. A method for optimizing communication link access for a multi-UAV formation according to claim 5, Features: In the S5, the parameters in the CSMA / CA protocol are adaptively adjusted according to the priority of the drone node. In the CSMA / CA protocol, the nodes randomly compete for the channel. When there is data to be sent and the channel is detected to be idle, a binary exponential backoff is used to wait for a random time slot. The backoff contention window CW is as follows: CW=Random(0,2 BE -1)·SlotTime Where Random represents a random integer in the range, BE is the backoff index, and its initial value is BEmin. When a channel contention conflict occurs each time, BE=BE+1, and the contention window is recalculated until the set BEmax is reached. When the data is successfully sent or the data packet is discarded, BE is initialized to BEmin, and SlotTime is the length of time occupied by each time slot; According to the priority of the node, the priority ρ is substituted into the calculation of the contention window, that is, BE min =ρ, by setting the BEmin of higher priority nodes to a smaller value, the high priority nodes can retreat smaller time slots in the initial state, so that they have a higher chance to access the channel and send data. At the beginning of each sampling cycle, the node will update the BEmin value according to the updated priority ρ, and only when the data packets of the B and C queues are ready to be sent, the priority ρ is not used for updating, but a value greater than ρ is used. 1 ,ρ 2 ,ρ 3 Greater than the value of 4 To update; The UAV formation network completes the communication interaction of status information through the data packet queue mechanism given by the double-layer priority and the node priority contention window, and continuously updates its own status through a given control rate until the formation control is finally completed.

Images