Method for maximizing average secrecy throughput of all users of UAV-assisted short packet communication system

Abstract

The invention relates to a method for maximizing the average secrecy throughput in a short packet communication system assisted by an unmanned aerial vehicle (UAV). The method comprises the following steps of: firstly, selecting a short packet communication system; the method maximizes the average confidential throughput of all users by jointly optimizing user scheduling, transmit power, transport block length, and UAV trajectory. Firstly, an original problem is converted into four sub-problems of user scheduling optimization, power distribution optimization, block length selection optimization and unmanned aerial vehicle trajectory optimization. In order to solve the non-convex subproblem of unmanned aerial vehicle trajectory optimization, an improved particle swarm optimization (PSO) algorithm is adopted to search for an optimal or suboptimal UAV flight path. According to the method, under the optimization framework of overall block coordinate descent (BCD), other sub-problems are solved by means of methods such as continuous convex approximation (SCA), and the like, and therefore, the method can be used for solving the sub-problems of the block coordinate descent (BCD) of the block coordinate descent (BCD) of the block coordinate descent (BCD) of the block coordinate descent (BCD) of the block coordinate descent (BCD) of the block coordinate descent (BCD) of the block coordinate descent (BCD) of the block coordinate descent (BCD) of the block coordinate descent (BCD). According to the method provided by the invention, the communication safety performance of all users in the UAV-assisted short packet communication system can be improved, and a new design thought and optimization method are provided for the application of the UAV in a time delay sensitive wireless communication scene.

Description

Technical Field

[0001] This invention relates to the field of wireless communication systems, and more specifically to a method for maximizing average secure throughput in a UAV-assisted short packet communication system. Background Technology

[0002] With the rapid development of mobile communications, the Internet of Things (IoT), and edge computing, future communication systems face higher demands for low latency, high reliability, and high security. Especially in latency-sensitive scenarios such as industrial control, autonomous driving, and telemedicine, traditional long-packet communication methods struggle to meet the demands for short data transmission and fast response, prompting widespread research into short-packet communication (SPC) technology. Unmanned aerial vehicle (UAV)-assisted communication, as a flexible and efficient aerial communication platform, plays an increasingly important role in ensuring communication for ground users due to its rapid deployment and strong line-of-sight link protection capabilities. Dynamically adjusting the UAV's trajectory can more effectively improve channel quality and enhance system robustness. However, in SPC, due to the short transmission latency and small data volume, traditional information theory methods struggle to accurately evaluate system performance, especially in scenarios with eavesdroppers, where system security is more vulnerable. Improving the average secure throughput of the system while maintaining latency and reliability has become a challenging optimization problem. Existing research typically focuses on UAV deployment or scheduling optimization, lacking a systematic approach to jointly optimize multiple system parameters such as user scheduling, transmit power, transmit block length, and UAV trajectory. Especially in scenarios with secure communication requirements, multi-dimensional collaborative design is needed to ensure secure transmission for all users even under worst-case conditions. Therefore, researching a method for maximizing the average secure throughput of UAV-assisted short-packet communication systems not only has significant theoretical value but also positively promotes the practical deployment of future low-latency secure communication systems. Summary of the Invention

[0003] This invention aims to solve the problems of the prior art. It proposes a method for maximizing the average secure throughput of all users in a UAV-assisted short packet communication system. The technical solution of this invention is as follows:

[0004] A method for maximizing the average secure throughput of all users in a UAV-assisted short packet communication system, comprising the following steps:

[0005] Step 1) Establish a model for maximizing the average secure throughput of all users in a UAV-assisted short packet communication system;

[0006] Step 2) Initialize user scheduling, transmit power, transmit block length, and UAV trajectory average security throughput decision threshold;

[0007] Step 3) Using the given transmit power, transmit block length, and UAV trajectory, solve for the user schedule and update it;

[0008] Step 4) Using the given user schedule, transport block length, and UAV trajectory, calculate the transmit power and update it;

[0009] Step 5) Using the given user schedule, transmit power, and UAV trajectory, calculate the transport block length and update it;

[0010] Step 6) Using the given user scheduling, transmit power, and transmission block length, an improved particle swarm optimization (PSO) algorithm is employed to solve and update the UAV trajectory;

[0011] Step 7) Determine if the average secure throughput update converges. Calculate the updated average secure throughput value. If the absolute value of the difference between the updated average secure throughput and the previous average secure throughput is not greater than the average secure throughput decision threshold, then the average secure throughput is determined to be converged, the maximum average secure throughput value is given, and the method ends. If the absolute value of the difference between the updated average secure throughput and the previous average secure throughput is greater than the average secure throughput decision threshold, then the newly calculated average secure throughput value is saved as the current average secure throughput value, and the process returns to Step 3) to update the system parameters until the average secure throughput meets the convergence condition and the maximum average secure throughput is given.

[0012] Furthermore, in step 1), the average secure throughput maximization model P1 for all users of the UAV-assisted short packet communication system is:

[0013]

[0014] Where N is the total time slot, B s [n] represents the secure throughput B. s [n] = min{R s,i [n],R b [n]}, For the secure throughput of the i-th legitimate user in the n-th time slot, C s,i [n] = log2(1+γ) i [n])-log2(1+γ e [n]) represents the safe rate with infinite block length in time slot n. The throughput from the ground source base station to the drone in the nth time slot, γ s For the signal-to-noise ratio from the base station to the drone, γ i and γ e ε represents the signal-to-noise ratio from the drone to user i and from the drone to the eavesdropper, respectively. s and εr η represents the decoding error probability of the drone and the user. e For the probability of information leakage, Q represents the channel dispersion parameter. -1 (·) is the right-tail function of the standard normal distribution. The inverse function of α i For user scheduling, P s P represents the base station's transmit power. i M is the transmit power of the UAV. s M is the block length sent by the base station. i Where q is the block length transmitted by the drone, N is the drone's location, K is the number of time slots, and K is the number of users. and This represents the maximum transmission power for base stations and drones. and M is the maximum available power budget for base stations and drones within the total time slot N. max S is the maximum transmit block length. max This represents the maximum horizontal distance a drone can fly in a time slot.

[0015] The optimization variables for problem P1 are user scheduling A, UAV and base station transmit power P, block length M, and UAV trajectory Q. The objective function for P1 is... Let C1a and C1b represent the average secure throughput, and let C1c, C1d, C1e, and C1f represent the base station and UAV transmit power constraints, C1g and C1h represent the block length constraints, and C1i and C1j represent the UAV trajectory constraints. The block coordinate descent method is used to transform the P1 problem into four subproblems for solution.

[0016] Furthermore, in step 3), the user scheduling A is optimized by fixing the UAV and base station transmit power P, block length M, and UAV trajectory Q. A slack variable τ[n] is introduced to replace the objective function of problem P1. Since the optimization variable A of user scheduling is binary, the above problem involves integer constraints, which is generally difficult to solve. Therefore, by relaxing the binary variable in constraint C1b of problem P1 to a continuous variable, the following sub-problem (P2) can be obtained:

[0017]

[0018] First, we introduce a slack variable τ[n] to rewrite the objective function of P1 as the objective function of P2. Problem P2 is a convex optimization problem, and its optimal solution can be found using convex optimization tools.

[0019] Furthermore, in step 4), the user scheduling A, block length M, and UAV trajectory Q are fixed to optimize the UAV and base station transmit power P. Sub-problem two (P3) is:

[0020]

[0021] Introducing slack variables and ν i [n], Problem P3 can be equivalent to P4:

[0022]

[0023] in, Because of the non-convex constraints C4e, C4f, C4g, C4h, and C4i, problem P4 remains non-convex. Therefore, to solve P4, we replace these constraints with corresponding convex approximations by applying Taylor expansion to a given point. Thus, at local points... We approximate this as question P5:

[0024]

[0025] in,

[0026]

[0027] Since P5 is a convex, optimal variable, it can be solved efficiently using standard convex optimization tools.

[0028] Furthermore, in step 5), the fixed user scheduling A, UAV and base station transmit power P, and UAV trajectory Q optimization block length M, sub-problem three (P6) is:

[0029]

[0030] P6 is a nonlinear integer programming problem caused by C6b and C6c. The solution is to convert the integer value vector L into a positive continuous vector. Let's relax (P6) into a convex optimization problem (P7):

[0031]

[0032] C7c and C7d in P7 are non-convex constraints that obey the second derivative law. Therefore, using the first-order Taylor approximation, we can give the local point P7. The convex reconstruction at point P8 is:

[0033]

[0034] P8 is a convex optimization problem, therefore it can be solved efficiently. To obtain an integer solution to problem three (P6) (using M...) ★ (This means that) we can simply round down the non-integer solutions of P8, so that in Indicates less than or equal to The largest integer. We note that since the objective function is a monotonically increasing function of the block length and does not violate any constraints, this rounding method provides a lower bound solution to the original integer problem.

[0035] Furthermore, in step 6), the fixed user scheduling A, the UAV and base station transmit power P, and the block length M are solved and updated using an improved particle swarm optimization (PSO) algorithm. Subproblem four (P9) is:

[0036]

[0037] The objective of subproblem four (P9) of UAV trajectory optimization is to find the optimal sequence of discrete trajectory points Q = {q[1],q[2],...,q[n]} to maximize the overall performance fitness function F(Q) of the system while satisfying the kinematic constraints of the UAV. Due to the non-convexity of the objective function and the complexity of the kinematic constraints, an improved particle swarm optimization algorithm is used to solve the problem. The specific steps of the algorithm are as follows:

[0038] Particle initialization: Generate P particles, where the position of each particle represents the trajectory vector of the UAV, defined as: Wherein, the trajectory of the Pth particle in the nth iteration is Q. (p,n) Each trajectory point is a two-dimensional coordinate, and the beginning and end points of the trajectory are fixed at the initial starting point q1 = q. N =q init The initial position of the particle satisfies the maximum velocity V. max Constraints: The intermediate point is randomly generated within the bounding box of the given user and base station locations; the initial velocity is V. (p,0) =Uniform(-αV) max ,αV max ), α∈(0,1), where α is the ratio of the initial velocity to the maximum velocity;

[0039] Fitness function construction: The trajectory of each particle is passed to the target function. Conduct an assessment;

[0040] Among them, B (p,n) Let λ be the system's average secure throughput calculated based on the trajectory of the Pth particle. s The acceleration term is used to limit severe trajectory jitter, serving as a smoothness penalty coefficient.

[0041] Individual optimality and global optimality update: The individual optimal trajectory of each particle is Q. pbest,(p) =argmax n f(Q (p ,n) The global optimal trajectory of the population is Q. gbest =argmax p,n f(Q (p,n) The particle's velocity and position are updated according to the following rules (for intermediate points):

[0042]

[0043] Where, ω n The inertial weights decrease dynamically with iteration; c1 and c2 are learning factors. To conform to a uniformly distributed random factor, the velocity is clipped to the maximum velocity constraint range; the updated trajectory must satisfy the following constraints: the start and end points are fixed at the initial point q1 = q N =q init The distance between any two adjacent points satisfies ||q|| t+1 -q t ||≤V max , If it exceeds the limit, it will automatically project back to within the limit; for the closed-loop segment q N-1 →q N Forward Repair q N-2 ,...,q2 ensures continuity.

[0044] When the number of iterations reaches its maximum value n = n max If the optimal fitness fails to improve over several consecutive generations, the process terminates prematurely, and the output trajectory Q is determined. * =Q gbest .

[0045] Furthermore, step 7) specifically involves determining the convergence of the average secure throughput update and calculating the updated average secure throughput value B. s (n) Compare |B s (n) -B s (n-1) |The magnitude of the throughput decision threshold ε, if |B s (n) -B s (n-1) If |B| is not greater than ε, then the average secure throughput is considered converged, the maximum average secure throughput value is given, and the method ends; if |B| is less than ε, then the average secure throughput converges. s (n) -B s (n-1)If | is greater than ε, then the newly calculated average secure throughput value is saved as the current average secure throughput value, and the process proceeds to step 3) to update the system parameters until the average secure throughput meets the convergence condition and the maximum average secure throughput is given.

[0046] The advantages and beneficial effects of this invention are as follows:

[0047] This invention maximizes the average secure throughput of a UAV-assisted short packet communication system under constraints of UAV transmit power, user scheduling, transmission block length, and UAV trajectory. Its innovation lies in jointly optimizing multi-dimensional resources to ensure system security performance under eavesdropping conditions, addressing the characteristics of short packet communication and overcoming the shortcomings of existing research on joint optimization of short packet secure communication. This invention employs a block coordinate descent framework, decomposing the original non-convex problem into four sub-problems: user scheduling, power allocation, block length selection, and UAV trajectory optimization. The trajectory optimization sub-problem is solved using an improved particle swarm optimization algorithm. The method provided by this invention helps improve the security robustness of UAV-assisted short packet communication and offers a new optimization strategy for the design of latency-sensitive, high-security integrated air-ground networks. Attached Figure Description

[0048] Figure 1 This invention provides a preferred embodiment of a UAV-assisted short packet communication system model;

[0049] Figure 2 These are convergence graphs of the present invention under different optimization conditions;

[0050] Figure 3 This is a graph showing the variation of the average security throughput of the present invention with the maximum transmission block length under different parameters.

[0051] Figure 4 This is a graph showing the variation of the average security throughput of the present invention with the maximum power budget under different parameters.

[0052] Figure 5 This is a flowchart illustrating the present invention. Detailed Implementation

[0053] The technical solutions of the embodiments of the present invention will be clearly and thoroughly described below with reference to the accompanying drawings. The described embodiments are merely some embodiments of the present invention.

[0054] The technical solution of the present invention to solve the above-mentioned technical problems is:

[0055] This invention Figure 1 A method for maximizing the average secure throughput of all users in a UAV-assisted short packet communication system is disclosed, comprising the following steps:

[0056] Step 1: Establish a model to maximize the average secure throughput of all users in a UAV-assisted short packet communication system;

[0057] Step 2: Initialize user scheduling, transmit power, transmission block length, and the decision threshold for average secure throughput of UAV trajectory;

[0058] Step 3: Using the given transmit power, transmit block length, and UAV trajectory, solve for the user schedule and update it;

[0059] Step 4: Using the given user schedule, transmission block length, and UAV trajectory, calculate the transmission power and update it;

[0060] Step 5: Using the given user scheduling, transmit power, and UAV trajectory, calculate the transport block length and update it;

[0061] Step 6: Using the given user scheduling, transmit power, and transmission block length, an improved particle swarm optimization (PSO) algorithm is employed to solve and update the UAV trajectory;

[0062] Step 7: Determine if the average secure throughput update converges. Calculate the updated average secure throughput value. If the absolute value of the difference between the updated average secure throughput and the previous average secure throughput is not greater than the average secure throughput decision threshold, then the average secure throughput is considered to have converged, the maximum average secure throughput value is given, and the method ends. If the absolute value of the difference between the updated average secure throughput and the previous average secure throughput is greater than the average secure throughput decision threshold, then the newly calculated average secure throughput value is saved as the current average secure throughput value, and the process returns to step 3) to update the system parameters until the average secure throughput meets the convergence condition and the maximum average secure throughput is given.

[0063] Furthermore, in the first step, the model P1 for maximizing the average secure throughput of all users in the UAV-assisted short packet communication system is:

[0064]

[0065] Where N is the total time slot, B s [n] represents the secure throughput B. s [n] = min{R s,i [n],R b [n]}, For the secure throughput of the i-th legitimate user in the n-th time slot, C s,i [n] = log2(1+γ) i [n])-log2(1+γ e [n]) represents the safe rate with infinite block length in time slot n. The throughput from the ground source base station to the drone in the nth time slot, γ s For the signal-to-noise ratio from the base station to the drone, γ i and γ e ε represents the signal-to-noise ratio from the drone to user i and from the drone to the eavesdropper, respectively. s and ε r η represents the decoding error probability of the drone and the user. e For the probability of information leakage, Q represents the channel dispersion parameter. -1 (·) is the right-tail function of the standard normal distribution. The inverse function of α i For user scheduling, P s P represents the base station's transmit power. i M is the transmit power of the UAV. s M is the block length sent by the base station. i Where q is the block length transmitted by the drone, N is the drone's location, K is the number of time slots, and K is the number of users. and This represents the maximum transmission power for base stations and drones. and M is the maximum available power budget for base stations and drones within the total time slot N. max S is the maximum transmit block length. max This represents the maximum horizontal distance a drone can fly in a time slot.

[0066] The optimization variables for problem P1 are user scheduling A, UAV and base station transmit power P, block length M, and UAV trajectory Q. The objective function for P1 is... Let C1a and C1b represent the average secure throughput, and let C1c, C1d, C1e, and C1f represent the base station and UAV transmit power constraints, C1g and C1h represent the block length constraints, and C1i and C1j represent the UAV trajectory constraints. The block coordinate descent method is used to transform the P1 problem into four subproblems for solution.

[0067] Furthermore, in the third step, the user scheduling A is optimized by fixing the transmit power P of the UAV and the base station, the block length M, and the UAV trajectory Q. A slack variable τ[n] is introduced to replace the objective function of problem P1. Since the optimization variable A of user scheduling is binary, the above problem involves integer constraints, which is generally difficult to solve. Therefore, by relaxing the binary variable in constraint C1b of problem P1 to a continuous variable, the following sub-problem (P2) can be obtained:

[0068]

[0069] First, we introduce a slack variable τ[n] to rewrite the objective function of P1 as the objective function of P2. Problem P2 is a convex optimization problem, and its optimal solution can be found using convex optimization tools.

[0070] Furthermore, in the fourth step, the user scheduling A, block length M, and UAV trajectory Q are fixed to optimize the UAV and base station transmit power P. Subproblem two (P3) is:

[0071]

[0072] Introducing slack variables and ν i [n], Problem P3 can be equivalent to P4:

[0073]

[0074] in, Because of the non-convex constraints C4e, C4f, C4g, C4h, and C4i, problem P4 remains non-convex. Therefore, to solve P4, we replace these constraints with corresponding convex approximations by applying Taylor expansion to a given point. Thus, at local points... We approximate this as question P5:

[0075]

[0076] in,

[0077]

[0078] Since P5 is a convex, optimal variable, it can be solved efficiently using standard convex optimization tools.

[0079] Furthermore, in step five, the optimization block length M for fixed user scheduling A, UAV and base station transmit power P, and UAV trajectory Q is: Subproblem three (P6) is:

[0080]

[0081] P6 is a nonlinear integer programming problem caused by C6b and C6c. The solution is to convert the integer value vector L into a positive continuous vector. Let's relax (P6) into a convex optimization problem (P7):

[0082]

[0083] C7c and C7d in P7 are non-convex constraints that obey the second derivative law. Therefore, using the first-order Taylor approximation, we can give the local point P7. The convex reconstruction at point P8 is:

[0084]

[0085] P8 is a convex optimization problem, therefore it can be solved efficiently. To obtain an integer solution to problem three (P6) (using M...) ★ (This means that) we can simply round down the non-integer solutions of P8, so that in Indicates less than or equal to The largest integer. We note that since the objective function is a monotonically increasing function of the block length and does not violate any constraints, this rounding method provides a lower bound solution to the original integer problem.

[0086] Furthermore, in step six, the user scheduling A, the UAV and base station transmit power P, and the block length M are fixed to optimize the UAV trajectory Q of block length M. Sub-problem four (P9) is:

[0087]

[0088] The objective of subproblem four (P9) of UAV trajectory optimization is to find the optimal sequence of discrete trajectory points Q = {q[1],q[2],...,q[n]} to maximize the overall performance fitness function F(Q) of the system while satisfying the kinematic constraints of the UAV. Due to the non-convexity of the objective function and the complexity of the kinematic constraints, an improved particle swarm optimization algorithm is used to solve the problem. The specific steps of the algorithm are as follows:

[0089] Particle initialization: Generate P particles, where the position of each particle represents the trajectory vector of the UAV, defined as: Wherein, the trajectory of the Pth particle in the nth iteration is Q. (p,n) Each trajectory point is a two-dimensional coordinate, and the beginning and end points of the trajectory are fixed at the initial starting point q1 = q. N =q init The initial position of the particle satisfies the maximum velocity V. max Constraints: The intermediate point is randomly generated within the bounding box of the given user and base station locations; the initial velocity is V. (p,0) =Uniform(-αV) max ,αV max ), α∈(0,1), where α is the ratio of the initial velocity to the maximum velocity;

[0090] Fitness function construction: The trajectory of each particle is passed to the target function. Conduct an assessment;

[0091] Among them, B (p,n) Let λ be the system's average secure throughput calculated based on the trajectory of the Pth particle. sThe acceleration term is used to limit severe trajectory jitter, serving as a smoothness penalty coefficient.

[0092] Individual optimality and global optimality update: The individual optimal trajectory of each particle is Q. pbest,(p) =argmax n f(Q (p ,n) The global optimal trajectory of the population is Q. gbest =argmax p,n f(Q (p,n) The particle's velocity and position are updated according to the following rules (for intermediate points):

[0093]

[0094] Where, ω n The inertial weights decrease dynamically with iteration; c1 and c2 are learning factors. To conform to a uniformly distributed random factor, the velocity is clipped to the maximum velocity constraint range; the updated trajectory must satisfy the following constraints: the start and end points are fixed at the initial point q1 = q N =q init The distance between any two adjacent points satisfies ||q|| t+1 -q t ||≤V max , If it exceeds the limit, it will automatically project back to within the limit; for the closed-loop segment q N-1 →q N Forward Repair q N-2 ,...,q2 ensures continuity.

[0095] When the number of iterations reaches its maximum value n = n max If the optimal fitness fails to improve over several consecutive generations, the process terminates prematurely, and the output trajectory Q is determined. * =Q gbest .

[0096] Furthermore, in step seven, specifically the determination of convergence of the average secure throughput update, the updated average secure throughput value B is calculated. s (n) Compare |B s (n) -B s (n-1) |The magnitude of the throughput decision threshold ε, if |B s (n) -B s (n-1) If |B| is not greater than ε, then the average secure throughput is considered converged, the maximum average secure throughput value is given, and the method ends; if |B| is less than ε, then the average secure throughput converges. s (n) -B s(n -1) If | is greater than ε, then the newly calculated average secure throughput value is saved as the current average secure throughput value, and the process proceeds to step 3) to update the system parameters until the average secure throughput meets the convergence condition and the maximum average secure throughput is given.

[0097] This invention maximizes the average secure throughput of a UAV-assisted short packet communication system under constraints of UAV transmit power, user scheduling, transmission block length, and UAV trajectory. Its innovation lies in jointly optimizing multi-dimensional resources to ensure system security performance under eavesdropping conditions, addressing the characteristics of short packet communication and overcoming the shortcomings of existing research on joint optimization of short packet secure communication. This invention employs a block coordinate descent framework, decomposing the original non-convex problem into four sub-problems: user scheduling, power allocation, block length selection, and UAV trajectory optimization. The trajectory optimization sub-problem is solved using an improved particle swarm optimization algorithm. The method provided by this invention helps improve the security robustness of UAV-assisted short packet communication and offers a new optimization strategy for the design of latency-sensitive, high-security integrated air-ground networks.

[0098] This invention studies a UAV-assisted short packet communication system. In this system, the base station is located at coordinates (-1000, 0) meters, and the positions of the six users are (150.077, 290.82), (-229.56, -205.29), (400, 200), (401, -116), (-204.92, 204.92), and (156.25, -252.821), respectively. The eavesdropper's position is (0, 0). The initial trajectory of the UAV is deployed as a circular trajectory with a fixed radius centered at the geometric center of all user positions. The channel power gain at a reference distance d0 = 1m is β0 = -60dB. Noise power σ 2 = -110dBm, time slot N is 100, decoding error probability ε s and ε r The probability of information leakage is η, which is 10⁻⁵. e The maximum transmission power of base stations and drones is 10⁻³. and The maximum available power budget for base stations and drones within total time slot N is 20dBm. and For 1KW, the maximum transmit block length is M. max The maximum horizontal distance S that a drone can fly in one time slot is 400. max It is 30 m / s.

[0099] In this embodiment, Figure 1This invention provides a preferred embodiment of a UAV-assisted short packet communication system model, employing TDMA to provide services to short packet communication users. The system model consists of a ground source base station, a relay UAV, multiple legitimate users, and an eavesdropper. All communication nodes are equipped with only one antenna. The ground source base station periodically transmits short data packets containing confidential information. These data packets are relayed to ground users via a trusted mobile UAV, which securely transmits the confidential information to multiple users on the ground by decoding and forwarding the data packets. Simultaneously, an eavesdropper attempts to eavesdrop on the ongoing confidential transmission. Figure 2 The convergence of the algorithm proposed in this chapter under different optimization conditions is presented. Figure 3 The variation of average secure throughput with maximum transport block length under different parameters is presented. Figure 2 As can be seen, the average secure throughput of the system gradually increases with the number of iterations. The average secure throughput under joint optimization of resource allocation and UAV trajectory is significantly higher than that under fixed circular trajectory optimization and fixed resource allocation optimization of UAV trajectory. However, when the maximum transmission block length decreases from 400 to 150, the average secure throughput of all three optimization cases decreases significantly. Figure 3 As can be seen from the data, under different parameters, the average secure throughput increases with the increase of the maximum transport block length, and decreases with the increase of the decoding error probability ε. s and ε r As the probability of information leakage η increases, the average secure throughput will increase, and when the probability of information leakage η increases... e As the length increases, the average secure throughput also gradually increases, and gradually approaches the system's performance at the point of wireless block length. From Figure 4 As can be seen from the data, under different parameters, the average secure throughput increases with the increase of the maximum power budget and with the increase of the maximum transport block length M. max and maximum transmit power P max With the increase of [data / capacity], the average secure throughput will increase.

Claims

1.A method for maximizing the average secrecy throughput of all users in a UAV-assisted short packet communication system, characterized in that, The method comprises the following steps: Step 1), establishing a model for maximizing the average secrecy throughput of all users of a UAV-assisted short packet communication system; Step 2), initializing user scheduling, transmission power, transmission block length, UAV trajectory and average secrecy throughput decision threshold; Step 3), solving user scheduling by using given transmission power, transmission block length and UAV trajectory, and updating the user scheduling; Step 4), solving transmission power by using given user scheduling, transmission block length and UAV trajectory, and updating the transmission power; Step 5), solving transmission block length by using given user scheduling, transmission power and UAV trajectory, and updating the transmission block length; Step 6), solving and updating UAV trajectory by using given user scheduling, transmission power and transmission block length by using an improved particle swarm optimization (PSO) algorithm; Step 7), judging whether the average secrecy throughput is updated to converge, calculating the updated average secrecy throughput value, and if the absolute value of the difference between the updated average secrecy throughput and the previous average secrecy throughput is not greater than the average secrecy throughput decision threshold, judging that the average secrecy throughput converges, giving the maximum average secrecy throughput value, and ending the method; If the absolute value of the difference between the updated average secrecy throughput and the previous average secrecy throughput is greater than the average secrecy throughput decision threshold, saving the newly calculated average secrecy throughput value as the average secrecy throughput value at this time, and updating the system parameters in step 3) until the average secrecy throughput meets the convergence condition, and giving the maximum average secrecy throughput. 2.The method of Claim 1, wherein, The model P1 for maximizing the average secrecy throughput of all users of the UAV-assisted short packet communication system in step 1) is: C1i:q[1]=q[N], C1j: ||q[n + 1] - q[n]|| < S max n = 1,..., N - 1. Where N is the total time slot, B s [n] represents the secure throughput B. s [n] = min{R s,i [n],R b [n]}, For the secure throughput of the i-th legitimate user in the n-th time slot, C s,i [n] = log2(1+γ) i [n])-log2(1+γ e [n]) represents the safe rate with infinite block length in time slot n. The throughput from the ground source base station to the drone in the nth time slot, γ s For the signal-to-noise ratio from the base station to the drone, γ i and γ e ε represents the signal-to-noise ratio from the drone to user i and from the drone to the eavesdropper, respectively. s and ε r η represents the decoding error probability of the drone and the user. e For the probability of information leakage, Q represents the channel dispersion parameter. -1 (·) is the right-tail function of the standard normal distribution. The inverse function of α i For user scheduling, P s P represents the base station's transmit power. i M is the transmit power of the UAV. s M is the block length sent by the base station. i Where q is the block length transmitted by the drone, N is the drone's location, K is the number of time slots, and P is the number of users. s max and P r max P represents the maximum transmit power for base stations and drones. s tot and M is the maximum available power budget for base stations and drones within the total time slot N. max S is the maximum transmit block length. max This represents the maximum horizontal distance a drone can fly in a single time slot. The optimization variables of the problem P1 are the user scheduling A, the UAV and base station transmit power P, the block length M and the UAV trajectory Q, and the objective function of P1 is denotes the average secrecy throughput, the constraints C1a and C1b are user scheduling constraints, which means that at most one user is scheduled to communicate with the UAV in each time slot, the constraints C1c, C1d, C1e and C1f are base station and UAV transmit power constraints, the constraints C1g and C1h are block length constraints, the constraints C1i and C1j are UAV trajectory constraints, and the problem P1 is converted into four sub-problems by using the block coordinate descent method. 3.The method of Claim 2, wherein, In step 3), the user scheduling A is optimized by fixing the UAV and base station transmission power P, the block length M and the UAV trajectory Q, a slack variable τ[n] is introduced, and the objective function of the problem P1 is replaced. Since the optimization variable A of the user scheduling is binary, the above problem involves integer constraints, which is generally difficult to solve. Therefore, the binary variable in the constraint C1b of the problem P1 is relaxed to a continuous variable, and the following subproblem one (P2) can be obtained: First, the slack variable τ[n] is introduced to rewrite the objective function of P1 as the objective function of P2. The problem P2 is a convex optimization problem, and its optimal solution can be solved by using a convex optimization tool. 4.The method of Claim 3, wherein, In step 4), the UAV and base station transmission power P is optimized by fixing the user scheduling A, the block length M and the UAV trajectory Q, and the subproblem two (P3) is: Introducing slack variables and v i [n], problem P3 can be equivalent to P4: where, Since the non-convex constraints C4e, C4f, C4g, C4h, C4i,, the problem P4 is still non-convex; therefore, to solve P4, we replace these constraints by the corresponding convex approximations by applying the Taylor expansion method at given points; thus, at the local point We approximately write the problem P5 as: wherein, Since P5 is convex, the optimization variable, one can use standard convex optimization tools to solve it efficiently. 5.The method of Claim 4, wherein, In step 5), the block length M is optimized by fixing the user scheduling A, the UAV and base station transmission power P and the UAV trajectory Q, and the subproblem three (P6) is: P6 is a nonlinear integer programming problem resulting from C6b, C6c; the method to solve this problem is to relax (P6) into a convex optimization problem P7 by converting the integer valued vector L into a positive continuous vector P7: C7c and C7d in P7 is a non-convex constraint, following the second derivative law; therefore, using the first order Taylor approximation, a convex reformulation of P7 at a local point P8 can be given as: P8 is a convex optimization problem and hence can be solved efficiently; to obtain an integer solution (denoted by M ★ ) to problem three (P6), we can simply round down the non-integer solution of P8 so that where denotes the largest integer less than or equal to ; we note that this rounding method provides a lower bound solution to the original integer problem since the objective function is a monotonic increasing function of block length and no constraints are violated. 6.The method of Claim 5, wherein, In step 6), the UAV trajectory Q is solved and updated by using an improved particle swarm optimization (PSO) algorithm by fixing the user scheduling A, the UAV and base station transmission power P and the block length M, and the subproblem four (P9) is: s.t.C9a:q[1]=q[N], C9b: ||q[n+1] - q[n]|| < S max n = 1,..., N - 1. The objective of the fourth subproblem(P9) of UAV trajectory optimization is to find the optimal discrete sequence of UAV trajectory points Q={q[1],q[2],...,q[n]} to maximize the overall performance fitness function F(Q) of the system while satisfying the kinematic constraints of the UAV; due to the non-convexity of the objective function and the complexity of the kinematic constraints, an improved particle swarm optimization algorithm is used to solve the problem; the specific steps of the algorithm are as follows: Particle initialization: P particles are generated, and the position of each particle represents the trajectory vector of the UAV, defined as: where the trajectory of the Pth particle at the nth iteration is Q (p,n) , each trajectory point is a two-dimensional coordinate, and the first and last points of the trajectory are fixed as the initial starting point q1=q N =q init ; the initial position of the particle satisfies the maximum speed V max limit; the intermediate points are randomly generated from the bounding box of the given user and base station positions; the initial speed is V (p,0) =Uniform(-αV max ,αV max ), α∈(0,1), and α is the proportion of the initial speed to the maximum speed. Fitness function construction: the trajectory of each particle is passed into the objective function Evaluation is performed; where B (p,n) is the system average secrecy throughput calculated based on the Pth particle trajectory, λ s is the smoothness penalty coefficient, and the acceleration term is used to limit the trajectory from shaking dramatically. Individual and global optimum update: the individual optimum trajectory of each particle is Q pbest,(p) = argmax n f(Q (p,n) ), and the global optimum trajectory of the population is Q gbest = argmax p,n f(Q (p,n) ), and the velocity and position of the particle are updated according to the following rules (for the intermediate point): where ω n is the inertia weight, which is dynamically decreased with iterations, c1 and c2 are learning factors, is a random factor subject to uniform distribution, the velocity is clipped to the maximum velocity constraint; the updated trajectory needs to satisfy the following constraints: the start and end points are fixed as the initial point q1 = q N = q init , the distance between any two adjacent points satisfies ‖q t+1 -q t ‖≤ V max , If it exceeds, it is automatically projected back within the limit; the closed loop segment q N-1 → q N is repaired forward to ensure continuity q N-2 ,...,q2 When the number of iterations reaches a maximum value n = n max , or if the optimal fitness does not improve for a number of consecutive generations, the optimization is terminated early and the trajectory Q * = Q gbest is output. 7.The method of Claim 6, wherein, The step 7) is specifically the judgment of the convergence of the average secret throughput update, calculating the updated average secret throughput value B s (n) , comparing |B s (n) -B s (n-1) | with the size of the throughput decision threshold ε, if |B s (n) -B s (n-1) | is not greater than ε, it is judged that the average secret throughput converges, the maximum average secret throughput value is given, and the method ends; if |B s (n) -B s (n -1) | is greater than ε, the newly calculated average secret throughput value is saved as the average secret throughput value at this time, the system parameters are updated in step 3), and the maximum average secret throughput is given until the average secret throughput meets the convergence condition.

Images