Unmanned aerial vehicle low interaction amount precoding anti-interference method based on strong and weak partition dynamic clustering

Abstract

The invention provides an unmanned aerial vehicle low-interaction amount pre-coding anti-interference method based on strong and weak partition dynamic clustering, relates to the field of unmanned aerial vehicle communication, and solves the problem that when an unmanned aerial vehicle cluster is used as an air distributed base station to communicate with a ground user in an external interference scene, the unmanned aerial vehicle cluster does not interact with the ground user. The problem of how to ensure the communication quality of interfered users while reducing the information interaction amount is solved. The method comprises the following steps: S1, performing unmanned aerial vehicle clustering operation based on clustering size dynamic control; s2, performing ground user confirmation operation based on cluster size dynamic coordination; s3, performing strong and weak cluster interference partition estimation in unmanned aerial vehicle clustering; s4, performing precoding matrix optimization iteration on each unmanned aerial vehicle cluster to realize an anti-interference process; s5, calculating and analyzing the information interaction amount in the communication process, and evaluating the anti-interference effect; the method can ensure and improve the signal to interference plus noise ratio of users interfered by the outside, and can significantly reduce the information interaction amount compared with a non-clustering scheme.

Description

Technical Field

[0001] The present invention relates to the field of unmanned aerial vehicle communication, and in particular to a low-interaction precoding anti-interference method for unmanned aerial vehicles based on strong and weak partition dynamic clustering. Background Art

[0002] With the rapid development of unmanned aerial vehicle (UAV) technology, distributed UAV communication systems are becoming a focus of attention as an emerging research field. In this system, multiple UAVs work together as a whole, and can effectively collect, process and transmit information while performing specific tasks. The core advantages of this technology lie in its excellent flexibility, powerful information processing capabilities and high adaptability to complex and changing environments. Especially when faced with the situation of failure of a single or a few UAVs, the distributed architecture ensures the stability and continuous operation of the entire system, allowing the remaining UAVs to continue working without affecting the completion of the overall mission.

[0003] The importance of distributed drone communication is not only reflected in its flexibility, but also in its powerful information processing capabilities and adaptability to complex environments. For example, in intelligent traffic management, by using drone swarms to monitor road traffic conditions in real time, it can provide support for vehicle navigation, thereby improving road use efficiency; in agricultural monitoring, drones can help farmers achieve precision planting by regularly inspecting crop growth and pest control status; and in the field of disaster response, drones can quickly assess the extent of damage after a natural disaster occurs, assist rescue teams in formulating action plans, and reduce losses. These application scenarios fully demonstrate the key role that distributed drone communication technology plays in the future development of society.

[0004] Despite this, current distributed UAV communications still face some challenges. Since wireless signal propagation relies on a straight-line line-of-sight path, it is susceptible to external malicious electromagnetic interference, resulting in unstable or even interrupted information transmission. In addition, in order to ensure effective collaboration between UAVs, various types of information, including channel status, must be exchanged frequently. However, this high-frequency data exchange not only consumes a lot of bandwidth resources, but also increases communication costs. In view of the existence of these problems, it is particularly important to develop a new method that can reduce unnecessary information exchange while enhancing anti-interference capabilities, which is of great significance for further improving the performance and reliability of distributed UAV communication systems. Summary of the invention

[0005] The purpose of the present invention is to solve the problem of how to reduce the amount of information interaction while ensuring the communication quality of the interfered users when the drone cluster communicates with the ground users as an aerial distributed base station in the external interference scenario; therefore, the present invention proposes a drone low-interaction precoding anti-interference method based on strong and weak partition dynamic clustering. In an external interference environment, the present invention can guarantee and improve the signal-to-interference-noise ratio of the users interfered by the outside world, and at the same time, it can significantly reduce the amount of information interaction compared with the non-clustering scheme, thereby effectively establishing a low-interaction drone communication scheme in an interference environment based on clustering.

[0006] The present invention adopts the following technical solutions to achieve the purpose:

[0007] A method for low-interaction precoding anti-interference of unmanned aerial vehicles based on strong and weak partition dynamic clustering, the method comprising the following steps:

[0008] S1, based on the dynamic control of cluster size, the UAV clustering operation is performed;

[0009] S2, based on the dynamic coordination of cluster size, to perform ground user confirmation operations;

[0010] S3, estimate the strong and weak cluster interference partitions in the UAV clustering;

[0011] S4, perform precoding matrix optimization iteration on each UAV cluster to achieve anti-interference process;

[0012] S5. Calculate and analyze the amount of information interaction during the communication process and evaluate the anti-interference effect.

[0013] Furthermore, step S1 includes the following sub-steps:

[0014] S11, assume that there are M aircrafts in the area, each carrying N t There are U drones with transmitting antennas and hovering at a height of h. At the same time, there are U ground users with single receiving antennas. There is also a jammer in the set area, which sends jamming signals covering the UAV communication frequency band.

[0015] S12, setting the number of clusters K, using the K-means clustering algorithm to cluster the drone group consisting of M drones, and obtaining the initial cluster center and cluster members;

[0016] S13. Based on the ant colony algorithm, a shortest TSP path connecting all cluster centers is generated, and the number of members in each cluster is checked. If the number does not meet the preset number requirement, the remaining drone clusters are checked and deployed in turn according to the shortest TSP path to realize the drone clustering process. After the drone clustering is completed, the number of drones in drone cluster i is calculated by M. i express;

[0017] Step S2 includes the following sub-steps:

[0018] S21, assigning ground users to the nearest drone cluster according to their distance from the drone cluster;

[0019] S22. If the number of ground users served by the drone cluster does not meet the preset number requirement, the ground users in the remaining drone clusters are checked and deployed in turn according to the shortest TSP path to realize the ground user confirmation process. After the ground user confirmation is completed, the number of ground users served by drone cluster i is calculated by U i express;

[0020] The content of step S3 is specifically as follows: based on the transmission power of the drone and the distance between the drone and the ground user, the interference intensity between the two drone clusters is estimated, a set of drone clusters with strong inter-cluster interference is divided, and at the same time, a drone cluster that does not constitute strong inter-cluster interference with other drone clusters is determined;

[0021] Step S4 includes the following sub-steps:

[0022] S41. If a drone cluster does not cause strong inter-cluster interference with other drone clusters, the drone cluster does not consider inter-cluster interference and independently performs optimization iteration of the intra-cluster precoding matrix;

[0023] S42. If there is a drone cluster set, and all members in the set have strong inter-cluster interference with at least one other member in the set, then first, without considering the inter-cluster interference, the optimization iteration of the intra-cluster precoding matrix is ​​performed independently for each member in the set, and then, considering the inter-cluster interference, the intra-cluster precoding matrix of each member in the set is cyclically iterated in turn;

[0024] Step S5 includes the following sub-steps:

[0025] S51, using the traditional SVD algorithm to analyze the amount of information interaction in the communication process under the preset scenario;

[0026] S52, analyzing the corresponding information interaction amount when no clustering operation is performed in a preset scenario;

[0027] S53, performing steps S1 to S4 in a preset scenario and analyzing the corresponding information interaction amount;

[0028] S54: Compare the information interaction amounts of step S52 and step S53 to evaluate the anti-interference effect.

[0029] Furthermore, the specific process of step S13 is as follows:

[0030] S131, traversing the K drone clusters generated by using the K-means clustering algorithm, and checking whether the number of drones in the drone cluster meets the preset number requirement;

[0031] S132. For the drone cluster i that does not meet the preset number requirement, the drone cluster i is numbered as 1, and the remaining K-1 drone clusters are numbered from 2 to K-1 in the order of the shortest TSP path;

[0032] S133, starting from the drone cluster numbered 2, in ascending order of numbering, until a drone available for adjustment is found in the drone cluster numbered P; if the number of drones in drone cluster i is too small, then the drone closest to the center of the drone cluster numbered P-1 in the drone cluster numbered P is assigned to the drone cluster numbered P-1, and then P=P-1 is set, and this operation is repeated until P=1; if the number of drones in drone cluster i is too large, then the drone closest to the center of the drone cluster numbered P in the drone cluster numbered P-1 is assigned to the drone cluster numbered P, and then P=P-1 is set, and this operation is repeated until P=1;

[0033] The specific process of step S22 is:

[0034] S221, traversing the K drone clusters generated by using the K-means clustering algorithm, and checking whether the number of ground users served by the drone cluster meets the preset number requirement;

[0035] S222. For the drone cluster i that does not meet the preset number requirement, the drone cluster i is numbered as 1, and the remaining K-1 drone clusters are numbered from 2 to K-1 in the order of the shortest TSP path;

[0036] S223. Check from the drone cluster numbered 2 in order of numbering from small to large until a ground user receiving service that can be adjusted is found in the drone cluster numbered P; if the number of ground users served by drone cluster i is too small, then the ground user receiving service that is closest to the center of the drone cluster numbered P-1 in the drone cluster numbered P is assigned to the drone cluster numbered P-1 to provide service, and then P=P-1 is set, and this operation is repeated until P=1; if the number of ground users served by drone cluster i is too large, then the ground user receiving service that is closest to the center of the drone cluster numbered P in the drone cluster numbered P-1 is assigned to the drone cluster numbered P to provide service, and then P=P-1 is set, and this operation is repeated until P=1.

[0037] In summary, due to the adoption of this technical solution, the beneficial effects of the present invention are as follows:

[0038] In the presence of external electromagnetic interference, the present invention uses precoding technology to optimize the signal processing, ensuring that the SINR of the interfered user can be effectively improved even under harsh conditions. This not only enhances the reliability of data transmission, but also improves the stability of the overall communication system.

[0039] Compared with the traditional non-clustering solution, the present invention divides the UAV network into multiple small-scale clusters and uses local collaboration within each cluster to replace frequent global information exchange. This method greatly reduces the amount of information interaction required by the system, thereby alleviating bandwidth pressure and communication costs.

[0040] By implementing a strong and weak zoning strategy, that is, taking differentiated countermeasures according to the degree of interference in each area, the present invention can alleviate the impact of external interference on the communication quality in a specific area to a certain extent. In addition, dynamically adjusting the cluster structure to adapt to environmental changes further improves the flexibility and robustness of the entire system.

[0041] For ground users, whether in densely populated urban areas or remote areas, they can enjoy more stable and smooth data connection services. Especially in emergencies or natural disasters, reliable drone communication systems can provide strong support for rescue work and improve response speed and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 The overall process diagram of the method of the present invention is briefly described;

[0043] Figure 2 Schematic diagram of UAV clustered multi-point cooperative communication under interference conditions;

[0044] Figure 3 A flowchart for an exemplary method implementation;

[0045] Figure 4 This is the location distribution diagram of drone clusters and users in the example;

[0046] Figure 5 This is a schematic diagram of the results of K-means clustering of drone clusters in the example;

[0047] Figure 6 This is a schematic diagram of the result of dynamic member adjustment of the drone cluster in the example;

[0048] Figure 7 This is a schematic diagram of the result of initial confirmation of the user in the example;

[0049] Figure 8 A schematic diagram showing the result of dynamic member adjustment for users in the example;

[0050] Fig. 9This is a schematic diagram comparing the simulation results of the interruption probability of the user in the example;

[0051] Fig.10 This is a comparative diagram of the reduction ratio of information interaction in the example. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0053] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. 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.

[0054] Example

[0055] like Figure 1 As shown, a method for low-interaction precoding anti-interference of UAV based on strong and weak partition dynamic clustering, the method comprises the following steps:

[0056] S1, based on the dynamic control of cluster size, the UAV clustering operation is performed;

[0057] S2, based on the dynamic coordination of cluster size, to perform ground user confirmation operations;

[0058] S3, estimate the strong and weak cluster interference partitions in the UAV clustering;

[0059] S4, perform precoding matrix optimization iteration on each UAV cluster to achieve anti-interference process;

[0060] S5. Calculate and analyze the amount of information interaction during the communication process and evaluate the anti-interference effect.

[0061] This embodiment will introduce the details of each step in the method in detail with specific examples.

[0062] Before the method is performed, the system model related content is initialized first, and the related parameters in the system model are determined. Figure 2 As an analogy, the UAV cluster multi-point cooperative communication architecture under the interference condition is designed to have M = 10 UAVs, and the number of transmitting antennas of each UAV is N t=1, and U = 5 single-antenna ground users. At the same time, the minimum number of drones in each drone cluster is designed to be 2, and the number of ground users served is at least 1 and not greater than the number of drones in the cluster; then the possible values ​​of the number of clusters K are 1, 2, 3, 4, and 5.

[0063] In this embodiment, the hovering height of the drone is set to 100m and the height of the ground user is set to 0m. The positions of the drone and the ground user are distributed in a square area with a side length of 6km. There is a jammer in the center of the area with a jamming power P J = 100W power is used to transmit Gaussian white noise covering the communication frequency band, and the transmission signal-to-interference-noise ratio of the drone is set to 0 to 20dB. In this embodiment, the communication channel between the drone and the ground user is a Rayleigh channel, where the calculation formula for the large-scale fading of the drone to the ground user channel is:

[0064]

[0065] Where ρ is the path loss, d is the distance between the UAV and the ground user, d0 is the reference distance, and ε is the path loss exponent. Small-scale parameters in the channel The elements in obey Rayleigh distribution, and the downlink channel of communication between UAV and ground users is expressed as

[0066] Based on the above initialization content, the method of this embodiment is as follows Figure 3 Follow the instructions.

[0067] Step 1: UAV clustering method under dynamic control of cluster size

[0068] Figure 4 The distribution diagram of drones and ground users in a simulation. The drones are clustered according to their locations. Taking the cluster number K = 3 as an example, the drones are initially clustered based on their locations using the K-means clustering algorithm. Figure 5 As shown, different colors are used to distinguish the results after clustering. The green drone cluster in the figure does not meet the restriction that the minimum number of drones is 2.

[0069] In order to ensure that the number of drones in each cluster meets the requirements, the number of drone members in each cluster is adjusted. Figure 5 The black cross represents the center of each cluster, and the red connecting line represents the TSP closed path connecting the centers of all drone clusters; the green drone cluster is numbered 1, and the red and blue drone clusters are numbered 2 and 3. There are drones available for adjustment in the red drone cluster, which corresponds to the drone cluster numbered 2. The drone in the drone cluster numbered 2 that is closest to the center of the drone cluster numbered 1 is assigned to the drone cluster numbered 1. At this point, the drone clustering is completed, and the results are as follows: Figure 6 shown.

[0070] Step 2: Ground user confirmation method under dynamic control of cluster size

[0071] First, the ground users in the area are identified with the nearest drone cluster. The results are as follows: Figure 7 . There is no ground user in the red drone cluster, so the ground users of each drone cluster are deployed. The red drone cluster is numbered 1, and the blue and green drone clusters are numbered 2 and 3. There is a ground user available for adjustment in the blue drone cluster, which corresponds to the drone cluster numbered 2. The ground user in the drone cluster numbered 2 that is closest to the center of the drone cluster numbered 1 is assigned to the drone cluster numbered 1. At this point, the ground user confirmation is complete, and the result is as follows Figure 8 shown.

[0072] Step 3: Strong and weak cluster interference distinction estimation method

[0073] Using the interference intensity factor δ i,j To estimate the degree of interference between the two clusters; the two drone clusters are respectively recorded as drone cluster i and drone cluster j, then the interference intensity coefficient δ between drone cluster i and drone cluster j is i,j As follows:

[0074]

[0075] In the formula, ρ k,i Represents the total number of M in drone cluster i i The path loss vector formed by the UAV and the ground user k, ρ k,j Similarly, corresponding to drone cluster j; ρ k,i The specific expression is as follows:

[0076]

[0077] In the formula, the coefficient ρ k,i,l represents the path attenuation between the lth UAV in the UAV cluster i and the ground user k; P i The total number of drones in cluster i is M i The transmission power vector of the UAV, P j Similarly, corresponding to drone cluster j; P i The specific expression is as follows:

[0078]

[0079] In the formula, the coefficient P i,l Represents the transmission power of the lth UAV in UAV cluster i.

[0080] Table 1 below shows the calculation results of the interference intensity coefficient of the example, and the calculation result value 3 is selected as the threshold for defining strong inter-cluster interference, so cluster 1 and cluster 2 constitute strong inter-cluster interference.

[0081] Table 1 Schematic diagram of calculation results of interference intensity coefficient

[0082]

[0083]

[0084] Step 4: Iterative optimization method of precoding matrix for each drone cluster

[0085] In this embodiment, for the drone cluster 3 that does not constitute strong inter-cluster interference with other clusters, inter-cluster interference is not considered, and the optimization iteration of the intra-cluster precoding matrix is ​​performed independently, as follows:

[0086] When the optimization iteration of the intra-cluster precoding matrix is ​​performed independently, the optimization objective of the precoding optimization problem is constructed as follows, with the SINR level of the interfered terrestrial users as the condition:

[0087]

[0088] Where W i is the optimization variable; k represents the SINR of terrestrial user k, which is calculated as follows:

[0089]

[0090] In the formula, the optimization variable It is the precoding matrix of the drones in drone cluster i to their ground users; represents the channel vector between UAV j in UAV cluster i and ground user k; represents the precoding vector designed by UAV j in UAV cluster i for ground user k; s k represents the data symbol of ground user k; h k,J represents the channel coefficient between the jammer and the ground user k; s J represents the malicious noise symbol sent by the jammer, and P J Represents the transmit power of the jammer; use n k Represents the interference signal sent by the jammer, that is, the mean is 0 and the variance is Gaussian white noise with complex Gaussian distribution;

[0091] Then, the transmission power of the UAV is also required to meet the following constraints:

[0092]

[0093] In the formula, P represents the maximum transmission power of the drone; (…) H stands for conjugate transpose;

[0094] Next, in order to prevent the SINR of all terrestrial users from falling into a generally poor situation as much as possible after optimization, the average SINR constraint of all terrestrial users is set as follows:

[0095]

[0096] In the formula, γ th Represents the set average receive SINR threshold.

[0097] For drone clusters 1 and 2 that form strong inter-cluster interference, first, inter-cluster interference is not considered, and the intra-cluster precoding matrix of each drone cluster is optimized and iterated independently. Then, inter-cluster interference is considered, and the intra-cluster precoding matrix of each drone cluster is iterated in turn, as follows:

[0098] For a set of drone clusters that form strong inter-cluster interference with other drone clusters, the process of optimization iteration and loop iteration from not considering inter-cluster interference to considering inter-cluster interference is as follows:

[0099] Solve the initial solution W of the precoding matrix 0 First, we use the optimization problem without considering inter-cluster interference to solve the initial solution of the precoding matrix of cluster i The optimization goal of the optimization problem is:

[0100]

[0101] In the formula, γ k represents the SINR of terrestrial user k, which is calculated as follows:

[0102]

[0103] In the formula, the optimization variable It is the initial precoding matrix of the UAVs in the UAV cluster i to their ground users; represents the channel vector between UAV j in UAV cluster i and ground user k; represents the initial precoding vector designed by UAV j in UAV cluster i for ground user k; s k represents the data symbol of ground user k; h k,J represents the channel coefficient between the jammer and the ground user k; s J represents the malicious noise symbol sent by the jammer, and P J Represents the transmit power of the jammer; use n kRepresents the interference signal sent by the jammer, that is, the mean is 0 and the variance is Gaussian white noise with complex Gaussian distribution;

[0104] Then, the transmission power of the UAV is also required to meet the following constraints:

[0105]

[0106] In the formula, P represents the maximum transmission power of the drone; (…) H stands for conjugate transpose;

[0107] Next, in order to prevent the SINR of all terrestrial users from falling into a generally poor situation as much as possible after optimization, the average SINR constraint of all terrestrial users is set as follows:

[0108]

[0109] In the formula, γ th Represents the set average receiving SINR threshold; used at the same time Represents the initial precoding matrix of drone m, using Represents the transmitted symbol vector of drone m.

[0110] Then, after considering the inter-cluster interference, the final precoding matrix is ​​solved by cyclic iteration; after obtaining the initial precoding matrix of each drone cluster in the drone set with strong inter-cluster interference, the influence of inter-cluster interference is considered in the SINR expression of ground user k, and the optimization objective of the new optimization problem is constructed as follows:

[0111]

[0112] At this time, γ k for:

[0113]

[0114] In the formula, the optimization variable It is the precoding matrix of the drones in drone cluster i to their ground users in the Lth iteration; represents the precoding vector designed by UAV j in UAV cluster i for ground user k in the Lth iteration; h k,m represents the channel vector between UAV m belonging to UAV cluster n and ground user k served by UAV cluster i; It is the precoding vector designed for UAV m in UAV cluster n for its ground user in the L-1th iteration; The data vector representing the ground user corresponding to UAV m in UAV cluster n;

[0115] Then, the transmission power of the UAV is also required to meet the following constraints:

[0116]

[0117] In the formula, P represents the maximum transmission power of the drone; (…) H stands for conjugate transpose;

[0118] Then, in the same way, in order to prevent the SINR of all terrestrial users from falling into a generally poor situation as much as possible after optimization, the average SINR constraint of all terrestrial users is set as follows:

[0119]

[0120] Use at the same time represents the precoding matrix of drone m in the Lth loop iteration, using represents the transmitted symbol vector of UAV m;

[0121] The precoding matrix of each drone cluster in the drone set is iteratively optimized in turn. When all drone clusters in the drone set have completed the precoding optimization in one round of iterative optimization, the norm difference and A of the precoding matrix are calculated as follows:

[0122]

[0123] The cyclic iteration process is continued until the number of iterations L reaches the maximum number of iterations, or the norm difference of the precoding matrix and A is less than the preset threshold, the iteration ends, and the final precoding matrix of each drone cluster after optimization is output. The process of the above strong inter-cluster interference precoding matrix cyclic iteration algorithm is briefly described as follows:

[0124] 1. Initialization: accuracy threshold, maximum number of iterations L max ;

[0125] 2. Input: Initial precoding matrix W of each drone cluster 0 , the number of iterations L = 0;

[0126] 3. Carry out the cycle;

[0127] 4. L = L + 1;

[0128] 5. Solve the optimization problem to obtain the precoding matrix W of each cluster in the Lth iteration process L ;

[0129] 6. Calculate the precoding matrix norm difference and A;

[0130] 7. Until the number of iterations L is greater than or equal to the maximum number of iterations L max Or the sum of the precoding matrix norm difference A is less than a preset accuracy threshold;

[0131] 8. Output: The final precoding matrix W of each drone cluster final .

[0132] In the simulation of this embodiment, a simulated annealing algorithm is used for the above process, and the method is described as follows:

[0133] First, initialize the parameters, including the maximum number of iterations L max , initial temperature T0, final temperature T f , temperature attenuation coefficient α and precoding update step length l; after assigning the initial temperature T0 to the real-time temperature variable T, randomly generate the initial precoding matrix, and then check whether the corresponding constraints are met. If they are met, it is used as the initial solution of the algorithm; calculate and record the corresponding fitness value γ according to the initial solution min1 , which is the minimum receiving SINR between ground users;

[0134] Determine whether the number of iterations has reached the maximum number of iterations. If so, output the result. If not, continue to update the initial solution, use the rand function to generate a random complex vector, and control the update degree of the precoding matrix by multiplying it with the precoding update step length l. Add it to the initial solution and perform power normalization to obtain a new solution. After calculating the new solution, obtain the new fitness value γ min2 ;

[0135] Then calculate the difference Δ between the new fitness value and the original fitness value, that is:

[0136] Δ=γ min2 -γ min1

[0137] The Metropolis criterion is used to accept the new fitness value corresponding to the new solution. The specific content of the Metropolis criterion is: if the difference Δ>0, the new solution is accepted and the new fitness value is assigned to the initial solution; if the difference Δ≤0, the new solution is accepted with a probability of exp(Δ / T), and the new fitness value is assigned to the initial solution when accepted;

[0138] Then, the temperature attenuation coefficient α is used, α∈(0,1), α is multiplied by the real-time temperature variable T to reduce the temperature, the reduced temperature is assigned to T, and it is determined whether the real-time temperature reaches the final temperature T f If it is reached, the result is output; if it is not reached, it returns to the number of iterations and repeats the above process.

[0139] Table 2 shows the parameter settings in the simulation, as follows:

[0140] Table 2 Simulation parameter setting diagram

[0141]

[0142]

[0143] Fig. 9 The simulation result diagram of the interruption probability of the optimization scheme used in the clustering and non-clustering cases in this embodiment and the interruption probability under the SVD algorithm in the non-clustering case is shown in FIG. The interruption probability is defined as the probability P that the user's received signal-to-noise ratio is lower than a preset threshold. out ,as follows:

[0144] P out =Pr(γ<γ TH )

[0145] Where Pr(…) represents the probability, γ is the receiving signal-to-noise ratio of the ground user; TH is the signal-to-noise ratio threshold, below which communication is interrupted. Fig. 9 It can be seen that when the number of clusters in this embodiment is 3, the interruption probability when clustering is slightly lower than that without clustering under the same transmission signal-to-noise ratio, and the interruption probability of the result of this embodiment is lower than that of the result using the SVD algorithm.

[0146] Step 5: Information Interaction Analysis

[0147] The information interaction volume of the communication process in the traditional SVD algorithm and the information interaction volume in the present embodiment without clustering are both in the distributed communication scenario of UAVs. The number of UAVs in the scenario is M, and the number of transmitting antennas carried by the UAVs is N. t , the number of single-antenna ground users is U; the number of information interactions generated in one communication process is as follows:

[0148] The drone cluster first determines the number and location information of the ground users it serves. The ground users send uplink pilot signals, and after receiving them, the drones feed back the location information of the ground users to the ground control center, which determines the number of ground users it serves. The number of ground users sending uplink pilot signals is recorded as In this process, the number of information interactions generated by the drone cluster receiving the uplink pilot signal is The number of information interactions when the UAV cluster transmits information back to the ground control center is M, and the number of information interactions when the ground control center transmits the confirmed relevant information back to the UAV cluster is also M; assuming that the ground control center determines U, The ground users are the objects of the UAV cluster service, and the process generates Information interaction;

[0149] The drone cluster measures the downlink channel state information at the same time; the number of information interactions when the drone sends the downlink pilot to the ground user for reception is MU, and the ground user feeds back the downlink CSI to the drone cluster, which also generates MU information interactions; this process generates a total of 2MU information interactions;

[0150] Then calculate the information exchange within the drone cluster; for each drone in the distributed drone cluster, it needs to send information to other drones in the cluster, and a total of M (M-1) information interactions are generated in this process;

[0151] Finally, the drone transmits the information to the ground user; M drones send information to U ground users, resulting in MU information interactions in total;

[0152] Through the above method, the corresponding information interaction volume is calculated and analyzed as a basis for subsequent comparison.

[0153] Similarly, in this embodiment, the scenario established in the clustering situation is a drone distributed communication scenario, the number of drones in the scenario is M, and the number of transmitting antennas carried by the drones is N. t , the number of single-antenna ground users is U, the number of clusters is K, and the number of drones in drone cluster i is M i , the number of ground users served by drone cluster i is U i ; The number of information interactions generated during one communication process is as follows:

[0154] The drone cluster first determines the number and location information of the ground users it serves. The ground users send uplink pilot signals, and after receiving them, the drones feed back the location information of the ground users to the ground control center, which determines the number of ground users it serves. The number of ground users sending uplink pilot signals is recorded as In this process, the number of information interactions generated by the drone cluster receiving the uplink pilot signal is The number of information interactions when the UAV cluster transmits information back to the ground control center is M, and the number of information interactions when the ground control center transmits the confirmed relevant information back to the UAV cluster is also M; assuming that the ground control center determines U, The ground users are the objects of the UAV cluster service, and the process generates Information interaction;

[0155] The drone cluster simultaneously measures the downlink channel state information; for drone cluster i, the number of information interactions for sending downlink pilot signals to the ground users it serves is M i U i , the ground user feeds downlink CSI back to the drone cluster i and also generates M i U i times of information interaction; for K drone clusters, this process generates Number of information interactions;

[0156] Then calculate the information exchange within the drone cluster; for drone cluster i, each drone sends information to other drones in the cluster, generating M i (M i -1) times of information interaction; for K drone clusters, this process generates Number of information interactions;

[0157] Finally, the UAV transmits the information to the ground user; for UAV cluster i, the M i The drone sends a message to the U i terrestrial users, generating a total of M i U i times of information interaction; for K drone clusters, this process generates Number of information interactions;

[0158] Through the above method, the corresponding information interaction volume is calculated and analyzed as a basis for subsequent comparison.

[0159] Table 3 shows a comparison table of information exchange amounts between clustering and SVD / non-clustering schemes in one communication process in this embodiment, and Table 4 shows a comparison table of information exchange amounts in each step of the communication process.

[0160] Table 3 Comparison of information interaction between clustering and SVD / non-clustering schemes

[0161]

[0162] Table 4 Comparison of information interaction in each step of the communication process

[0163]

[0164] The analysis of the reduction in the amount of information interaction in a communication process with clustering and comparison between using the SVD algorithm and without clustering is as follows:

[0165] In the (2) stage of Table 4, the amount of information interaction using the SVD algorithm / non-clustering scheme is 2MU, and the amount of information interaction using the clustering scheme is satisfy When Substitute it into the information interaction formula before clustering, and we get:

[0166]

[0167] Due to M i and U iare all positive integers. Obviously, the amount of information interaction before clustering in stage (2) is greater than that after clustering. Similarly, it can be proved that the amount of information interaction before clustering in stage (4) is greater than that after clustering.

[0168] The amount of information interaction of clustering in stage (3) in Table 4 is The amount of information interaction using the SVD algorithm / non-clustering solution is M(M-1)=M 2 -M, so just judge The size of Substitute it in and simplify it to get 2∑ 1≤j≤l≤K M j M l . Due to M i are all positive integers, which proves that the amount of information interaction before clustering in stage (3) is greater than that after clustering.

[0169] Fig.10 The figure shows the amount of information interaction in a communication process with different clustering numbers under a specific number of ground users and number of drones. Combined with the above analysis, in a communication process with ground users, the amount of information interaction after clustering is reduced by 1.1% compared with that before clustering. Second-rate.

Claims

1. A UAV low-interaction precoding anti-interference method based on strong and weak partition dynamic clustering, characterized in that: The method comprises the following steps: S1, based on the dynamic control of cluster size, the UAV clustering operation is performed; S2, based on the dynamic coordination of cluster size, to perform ground user confirmation operations; S3, estimate the strong and weak cluster interference partitions in the UAV clustering; S4, perform precoding matrix optimization iteration on each UAV cluster to achieve anti-interference process; S5. Calculate and analyze the amount of information interaction during the communication process and evaluate the anti-interference effect.

2. The method for low-interaction precoding anti-interference of unmanned aerial vehicles based on strong and weak partition dynamic clustering according to claim 1 is characterized by: Step S1 includes the following sub-steps: S11, assume that there are M aircrafts in the area, each carrying N t There are U drones with transmitting antennas and hovering at a height of h. At the same time, there are U ground users with single receiving antennas. There is also a jammer in the set area, which sends jamming signals covering the UAV communication frequency band. S12, setting the number of clusters K, using the K-means clustering algorithm to cluster the drone group consisting of M drones, and obtaining the initial cluster center and cluster members; S13. Based on the ant colony algorithm, a shortest TSP path connecting all cluster centers is generated, and the number of members in each cluster is checked. If the number does not meet the preset number requirement, the remaining drone clusters are checked and deployed in turn according to the shortest TSP path to realize the drone clustering process. After the drone clustering is completed, the number of drones in drone cluster i is calculated by M. i express; Step S2 includes the following sub-steps: S21, assigning ground users to the nearest drone cluster according to their distance from the drone cluster; S22. If the number of ground users served by the drone cluster does not meet the preset number requirement, the ground users in the remaining drone clusters are checked and deployed in turn according to the shortest TSP path to realize the ground user confirmation process. After the ground user confirmation is completed, the number of ground users served by drone cluster i is calculated by U i express; The content of step S3 is specifically as follows: based on the transmission power of the drone and the distance between the drone and the ground user, the interference intensity between the two drone clusters is estimated, a set of drone clusters with strong inter-cluster interference is divided, and at the same time, a drone cluster that does not constitute strong inter-cluster interference with other drone clusters is determined; Step S4 includes the following sub-steps: S41. If a drone cluster does not cause strong inter-cluster interference with other drone clusters, the drone cluster does not consider inter-cluster interference and independently performs optimization iteration of the intra-cluster precoding matrix; S42. If there is a drone cluster set, and all members in the set have strong inter-cluster interference with at least one other member in the set, then first, without considering the inter-cluster interference, the optimization iteration of the intra-cluster precoding matrix is ​​performed independently for each member in the set, and then, considering the inter-cluster interference, the intra-cluster precoding matrix of each member in the set is cyclically iterated in turn; Step S5 includes the following sub-steps: S51, using the traditional SVD algorithm to analyze the amount of information interaction in the communication process under the preset scenario; S52, analyzing the corresponding information interaction amount when no clustering operation is performed in a preset scenario; S53, performing steps S1 to S4 in a preset scenario and analyzing the corresponding information interaction amount; S54: Compare the information interaction amounts of step S52 and step S53 to evaluate the anti-interference effect.

3. The method for low-interaction precoding anti-interference of unmanned aerial vehicles based on strong and weak partition dynamic clustering according to claim 2 is characterized in that: The specific process of step S13 is: S131, traversing the K drone clusters generated by using the K-means clustering algorithm, and checking whether the number of drones in the drone cluster meets the preset number requirement; S132. For the drone cluster i that does not meet the preset number requirement, the drone cluster i is numbered as 1, and the remaining K-1 drone clusters are numbered from 2 to K-1 in the order of the shortest TSP path; S133, starting from the drone cluster numbered 2, in ascending order of numbering, until a drone available for adjustment is found in the drone cluster numbered P; if the number of drones in drone cluster i is too small, then the drone closest to the center of the drone cluster numbered P-1 in the drone cluster numbered P is assigned to the drone cluster numbered P-1, and then P=P-1 is set, and this operation is repeated until P=1; if the number of drones in drone cluster i is too large, then the drone closest to the center of the drone cluster numbered P in the drone cluster numbered P-1 is assigned to the drone cluster numbered P, and then P=P-1 is set, and this operation is repeated until P=1; The specific process of step S22 is: S221, traversing the K drone clusters generated by using the K-means clustering algorithm, and checking whether the number of ground users served by the drone cluster meets the preset number requirement; S222. For the drone cluster i that does not meet the preset number requirement, the drone cluster i is numbered as 1, and the remaining K-1 drone clusters are numbered from 2 to K-1 in the order of the shortest TSP path; S223. Check from the drone cluster numbered 2 in order of numbering from small to large until a ground user receiving service that can be adjusted is found in the drone cluster numbered P; if the number of ground users served by drone cluster i is too small, then the ground user receiving service that is closest to the center of the drone cluster numbered P-1 in the drone cluster numbered P is assigned to the drone cluster numbered P-1 to provide service, and then P=P-1 is set, and this operation is repeated until P=1; if the number of ground users served by drone cluster i is too large, then the ground user receiving service that is closest to the center of the drone cluster numbered P in the drone cluster numbered P-1 is assigned to the drone cluster numbered P to provide service, and then P=P-1 is set, and this operation is repeated until P=1.

4. The method for low-interaction precoding anti-interference of unmanned aerial vehicles based on strong and weak partition dynamic clustering according to claim 2 is characterized in that: In step S3, the two drone clusters are recorded as drone cluster i and drone cluster j respectively, and the interference intensity coefficient δ between drone cluster i and drone cluster j is i，j As follows: In the formula, ρ k，i Represents the total number of M in drone cluster i i The path loss vector formed by the UAV and the ground user k, ρ k，j Similarly, corresponding to drone cluster j; ρ k，i The specific expression is as follows: In the formula, the coefficient ρ k,i，l represents the path attenuation between the lth UAV in UAV cluster i and the ground user k; P i The total number of drones in cluster i is M i The transmitting power vector of the UAV, P j Similarly, corresponding to drone cluster j; P i The specific expression is as follows: In the formula, the coefficient P i，l Represents the transmission power of the lth UAV in UAV cluster i.

5. The method for low-interaction precoding anti-interference of unmanned aerial vehicles based on strong and weak partition dynamic clustering according to claim 4 is characterized in that: In step S3, the path loss calculation formula between the UAV and the ground user is as follows: Where ρ is the path loss, d is the distance between the UAV and the ground user, d0 is the reference distance, and ε is the path loss exponent.

6. The method for low-interaction precoding anti-interference of unmanned aerial vehicles based on strong and weak partition dynamic clustering according to claim 2 is characterized by: In step S41, for the drone cluster that does not constitute strong inter-cluster interference, when independently optimizing the intra-cluster precoding matrix, the optimization objective of the precoding optimization problem is constructed as follows, with the SINR level of the interfered ground users as the condition: Where W i is the optimization variable; k represents the SINR of terrestrial user k, which is calculated as follows: In the formula, the optimization variable It is the precoding matrix of the drones in drone cluster i to their ground users; represents the channel vector between UAV j in UAV cluster i and ground user k; represents the precoding vector designed by UAV j in UAV cluster i for ground user k; s k represents the data symbol of ground user k; h k，J represents the channel coefficient between the jammer and the ground user k; s J represents the malicious noise symbol sent by the jammer, and P J Represents the transmit power of the jammer; use n k Represents the interference signal sent by the jammer, that is, the mean is 0 and the variance is Gaussian white noise with complex Gaussian distribution; Then, the transmission power of the UAV is also required to meet the following constraints: In the formula, P represents the maximum transmission power of the drone; (…) H stands for conjugate transpose; Next, the average SINR constraint of all terrestrial users is set as follows: In the formula, γ th Represents the set average receive SINR threshold.

7. The method for low-interaction precoding anti-interference of unmanned aerial vehicles based on strong and weak partition dynamic clustering according to claim 2 is characterized by: In step S42, for the set of drone clusters consisting of drone clusters that form strong inter-cluster interference with other drone clusters, the process of optimizing iteration and loop iteration from not considering inter-cluster interference to considering inter-cluster interference is as follows: S421, solving the initial solution W of the precoding matrix o First, we use the optimization problem without considering inter-cluster interference to solve the initial solution of the precoding matrix of cluster i The optimization goal of the optimization problem is: In the formula, γ k represents the SINR of terrestrial user k, which is calculated as follows: In the formula, the optimization variable It is the initial precoding matrix of the UAVs in the UAV cluster i to their ground users; represents the channel vector between UAV j in UAV cluster i and ground user k; represents the initial precoding vector designed by UAV j in UAV cluster i for ground user k; s k represents the data symbol of ground user k; h k，J represents the channel coefficient between the jammer and the ground user k; s J represents the malicious noise symbol sent by the jammer, and P J Represents the transmit power of the jammer; use n k Represents the interference signal sent by the jammer, that is, the mean is 0 and the variance is Gaussian white noise with complex Gaussian distribution; Then, the transmission power of the UAV is also required to meet the following constraints: In the formula, P represents the maximum transmission power of the drone; (…) H stands for conjugate transpose; Next, the average SINR constraint of all terrestrial users is set as follows: In the formula, γ th Represents the set average receiving SINR threshold; used at the same time Represents the initial precoding matrix of drone m, using represents the transmitted symbol vector of UAV m; S422, loop iteratively solve the final precoding matrix; after obtaining the initial precoding matrix of each drone cluster in the drone set with strong inter-cluster interference through step S421, consider the impact of inter-cluster interference in the SINR expression of ground user k, and construct the optimization objective of the new optimization problem as follows: At this time, γ k for: In the formula, the optimization variable It is the precoding matrix of the drones in drone cluster i to their ground users in the Lth iteration; represents the precoding vector designed by UAV j in UAV cluster i for ground user k in the Lth iteration; h k，m represents the channel vector between UAV m belonging to UAV cluster n and ground user k served by UAV cluster i; It is the precoding vector designed for UAV m in UAV cluster n for its ground user in the L-1th iteration; The data vector representing the ground user corresponding to UAV m in UAV cluster n; Then, the transmission power of the UAV is also required to meet the following constraints: In the formula, P represents the maximum transmission power of the drone; (…) H stands for conjugate transpose; Then, in the same manner as step S41, the average SINR constraint of all terrestrial users is set as follows: Use at the same time represents the precoding matrix of drone m in the Lth loop iteration, using represents the transmitted symbol vector of UAV m; The precoding matrix of each drone cluster in the drone set is iteratively optimized in turn. When all drone clusters in the drone set have completed the precoding optimization in one round of iterative optimization, the norm difference and A of the precoding matrix are calculated as follows: The iterative process is continued until the number of iterations L reaches the maximum number of iterations, or the norm difference of the precoding matrix and A is less than the preset threshold, then the iteration ends and the final precoding matrix of each drone cluster after optimization is output.

8. The method for low-interaction precoding anti-interference of unmanned aerial vehicles based on strong and weak partition dynamic clustering according to claim 7 is characterized in that: The precoding optimization algorithm within the drone cluster uses the simulated annealing algorithm in the following way: First, initialize the parameters, including the maximum number of iterations L max , initial temperature T0, final temperature T f , temperature attenuation coefficient α and precoding update step length l; after assigning the initial temperature T0 to the real-time temperature variable T, randomly generate the initial precoding matrix, and then check whether the corresponding constraints are met. If so, it is used as the initial solution of the algorithm; Calculate and record the corresponding fitness value γ according to the initial solution min1 , which is the minimum receiving SINR between ground users; Determine whether the number of iterations has reached the maximum number of iterations. If so, output the result. If not, continue to update the initial solution, use the rand function to generate a random complex vector, and control the update degree of the precoding matrix by multiplying it with the precoding update step length l. Add it to the initial solution and perform power normalization to obtain a new solution. After calculating the new solution, obtain the new fitness value γ min2 ; Then calculate the difference Δ between the new fitness value and the original fitness value, that is: Δ=γ min2 -c min1 The Metropolis criterion is used to accept the new fitness value corresponding to the new solution. The specific content of the Metropolis criterion is: if the difference Δ>0, the new solution is accepted and the new fitness value is assigned to the initial solution; if the difference Δ≤0, the new solution is accepted with a probability of exp(Δ / T), and the new fitness value is assigned to the initial solution when accepted; Then, the temperature attenuation coefficient α is used, α∈(0,1), α is multiplied by the real-time temperature variable T to reduce the temperature, the reduced temperature is assigned to T, and it is determined whether the real-time temperature reaches the final temperature T f If it is reached, the result is output; if it is not reached, it returns to the number of iterations and repeats the above process.

9. The method for low-interaction precoding anti-interference of unmanned aerial vehicles based on strong and weak partition dynamic clustering according to claim 2 is characterized by: The preset scenario established in step S51 and step S52 is a UAV distributed communication scenario, the number of UAVs in the scenario is M, and the number of transmitting antennas carried by the UAVs is N. t , the number of single-antenna ground users is U; the number of information interactions generated in one communication process is as follows: The drone cluster first determines the number and location information of the ground users it serves. The ground users send uplink pilot signals, and after receiving them, the drones feed back the location information of the ground users to the ground control center, which determines the number of ground users it serves. The number of ground users sending uplink pilot signals is recorded as In this process, the number of information interactions generated by the drone cluster receiving the uplink pilot signal is The number of information interactions when the UAV cluster transmits information back to the ground control center is M, and the number of information interactions when the ground control center transmits the confirmed relevant information back to the UAV cluster is also M; assuming that the ground control center determines U, The ground users are the objects of the UAV cluster service, and the process generates Information interaction; The drone cluster measures the downlink channel state information at the same time; the number of information interactions when the drone sends the downlink pilot to the ground user for reception is MU, and the ground user feeds back the downlink CSI to the drone cluster, which also generates MU information interactions; This process generates a total of 2MU information interactions; Then calculate the information exchange within the drone cluster; for each drone in the distributed drone cluster, it needs to send information to other drones in the cluster, and a total of M (M-1) information interactions are generated in this process; Finally, the drone transmits the information to the ground user; M drones send information to U ground users, resulting in MU information interactions in total; Through the above method, the corresponding information interaction amount is calculated and analyzed as a comparison basis in step S54.

10. The method for low-interaction precoding anti-interference of unmanned aerial vehicles based on strong and weak partition dynamic clustering according to claim 2 is characterized by: In step S53, the scenario established in the clustering situation is a UAV distributed communication scenario, the number of UAVs in the scenario is M, and the number of transmitting antennas carried by the UAVs is N. t , the number of single-antenna ground users is U, the number of clusters is K, and the number of drones in drone cluster i is M i , the number of ground users served by drone cluster i is U i ; The number of information interactions generated during one communication process is as follows: The drone cluster first determines the number and location information of the ground users it serves. The ground users send uplink pilot signals, and after receiving them, the drones feed back the location information of the ground users to the ground control center, which determines the number of ground users it serves. The number of ground users sending uplink pilot signals is recorded as In this process, the number of information interactions generated by the drone cluster receiving the uplink pilot signal is The number of information interactions when the UAV cluster transmits information back to the ground control center is M, and the number of information interactions when the ground control center transmits the confirmed relevant information back to the UAV cluster is also M; assuming that the ground control center determines U, The ground users are the objects of the UAV cluster service, and the process generates Information interaction; The drone cluster simultaneously measures the downlink channel state information; for drone cluster i, the number of information interactions for sending downlink pilot signals to the ground users it serves is M i U i , the ground user feeds downlink CSI back to the drone cluster i and also generates M i U i times of information interaction; for K drone clusters, this process generates Number of information interactions; Then calculate the information exchange within the drone cluster; for drone cluster i, each drone sends information to other drones in the cluster, generating M i (M i -1) times of information interaction; for K drone clusters, this process generates Number of information interactions; Finally, the UAV transmits the information to the ground user; for UAV cluster i, the M i The drone sends a message to the U i terrestrial users, generating a total of M i U i times of information interaction; for K drone clusters, this process generates Number of information interactions; Through the above method, the corresponding information interaction amount is calculated and analyzed as a comparison basis in step S54.

Images