Methods for maximizing user experience quality in multi-drone aerial video transmission

Abstract

This invention relates to the field of cellular-connected drone video transmission technology, specifically disclosing a method for maximizing user experience quality in multi-drone aerial video transmission. This method is used in aerial video streaming systems that utilize multiple cellular-connected drones to capture video from different points of interest (PoI) regions and transmit the video to ground base stations (BS) and users, allowing ground users to share the drones' field of view. Based on this aerial video streaming system, this invention constructs an optimization problem. By jointly optimizing transmission scheduling, video playback rate, and drone trajectory design, it maximizes the minimum quality of experience (QoE) for all users, considering uplink interference and the trade-off between video quality and smoothness. Simulation results show that, compared to benchmark solutions, the proposed solution achieves a significant performance improvement and achieves a trade-off between video quality and smoothness.

Description

Technical Field

[0001] This invention relates to the field of cellular network-connected drone video transmission technology, and more particularly to a method for maximizing user experience quality in multi-drone aerial video transmission. Background Technology

[0002] In recent years, mobile video streaming has developed rapidly and covered a wide range of applications, including live television broadcasts or sporting events, interactive games, and video surveillance. The rapid growth of high-quality video services has placed enormous transmission pressure on current cellular network infrastructure, becoming a bottleneck for improving Quality of Experience (QoE). The sixth-generation (6G) mobile communication network is expected to provide wider coverage, seamless connectivity, and lower latency, with the integration of terrestrial and non-terrestrial network architectures becoming a crucial component. On the one hand, due to their high mobility and flexibility, drones can be used as aerial base stations or relay stations to provide flexible video streaming services to ground users. On the other hand, drones equipped with high-definition cameras can be integrated into cellular networks as a new type of aerial user, supporting the booming development of multimedia services such as aerial video surveillance and tracking, and geographic photography. Thanks to the extensive coverage of existing cellular networks and the high capacity of cellular wireless links, cellular-connected drones can achieve remote control beyond line-of-sight and further improve communication transmission performance.

[0003] However, many challenges remain in using cellular-connected drones for aerial video streaming. First, the frequent variations in air-to-ground channel quality due to the high mobility of drones make it difficult to achieve higher speeds and lower latency / jitter during aerial video streaming. Therefore, dynamically adjusting the video rate while maintaining user experience quality is crucial. Furthermore, while line-of-sight (LoS) air-to-ground links present both opportunities and significant challenges for cellular-connected drones, although they allow for higher uplink data rates at associated base stations, they also severely interfere with uplink transmissions at unassociated base stations.

[0004] Current research often uses only single drones for monitoring and transmission, without considering co-channel interference between multiple drones. In reality, transmitting aerial video streams simultaneously from multiple cellular-connected drones while considering co-channel interference is challenging and has not been thoroughly investigated in previous work. Summary of the Invention

[0005] This invention provides a method for maximizing user experience quality in multi-drone aerial video transmission, addressing the technical problem of how to maximize the minimum user experience quality in cellular network-connected multi-drone aerial video transmission.

[0006] To address the above technical problems, this invention provides a method for maximizing user experience quality in multi-UAV aerial video transmission, comprising the following steps:

[0007] S1. Construct an aerial video streaming transmission system and determine the trajectory constraints of the UAV. The aerial video streaming transmission system consists of a U-shaped cellular network UAV and K ground base stations. Each UAV captures video from its own area of ​​interest (PoI) and then transmits it to the corresponding base station. The base station transmits the video to ground users through a high-capacity cellular link.

[0008] S2. Determine the communication link constraints and video stream constraints of the over-the-air video stream transmission system;

[0009] S3. Calculate the cumulative user experience quality;

[0010] S4. With the goal of maximizing the minimum cumulative user experience quality, and using the constraints of the drone trajectory, the communication link, and the video stream as constraints, construct a joint optimization problem for transmission scheduling, video playback rate, and drone trajectory.

[0011] S5. Solve the optimization problem to obtain the optimal solutions for transmission scheduling, video playback rate, and drone trajectory.

[0012] Further, in step S3, the user's cumulative experience quality is calculated as follows:

[0013]

[0014] Where N represents the total number of time slots that discretize the total time T into time slots, and the length of each time slot is... θ and β are fixed parameters determined by the preset application, r u [n] represents the ground user g u The video playback rate at time slot n, User g u Required playback rate; ρ is the weighting factor; This means that for any drone v u , 1≤u≤U.

[0015] Furthermore, the video stream constraint condition is expressed as follows:

[0016]

[0017] Where, x u,k [n] is a defined binary variable representing the drone v. u The transmission scheduling strategy at time slot n, if the drone v u It is decided to direct the signal from time slot n to base station s kSending video, then x u,k [n] = 1, otherwise x u,k [n] = 0; R u,k [n] represents the time slot n for the drone v u and base station s k The achievable rates between them.

[0018] Furthermore, the communication link constraints are expressed as follows:

[0019]

[0020]

[0021]

[0022] The communication link constraint is interpreted as follows: in each time slot, each base station serves at most one drone, and each drone sends video to at most one base station.

[0023] Furthermore, the trajectory constraints of the drone are expressed as follows:

[0024]

[0025]

[0026]

[0027] Where, q u [n] represents the drone v u At position n in time slot, V max d represents the maximum speed of the drone, δ represents the length of each time slot after time discretization, and d represents the maximum speed of the drone. min This indicates the safe distance between drones to ensure collision avoidance. u Indicates drone v u The radius of the monitored PoI region, a u Indicates drone v u The center of the monitored PoI area.

[0028] Furthermore, R u,k [n] is calculated as follows:

[0029] R u,k [n] = B log2(1+γ) u,k [n]),

[0030] B represents bandwidth, γ u,k [n] represents the drone v u Select base station s in time slot n k Send video at base station s kThe ratio of received signal to interference plus noise at the location;

[0031] γ u,k [n] is calculated as follows:

[0032]

[0033] Among them, h u,k [n] represents the time slot n for the drone v u and base station s k The channel gain between P u Indicates drone v u uplink transmit power, σ 2 Indicates noise power;

[0034] h u,k [n] is calculated as follows:

[0035]

[0036] Where β0 is the channel power gain when the spacing is 1m. This indicates that the drone v is in time slot n. u and base station s k The distance between them, α≥2 is the channel loss factor, H represents the distance between them, and w represents the distance between them. k Indicates base station s k The horizontal coordinates.

[0037] Furthermore, the optimization problem is described as follows:

[0038]

[0039]

[0040]

[0041]

[0042]

[0043]

[0044]

[0045]

[0046] Where μ is defined as the user's minimum cumulative experience quality.

[0047] Furthermore, step S5 specifically includes the following steps:

[0048] S51. Add an additional penalty term to the objective function, transforming the optimization problem into optimization problem P2:

[0049]

[0050]

[0051]

[0052] Where λ is a penalty parameter that is iteratively increased to ensure that the balance constraint is satisfied, and b u,k [n] is the introduced auxiliary penalty variable, defined as x = {x u,k [n]}, b={b u,k [n]}, b satisfies:

[0053] D = UKN, (2x - 1) T (2b-1)=D,0≤x≤1,||2b-1|| 2 ≤D,

[0054] Here, 0 and 1 represent D-dimensional vectors that are all zero and all one, respectively;

[0055] S52, Definition and After iteration l, x u,k [n],r u [n],q u [n] and b u,k The solution obtained for [n] is updated in the (l+1)th iteration by the following operations;

[0056] S53, Given and The optimization problem P2 is transformed into an optimization problem P3. Solving this optimization problem P3 yields the optimal solution in the (l+1)th iteration.

[0057] S54, Based on the optimization obtained in step S53 And given The optimization problem P2 is transformed into the optimization problem P4. Then, by introducing slack variables, the optimization problem P4 is transformed into the optimization problem P5. Finally, by considering the local points... By using an approximate substitution with a lower bound, optimization problem P5 is transformed into optimization problem P6. Solving optimization problem P6 yields the optimal solution in the (l+1)th iteration.

[0058] S56, based on optimization The optimization problem P2 is transformed into an optimization problem P7. Solving this optimization problem P7 yields the optimal solution in the (l+1)th iteration.

[0059] S57. Perform iterative optimization using the same steps as S52-S56 until the objective value of optimization problem P1 converges, and output {x} at this point. u,k [n],r u [n],q u The optimal solution for [n]}.

[0060] Furthermore, in step S56, the optimal solution The calculation is as follows:

[0061]

[0062] This invention provides a method for maximizing the quality of user experience in multi-UAV aerial video transmission, applicable to aerial video streaming systems. This system uses multiple cellular-connected UAVs to capture video from different Points of Interest (PoI) regions and transmits the video to ground base stations (BS) and users, allowing ground users to share the UAVs' field of view. Based on this aerial video streaming system, this invention constructs an optimization problem. By jointly optimizing transmission scheduling, video playback rate, and UAV trajectory design, it maximizes the minimum quality of experience (QoE) for all users, considering uplink interference and the trade-off between video quality and smoothness. This optimization problem is a difficult-to-solve mixed-integer non-convex optimization problem. This invention proposes a block coordinate descent method with overlapping variable partitioning to address this problem, increasing the optimization space. To handle the binary constraints in the problem, a precise penalty method with balance constraints is also proposed to ensure the accuracy of the penalty function. Furthermore, this invention employs a successive convex approximation method to handle the non-convex constraints of the optimization problem. Simulation results show that, compared with the baseline scheme, the proposed scheme achieves a significant performance improvement and achieves a trade-off between video quality and smoothness. Attached Figure Description

[0063] Figure 1 This is a flowchart of a method for maximizing user experience quality in multi-UAV aerial video transmission provided in an embodiment of the present invention;

[0064] Figure 2 This is a schematic diagram of drone trajectory optimization (a) and playback rate allocation (b) under different ρ values ​​provided in the embodiments of the present invention;

[0065] Figure 3 This is a schematic diagram of the transmission scheduling of UAV v1(a) and UAV v2(b) provided in an embodiment of the present invention;

[0066] Figure 4This is a schematic diagram of the maximum-minimum QoE under different P values ​​(a) and T values ​​(b) provided in the embodiments of the present invention. Detailed Implementation

[0067] The embodiments of the present invention are described in detail below with reference to the accompanying drawings. The embodiments are given for illustrative purposes only and should not be construed as limiting the present invention. The accompanying drawings are for reference and illustration only and do not constitute a limitation on the scope of patent protection of the present invention, because many changes can be made to the present invention without departing from the spirit and scope of the present invention.

[0068] like Figure 1 As shown, the method for maximizing user experience quality in multi-UAV aerial video transmission provided by this embodiment of the invention includes the following steps:

[0069] S1. Construct an aerial video streaming system and determine the trajectory constraints of the UAV;

[0070] S2. Determine the communication link constraints and video stream constraints of the over-the-air video stream transmission system;

[0071] S3. Calculate the cumulative user experience quality;

[0072] S4. With the goal of maximizing the minimum cumulative user experience quality, and with constraints on drone trajectory, communication link, and video stream, construct a joint optimization problem for transmission scheduling, video playback rate, and drone trajectory.

[0073] S5. Solve the optimization problem to obtain the optimal solutions for transmission scheduling, video playback rate, and drone trajectory.

[0074] This example considers an aerial video streaming system consisting of U>1 cellular-connected drones and K ground base stations. Each drone captures video from its respective Area of ​​Interest (PoI) and transmits it to the corresponding base station. The base station then transmits the video to ground users via a high-capacity cellular link. The set of base stations consists of... The horizontal coordinate is represented as 1≤k≤K, the set of drones consists of This is represented as follows. Since horizontal flight is more energy-efficient, this example considers a drone flying at a fixed altitude H. This example assumes there are U PoI regions, defined by the set {ζ1,ζ2,…,ζ}. U} indicates that each drone v u Assigned to use from PoI region ζ u Capture video, where ζ u Therefore A circular region centered at o has a radius of o. u>0,1≤u≤U. For simplicity, this example discretizes the total time T into segments with sufficiently small time slots. N time slots. Therefore, the drone v u The trajectory can be approximated as: q u [n] represents the drone v u The position at time slot n.

[0075] To periodically monitor the PoI region, this example assumes that each drone v u In their respective PoI regions ζ u Internal flight, i.e., ||q u [n]-a u ‖≤o u , n, and returns to its initial position, q, at the end of time T. u [1] = q u [N], Furthermore, due to the mechanical limitations of drones, this example has the following constraint: ||q|| u [n+1]-q u [n]‖≤V max δ, n, where V max This represents the maximum speed of the drone. Additionally, this example includes the constraint ||q. i [n]-q j [n]||≥d min , 1≤i≠j≤U, where d min This indicates the safe distance between drones to ensure collision avoidance. (Drone v) u and base station s k The distance between them in time slot n can be expressed as

[0076] This example assumes that the communication link between the drone and the base station is a line-of-sight link, i.e., the drone v u and base station s k The channel gain between them is expressed at time slot n as: Where β0 is the channel power gain at a spacing of 1m, and α≥2 is the channel loss factor. Define a binary variable x. u,k [n]∈{0,1} represents the transmission scheduling strategy for the UAV. Specifically, if UAV v u It is decided to direct the signal from time slot n to base station s k Sending video, then x u,k [n] = 1; otherwise x u,k [n] = 0. Therefore, x u,k[n] specifies not only the drone's transmission scheduling but also the association strategy between the drone and the base station in different time slots. In each time slot, this example assumes that each base station serves at most one drone, and each drone sends video to at most one base station. Therefore, this example has... In addition, P is defined in this example. u Indicates drone v u The uplink transmit power corresponds to the drone's maximum transmit power. Therefore, if the drone v u Select base station s in time slot n k If video is sent, then base station s k The received signal-to-interference-plus-noise ratio (SINR) at the location can be obtained from... The calculation yields σ. 2 This represents the noise power. In the expression above, for any time slot n, except for the UAV v... u All other drones transmit signals in v mode. u Transmitted to base station s k Co-channel interference can be caused by Therefore, the drone v at time slot n. u and base station s k The reachable rate between , can be expressed as R u,k [n] = Blog2(1+γ) u,k [n]), where B represents bandwidth.

[0077] PoI region ζ u The surveillance video was generated by a drone. u The video is captured and then transmitted via cellular network to a base station, which then provides video services to ground users. For example, users wearing virtual reality (VR) headsets can share the drone's view within the Point of Interest (PoI) area. Assuming the cellular link capacity from the base station to the ground user is large enough, the bottleneck for video streaming lies in the air-to-ground communication link. u [n] represents the ground user g u At time slot n, the video playback rate is achieved by the user via drone v. u The transmitted video can be viewed in real time within the PoI area. u The situation within. To avoid video playback interruption, this example introduces an information causal relationship constraint, that is... This example assumes a time slot is allocated for video processing and preparation, such as decoding or playback preparation. In this paper, the example uses user experience quality as a performance metric; specifically, it uses a logarithmic function with diminishing returns. To evaluate user g u The video playback quality, where θ and β are fixed parameters determined by a specific application. User gu The required playback rate is related to the screen size. Furthermore, this example uses the variance between the current playback rate and the average playback rate to evaluate the fluctuation of video playback at time slot n, i.e., video smoothness. In short, user g u The cumulative quality of experience is given by the following formula. ρ is used as a weighting factor to balance the trade-off between video quality and smoothness.

[0078] In this paper, we jointly optimize transmission scheduling, playback rate allocation, and drone trajectory to maximize the minimum cumulative experience quality for all users, thereby ensuring fairness among all users. The optimization problem can be described as follows:

[0079]

[0080]

[0081]

[0082]

[0083]

[0084]

[0085]

[0086]

[0087] Where μ is defined as the user's minimum cumulative experience quality. Equation (2) is the video stream constraint, Equations (3)-(5) are the communication link constraints, and Equations (6)-(8) are the UAV trajectory constraints.

[0088] It is worth noting that, since constraint (5) contains a binary constraint and constraints (2) and (8) are both non-convex constraints, problem P1 is a mixed integer non-convex optimization problem, which is difficult to solve directly. In the following text, this example proposes an efficient algorithm for finding suboptimal solutions using the nondeterministic block coordinate descent method, the exact penalty method, and the successive convex approximation technique.

[0089] To handle the binary constraints in (5), this example employs an exact penalty method with balance constraints to ensure the accuracy of the penalty function. Specifically, this example first transforms the binary constraints in (5) into some equivalent constraints and applies an additional penalty to make the solution approximate 0 and 1. The main idea comes from the following lemma, where the Euclidean inner product between x and b is obtained by x T b or<x,b> express.

[0090] Lemma 1: In this example, we define a vector. Assume 0 ≤ x ≤ 1, ||2b-1|| 2 ≤D, and (2x-1) T (2b-1) = D. Here, 0 and 1 represent D-dimensional vectors of all zeros and all ones. Therefore, in this example, x∈{0,1} D b∈{0,1} D And x = b.

[0091] Based on Lemma 1, a new auxiliary penalty variable is introduced. Binary constraint condition x∈{0,1} D This is equivalent to the following three constraints: 0 ≤ x ≤ 1, ||2b-1|| 2 The constraint <2x-1, 2b-1> = D holds. Furthermore, it can be verified that <2x-1, 2b-1> ≤ D holds for any feasible x and b. Assume D = UKN, x = {x...} u,k [n]}, b={b u,k [n]}. In this example, by adding an extra penalty term to the objective function, the balance constraint is made easier to satisfy. Therefore, problem (P1) can be transformed into:

[0092]

[0093]

[0094]

[0095] Here, λ is a penalty parameter that iteratively increases to ensure the balance constraint is satisfied. It can be proven that when λ is sufficiently large, problem P1 is equivalent to problem P2. Unlike traditional block coordinate descent algorithms, this example designs a nondeterministic block coordinate descent algorithm to solve problem P2, where variables are defined... and The optimization process is performed sequentially until convergence. It's worth noting that the playback rate variable {r}... u [n]} is also included in and This avoids strong locality in the optimization solution. This example defines... and As the solution obtained after l iterations, the updates in the next l+1 iterations can be obtained by the following operations.

[0096] This example considers the given... and Optimize transmission scheduling in the case of {x u,k [n]} and playback rate allocation {r uThe subproblem of [n] can be represented as:

[0097]

[0098] It can be proven that problem P3 is a standard convex optimization problem, and its optimal solution can be obtained by using optimization tools such as CVX using the interior point method.

[0099] In a given transmission schedule and penalty variables In this case, problem P2 can be written as:

[0100]

[0101] Due to the non-convex constraints in (2) and (8), problem P4 is neither a non-concave maximization problem nor a quasi-concave maximization problem. To handle these constraints, this example employs a successive convex approximation method, where non-convex constraints are replaced by approximate convex constraints obtained in the previous iteration, and choosing a reasonable approximation value can produce a solution that is equally applicable to the original problem. In particular, R in equation (2) u,k [n] can be equivalently represented as in Next, this example introduces a slack variable {y} j,k [n]} Restate problem P4 as:

[0102]

[0103]

[0104]

[0105]

[0106] This example verifies that the equation in (12) always holds true in the optimal solution to problem P5, because y j,k [n] can be increased incrementally without reducing the objective value of problem P5, and all other constraints are still satisfied.

[0107] To solve the non-concave terms in equation (11) This example uses a given local point. Applying a first-order Taylor expansion to the logarithmic function yields the following... The inequality is, in other words:

[0108]

[0109] in and in Similarly, through local points and ‖q j [n]-w k || 2 and‖q i [n]-q j [n]‖ 2 Applying a first-order Taylor expansion, this example yields... as well as By a given local point Using the lower bound derived above for approximate replacement, problem P5 can be approximated as:

[0110]

[0111]

[0112]

[0113]

[0114] Problem (P6) is a standard convex optimization problem, which can be solved efficiently using standard convex optimization solvers such as CVX.

[0115] For a given user transport schedule Update penalty variable {b u,k The problem of [n] can be represented as:

[0116]

[0117] In problem P7, if the following conditions are met... Then any feasible solution will be the optimal solution; otherwise, the problem will obtain the optimal solution at the constraint boundary (that is, when the equation in equation (17) holds), which can be expressed as:

[0118]

[0119] Based on the above process, step S5 specifically includes the following steps:

[0120] S51. Add an additional penalty term to the objective function to transform the optimization problem into optimization problem P2;

[0121] S52, Definition and After iteration l, x u,k [n],r u [n],q u [n] and b u,kThe solution obtained for [n] is updated in the (l+1)th iteration by the following operations;

[0122] S53, Given and The optimization problem P2 is transformed into an optimization problem P3. Solving this optimization problem P3 yields the optimal solution in the (l+1)th iteration.

[0123] S54, Based on the optimization obtained in step S53 And given The optimization problem P2 is transformed into the optimization problem P4. Then, by introducing slack variables, the optimization problem P4 is transformed into the optimization problem P5. Finally, by considering the local points... By using an approximate substitution with a lower bound, optimization problem P5 is transformed into optimization problem P6. Solving optimization problem P6 yields the optimal solution in the (l+1)th iteration.

[0124] S56, based on optimization The optimization problem P2 is transformed into an optimization problem P7. Solving this optimization problem P7 yields the optimal solution in the (l+1)th iteration.

[0125] S57. Perform iterative optimization using the same steps as S52-S56 until the objective value of optimization problem P1 converges, and output {x} at this point. u,k [n],r u [n],q u The optimal solution for [n]}.

[0126] This example uses a nondeterministic block descent method to iteratively solve three subproblems until the objective value cannot be further improved. Algorithm 1 describes the entire iterative process for solving problem P1, where the penalty parameter λ is continuously updated with the number of iterations until it reaches λ. max , where the update multiplier for each iteration is denoted by c. It is worth noting that the initial λ is set relatively small here, thus providing more flexible user transport scheduling; when λ is sufficiently large, it makes {x} u,k The solutions to [n]} are close to 0 or 1. In Algorithm 1, this example transforms the mixed-integer non-convex optimization problem P1 into a series of convex optimization problems, which are solved using the interior-point method, where the computational complexity depends on the number of optimization variables. Since problem P7 obtains a closed-form solution with negligible complexity, the total computational complexity of Algorithm 1 is reduced by... Given, where L is a denoted , The order of magnitude is the number of iterations, and ∈ represents the precision of the solution.

[0127]

[0128]

[0129] The following simulation experiments demonstrate the effectiveness of the proposed algorithm. This example considers an aerial video streaming system where U = 2 UAVs perform video surveillance tasks in their respective Point of Interest (PoI) areas, and K = 6 base stations are uniformly and randomly distributed over a 1.5 × 1 km area. 2 Within a rectangular area on the ground. This example is based on... Figure 2 The base station locations shown in (a) yielded the following results. It is assumed that all PoI regions have a radius of o. u A circular area of ​​300m is defined, with the flight altitude H set to 100m. Channel-related parameters are set to σ. 2 = -110dBm, β0 = -60dB, α = 2, P = 0.1W and B = 1MHz, and the parameters related to user experience quality are set to θ = 0.8 and β = 400. The parameters related to the algorithm are set to the initial value of λ. 0 =0.01, and ∈=10 -3 Other parameters are set to δ = 1s, V max =50m / s.

[0130] Figure 2 The example demonstrates the optimized drone trajectory and playback rate allocation results at T=120s, where the weighting factor ρ is used to balance the trade-off between video quality and video smoothness. This example observes that when ρ is small, drones v1 and v2 tend to maintain a greater distance from each other, thereby improving the user experience by avoiding co-channel interference, especially when they are associated with two base stations that are close to each other. For example, when drones v1 and v2 are associated with base stations s2 and s1 respectively, as shown... Figure 2 As shown in (a). In this case, the reduction in co-channel interference avoidance comes at the cost of sacrificing the beneficial communication link quality of associated base stations. The optimized trajectory design in this example can balance the trade-off between these two aspects. When ρ is large, minimizing video quality variations becomes more important, where the corresponding optimized trajectory should lead to a more stable playback rate allocation. This is also... Figure 2 This was verified in (b), where the larger ρ is, the more stable the playback rate distribution.

[0131] Figure 3 The transmission scheduling and correlation of different UAVs during flight are shown. This example observes that the value of the transmission scheduling indicator variable is either 0 or 1, which can be efficiently obtained using the proposed Algorithm 1 with precise penalty method.

[0132] To demonstrate the superiority of the algorithm in this example, three benchmark schemes are considered: the maximum throughput benchmark scheme, the circular trajectory benchmark scheme, and the static UAV benchmark scheme. In the maximum throughput benchmark scheme, the uplink transmission throughput from the UAV to the base station is maximized without considering co-channel interference. In the circular trajectory benchmark scheme, all UAVs fly along the boundary of their corresponding Point of Interest (PoI). In the static UAV benchmark scheme, each UAV hovers over the center of its corresponding PoI throughout the entire time period. Figure 4 (a) shows the relationship between the optimized minimum user experience quality and the transmit power P when T = 200s. In this example, we observe that the proposed Algorithm 1 outperforms the three benchmark schemes, and the performance is more significant when P is small. This is because as the transmit power of each drone increases, the corresponding interference signal to other drones also increases. Figure 4 (b) shows the relationship between the optimized minimum user experience quality and time T when P = 0.1W. This example shows that the minimum user experience quality increases over time, as expected. The performance improvement is more significant when T is large, and this improvement stems from the flexible interference-aware trajectory design and transmission scheduling in Algorithm 1.

[0133] In summary, the method for maximizing user experience quality in multi-UAV aerial video transmission provided by this invention is applied to an aerial video streaming system. This system uses multiple cellular-connected UAVs to capture video from different Points of Interest (PoI) regions and transmits the video to a ground base station (BS) and users, allowing ground users to share the UAV's field of view. This invention constructs an optimization problem based on this aerial video streaming system. By jointly optimizing transmission scheduling, video playback rate, and UAV trajectory design, it maximizes the minimum quality of experience (QoE) for all users, considering uplink interference in video transmission and the trade-off between video quality and smoothness. This optimization problem is a difficult-to-solve mixed-integer non-convex optimization problem. This invention proposes a block coordinate descent method with overlapping variable blocks to solve this problem, increasing the optimization space. To handle the binary constraints in the problem, a precise penalty method with balance constraints is also proposed to ensure the accuracy of the penalty function. Furthermore, this invention employs a successive convex approximation method to handle the non-convex constraints of the optimization problem. Simulation results show that, compared with the baseline scheme, the proposed scheme achieves a significant performance improvement and achieves a trade-off between video quality and smoothness.

[0134] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A method for maximizing user experience quality in multi-UAV aerial video transmission, characterized in that, Including the following steps: S1. Construct an aerial video stream transmission system and determine the trajectory constraints of the UAV. This aerial video stream transmission system consists of... Cellular connected drones and It consists of several ground base stations, in which each drone captures video from its respective Area of ​​Interest (PoI) and then transmits it to the corresponding base station, which then transmits the video to ground users via a high-capacity cellular link. S2. Determine the communication link constraints and video stream constraints of the over-the-air video stream transmission system; S3. Calculate the cumulative user experience quality; S4. With the goal of maximizing the minimum cumulative user experience quality, and using the constraints of the drone trajectory, the communication link, and the video stream as constraints, construct a joint optimization problem for transmission scheduling, video playback rate, and drone trajectory. The optimization problem is described as follows: ， ，(1) ，(2) ， (3) ， (4) ， (5) ， (6) ， (7) ， (8) in, Defined as the minimum cumulative quality of user experience. Indicates the total time The total number of time slots discretized into time slots, and the length of each time slot. ; and These are fixed parameters determined by the default application. Represented as ground users In the time slot The video playback speed at that time User Required playback rate; As a weighting factor; This means that for any drone , Equation (2) represents the video stream constraints. The defined binary variable represents the drone. In the time slot The transmission scheduling strategy at that time, if the drone Decision in time slot to base station Sending a video, then ,otherwise ; Indicates time slot drones and base stations The achievable rates between; Equations (3), (4), and (5) are communication link constraints, which are interpreted as follows: in each time slot, each base station serves at most one UAV, and each UAV sends video to at most one base station; Equations (6), (7), and (8) are UAV trajectory constraints. Indicates drone In the time slot The position at that time Indicates the maximum speed of the drone. This represents the length of each time slot after time discretization. This indicates the safe distance between drones to ensure collision avoidance. Indicates drone The radius of the monitored PoI area, Indicates drone The center of the monitored PoI area; S5. Solve the optimization problem to obtain the optimal solutions for transmission scheduling, video playback rate, and drone trajectory; Step S5 specifically includes the following steps: S51. Add an additional penalty term to the objective function, transforming the optimization problem into optimization problem P2: ， ， ， (9) ， (10) in It is a penalty parameter that is iteratively increased to ensure that the balance constraint is satisfied. Define the auxiliary penalty variable to be introduced. , satisfy: , ,here and They represent dimensional vectors consisting of all zeros and all ones; S52, Definition and As an iteration After that and The corresponding solution obtained, The update in the next iteration is obtained by the following operation; S53, Given and The optimization problem P2 is transformed into an optimization problem P3, and solving this optimization problem P3 yields the results. The optimal solution in the next iteration ; S54, Based on the optimization obtained in step S53 And given , The optimization problem P2 is transformed into optimization problem P4. Then, by introducing slack variables, optimization problem P4 is transformed into optimization problem P5. Finally, by considering local points... Using the lower bound for approximate substitution, the optimization problem P5 is transformed into optimization problem P6. Solving optimization problem P6 yields... The optimal solution in the next iteration ; S56, based on optimization The optimization problem P2 is transformed into the optimization problem P7. Solving the optimization problem P7 yields the following results. The optimal solution in the next iteration ; S57. Perform iterative optimization using the same steps as S52-S56 until the objective value of optimization problem P1 converges, and output the result at this point. The optimal solution.

2. The method for maximizing user experience quality in multi-UAV aerial video transmission according to claim 1, characterized in that, The calculation is as follows: ， Indicates bandwidth. Indicates drone Choose in time slot to base station Sending video at the base station The ratio of received signal to interference plus noise at the location; The calculation is as follows: ， in, Indicates time slot drones and base stations Channel gain between Indicates drone uplink transmit power, Indicates noise power; The calculation is as follows: ， in, This is the channel power gain with a spacing of 1m. Indicates in time slot drones and base stations The distance between them It is the channel loss factor. Indicates that the drone is at a fixed altitude flight, Indicates base station The horizontal coordinates.

3. The method for maximizing user experience quality in multi-UAV aerial video transmission according to claim 1, characterized in that, In step S56, the optimal solution The calculation is as follows: 。

Images