UAV-assisted task offloading and resource allocation method, device and medium based on NOMA

Abstract

The present invention discloses a method, device and medium for unmanned aerial vehicle (UAV)-assisted task offloading and resource allocation based on NOMA. The method is aimed at a multi-UAV-assisted MEC system. The method comprises: under a given computing task and time, establishing a joint optimization of computing resources and UAV trajectory planning to minimize the energy consumption of a weighted system; decomposing the optimization problem of minimizing the energy consumption of a weighted system into three sub-problems, including ground terminal equipment grouping, optimal allocation of computing resources of a UAV and a ground terminal equipment given a UAV trajectory, and trajectory planning based on a given resource allocation scheme; respectively solving the ground terminal equipment grouping, optimal allocation of computing resources of a UAV and a ground terminal equipment given a UAV trajectory, and trajectory planning based on a given resource allocation scheme, and obtaining ground terminal equipment grouping, ground terminal equipment computing resource allocation, UAV resource allocation and UAV trajectory planning.

Description

Technical Field

[0001] The present invention relates to a NOMA-based unmanned aerial vehicle (UAV) assisted task offloading and resource allocation method, device and medium, and belongs to the field of communication technology. Background Art

[0002] With the development of information technology, many intelligent applications are emerging, such as face recognition, interactive games, and virtual reality. These applications are usually computationally intensive and delay-sensitive. The emergence of Mobile Edge Computing (MEC) facilitates the implementation of these computing tasks. In practical applications, due to shadow fading and multipath effects, the signal attenuation of ground communications is serious. Therefore, the MEC system assisted by Unmanned Aerial Vehicle (UAV) is valued for its flexible mobility and low cost. In the mobile edge computing system, edge service cache refers to the pre-storage of programs required to perform computing tasks on the MEC server.

[0003] Since ground terminal devices offload data to MEC servers and then return the calculation results via wireless communication, MEC networks impose a heavy burden on scarce radio spectrum. In order to increase the task offloading rate, the transmission power must be increased, which intensifies the interference with other communications and shortens the battery life of local devices. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a method for unloading and allocating resources of unmanned aerial vehicles (UAVs) assisted by the NOMA system. The method adopts partial unloading for the weighted energy consumption of the system, and the ground terminal equipment reduces the system energy consumption by grouping.

[0005] To achieve the above object, the present invention is implemented by adopting the following technical solutions:

[0006] In the first aspect, the present invention provides a method for UAV-assisted task offloading and resource allocation based on the NOMA system, based on an energy-saving UAV (UAV) MEC framework combined with non-orthogonal multiple access (NOMA), including connected UAVs and ground terminal devices, wherein the UAV acts as an edge server to provide computing assistance for the ground terminal devices, and adopts NOMA to reduce the energy consumption of task offloading.

[0007] The method comprises:

[0008] Given the computing task and time, the joint optimization of computing resources and UAV trajectory planning is established to minimize the energy consumption of the weighted system.

[0009] The optimization problem of minimizing the energy consumption of the weighted system is decomposed into three sub-problems, including the grouping of ground terminal devices, the optimal allocation of computing resources of the UAV and ground terminal devices given the UAV trajectory, and trajectory planning based on a given resource allocation scheme.

[0010] The ground terminal equipment grouping, the optimal allocation of computing resources of UAV and ground terminal equipment for a given UAV trajectory, and the trajectory planning based on a given resource allocation scheme are solved respectively, and the ground terminal equipment grouping, ground terminal equipment computing resource allocation, UAV resource allocation and UAV trajectory planning are obtained.

[0011] In some embodiments, the method further comprises:

[0012] S1. The ground terminal device sends a task request. Each ground terminal device performs local calculations based on the ground terminal device computing resource allocation and the UAV resource allocation, and offloads part of its tasks to the MEC server on the UAV.

[0013] S2. The ground terminal device selects to access a drone, and the drone selects to accept or reject the application according to the grouping of the ground terminal device, thereby grouping the ground terminal devices and determining the pairing relationship between the drone and the ground terminal devices.

[0014] S3. The drone flies according to the optimized drone trajectory. The drone and ground terminal equipment allocate resources to the ground terminal equipment according to the optimized ground terminal equipment computing resource allocation and drone resource allocation. After the MEC server on the drone and the ground terminal equipment receive the unloaded data, they start to allocate computing resources and execute this task from the next time slot.

[0015] In some embodiments, the weighted system energy consumption minimization optimization problem P1 includes:

[0016] P1:

[0017]

[0018]

[0019]

[0020]

[0021]

[0022]

[0023]

[0024] Where E is the weighted total energy consumption of the system, Q = (Q k [n]: k∈K, n∈N) is the set sum of the horizontal position coordinates of UAV k; is the sum of the CPU frequencies of the ground terminal device m in the nth time slot; is the sum of the CPU frequencies of the tasks assigned by UAVk to ground terminal device m in the nth time slot, C1 represents the maximum computing capacity of the ground terminal device m, K = {1,2,...,K} and M = {1,2,...,M} are the sets of UAV and ground terminal devices respectively, N = {1,2,...,N} is the time slot interval, is the CPU frequency of the ground terminal device m in the nth time slot, The maximum CPU frequency of the ground terminal device m, C2 represents the maximum computing power of the drone k, is the CPU frequency of the task assigned to the ground terminal device m by UAVk in the nth time slot, is the highest CPU frequency of the MEC server on drone k; C3 represents the causal relationship between task offloading and edge computing, R m,k [n] is the number of bits unloaded from the ground terminal device m to the drone k in the nth time slot, is the number of bits completed by drone k for ground terminal device m in time slot n. C4 is to ensure that the system completes all computing tasks within a given time interval. is the number of bits completed by ground terminal device m in time slot n, D m is the number of bits to be executed by each group of ground terminal equipment m. C5 is the transmission power limit of the ground terminal equipment, p m.k [n] is the transmission power used by the ground terminal device m to offload the task to the drone k, is the maximum power of the ground terminal device m. C6 is the speed limit of the drone k, Q k [n] is the horizontal position coordinate of UAV k, is the maximum speed of drone k; C7 is the limit of the initial and final positions of drone k. I , is the initial position of the drone, Q F is the final position of the drone.

[0025] In some embodiments, given a computing task and time, a joint optimization of computing resources and UAV trajectory planning is established to minimize the weighted system energy consumption, including:

[0026] Due to the coverage of the base station, the ground terminal devices in the area cannot communicate directly with the BS. NOMA allows multiple ground terminal devices to share the same frequency band when offloading tasks to the same server. For the front-end link, the ground terminal devices within the coverage of each drone use orthogonal frequency division multiple access, and the bandwidth is evenly divided among multiple ground terminal devices. Assuming that the total uplink / downlink communication bandwidth is B, the two-way wireless communication spectrum bandwidth from drone k to the ground terminal device is B / |M k |, where M k The total number of ground terminal devices serving UAV k.

[0027] Each ground terminal device has a computing task to complete in a time interval T. Although the entire computing process is continuous in time, the interval can be discretized into N (N>1) time slots of equal length τ. Define N = {1, 2, ..., N} as the time slot interval.

[0028] Assume M k The number of ground terminal devices serving UAV k. Each ground terminal device has a computing task requirement.

[0029] The three-dimensional Cartesian coordinate system is used to describe the positions of the UAV and the ground terminal equipment. The horizontal position coordinates of the UAV k are Q k [n]=(x k [n],y k [n], H), the height is fixed H, and the horizontal position coordinate of the ground terminal equipment m is Q m [n]=(x m [n],y m [n],0), then the distance between the ground terminal device m and the drone k is The channel gain between the ground terminal device m and the drone k is Where h0 represents the channel gain when the reference distance is 1m.

[0030] Assume that the number of bits to be executed by each group of ground terminal devices m is D m A partial offloading strategy is adopted, that is, each ground terminal device m can perform local computation while offloading part of its tasks to the MEC server on the drone. After the MEC servers receive the offloaded data, they will start to allocate computing resources to execute this task starting from the next time slot. The number of bits offloaded by the ground terminal device m to the drone k in the nth time slot is R k,m [n].

[0031] The number of bits completed by ground terminal device m in time slot n is

[0032]

[0033] in, is the CPU frequency of the ground terminal device m in the nth time slot, The maximum CPU frequency of the ground terminal device m, Indicates the number of CPU cycles required to complete a computing task of the ground terminal device m.

[0034] In the nth time slot, the CPU frequency of the task assigned by UAVk to the ground terminal device m is The maximum CPU frequency of the MEC server on drone k is in

[0035] The number of bits completed by drone k for ground terminal device m in time slot n is

[0036]

[0037] in, It represents the number of CPU cycles required for UAV k to complete a computing task.

[0038] The system energy consumption consists of three parts: task calculation, task offloading and UAV flight energy consumption.

[0039] Since both the UAV and the ground terminal equipment need to perform calculations, both must be considered. The local energy consumed by the ground terminal equipment m in time slot n is

[0040]

[0041] in, is the constant of the effective switching capacitance of the CPU at the ground terminal device m

[0042] UAV k calculates the energy consumption of ground terminal device m in time slot n

[0043]

[0044] Therefore, the total computing energy consumption of the ground terminal equipment and the UAV is

[0045]

[0046]

[0047] Due to the motion of the UAVs, the channel conditions change over time. To make the continuous process feasible for resource allocation, it is assumed that the position of the UAV remains unchanged within a time slot. Therefore, the ground terminal devices of each group are arranged by each UAV in ascending order of the channel gain in each time slot. Therefore, the MEC server of each UAV has a timely updated list.

[0048] definition is the ground terminal device of UAV k in time slot n, where is the channel gain of the mth ground terminal device in the UAV k link during time slot n.

[0049] In each time slot, the UAV will perform SIC demodulation on the signals from multiple ground terminal devices. The sequence of SIC is arranged in descending order of the current channel gain of each ground terminal device. Note that only signals transmitted to the same UAV will cause interference. Each UAV will not be interfered by signals targeting other UAVs because they operate in different sub-bands. When the signal is decoded, it is received by the ground terminal device r1 k [n] To ground terminal equipment The signal is considered as interference. Ground terminal equipment After decoding the signal, it is removed from the superimposed signal. Then, the server starts decoding the signal from the ground terminal device. This iterative process continues until the ground terminal device r1 k The signal [n] is decoded. When the decoded signal comes from the ground terminal device r1 k [n] signal, the server is no longer interfered. Assuming that the ground terminal equipment The transmission power used to offload the task to UAV k is p rm,k [n], according to Shannon’s formula, at time slot n, the amount of computing tasks unloaded by ground terminal device m should satisfy Ground Terminal Equipment The maximum amount of data that can be unloaded to UAV k in the nth time slot is

[0050]

[0051] Where N0 is the power spectral density of the additive white Gaussian noise at the receiver.

[0052] To ensure the unloading task Can be offloaded to UAV k for execution. According to Shannon's formula, the transmission power of ground terminal device m in the nth time slot is The total energy consumption of task offloading is

[0053]

[0054] The initial and final positions of the drone are specified, and the maximum speed of drone k is The trajectory is not further constrained. To obtain the flight energy consumption, the UAV is assumed to fly uniformly in a straight line between the two locations, so acceleration and turning radius are not considered in the model.

[0055] The average speed of drone k in the nth time slot is v k [n] Use W k represents the mass of UAV k, then the energy consumption of UAV k’s flight is:

[0056]

[0057] Weighted total energy consumption of the system:

[0058]

[0059] Among them, α is the UAV flight energy consumption coefficient. user +w uav =1.

[0060] In some embodiments, solving the ground terminal device grouping, the optimal allocation of computing resources of the drone and the ground terminal device for a given drone trajectory, and the trajectory planning based on a given resource allocation scheme respectively, and obtaining the ground terminal device grouping, the ground terminal device computing resource allocation, the drone resource allocation, and the drone trajectory planning includes:

[0061] The matching game algorithm is used to solve the grouping problem and obtain the ground terminal equipment grouping;

[0062] Through CVX and quadratic approximation, the optimal allocation of computing resources for a given UAV trajectory and ground terminal equipment and the trajectory planning based on a given resource allocation scheme are transformed into a convex problem. Through an iterative algorithm, the optimal allocation of computing resources for a given UAV trajectory and ground terminal equipment and the trajectory planning based on a given resource allocation scheme are alternately solved, gradually approaching the optimal resource management of the system, and obtaining the computing resource allocation of ground terminal equipment, UAV resource allocation and UAV trajectory planning.

[0063] In some embodiments, a matching game algorithm is used to solve the grouping problem to obtain ground terminal device groups, including:

[0064] The problem of ground terminal devices accessing drones is formulated as a bilateral many-to-one matching problem.

[0065] In the matching, the ground terminal devices are set M and the drones are set K. The matching can be formulated as The preference relationship between ground terminal equipment and drones is: The preference relationship between drones and ground terminal equipment is:

[0066] A matching μ is defined as a function mapping from M∪K to K∪M, where k∈K and m∈M; μ(m)∈K, |μ(m)|=1

[0067] 0≤|μ(k)|≤a

[0068] Among them, a is the maximum number of ground terminal devices that can be matched by a drone. The first point means that each ground terminal device cannot be matched with more than one drone, and the second point means that the total number of ground terminal devices matched by each drone cannot exceed a.

[0069] Since this method takes into account that multiple ground terminal devices access the UAV at the same time, the utility function between the ground terminal device and the UAV needs to be carefully designed. Taking the weighted energy consumption obtained by the ground terminal device accessing the UAV as the utility function, the utility function of the ground terminal device m is

[0070]

[0071] The utility function of drone k is

[0072]

[0073] According to the above formula, the utility function of the ground terminal device and the UAV can be calculated and arranged in ascending order to obtain the preference list of the ground terminal device for the UAV {PLU m} m∈M and the ground terminal device's preference list for drones {PLU k} k∈K .

[0074] If drone k does not match any ground terminal device m, drone k matches the current highest-ranked ground terminal device m on the preference list. If this ground terminal device m does not match any drone at this time, the match is successful. But if this ground terminal device m has been matched with other drones k', consider the following situation: if in the preference list of ground terminal device m, ground terminal device m prefers drone k' rather than drone k, this ground terminal device m matches drone k'; conversely, if in the preference list of ground terminal device m, ground terminal device m prefers drone k rather than drone k', this ground terminal device m matches drone k.

[0075] The matching ends when all ground terminal devices are matched successfully or there is a ground terminal device without a match, but the ground terminal device sends a request to all applicable drones and is rejected.

[0076] In some embodiments, the optimization problem of minimizing the weighted system energy consumption is decomposed into three more manageable sub-problems, namely, grouping of ground terminal devices, optimal allocation of computing resources of drones and ground terminal devices for a given drone trajectory, and trajectory planning based on a given resource allocation scheme, including:

[0077] Given the UAV's trajectory, formulate subproblem P1.1 to find the optimal allocation strategy for computing resources

[0078] P1.1:

[0079] stC1-C4

[0080] The objective function of P1.1 is the weighted sum of the energy cost of local computing, task offloading, edge computing, and drone flight. In P1.1, and is a polynomial function of the CPU frequency with positive coefficients.

[0081] Therefore, it can be seen that the optimization problem P1.1 is a linear programming problem. Formulas C1-C4 are all linear, so problem P1.1 is a convex problem and is solved using CVX in the MATLAB toolbox.

[0082] When the allocation of ground terminal equipment and UAV resources is determined, the UAV trajectory problem can be solved. Problem P1 can be transformed into sub-problem P1.2:

[0083] P1.2:

[0084] stC5-C6

[0085] For the variable Q k , formula ||Q k -Q m || 2 is convex, and a continuous convex approximation algorithm is used. The transmission rate from the ground terminal device to the drone can be approximated as:

[0086]

[0087] in,

[0088] The P1.1 and P1.2 subproblems are solved by numerical iteration and gradually approaching the optimal point to obtain an approximate solution, and the corresponding solutions are updated to obtain the solution to P1: ground terminal equipment resource allocation, UAV resource allocation and UAV trajectory.

[0089] In a second aspect, the present invention provides a UAV-assisted task offloading and resource allocation device based on a NOMA system, including a processor and a storage medium;

[0090] The storage medium is used to store instructions;

[0091] The processor is configured to operate according to the instructions to execute the method according to the first aspect.

[0092] In a third aspect, the present invention provides a device, comprising:

[0093] Memory;

[0094] processor;

[0095] as well as

[0096] Computer programs;

[0097] The computer program is stored in the memory and is configured to be executed by the processor to implement the method described in the first aspect above.

[0098] In a fourth aspect, the present invention provides a storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method described in the first aspect.

[0099] Compared with the prior art, the present invention has the following beneficial effects:

[0100] 1. The present invention comprehensively considers introducing the NOMA concept into drone-assisted mobile edge computing, and uses non-orthogonal multiple access to more effectively manage task offloading. By adopting superposition coding (SIC) and successive interference cancellation (SIC), NOMA can make full use of the limited spectrum during the offloading / downloading process of MEC.

[0101] 2. The present invention jointly considers the grouping of ground terminal devices, which can effectively reduce the weighted energy consumption of the system.

[0102] 3. Different from the traditional computing task offloading method, this method makes full use of the mobility of drones and jointly optimizes the three sub-problems of computing resource allocation and drone placement, so that the weighted energy consumption of the system is effectively reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0103] Figure 1 is a schematic diagram of a method flow chart provided according to an embodiment of the present invention;

[0104] Figure 2 Schematic diagram of a system model provided according to an embodiment of the present invention. DETAILED DESCRIPTION

[0105] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0106] Embodiment 1:

[0107] This embodiment provides a method for unmanned aerial vehicle-assisted task offloading and resource allocation based on service cache, using non-orthogonal multiple access (NOMA) to more effectively manage task offloading. By adopting superposition coding (SIC) and successive interference cancellation (SIC), NOMA can make full use of limited spectrum in the offloading / downloading process of MEC.

[0108] include:

[0109] In some embodiments, reference Figure 1 , Figure 2 , a multi-UAV assisted MEC system provided in this embodiment includes a base station, K rotary-wing UAVs and M ground terminal devices, where K = {1,2,...,K} and M = {1,2,...,M} are sets of UAVs and ground terminal devices, respectively. Each UAV is equipped with a small server to provide computing and caching services for the ground terminal. Each UAV is equipped with a small edge server to provide computing and caching services for the ground terminal. And each UAV simultaneously serves its associated NOMA ground terminal device through non-orthogonal multiple access on its own bandwidth. Due to the coverage problem of the base station, the base station is based at a relatively far distance, and the device cannot communicate directly with the BS. The ground terminal devices are grouped according to certain rules and matched with the UAVs. The ground terminal device group served by the kth UAV is recorded as M k .

[0110] This paper proposes an energy-saving unmanned aerial vehicle (UAV) MEC framework combined with non-orthogonal multiple access (NOMA), in which multiple drones are deployed as edge servers to provide computing assistance for ground terminal devices on the ground, and NOMA is used to reduce the energy consumption of task offloading. Given a computing task and time, the optimization problem of minimizing the energy consumption of the weighted system is considered by considering the joint optimization of computing resources and drone trajectory planning. Due to the coupling of parameters, minimizing energy consumption is a highly non-convex problem. Therefore, the problem is decomposed into three more manageable sub-problems, namely, grouping of ground terminal devices, optimal allocation of computing resources for given drone trajectories, and trajectory planning based on a given resource allocation scheme. A matching game algorithm is used to solve the grouping problem separately, and then the two problems are transformed into convex problems through CVX and quadratic approximation. Then, an effective iterative algorithm is proposed to alternately solve the two sub-problems, gradually approaching the optimal resource management of the system.

[0111] S1. Based on the task request sent by the ground terminal device, the UAV and the ground terminal device adopt a partial offloading strategy. Each ground terminal device can perform local computing and offload part of its tasks to the MEC server on the UAV.

[0112] S2. The ground terminal device chooses to access a drone, and the drone can choose to accept or reject the application according to the grouping criteria, thereby grouping the ground terminal devices.

[0113] S3. After receiving the unloaded data, the MEC server on the drone and the ground terminal device start to allocate computing resources to execute the task from the next time slot;

[0114] S4. The ground terminal device and the UAV respond to the task request sent by the ground terminal device, and use the pre-built minimum weighted system energy consumption optimization problem to jointly optimize the ground terminal device grouping, ground terminal device computing resource allocation, UAV resource allocation and UAV trajectory planning;

[0115] S5. Determine which drone the ground terminal equipment can be partially offloaded to for calculation according to the optimized ground terminal equipment grouping;

[0116] The UAV flies according to the optimized UAV trajectory, allocates resources to the ground terminal equipment according to the optimized ground terminal equipment calculation resource allocation and UAV resource allocation, and executes the task.

[0117] In some embodiments, in step S4, minimizing the weighted system energy consumption optimization P1 includes:

[0118] P1:

[0119]

[0120]

[0121]

[0122]

[0123]

[0124]

[0125]

[0126] Where E is the weighted total energy consumption of the system, Q = (Q k [n]: k∈K, n∈N) is the set sum of the horizontal position coordinates of UAV k; is the sum of the CPU frequencies of the ground terminal device m in the nth time slot; is the sum of the CPU frequencies of the tasks assigned by UAVk to ground terminal device m in the nth time slot, C1 represents the maximum computing capacity of the ground terminal device m, K = {1,2,...,K} and M = {1,2,...,M} are the sets of UAV and ground terminal devices respectively, N = {1,2,...,N} is the time slot interval, is the CPU frequency of the ground terminal device m in the nth time slot, The maximum CPU frequency of the ground terminal device m, C2 represents the maximum computing power of the drone k, is the CPU frequency of the task assigned to the ground terminal device m by UAVk in the nth time slot, is the highest CPU frequency of the MEC server on drone k; C3 represents the causal relationship between task offloading and edge computing, R m,k [n] is the number of bits unloaded from the ground terminal device m to the drone k in the nth time slot, is the number of bits completed by drone k for ground terminal device m in time slot n. C4 is to ensure that the system completes all computing tasks within a given time interval. is the number of bits completed by ground terminal device m in time slot n, D m is the number of bits to be executed by each group of ground terminal equipment m. C5 is the transmission power limit of the ground terminal equipment, p m.k [n] is the transmission power used by the ground terminal device m to offload the task to the drone k, is the maximum power of the ground terminal device m. C6 is the speed limit of the drone k, Q k [n] is the horizontal position coordinate of UAV k, is the maximum speed of drone k; C7 is the limit of the initial and final positions of drone k. I , is the initial position of the drone, Q F is the final position of the drone.

[0127] In some embodiments, a method for constructing a weighted system energy consumption optimization problem includes:

[0128] Due to the coverage of the base station, the ground terminal devices in the area cannot communicate directly with the BS. NOMA allows multiple ground terminal devices to share the same frequency band when offloading tasks to the same server. For the front-end link, the ground terminal devices within the coverage of each drone use orthogonal frequency division multiple access, and the bandwidth is evenly divided among multiple ground terminal devices. Assuming that the total uplink / downlink communication bandwidth is B, the two-way wireless communication spectrum bandwidth from drone k to the ground terminal device is B / |M k |, where M k The total number of ground terminal devices serving UAV k.

[0129] Each ground terminal device has a computing task to complete in a time interval T. Although the entire computing process is continuous in time, the interval can be discretized into N (N>1) time slots of equal length τ. Define N = {1, 2, ..., N} as the time slot interval.

[0130] Assume M k The number of ground terminal devices serving UAV k. Each ground terminal device has a computing task requirement.

[0131] The three-dimensional Cartesian coordinate system is used to describe the positions of the UAV and the ground terminal equipment. The horizontal position coordinates of the UAV k are Q k [n]=(x k [n],y k [n], H), the height is fixed H, and the horizontal position coordinate of the ground terminal equipment m is Q m [n]=(x m [n],y m [n],0), then the distance between the ground terminal device m and the drone k is The channel gain between the ground terminal device m and the drone k is Where h0 represents the channel gain when the reference distance is 1m.

[0132] Assume that the number of bits to be executed by each group of ground terminal devices m is D m A partial offloading strategy is adopted, that is, each ground terminal device m can perform local computation while offloading part of its tasks to the MEC server on the drone. After the MEC servers receive the offloaded data, they will start to allocate computing resources to execute this task starting from the next time slot. The number of bits offloaded by the ground terminal device m to the drone k in the nth time slot is R k,m [n].

[0133] The number of bits completed by ground terminal device m in time slot n is

[0134]

[0135] in, is the CPU frequency of the ground terminal device m in the nth time slot, The maximum CPU frequency of the ground terminal device m, Indicates the number of CPU cycles required to complete a computing task of the ground terminal device m.

[0136] In the nth time slot, the CPU frequency of the task assigned by UAVk to the ground terminal device m is The maximum CPU frequency of the MEC server on drone k is in

[0137] The number of bits completed by drone k for ground terminal device m in time slot n is

[0138]

[0139] in, It represents the number of CPU cycles required for UAV k to complete a computing task.

[0140] The system energy consumption consists of three parts: task calculation, task offloading and UAV flight energy consumption.

[0141] Since both the UAV and the ground terminal equipment need to perform calculations, both must be considered. The local energy consumed by the ground terminal equipment m in time slot n is

[0142]

[0143] in, is the constant of the effective switching capacitance of the CPU at the ground terminal device m

[0144] UAV k calculates the energy consumption of ground terminal device m in time slot n

[0145]

[0146] Therefore, the total computing energy consumption of the ground terminal equipment and the UAV is

[0147]

[0148]

[0149] Due to the motion of the UAVs, the channel conditions change over time. To make the continuous process feasible for resource allocation, it is assumed that the position of the UAV remains unchanged within a time slot. Therefore, the ground terminal devices of each group are arranged by each UAV in ascending order of the channel gain in each time slot. Therefore, the MEC server of each UAV has a timely updated list.

[0150] definition is the ground terminal device of UAV k in time slot n, where is the channel gain of the mth ground terminal device in the UAV k link during time slot n.

[0151] In each time slot, the UAV will perform SIC demodulation on the signals from multiple ground terminal devices. The sequence of SIC is arranged in descending order of the current channel gain of each ground terminal device. Note that only signals transmitted to the same UAV will cause interference. Each UAV will not be interfered by signals targeting other UAVs because they operate in different sub-bands. When the signal is decoded, it is received by the ground terminal device r1 k [n] To ground terminal equipment The signal is considered as interference. Ground terminal equipment After decoding the signal, it is removed from the superimposed signal. Then, the server starts decoding the signal from the ground terminal device. This iterative process continues until the ground terminal device r1 k The signal of [n] is decoded. The server here refers to the mec on the drone, which is responsible for the task calculation proposed by the terminal device and will allocate computing resources. Because it is noma, there is interference.

[0152] When decoding from the ground terminal equipment r1 k [n] signal, the server is no longer interfered. Assuming that the ground terminal equipment The transmission power used to offload the task to UAV k is p rm,k [n], according to Shannon’s formula, at time slot n, the amount of computing tasks unloaded by ground terminal device m should satisfy Ground Terminal Equipment The maximum amount of data that can be unloaded to UAV k in the nth time slot is

[0153]

[0154] Where N0 is the power spectral density of the additive white Gaussian noise at the receiver.

[0155] To ensure the unloading task Can be offloaded to UAV k for execution. According to Shannon's formula, the transmission power of ground terminal device m in the nth time slot is The total energy consumption of task offloading is

[0156]

[0157] The initial and final positions of the drone are specified, and the maximum speed of drone k is The trajectory is not further constrained. To obtain the flight energy consumption, the UAV is assumed to fly uniformly in a straight line between the two locations, so acceleration and turning radius are not considered in the model.

[0158] The average speed of drone k in the nth time slot is v k [n] Use W k represents the mass of UAV k, then the energy consumption of UAV k’s flight is:

[0159]

[0160] Weighted total energy consumption of the system:

[0161]

[0162] Among them, α is the UAV flight energy consumption coefficient. user +w uav =1.

[0163] In some embodiments, in step S2, the optimization problem of ground terminal device grouping includes:

[0164] S21: This method constructs the problem of ground terminal devices accessing drones as a bilateral many-to-one matching problem, in which multiple ground terminal devices can access the same drone at the same time. Each drone can access at most several ground terminal devices, and each ground terminal device can only select This method constructs the problem of ground terminal devices accessing drones as a bilateral many-to-one matching problem, in which multiple ground terminal devices can access the same drone at the same time.

[0165] Each ground terminal device can only choose to access one drone.

[0166] The drone can choose to accept or reject the application of the ground terminal device according to its own preference. The matching ends when all ground terminal devices are matched successfully or there is no matching ground terminal device, but the ground terminal device sends requests to all applicable drones and is rejected.

[0167] Choose to visit any drone. The drone can choose to accept or reject the application of the ground terminal device according to its own preference. The matching ends when all ground terminal devices are matched successfully or there is a ground terminal device without a match, but the ground terminal device sends a request to all applicable drones and is rejected.

[0168] S22: In the matching, the ground terminal devices are set M and the drones are set K. The matching can be formulated as The preference relationship between ground terminal equipment and drones is: The preference relationship between drones and ground terminal equipment is:

[0169] S23: A matching μ is defined as a function mapping from M∪K to K∪M, where k∈K and m∈M.

[0170] (1)μ(m)∈K, |μ(m)|=1

[0171] (2) 0≤|μ(k)|≤a

[0172] Among them, a is the maximum number of ground terminal devices that can be matched by a drone. The first point means that each ground terminal device cannot be matched with more than one drone, and the second point means that the total number of ground terminal devices matched by each drone cannot exceed a.

[0173] S24: Since this method takes into account that multiple ground terminal devices access the UAV at the same time, the utility function between the ground terminal device and the UAV needs to be carefully designed. Taking the weighted energy consumption obtained by the ground terminal device accessing the UAV as the utility function, the utility function of the ground terminal device m is

[0174]

[0175] S25: Similarly, the utility function of drone k is

[0176]

[0177] S26: According to the above formula, the ground terminal device and the UAV can calculate the utility function and arrange them in ascending order to obtain the preference list of the ground terminal device for the UAV {PLU m} m∈M and the ground terminal device's preference list for drones {PLU k} k∈K .

[0178] S27: If drone k does not match any ground terminal device m, drone k matches the current highest-ranked ground terminal device m on the preference list. If this ground terminal device m does not match any drone at this time, the match is successful. However, if this ground terminal device m has continued to match with other drones k', consider the following situation: if in the preference list of ground terminal device m, ground terminal device m prefers drone k' rather than drone k, this ground terminal device m matches drone k'; conversely, if in the preference list of ground terminal device m, ground terminal device m prefers drone k rather than drone k', this ground terminal device m matches drone k.

[0179] In some embodiments, S4. using a pre-constructed optimization problem of minimizing weighted energy consumption to jointly optimize the ground terminal device resource allocation, the UAV resource allocation and the UAV trajectory includes:

[0180] S41. Given the UAV’s trajectory, formulate subproblem P1.1 to find the optimal allocation strategy for computing resources:

[0181] P1.1:

[0182] stC1-C4

[0183] The objective function of P1.1 is the weighted sum of the energy cost of local computing, task offloading, edge computing, and drone flight. Since the flight trajectory of the drone is predetermined, the energy cost of drone flight becomes a negligible constant in P1.1. and is a polynomial function of the CPU frequency with positive coefficients. Therefore, these two terms are convex.

[0184] Therefore, it can be seen that the optimization problem P1.1 is a linear programming problem. Formulas C1-C4 are all linear, so problem P1.1 is a convex problem and is solved using CVX in the MATLAB toolbox.

[0185] S42. When the ground terminal equipment and UAV resource allocation are determined, the UAV trajectory problem is solved. Problem P1 can be transformed into sub-problem P1.2:

[0186] P1.2:

[0187] stC5-C6

[0188] Since the computing resources in P1.2 are fixed, there is no need to consider the energy cost of the computation. k , formula ||Q k -Q m || 2 is convex, so a continuous convex approximation algorithm can be used. Because the first-order Taylor expansion of any point of any convex function is the global lower bound of the convex function, the convex function can be approximated by the first-order Taylor expansion. The transmission rate from the ground terminal device to the drone can be approximated as:

[0189]

[0190] in,

[0191] S43. Solve the three sub-problems of P1.1, P1.2 and ground terminal device grouping by numerical iteration-progressively approaching the optimal point to obtain an approximate solution, and update the corresponding solutions to obtain the solution of P1: ground terminal device grouping, ground terminal device resource allocation, UAV resource allocation and UAV trajectory;

[0192] Example 2

[0193] Based on Embodiment 1, this embodiment provides a drone-assisted task offloading and resource allocation device based on service cache, including a processor and a storage medium;

[0194] The storage medium is used to store instructions;

[0195] The processor is configured to operate according to the instructions to execute the method according to embodiment 1.

[0196] Example 3

[0197] Based on embodiment 1, this embodiment provides a device, including:

[0198] Memory;

[0199] processor;

[0200] as well as

[0201] Computer programs;

[0202] The computer program is stored in the memory and is configured to be executed by the processor to implement the method described in Example 1.

[0203] Example 4

[0204] Based on Embodiment 1, this embodiment provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the method described in Embodiment 1 is implemented.

[0205] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0206] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0207] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0208] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0209] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for unmanned aerial vehicle assisted task offloading and resource allocation based on NOMA system, characterized in that: Based on an energy-saving UAV MEC framework combined with non-orthogonal multiple access, including connected UAVs and ground terminal devices, the UAV acts as an edge server to provide computing assistance for the ground terminal device, and NOMA is used to reduce the energy consumption of task offloading; The method comprises: Given the computing task and time, establish the joint optimization of computing resources and UAV trajectory planning to minimize the energy consumption of the weighted system; Decomposing the optimization problem of minimizing the energy consumption of the weighted system into three sub-problems, including grouping of ground terminal devices, optimal allocation of computing resources of the UAV and ground terminal devices given the UAV trajectory, and trajectory planning based on a given resource allocation scheme; Solve the ground terminal equipment grouping, the optimal allocation of computing resources of the UAV and ground terminal equipment for a given UAV trajectory, and the trajectory planning based on a given resource allocation scheme, and obtain the ground terminal equipment grouping, ground terminal equipment computing resource allocation, UAV resource allocation, and UAV trajectory planning; The method further comprises: S1. The ground terminal device sends a task request. Each ground terminal device performs local calculations based on the ground terminal device computing resource allocation and the UAV resource allocation, and offloads part of its tasks to the MEC server on the UAV. S2. The ground terminal device selects to access a drone, and the drone chooses to accept or reject the request based on the grouping of the ground terminal device, thereby grouping the ground terminal device and determining the grouping relationship between the drone and the ground terminal device; S3. The drone flies according to the optimized drone trajectory. The drone and the ground terminal equipment allocate resources to the ground terminal equipment according to the optimized ground terminal equipment computing resource allocation and drone resource allocation. After the MEC server on the drone and the ground terminal equipment receive the unloaded data, they start to allocate computing resources and execute this task from the next time slot. The optimization problem P1 of minimizing weighted system energy consumption includes: Where E is the weighted total energy consumption of the system, Q = (Q k [n]: k∈K,n∈N) is the set of horizontal position coordinates of UAV k; is the set of CPU frequencies of the ground terminal device m in the nth time slot; is the set of CPU frequencies of the tasks assigned by UAVk to the ground terminal device m in the nth time slot, It means any; C1 represents the maximum computing capacity of the ground terminal device m, K = {1, 2, ..., K} and M = {1, 2, ..., M} are the sets of UAV and ground terminal devices respectively, N = {1, 2, ..., N} is the time slot interval, is the CPU frequency of the ground terminal device m in the nth time slot, The maximum CPU frequency of the ground terminal device m, C2 represents the maximum computing capacity of drone k, is the CPU frequency of the task assigned to the ground terminal device m by UAVk in the nth time slot, is the maximum CPU frequency of the MEC server on drone k; C3 represents the causal relationship between task offloading and edge computing, R m,k [n] is the number of bits unloaded from the ground terminal device m to the drone k in the nth time slot, is the number of bits completed by drone k for ground terminal device m in time slot n; C4 ensures that the system completes all computing tasks within a given time interval. is the number of bits completed by ground terminal device m in time slot n, D m The number of bits to be executed by the ground terminal device m of each group; C5 is the transmission power limit of the ground terminal equipment, p m.k [n] is the transmission power used by the ground terminal device m to offload the task to the drone k, is the maximum power of the ground terminal device m; C6 is the speed limit of drone k, Q k [n] is the horizontal position coordinate of UAV k, is the maximum speed of UAV k; C7 is the limit of the initial and final positions of drone k; Q I , is the initial position of the drone, Q F is the final position of the drone.

2. The method for unloading and allocating resources of unmanned aerial vehicle-assisted tasks based on the NOMA system according to claim 1 is characterized in that: Given the computing task and time, the joint optimization of computing resources and UAV trajectory planning is established to minimize the energy consumption of the weighted system, including: For the front-end link, the ground terminal equipment within the coverage of each drone adopts orthogonal frequency division multiple access, and the bandwidth is evenly divided among multiple ground terminal equipment. Assuming that the total uplink / downlink communication bandwidth is B, the two-way wireless communication spectrum bandwidth from drone k to ground terminal equipment is B / |M k |, where M k The total number of ground terminal devices serving UAV k; Each ground terminal device has a computing task to be completed in a time interval T; the interval can be discretized into N, N>1, time slots of equal length τ; define N = {1, 2, ..., N} as the time slot interval; M k The number of ground terminal devices serving UAV k, each of which has a computing task requirement; The three-dimensional Cartesian coordinate system is used to describe the positions of UAV and ground terminal equipment; the horizontal position coordinates of UAV k are Q k [n]=(x k [n],y k [n], H), the height is fixed H, and the horizontal position coordinate of the ground terminal equipment m is Q m [n]=(x m [n],y m [n],0), then the distance between the ground terminal device m and the drone k is The channel gain between the ground terminal device m and the drone k is Where h0 represents the channel gain when the reference distance is 1m; Assume that the number of bits to be executed by each group of ground terminal devices m is D m ; A partial offloading strategy is adopted, that is, each ground terminal device m can perform local computing and offload part of its tasks to the MEC server on the drone; after the MEC servers receive the offloaded data, they will start to allocate computing resources to execute this task from the next time slot; the number of bits that the ground terminal device m offloads to the drone k in the nth time slot is R m,k [n]; The number of bits completed by ground terminal device m in time slot n is in, is the CPU frequency of the ground terminal device m in the nth time slot, The maximum CPU frequency of the ground terminal device m, It represents the number of CPU cycles required to complete a computing task of the ground terminal device m; In the nth time slot, the UAV k The CPU frequency of the task assigned to the ground terminal device m is The maximum CPU frequency of the MEC server on drone k is in The number of bits completed by drone k for ground terminal device m in time slot n is in, represents the number of CPU cycles required for drone k to complete a computing task; The system energy consumption consists of three parts: task calculation, task offloading, and UAV flight energy consumption; The local energy consumed by ground terminal device m in time slot n is in, is a constant called the effective switching capacitance of the CPU at the ground terminal device m; UAV k calculates the energy consumption of ground terminal device m in time slot n: Therefore, the total energy consumption of the ground terminal equipment and the UAV is: The ground terminal devices of each group are arranged by each drone in ascending order according to the channel gain of each time slot; the MEC server of each drone has a timely updated list; definition is the ground terminal device of UAV k in time slot n, where is the channel gain of the mth ground terminal device in the UAVk link during time slot n; In each time slot, the UAV will perform SIC demodulation on the signals from multiple ground terminal devices. The SIC sequence is arranged in descending order of the current channel gain of each ground terminal device. When the server on the UAV k receives the signal from the ground terminal device, When the signal is decoded, it is received by the ground terminal equipment To ground terminal equipment The signal is considered as interference; ground terminal equipment After decoding the signal, it is removed from the superimposed signal; then, the server on drone k starts decoding the signal from the ground terminal device This iterative process continues until the ground terminal equipment The signal is decoded; when the decoded signal comes from the ground terminal equipment The server is no longer interfered when the signal is sent from the ground terminal device. The transmission power used to offload the task to UAV k is p rm,k [n], according to Shannon’s formula, at time slot n, the amount of computing tasks unloaded by ground terminal device m should satisfy Ground Terminal Equipment The maximum amount of data that can be unloaded to UAV k in the nth time slot is Where N0 is the power spectral density of additive white Gaussian noise at the receiver; To ensure the unloading task Can be offloaded to UAV k for execution. According to Shannon's formula, the transmission power of ground terminal device m in the nth time slot is The total energy consumption of task offloading is The maximum speed of UAV k is The average speed of drone k in the nth time slot is v k [n]; Use W k represents the mass of UAV k, then the energy consumption of UAVk’s flight is: Weighted total energy consumption of the system: Among them, α is the UAV flight energy consumption coefficient; w user Represents the energy consumption coefficient of ground terminal equipment, w uav represents the energy consumption coefficient of the UAV, w user +w uav =1.

3. The method for unloading and allocating resources of unmanned aerial vehicle-assisted tasks based on the NOMA system according to claim 2 is characterized in that: Solve the ground terminal equipment grouping, the optimal allocation of computing resources of the UAV and ground terminal equipment for a given UAV trajectory, and the trajectory planning based on a given resource allocation scheme, and obtain the ground terminal equipment grouping, ground terminal equipment computing resource allocation, UAV resource allocation, and UAV trajectory planning, including: The matching game algorithm is used to solve the grouping problem and obtain the ground terminal equipment grouping; Through quadratic approximation and CVX in MATLAB toolbox, the optimal allocation of computing resources for UAV and ground terminal equipment for a given UAV trajectory and the trajectory planning based on a given resource allocation scheme are transformed into convex problems. Through an iterative algorithm, the optimal allocation of computing resources for UAV and ground terminal equipment for a given UAV trajectory and the trajectory planning based on a given resource allocation scheme are alternately solved, gradually approaching the optimal resource management of the system, and obtaining the computing resource allocation of ground terminal equipment, UAV resource allocation and UAV trajectory planning.

4. The method for unloading and allocating resources of unmanned aerial vehicle-assisted tasks based on the NOMA system according to claim 3 is characterized in that: The matching game algorithm is used to solve the grouping problem and obtain the ground terminal equipment grouping, including: The problem of ground terminal devices accessing drones is formulated as a bilateral many-to-one matching problem. In the matching, the ground terminal devices are set M and the drones are set K. The matching can be formulated as The preference relationship between ground terminal equipment and drones is: The preference relationship between drones and ground terminal equipment is: A matching μ is defined as a function mapping from M∪K to K∪M, where k∈K and m∈M; μ(m)∈K, |μ(m)|=1 0≤|μ(k)|≤a Among them, a is the maximum number of ground terminal devices matched by the drone; the first point means that each ground terminal device cannot match more than one drone, and the second point means that the total number of ground terminal devices matched by each drone cannot exceed a; Since this method takes into account that multiple ground terminal devices access the UAV at the same time, it is necessary to carefully design the utility function between the ground terminal device and the UAV; taking the weighted energy consumption obtained by the ground terminal device accessing the UAV as the utility function, the utility function of the ground terminal device m is: The utility function of drone k is According to the above formula, the utility function of the ground terminal device and the UAV can be calculated and arranged in ascending order to obtain the preference list of the ground terminal device for the UAV {PLU m } m∈M and the ground terminal device's preference list for drones {PLU k } k∈K ; If drone k does not match any ground terminal device m, drone k matches the ground terminal device m with the highest current ranking in the preference list. If the ground terminal device m does not match any drone at this time, the match is successful; but if the ground terminal device m has continued to match with other drones k', consider the following situation: if in the preference list of ground terminal device m, the ground terminal device m prefers drone k' rather than drone k, the ground terminal device m matches drone k'; conversely, if in the preference list of ground terminal device m, the ground terminal device m prefers drone k rather than drone k', the ground terminal device m matches drone k; The matching ends when all ground terminal devices are matched successfully or there is a ground terminal device without a match, but the ground terminal device sends a request to all applicable drones and is rejected.

5. The method for unloading and allocating resources of unmanned aerial vehicle-assisted tasks based on the NOMA system according to claim 1 is characterized in that: The optimization problem of minimizing the energy consumption of the weighted system is decomposed into three more manageable sub-problems, namely, the grouping of ground terminal devices, the optimal allocation of computing resources of the UAV and ground terminal devices given the UAV trajectory, and trajectory planning based on a given resource allocation scheme, including: Given the UAV's trajectory, formulate subproblem P1.1 to find the optimal allocation strategy for computing resources The objective function of P1.1 is the weighted sum of the energy cost of local computing, task offloading, edge computing, and drone flight. In P1.1, and is a polynomial function of CPU frequency with positive coefficients; Therefore, it can be seen that the optimization problem P1.1 is a linear programming problem. Formulas C1-C4 are all linear, so problem P1.1 is a convex problem and is solved using CVX in the MATLAB toolbox. When the allocation of ground terminal equipment and UAV resources is determined, the UAV trajectory problem can be solved. Problem P1 can be transformed into sub-problem P1.2: For the variable Q k , formula ||Q k -Q m || 2 It is convex, and a continuous convex approximation algorithm is used; the transmission rate from the ground terminal device to the drone can be approximated as: in, The P1.1 and P1.2 subproblems are solved by numerical iteration and gradually approaching the optimal point to obtain an approximate solution, and the corresponding solutions are updated to obtain the solution to P1: ground terminal equipment resource allocation, UAV resource allocation and UAV trajectory.

6. A drone-assisted task offloading and resource allocation device based on the NOMA system, characterized in that: including processor and storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the method according to any one of claims 1 to 5.

7. An electronic device, characterized in that: include: Memory; processor; as well as Computer programs; The computer program is stored in the memory and is configured to be executed by the processor to implement the method according to any one of claims 1 to 5.

8. A storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images