CN114339660B - Unmanned aerial vehicle cluster random access method - Google Patents

Abstract

The invention discloses a random access method for an unmanned aerial vehicle cluster, and belongs to the field of unmanned aerial vehicle communication. The invention comprises two parts of designing standardized message mutual acknowledgement format and designing random access mechanism. The invention solves the problem of uncertainty generated when the sub-nodes of the unmanned aerial vehicle cluster are accessed and exited by designing the distributed high-dynamic random access protocol distributed on demand at the top layer of the unmanned aerial vehicle cluster, the dynamic resource management method based on the proportional fairness algorithm and the random access method, and effectively avoids the problem of a plurality of access request conflicts in the communication process. The invention can use the efficient time slot allocation method along with the change of task plan, task progress and strategic tactics to improve the performance of the TDMA protocol, quickly adjust the role played by each unmanned aerial vehicle and fairly and efficiently redistribute the occupied resources. The invention can control unmanned aerial vehicle cluster collaborative operation, and is suitable for the fields of military, agriculture, urban rescue, environmental survey, smart city and the like.

Description

A random access method for UAV swarms

technical field

The invention relates to a method for randomly accessing an unmanned aerial vehicle cluster, which belongs to the field of unmanned aerial vehicle communication.

Background technique

The UAV cluster is composed of a large number of miniaturized, cheap, and highly maneuverable UAV nodes. The nodes are interconnected through wireless self-organizing network technology to build an intelligent system with distributed functions. Its remarkable combat effectiveness, low cost, The advantages of small damage loss and easy mass equipment are the main trends in the development of UAVs in the future. This distributed control mode of UAV cluster cooperative confrontation breaks through the limitations of local perception and task execution of a single UAV, and plays an important role in military, agricultural systems, urban rescue, environmental survey, smart cities and other application fields. .

At present, the mainstream UAV cluster networking technologies include relay network, star network and mesh ad hoc network. There are many types of aircraft scattered in the relay network, which can be flown by a single aircraft or formed into a network. The isolated nodes are integrated into the combat network through the relay drone; in the star network, the communication is based on ground The base station is the center, and the UAV terminals are scattered around the ground base station, and the central ground base station completes the control of the surrounding UAVs to realize networking, which is suitable for scenarios where the number of UAVs is small and the scope of tasks is small; In the network, each drone is a node with the same function. When communicating between drones, one-hop or multi-hop routing is required. When the drone point cannot be linked to the ground central station in one hop , it can realize the interconnection of all nodes in the entire network through multi-hop routing to the central station, but due to the semi-centralized nature of this network, the communication architecture is not very robust.

Due to the complexity of UAV swarm structure and the diversity of behaviors, the load limit of a single UAV cannot complete swarm cooperative control, and the swarm cooperative decision-making control method based on centralized or semi-centralized hierarchical architecture is no longer applicable. In addition, because UAV clusters rely on the local communication capabilities of UAVs to achieve collaborative information sharing to complete confrontation tasks, lightweight requirements for communication protocols between cluster individuals are also put forward. The current access method of UAV swarms tends to be complicated, with a large amount of calculation and high demonstration cost, which makes it impossible for the unmanned swarm cooperative confrontation algorithm to be demonstrated and verified on the actual UAV swarm, and the miniaturized random access This method cannot effectively solve the frequent access problem under a large number of nodes. Therefore, designing a method that can meet the frequent access of a large number of UAVs, reduce access conflicts, and complete the optimal configuration of roles has become the key to realize the application of UAV clusters.

Contents of the invention

Aiming at the requirement of frequent access and exit of unit drones during the working process of drone clusters, and solving the dynamic redistribution problems caused by role changes and task changes in drone cluster operations and the problem of multiple nodes sending access requests at the same time Conflict avoidance problem, the present invention discloses a random access method of UAV cluster, the purpose is to solve the uncertainty of the deployment, use, network access, withdrawal and withdrawal of UAV cluster nodes, and clarify the node task type and role division , realize real-time adjustment, and avoid multiple access request conflicts in the communication process.

The purpose of the present invention is achieved by the following technical solutions:

The invention discloses a random access method for unmanned aerial vehicle clusters, which is a distributed and highly dynamic random access technology allocated on demand, including two parts: one is to design a standardized message mutual recognition format, which is centerless for unmanned aerial vehicle clusters The interconnection network uses time division multiple access technology (TDMA) to divide time into different time segments (frames), then divide the frame into different time slots, and then assign each time slot uniquely to an unmanned Each node can only operate in the time slot allocated to it; the second is to design a random access mechanism to avoid access conflicts. Sending and receiving adopts a response mechanism. During the transmission process, the number of currently accessed platforms and the position of the reserved time slot are notified through the broadcast time slot allocation table. The node to be accessed can send an access request in the reserved time slot. If the access is allowed, the updated time slot allocation table will be notified by broadcast, and if the access is prohibited, the node to be accessed will be notified by broadcast to continue waiting or perform other operations. During the mission process, along with the changes in the mission plan, mission process, and strategy and tactics, use an efficient time slot allocation method to improve the performance of the TDMA protocol, quickly adjust the role of each UAV, and redistribute the resources it occupies fairly and efficiently. distribute.

A method for randomly accessing an unmanned aerial vehicle cluster disclosed by the present invention comprises the following steps:

Step 1: The random access protocol establishes the wireless link of the cluster nodes.

The sub-cluster network adopts a single-channel mode, which integrates the control commands of the control channel into the data channel, that is, when sending time slots, it is necessary to send the clock reference, time slot allocation table, status data of the current node, control commands of other nodes, current Task data for the node. When one node is sending, all other nodes are in the receiving state.

According to the network transmission requirements of remote control, telemetry and reconnaissance data of different UAV types in the cluster network in different application environments, plan the message format of the heterogeneous platform from the top level, unify the information interaction mechanism, and decouple the physical interface. In order to be compatible with different data occurrence rates and packet lengths, it is applicable to multiple types of protocols, adopts packetization methods such as packet services and encapsulation services, and uses a flexible encapsulation frame protocol that combines fixed length and variable length. According to the VMF message (Varible Message Foema, Variable message format) and CCSDS (Consultative Committee for Space Data Systems, International Space Data System Advisory Committee) protocol standard, using the general frame frame method, put the fixed-length data frame into the general transmission frame of the data link.

Step 2: Assign random access criteria as needed to ensure the smooth access of UAV nodes to the network.

Sending and receiving adopts a response mechanism. During the transmission process, the number of currently accessed platforms and the position of the reserved time slot are notified through the broadcast time slot allocation table. The node to be accessed can send an access request in the reserved time slot. If the access is allowed, the updated time slot allocation table will be notified by broadcast, and if the access is prohibited, the node to be accessed will be notified by broadcast to continue waiting or perform other operations. In the receiving frame structure design, the last time slot of each multiframe is a fixed allocation time slot, which is used to receive the time slot reservation request, and in the sending frame structure design, the first time slot is used to send time slot allocation information signaling. Each time slot is divided into It is 1-k sub-slots, and the value of k is dynamically allocated by the central node in the cluster according to the network load.

Step 3: For the UAV swarm cooperative combat system, an improved generalized proportional fairness (GPF) method is used to realize the resource scheduling strategy of the UAV swarm network communication system.

Scale factors with different ratios are introduced in the system initialization, and the GPF algorithm can be used to flexibly adjust the compromise between system throughput and user rate fairness. In addition, the GPF algorithm can also overcome the problem that the traditional PF algorithm can only achieve short-term rate fairness among users. Shortcomings, to achieve long-term fairness.

The proportional factor introduced by the GPF algorithm is and, by adjusting the values of a and b to flexibly adjust the compromise between system throughput and user rate fairness. The channel is divided into S RUs, and each RU is composed of an equal number of unit time slots, and multi-user transmission in the time domain takes frames as the unit. The user selects the user m with the largest generalized proportional fairness factor for transmission on each RU. The generalized proportional fairness factor is defined as follows:

The selection rules of the best user m ^* on each RU are as follows:

in, Indicates the instantaneous rate achieved by user m on RU s in the tth frame, which is related to the channel condition of user m in the tth frame, R _m (t) represents the history of user m in the sliding window T duration until the tth frame The average rate, R _m (t), is expressed as follows:

In the formula, r _m (t-1) represents the actual transmission rate of user m in the last frame, and T represents the duration of the sliding window. When the system is initialized, the selection of the ratio of scaling factors a and b affects the scheduling results of each frame, and different a/b values reflect different degrees of compromise between system throughput and user rate fairness.

Step 4: Optimize throughput performance by setting a and b configuration values in the GPF scheduling method.

Step 5: Apply the time slot allocation method to complete the time slot scheduling in a relatively short period of time to achieve conflict-free data transmission and maximize channel utilization.

The complete process of TDMA protocol time slot allocation algorithm includes time slot selection, time slot application, time slot confirmation and time slot release.

Time slot selection: In TDMA, time slot resources are marked as four states and expressed in binary, where 00 indicates that the current time slot is idle, 01 indicates that the current time slot is occupied by the node, and 10 indicates that the current time slot is occupied by the node. Occupied by the master node, 11 indicates that the current time slot is occupied by other child nodes.

The maximum number of nodes allowed to be accommodated is N, and the total amount of data slots in a single time frame is M, then each node can use an N×M two-dimensional matrix T=[t _ij ] _N*M to store slot allocation information:

When a node selects a time slot, it only needs to query the local time slot allocation matrix, and then apply to occupy the time slot.

Time slot application and confirmation: The time slot allocation of the TDMA protocol is a dynamic interactive process, and its dynamicity is mainly reflected in the two stages of time slot application and time slot confirmation. In the time slot application phase, TDMA cyclically monitors the traffic load of each node and other nodes on the communication link, and then calculates the number of data time slots required by the node in the current time frame. The calculation formula is as follows:

b=(1-θ)*d*r

In the formula, θ is the data slot overhead ratio, d is the duration of a single data slot, r is the node data transmission rate, and b is the number of bits that can be transmitted in a single data slot. In the formula, n is the total number of nodes, t _i is the business load sent by the node to the master node i, and sn is the total amount of data time slots required by the node.

If the number of data time slots required by the node is sn>0, the node needs to apply for m free data time slots in the local two-dimensional time slot allocation matrix in turn, and send its own time slot application information through the REQ packet to the master node. The master node that receives the REQ packet stores the time slot application information in the local two-dimensional time slot matrix.

Time slot release: The dynamic time slot allocation protocol is applied to meet the non-fixed requirements of nodes for occupying time slot resources. When a node in the network no longer uses the time slot resource due to failure or service interruption, the time slot occupied by it is released, so that other nodes sending services can obtain more time slot resources. The TDMA protocol active time slot release and passive time slot release are respectively used to deal with the two types of situations of service interruption and node failure. Active time slot release solves the problem of service interruption, and passive time slot release solves node failure. The release of the time slot is accomplished through two cases where the child node actively sends the release information and the master node directly sends the release information.

Beneficial effect:

1. A UAV cluster random access method disclosed in the present invention is compatible with the universal dynamic time slot allocation protocol of the fixed time slot TDMA mode and the dynamic time slot TDMA mode, and designs the minimum fixed time slot time frame structure, which includes fixed The number of time slots forms a time frame module with a fixed structure. By simple stacking of time frame modules, the dynamic time slot method can be realized, thereby significantly reducing the complexity of the dynamic time slot allocation algorithm, and at the same time avoiding the problem of inflexible topological structure caused by fixed time slots

2. A random access method for UAV clusters disclosed in the present invention performs random oscillation processing on the access request time of each node, that is, the node randomly delays sending requests by n multi-frame periods from the time of preparing to access the network, so Even if a collision occurs at a certain moment, the collision can be avoided in the next access with a high probability, greatly reducing the probability of continuous collision and failure to access.

3. The general PF algorithm only considers the historical average rate of users in a relatively short period of time, but the GPF algorithm of the present invention considers a sliding window whose duration is much longer than one transmission cycle, thereby ensuring long-term rate fairness among users.

4. Aiming at the network transmission requirements of remote control, telemetry and reconnaissance data of different UAV types in the cluster network in different application environments, a random access method for UAV clusters disclosed in the present invention plans the message format of heterogeneous platforms from the top layer, and unifies The information exchange mechanism decouples the physical interface, and improves the system, compatibility, scalability, inheritance and interoperability of group communication.

Description of drawings

FIG. 1 is a schematic diagram of time slot reserved random access according to the present invention;

Fig. 2 is a schematic diagram of the random network entry process of UAV nodes in the present invention;

Fig. 3 is a schematic diagram of multi-node network access conflict resolution in the example of the present invention;

Fig. 4 is the system throughput diagram under different a, b configurations in the example of the present invention;

Fig. 5 is a figure of long-term rate fairness VS system throughput in the example of the present invention;

Fig. 6 is a figure of short-term rate fairness VS frame number in the example of the present invention;

Fig. 7 is a flow chart of dynamic adjustment of UAV node resources in the example of the present invention;

Fig. 8 is a flowchart of network operation in the example of the present invention.

Detailed ways

The present invention will be described in detail below with reference to the drawings and embodiments. At the same time, the technical problems and beneficial effects solved by the technical solution of the present invention are also described. It should be pointed out that the described embodiments are only intended to facilitate the understanding of the present invention, and have no limiting effect on it.

As shown in FIG. 2 , this embodiment discloses a random access method for UAVs, which effectively solves the problem of cluster multi-point access conflicts and optimal resource allocation in the current cooperative confrontation of UAVs. The main process of the overall operation of the UAV cluster communication network in the example disclosed in this embodiment is shown in Figure 8, which mainly includes the stages of network initialization, application, authentication, network access, and network shutdown.

Network initialization is the process of providing sufficient initialization information for child nodes to enable them to exchange information on the network and run processing. The network initialization process is completed by loading initialization parameters. The main initialization parameters include transmission parameters, network parameters, and platform parameters. After the initialization of each child node is completed, the network operation is started according to the set network parameters. The process of ordinary UAV nodes joining the network operation: after the network initialization, the child node receives the time message and completes the network synchronization; when the child node is ready to join the network, it sends a network application message to the master node; after receiving the application message, the master node judges whether The child node is allowed to enter the network and respond; the child node processes according to the response result: if the response message is not allowed to enter the network, try to enter the network again or give up the network access; if the response message is to allow network access, send and receive messages according to the network protocol, and the network access is successful.

Network shutdown: After the task is completed, the master control node of the UAV cluster performs the network shutdown operation and sends a network management message to the sub-nodes in the network. After the network child node receives the message, it stops all communication in the network at the specified stop time.

As shown in Figure 2, the random access method of the UAV cluster disclosed in this embodiment, the specific implementation steps are as follows:

Step 1: The random access protocol establishes the wireless link of the cluster nodes.

A universal dynamic time slot allocation protocol compatible with fixed time slot TDMA mode and dynamic time slot TDMA mode is used, and a minimum fixed time slot time frame structure is adopted, which includes a fixed number of time slots to form a time frame module with a fixed structure. Through the simple stacking of time frame modules, the dynamic time slot mode can be realized, which greatly reduces the complexity of the dynamic time slot allocation algorithm, and at the same time avoids the problem of inflexible topological structure caused by fixed time slots. In order to be compatible with different data occurrence rates and packet lengths, and to be applicable to multiple types of protocols, an elastic encapsulation frame protocol combining fixed length and variable length is designed by using subpackage methods such as packet service and encapsulation service. According to the VMF message and the CCSDS (Consultative Committee for Space DataSystems, International Space Data System Advisory Committee) protocol standard, the fixed-length data frame is put into the general transmission frame of the data link by using the general frame frame.

The packet transmission frame includes a preamble, a frame identification code, a password synchronization word and a data frame source packet. A source packet consists of a packet-leading header and a packet-data field, both of which are required and lined up seamlessly. The length of the packet leading header is fixed, and the length of the packet data field is variable. In practical applications, the length of the source packet should be appropriate. If the source packet is too short, the transmission efficiency will be low; Three transmit frame data fields.

General-purpose time slots are mainly used to transmit general-purpose data such as business data, control instructions, and status information. The public time slots are mainly used for emergency signaling and random network access applications. Therefore, the frame structure of the public time slots is much shorter than that of the general time slots.

Step 2: Assign random access criteria as needed to ensure the smooth access of UAV nodes to the network.

The state of each time slot node in Figure 1 is represented by (state (state), target (target)), state indicates the state of the node in time slot s, and target indicates that the node will send or receive data packets in time slot s The one-hop neighbor node of . Therefore, in the time slot s, the state of the node can be divided into seven types:

Transport state (Transport)—at time slot s, the node sends a data packet to neighbor node a: (state=Transport, target=a). If the packet transmitted by the node is a broadcast packet, then target=Broadcast;

Receive state (Receive)——at time slot s, receive data packets from neighbor node b: (state=Receive, target=b);

Transmission blocking state (Block_t) - at time slot s, at least one node among neighbor nodes is receiving data packets from other nodes, and no neighbor node is transmitting packets;

Reception blocking state (Block_r)——At time slot s, at least one node among neighbor nodes is sending packets to other nodes, and no neighbor node is receiving packets;

Transmission and reception blocking state (Block_tr)——in time slot s, at least one node among neighbor nodes is sending packets to other nodes, and besides, at least one neighbor node is receiving packets from other nodes;

Collision - the node detects a collision while receiving the packet;

Idle state (Idle)—the node is idle, and no neighbor node is in the state of sending and receiving data packets.

It should be noted that the target is only meaningful in the first two states, that is, in the transmission state and the receiving state. In addition, these node states are independent of each other, and a node can only be in one of them in a time slot. A state, it can be considered that the time slots in which the node does not send data to the outside can be called passive time slots.

Figure 2 Aiming at the conflict problem of multiple UAV nodes submitting access applications at the same time, the access request time of each node is randomly oscillated, that is, nodes cannot send access applications every time, but start at the moment of preparing to access the network , to randomly delay sending requests for n multiframe periods (n is a random number within a certain range). The flowchart is shown in Figure 3.

Step 3: For the UAV swarm cooperative combat system, an improved generalized proportional fairness (GPF) method is used to realize the resource scheduling strategy of the UAV swarm network communication system.

The working steps of the cluster network improved generalized proportional fair scheduling method are as follows:

(1) When the UAV cluster node has resource application requirements, calculate the communication quality requirement index value of the node, and include it in the information sent by the UAV;

(2) The scheduler of the cluster network communication system resources evaluates the quality of each channel according to the channel detection reference signal transmitted by the UAV terminal;

(3) The scheduler of the cluster network communication system resources estimates the spectrum allocation system according to the channel quality information;

(4) Calculate the data transmission rate of each cluster node according to the spectrum allocation coefficient;

(5) The resource scheduler of the cluster network communication system comprehensively considers the instantaneous data transmission rate of the drone and the fairness of the drone to estimate the priority of each cluster node in each channel;

(6) Select the cluster node with the highest priority to schedule the resources of the cluster network communication system, and assign the channel to the UAV node with the highest priority;

(7) Delete the allocated channel from the list of candidate channel sets;

(8) If the list of candidate channel sets is empty, it means that all cluster network communication system resource scheduling has been allocated, then end the support of the algorithm, otherwise return to step (3) and continue the cluster network communication system for resource scheduling and allocation.

According to the above steps, according to the different resource types, different priorities, different delays and different packet loss rates of the UAV cluster nodes, the corresponding service data packet transmission characteristic parameter values are adopted. For services with high real-time requirements, the scheduler guarantees the lowest bit rate for the bearer of the service; for services with low real-time requirements, the scheduler does not need to guarantee the lowest bit rate for the bearer, so it is in the case of network congestion Under the circumstances, the business needs to bear the arrangement of reducing the rate. The default bearer and dedicated bearer will be designed here. The default bearer is used for small data volume and real-time business data; when the default bearer cannot meet the real-time requirements, the dedicated bearer is enabled to meet the speed and delay requirements.

Step 4: Optimize throughput performance by setting a and b configuration values in the GPF scheduling method.

Through continuous optimization of the GPF scheduling algorithm guidelines, combined with Figure 4 and Figure 5, the relationship between system throughput and long-term (ΔT=30sec) user rate fairness is shown when the number of UAV cluster subgroup nodes is 30. It can be seen that the rate fairness of the GPF scheduler decreases with the increase of the ratio of a/b. When a=0 and b=1, the GPF algorithm guarantees the lowest average rate of user transmission, so the system has high rate fairness, but The throughput is very low; at the same time, it can also be observed that the traditional PF (a=1, b=1) scheduler is also unfair in terms of user rate fairness. In order to achieve a compromise between throughput and user rate fairness, it is more appropriate to set the GPF algorithm parameters a=1, b=3 or a=1, b=2.

Figure 6 shows the change of short-term user rate fairness with the number of frames in the UAV cluster networking communication network, when the number of UAV cluster sub-group nodes is 30. It can be seen from the figure that compared with the MT scheduler (that is, when a=1, b=0), the GPF scheduler can greatly improve the short-term rate fairness of users, and the rate fairness performance between aircraft is even better than that of the RR scheduler. Algorithms are even better. In addition, the performance of the GPF scheduler under different a and b configurations is different. The performance of the traditional PF (a=1, b=1) scheduler is slightly better than that of the GPF scheduler under the a=1, b=2 configuration. Poor device.

Step 5: Apply the time slot allocation algorithm to complete the time slot scheduling in a relatively short period of time to achieve conflict-free data transmission and maximize channel utilization.

In the time slot confirmation stage, in order to ensure that there is no conflict in the time slot occupancy, the master node needs to conduct a unified arbitration for the time slot applications of each node. The arbitration rules are as follows:

1) If the data time slot s is only occupied by a child node i application, the time slot s is allocated to the node i for use.

2) If the data time slot s is occupied by multiple child nodes at the same time, the time slot s is preferentially allocated to the child nodes that transmit high-priority services; if the business priorities transmitted by the child nodes are the same, the time slot s is preferentially allocated to the node ID Smaller child nodes are used.

After the arbitration is over, the master node broadcasts the result of the arbitration to each child node through an ACK packet within the confirmation time slot. The node receiving the ACK packet needs to update its local two-dimensional time slot allocation matrix to prepare for the time slot application of the next time frame.

slot release phase

The TDMA protocol adopts active time slot release and passive time slot release respectively to deal with the two types of situations of service interruption and node failure.

1) Active slot release

A service timer is set for each node. Assuming that node i still has no data transmission when the service timer expires, all data time slots previously occupied by the node need to be released. At this time, node i polls the local two-dimensional time slot allocation matrix, and changes all the time slot states with value 1 in the i-th row to 0, and the remaining time slot states remain unchanged. Then, when the next control time slot arrives, the local two-dimensional time slot allocation matrix is sent to the master node through the REQ packet.

2) Passive slot release

If the master node does not receive the HELLO packet sent by it within a specific time, the default node i exits the system. At this time, each node needs to do the following operations: delete node i from the information table; poll their respective two-dimensional time slot allocation matrix, and reset all the time slot statuses in row i to 0; wait for the next control When the time slot arrives, the respective two-dimensional time slot allocation matrix is sent to the master node through the REQ packet.

Once the data slot is released, it returns to the idle state, so it can be reapplied by nodes in the network. TDMA effectively improves the time slot reuse rate through the time slot release process, thereby avoiding the waste of channel resources. The time slot adjustment process is shown in Figure 7.

So far, the UAV swarm access method is completed.

The specific description above is to further describe the purpose, technical solution and beneficial effect of the invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not used to limit the protection scope of the present invention. , Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims (3)

1. An unmanned aerial vehicle cluster random access method is characterized in that: comprises the following steps of the method,

step 1: establishing a cluster node wireless link by a random access protocol;

step 2: random access criteria are distributed according to the need to ensure that unmanned plane nodes smoothly access the network;

step 3: aiming at the unmanned aerial vehicle cluster cooperative combat system, an improved GPF method is used for realizing a resource scheduling strategy of an unmanned aerial vehicle cluster network communication system;

the implementation method of the step 3 is as follows:

the GPF algorithm can flexibly adjust the trade-off between the throughput of the system and the fairness of the user rate by introducing the scale factors with different ratios in the system initialization, and in addition, the GPF algorithm can overcome the defect that the traditional PF algorithm can only realize the fairness of the rate among short-term users and realize the fairness of the long-term;

the scale factors introduced by the GPF algorithm are a and b, and the compromise between different degrees of system throughput and user rate fairness is flexibly adjusted by adjusting the values of a and b; the channel is divided into S RUs, each RU is composed of equal number of unit time slots, and the time domain multi-user transmission is carried out by taking a frame as a unit; the user selects a user m with the maximum generalized proportional fairness factor on each RU for transmission; the generalized proportional fairness factor is defined as follows:

the selection rule for the best user m on each RU is as follows:

wherein ,representing the instantaneous rate that user m achieves over the RUs for the t-th frame, R, in relation to the channel conditions of user m during the t-th frame _m (T) represents the historical average rate of user m over the duration of sliding window T, R, when ending at the T-th frame _m The expression of (t) is as follows:

in the formula ,r_m (T-1) represents the actual transmission rate of the last frame by the user m, and T represents the duration of the sliding window; system primaryDuring initialization, the ratio of the scaling factors a and b is selected to influence the scheduling result of each frame, and different a/b values reflect the trade-off of different degrees between the throughput of the system and the fairness of the user rate;

step 4: optimizing throughput performance by setting a, b configuration values in a GPF scheduling method;

step 5: the time slot scheduling is completed in a short time by applying the time slot allocation method, so that collision-free data transmission is realized, and the channel utilization rate is improved to the maximum extent;

the implementation method of the step 5 is that,

the complete process of the TDMA protocol time slot allocation algorithm comprises time slot selection, time slot application, time slot confirmation and time slot release;

time slot selection: in TDMA, the time slot resources are marked in four states and expressed in binary, where 00 indicates that the current time slot is in an idle state, 01 indicates that the current time slot is occupied by the node, 10 indicates that the current time slot is occupied by the master node, and 11 indicates that the current time slot is occupied by other child nodes;

the maximum number of nodes allowed to be accommodated is N, the total data time slot amount of a single time frame is M, and each node can use an N×M two-dimensional matrix T= [ T ] _ij ] _N*M To store slot allocation information:

when a node selects a time slot, only a local time slot allocation matrix is required to be inquired, and then the time slot is applied for occupation;

time slot application and acknowledgement: the time slot allocation of the TDMA protocol is a dynamic interactive process, and the dynamic property is mainly reflected in two stages of time slot application and time slot confirmation; in the time slot application stage, the TDMA cycle monitors the traffic load of each node and other nodes on the communication link, and then calculates the number of data time slots required by the node in the current time frame, and the calculation formula is as follows:

b＝(1-θ)*d*r

in the formula, theta is the data time slot overhead ratio, d is the duration of a single data time slot, r is the data transmission rate of a node, and b is the number of bits which can be transmitted by the single data time slot; in the formula, n is the total node number, t _i Service load capacity is sent to a main node i by a node, and sn is the total amount of data time slots required by the node;

if the number sn of the data time slots required by the node is more than 0, the node needs to sequentially apply for m idle data time slots in a local two-dimensional time slot allocation matrix, and send self time slot application information to the master node through a REQ packet; the master node receiving the REQ packet stores the time slot application information in a local two-dimensional time slot matrix;

time slot release: the dynamic time slot allocation protocol is applied, so that the occupation non-fixed requirement of the node on time slot resources is met; when the node in the network does not use the time slot resources any more due to failure or service interruption, the time slot occupied by the node is released, so that other nodes sending the service can acquire more time slot resources; the two conditions of service interruption and node failure are respectively dealt with by adopting a mode of TDMA protocol active time slot release and passive time slot release; active time slot release solves the problem of service interruption, and passive time slot release solves the node failure condition; and completing time slot release by actively transmitting release information through the child node and directly transmitting the release information through the master node.

2. The unmanned aerial vehicle cluster random access method of claim 1, wherein: the implementation method of the first step is that,

the sub-group network adopts a single channel mode, and fuses control instructions of a control channel into a data channel, namely, when a time slot is transmitted, a clock reference, a time slot allocation table, state data of a current node, control instructions of other nodes and task data of the current node are required to be transmitted; when one node transmits, all other nodes are in a receiving state;

aiming at networking transmission requirements of different unmanned aerial vehicle types of remote control, telemetry and reconnaissance data of a cluster network under different application environments, a special-shaped platform message format is planned from the top layer, an information interaction mechanism is unified, and a physical interface is decoupled; in order to be compatible with different data occurrence rates and packet lengths, the method is suitable for multiple types of protocols, an elastic encapsulation frame protocol combining fixed length and variable length is used in a packet mode of packet service, encapsulation service and the like, and a data frame with fixed length is put into a data chain universal transmission frame in a universal frame mode according to a variable message format message and an international space data system consultation committee protocol standard.

3. The unmanned aerial vehicle cluster random access method of claim 1, wherein: the implementation method of the second step is that,

the sending and receiving adopt a response mechanism, the sending process announces the number of platforms which are currently accessed and the position of the reserved time slot through a broadcast time slot allocation table, the node to be accessed can send an access request in the reserved time slot, the system node judges the access request, if the access is permitted, the updated time slot allocation table is announced through the broadcast, and if the access is forbidden, the node to be accessed is announced through the broadcast to continue waiting or executing other operations; the last time slot of each multi-frame in the receiving frame structure design is a fixed allocation time slot and is used for receiving a time slot reservation request, the first time slot in the sending frame structure design sends a time slot allocation information signaling, each time slot is divided into 1-k sub-time slots, and the k value is dynamically allocated by a central node in a cluster according to network load.