Resource optimization method for decellularized massive MIMO system based on drone assistance

Abstract

The present invention discloses a resource optimization method for a decellularized large-scale MIMO system assisted by a drone. In the scenario of a decellularized large-scale MIMO system assisted by a drone, the drone acts as an aerial access point and a ground AP point to jointly provide services for ground users. The specific design of the method includes designing a user scheduling rule based on air-ground AP joint service; designing a communication model based on air-ground joint service according to the scheduling rule; designing a fair resource allocation scheme, including a user scheduling scheme, a drone location deployment scheme, and a user power allocation scheme, to maximize the user minimum downlink rate of the above communication model, and finally verifying the feasibility of the proposed utility model through optimization solution methods such as block coordinate descent and continuous convex optimization technology. The present invention can propose an optimal resource allocation strategy including drones according to the current user and ground resource distribution, so as to achieve blind spot-free coverage of ground user services.

Description

Technical Field

[0001] The present invention belongs to the field of communication resource management, and in particular relates to a resource optimization method for a de-cellularized large-scale MIMO system assisted by a drone. Background Art

[0002] Decellularized massive MIMO (Massive Multiple Input Multiple Output) is expected to outperform 5G (B5G) wireless technology as it provides higher spectral efficiency, higher energy efficiency, and better spatial diversity. However, the large number of long cables (front-end transport requirements) between each access point (AP) and the central processing unit (CPU) hinders its practical application. In recent years, UAVs have been often considered as aerial communication aids due to their absolute advantages in rapid deployment, controllable maneuverability, low cost, and high probability of air-to-ground link line of sight. They are used for rapid service restoration after partial or complete infrastructure damage caused by natural disasters and in communication congested areas. Therefore, the addition of drones as aerial APs can effectively improve the challenges brought by the large number of long cables in decellularized massive MIMO systems. Although UAVs can break through the limitations of traditional decellularized massive MIMO, there are still some remaining issues that need to be addressed.

[0003] One of the most critical issues in the decellularized massive MIMO system with UAVs is resource allocation. First, ground APs are already able to provide high-quality services to neighboring users. Improper UAV scheduling will cause communication switching of neighboring users, which will increase the system burden and reduce the user service quality. In addition, the performance and operation time of the UAV system are fundamentally limited by the limited onboard energy. If the UAV is not well designed with a good stop point, it will cause a lot of UAV maneuvering energy consumption, reducing its communication efficiency. In addition, when radio resources are insufficient, the deployment of UAVs covering multiple farthest users cannot meet the service quality requirements (QoE) of some users. It is easy to design multiple drones to meet the QoE of users, but the entire system may generate a lot of resource consumption. Finally, the QoE requirements of users are diverse and randomly distributed. It is very challenging for the CPU to make reasonable resource scheduling and allocation to improve user experience and achieve fair performance among users. Summary of the invention

[0004] Purpose of the invention: The purpose of the present invention is to provide a resource optimization method for a UAV-assisted decellularized massive MIMO system that meets the system equipment limitations and the QoE requirements of all users, that is, the CPU can achieve optimal UAV deployment and user scheduling based on user requests, while considering system resource consumption and reasonably allocating resources.

[0005] Technical solution: The resource optimization method of the UAV-assisted decellularized massive MIMO system of the present invention deploys UAVs as aerial APs in the target area, constructs a joint service communication model by executing the following steps, and optimizes the configuration of communication resources based on the model;

[0006] Step 1: Deploy drones as aerial APs in the target area to provide services to users in signal coverage blind areas, and design a user scheduling solution for joint services of ground APs and aerial APs.

[0007] Step 2: Based on the user scheduling scheme of the joint service of ground AP and aerial AP, a joint service communication model is constructed;

[0008] Step 3: Design a resource allocation scheme including user scheduling, UAV location deployment and user power allocation to maximize the minimum downlink rate of users in the above joint service communication model;

[0009] Step 4: Verify the feasibility of the joint service communication model through optimization solution method.

[0010] Furthermore, the step 1 specifically includes:

[0011] The AP selection method is to connect multiple APs within the service range for each user at a time, using binary variables {χ mk , χ uk} respectively represent the scheduling of user ground AP and aerial AP, and define the user scheduling matrix If user K is served by a ground AP point, then the binary variable χ mk The coefficient is 1, otherwise it is 0. If the user is served by an aerial AP point, the binary variable χ uk =1, otherwise 0; According to the resource heterogeneity and the restrictions of UAV deployment, the coordinated access rules of aerial and ground AP points are considered: during each resource allocation period, each user communicates with a single service point in the air or on the ground; the ground AP provides services to all users in the target area; the aerial AP provides services to users within the wireless coverage of the target area, which can be expressed as binary constraints:

[0012]

[0013]

[0014]

[0015] The set of users and ground APs are represented by K and M respectively. represents the horizontal coordinate of the kth user, w a Indicates the horizontal coordinate of the ground AP, represents the horizontal position of the UAV u, Ra , R u Respectively represent the service range of ground AP and aerial AP, R a Affected by the cable length, R u It is determined by the flight altitude and battery level of the drone.

[0016] Furthermore, the step 2 specifically includes:

[0017] Step 2-1: Design a channel model that combines the effects of small-scale fading and large-scale fading. Let g mk ∈￡ represents the channel gain between the mth AP and the kth user, and the channel transmission model is Among them, β mk 、h mk They represent the large-scale fading coefficient and small-scale fading factor between the m-th AP and the k-th user, respectively, where the small-scale fading is Rayleigh fading, i.e., h mk ~CN(0,1), large-scale fading coefficient β of ground AP point mk Modeled as the product of path loss and shadow fading, β mk =pl mk ·s mk , where s mk represents the lognormal shading, pl mk The three-stage path loss is expressed as:

[0018]

[0019] Where L is a constant that depends on the carrier frequency, user and AP heights, and d mk is the horizontal distance between AP m and user k, d0 and d1 are reference value distances;

[0020] Step 2-2: The communication link between the aerial AP and the user is expressed as:

[0021]

[0022] Among them, a1 and a2 are the path loss indexes of line-of-sight LoS and non-line-of-sight NLoS respectively. are the probabilities of LoS links and NLoS links, respectively. is the Euclidean distance between the u-th UAV and the k-th user, and H is the fixed flight altitude of the UAV. When designing a communication system based on aerial AP points, considering the randomness of LoS and NLoS links, the LoS transmission probability between the u-th aerial AP and the k-th user is:

[0023]

[0024] Where a and b are environmental parameters, and U is the set of aerial APs. is the elevation angle between the kth user and the uth aerial AP. The LoS probability model will increase with the increase of AP height and elevation angle. The average large-scale fading between the kth user and the uth aerial AP is expressed as The channel model is derived in large-scale form as E{g uk}=β uk ;

[0025] Step 2-3: In the scenario where the channel conditions are known, the downlink uses conjugate beamforming to send signals to the user; the received signal of user k under user scheduling is where x ik Indicates the signal sent by the AP; if the user is served by a ground AP, then:

[0026]

[0027] Among them, p mk s is the power allocation parameter for the terrestrial AP in the downlink of the cellular massive MIMO. k The signal transmitted to the user satisfies E{|s k | 2}=1,ω k is additive white Gaussian noise (AWGN);

[0028] In the scenario where the aerial AP only provides business services to users within its fixed range, when the aerial AP is hovering at a fixed position, the signal received by users within its communication range is:

[0029]

[0030] Among them, ρ uk Allocate parameters for power;

[0031] The entire received signal of the user is expressed as

[0032] Furthermore, the step 3 specifically includes:

[0033] Step 3-1: User scheduling scheme based on joint service of ground AP and aerial AP. In the scenario where ground AP shares channel and aerial AP shares channel, ground users are served by one AP at a time, and the user scheduling constraint satisfies

[0034] Step 3-2. System resources are limited, and the AP's transmit power cannot exceed its maximum transmit power limit. The constraints are as follows:

[0035]

[0036] Among them, P max , max are the maximum transmit powers of AP and UAV respectively;

[0037] Step 3-3: Maximize the minimum downlink communication rate of users under the joint service communication model, combine user scheduling constraints, drone deployment constraints, and power allocation constraints, and construct a resource allocation optimization problem for the traditional decellularized MIMO system assisted by drones to achieve blind spot-free coverage of users in the system. The optimization problem is as follows:

[0038]

[0039]

[0040] Among them, (1) and (2) in formula (8) are constraints on user scheduling, constraint (3) is the restriction on the deployment location of the UAV, (4) and (5) are constraints on power allocation, and S af The maximum safe distance between multiple drones.

[0041] Furthermore, the step 4 specifically includes:

[0042] Step 4-1, the above optimization problem (8) is a mixed integer non-convex problem, and the block coordinate descent and continuous convex approximation technology are used to transform the non-convex problem into a convex problem for iterative solution;

[0043] Step 4-2: First, adjust the binary variable of user scheduling to a continuous variable of 0 to 1. According to the optimization variables χ, q, and P(p,ρ) contained in the problem, use the block coordinate descent method to decompose the new non-convex problem into three sub-problems: user scheduling (χ), aerial AP deployment (q), and power allocation (P(p,ρ)). Based on the continuous convex optimization technology, optimize the aerial AP deployment for a given user scheduling and power allocation by optimizing the user scheduling, and combine the variable fast decomposition to optimize the complex power allocation for a given user scheduling and aerial AP deployment.

[0044] Step 4-3: Let the problem of maximizing the minimum user rate be a function Given the UAV deployment location and power allocation, the user scheduling convex optimization problem is:

[0045]

[0046]

[0047] in, for The first-order Taylor expansion upper bound of the expanded interference term is that in each iteration, the original function is approximated by a more tractable function at a given local point, λ r is the target result in the rth iteration, χ r is defined as the user result obtained in the rth iteration, then

[0048]

[0049] When q, p, ρ are given, the problem is a convex problem and is solved by the solver;

[0050] Step 4-4: For given χ, p, ρ and any given local point q r , the convex optimization problem of aerial AP deployment is:

[0051]

[0052]

[0053] in, YesR u The scaled value of They are The first-order coefficients and constants of is the aerial AP deployment result obtained in the rth iteration;

[0054] Step 4-5: Introduce a variable block p for representing the transmission power at AP point m m ,m=1...M, with p -m represents the set of other variable blocks except m variable blocks; for given χ, q, the power allocation convex optimization problem is:

[0055]

[0056]

[0057] in, They are The first-order Taylor expansion upper bound;

[0058] Step 4-6: The feasible set of each subproblem is a subset of the feasible set, then

[0059] λ(χ r ,q r ,{p r , ρ r})<λ(χ r+1 ,q r+1 ,{p r+1 , ρ r+1})

[0060] Therefore, the problem-solving process is convergent. Therefore, the resource optimization method of the drone-assisted decellularized massive MIMO system proposed in the present invention improves the system coverage performance and user communication performance through user scheduling, drone deployment, power allocation and other solutions.

[0061] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages:

[0062] 1. This method adds aerial AP point UAV to the traditional decellularized massive MIMO to improve the system's wireless coverage performance;

[0063] 2. This method can reasonably arrange the deployment of UAVs, preventing random switching of user connections and waste of system resources;

[0064] 3. This method designs a fair resource allocation plan, including user scheduling plan, drone location deployment plan, and user power allocation plan. The iterative solution between each plan can ensure the rationality and stability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 Flowchart of the resource optimization method for UAV-assisted decellularized massive MIMO system.

[0066] Figure 2 System model diagram for resource optimization method based on drone-assisted decellularized massive MIMO system. DETAILED DESCRIPTION

[0067] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0068] A resource optimization method for a drone-assisted decellularized massive MIMO system, such as Figure 1 As shown, the following steps are included:

[0069] Since each ground AP and CPU are deployed using long cables, due to the limitation of cable length, remote users cannot access the AP. The easily deployable UAV as an aerial AP can provide services for users in blind areas, but the addition of UAV increases the system resource burden to a certain extent. In addition, if the UAV is not deployed properly, it will interfere with other ground users and affect the overall stability of the system. Therefore, in the following section, the user scheduling scheme and joint communication model are designed from the perspective of efficient and fair resource allocation to ensure that edge users can connect and communicate normally.

[0070] Step 101: Design a user scheduling rule based on joint service of ground AP and aerial AP.

[0071] Step 101-1. Assume that the AP selection method is to connect multiple APs within the service range for each user at a time, using binary variables {χ mk , χ uk} respectively represent the user's scheduling situation, and define the user scheduling matrix If user K is served by a ground AP point, then the binary variable χ mk The coefficient is 1, otherwise it is 0, that is,

[0072]

[0073] If the user is served by an aerial AP point, the binary variable χ uk =1, otherwise 0, that is

[0074]

[0075] Step 101-2. To achieve the synergy between aerial and ground AP points, further considering resource heterogeneity and the limitations of drone deployment, consider the following access rules: During each resource allocation period, 1) each user can only communicate with a single aerial or ground service point at a time; 2) the ground AP can provide services to all users within a limited range; 3) the aerial AP provides services to users within its wireless coverage range. These principles are expressed as binary constraints:

[0076]

[0077]

[0078]

[0079] The set of users and ground APs are represented by K and M respectively. represents the horizontal coordinate of the kth user, w a Indicates the horizontal coordinate of the ground AP, represents the horizontal position of the UAV u, R a , R u Respectively represent the service range of ground AP and aerial AP, R a Affected by the cable length, R u It is determined by the drone’s flight altitude and battery level.

[0080] Step 102: The large-scale fading coefficient of ground communication is mainly affected by path loss and shadow fading. The communication between UAV and ground users is generally line-of-sight propagation. The large-scale fading is related to the UAV height and the angle with the ground user. According to the above user scheduling rules, the user equipment only receives one service at a time. Therefore, the following design is a communication model of joint services affected by user scheduling rules.

[0081] Step 102-1. The channel model combines the effects of small-scale fading and large-scale fading. Let g mk ∈￡ represents the channel gain between the mth AP and the kth user, and the channel transmission model is Among them, β mk 、h mk They represent the large-scale fading coefficient and small-scale fading factor between the m-th AP and the k-th user, respectively, where the small-scale fading is Rayleigh fading, i.e., h mk ~CN(0,1), large-scale fading coefficient β of ground AP point mk Modeled as the product of path loss and shadow fading, β mk =pl mk ·s mk Among them, s mk represents the lognormal shading, pl mk The three-stage path loss is expressed as:

[0082]

[0083] Where L is a constant that depends on the carrier frequency, user and AP heights, and d mk is the horizontal distance between AP m and user k, d0 and d1 are reference value distances;.

[0084] Step 102-2. For the communication link between the aerial AP point and the user, the channel gain can be obtained in a similar manner, that is,

[0085]

[0086] Where a1 and a2 are the path loss exponents of line-of-sight (LoS) and non-line-of-sight (NLoS), respectively. are the probabilities of LoS links and NLoS links, respectively. is the Euclidean distance between the u-th UAV and the k-th user, and H is the fixed flight altitude of the UAV. When designing a communication system based on aerial AP points, the randomness of LoS and NLoS links should be considered. The LoS transmission probability between the u-th aerial AP and the k-th user is

[0087]

[0088] Where a and b are environmental parameters, is the elevation angle between the kth user and the uth aerial AP, so the LoS probability model will become larger as the UAV altitude and elevation angle increase. Further, the average large-scale fading between the kth user and the uth aerial AP can be expressed as

[0089] Since the time scale of aerial AP deployment is much larger than the channel coherence time, leakage interference, network metrics and practical constraints should be derived in a large-scale form, i.e., E{g uk}=β uk .

[0090] Step 102-3. Assuming that the channel conditions are known, the downlink uses conjugate beamforming to send signals to the user. Therefore, the received signal of user k under user scheduling is where x ik Indicates the signal sent by the AP. Specifically, if the user is served by a ground AP, then

[0091]

[0092] Among them, p mk s is the power allocation parameter for the terrestrial AP in the downlink of the cellular massive MIMO. k The signal transmitted to the user satisfies E{|s k | 2}=1,ω k is additive white Gaussian noise (AWGN).

[0093] Assume that the aerial AP only provides business services to users within the communication coverage area and does not interfere with other users' signals. When the aerial AP is hovering at a fixed position, the signal received by users within its communication range is

[0094]

[0095] Among them, ρ uk Assign parameters to power.

[0096] The entire received signal of the user is expressed as

[0097] Step 103: The scheduling of aerial APs mainly provides services for long-distance users. UAVs should be deployed between multiple long-distance users, and service resources should not be preempted by close-range users. In addition, UAVs should maintain a safe distance to avoid collisions. In this case, a fair resource allocation scheme needs to be designed to maximize user coverage using limited resources.

[0098] Step 103-1. Assume that the ground APs share channels, the air APs share channels, and there is no interference between the two. The ground users are served by only one AP at a time. Therefore, user scheduling should satisfy

[0099] Step 103-2. The deployment of aerial APs is limited to a certain extent by the geographical environment. Multiple aerial APs should maintain a safe physical distance between them.j -q u ||≤S af ,u≠j. System resources are limited,

[0100] The AP's transmit power cannot exceed its maximum transmit power limit and must meet the following constraints:

[0101]

[0102] Among them, P max , max are the maximum transmit powers of AP and UAV respectively;

[0103] Step 103-3. Maximize the user downlink communication rate under the above communication model, combine the user scheduling constraints, drone deployment constraints and power allocation constraints, and construct a resource allocation optimization problem for the traditional decellularized MIMO system assisted by drones to achieve blind spot-free coverage of users in the system. The optimization problem is as follows:

[0104]

[0105]

[0106] Among them, (1)(2) are constraints on user scheduling, constraint (3) is a restriction on the deployment location of UAVs, and (4)(5) are constraints on power allocation.

[0107] Step 104: Verify the feasibility of the proposed UAV-assisted model through an optimization solution method.

[0108] Step 104-1. Since user scheduling and associated optimization variables are binary, and the drone deployment variable q and the transmit power variable are non-convex constraints, the optimization problem is a mixed integer non-convex problem. Block coordinate descent and continuous convex approximation techniques are used to transform the non-convex problem into a convex problem for iterative solution. Specifically, based on the continuous convex optimization technology, the user scheduling is optimized for a given aerial AP deployment and power allocation, and the aerial AP deployment is optimized for a given user scheduling and power allocation. The complex power allocation is optimized for a given user scheduling and aerial AP deployment by combining variable fast decomposition.

[0109] Step 104-2. To make the problem easier to handle, first relax the binary variable of user scheduling to a continuous variable of 0 to 1, and further use the block coordinate descent method to decompose the new non-convex problem into three sub-problems: user scheduling, aerial AP deployment, and power allocation.

[0110] Step 104-3. Let the problem of maximizing the minimum user rate be a function Under the given UAV deployment location and power allocation, the convex optimization problem obtained by the user scheduling optimization process is:

[0111]

[0112]

[0113] in, for The first-order Taylor expansion upper bound of the expansion interference term is as follows: in each iteration, the original function is approximated by a more tractable function at a given local point, and χ r Defined as the user result obtained in the rth iteration, is the target value in the rth solution, then

[0114]

[0115] When q, p, ρ are given, the problem is convex and can be solved by the solver.

[0116] Step 104-4. For given x, p, ρ and any given local point q r , the original aerial AP deployment problem is approximated as follows:

[0117]

[0118]

[0119] in, YesR u The scaled value of They are The first-order coefficients and constants of is the aerial AP deployment result obtained in the rth iteration.

[0120] Step 104-5. The power optimization problem at the ground AP is more complicated, so a variable block p is introduced to represent the transmission power at AP point m. m ,m=1...M, with p -m represents the set of other variable blocks except m variable blocks. For given χ, q, the approximate convex problem of the original power optimization problem is:

[0121]

[0122]

[0123] in, They are First-order Taylor expansion upper bound.

[0124] Step 104-5. In the classic block coordinate descent method, the subproblem of updating each variable block needs to be solved accurately with optimality in each iteration, and user scheduling, deployment optimization, and transmit power optimization are all lower bounds of the overall constraint function, that is, the following three formulas are satisfied:

[0125] λ(χ r ,q r ,{p r , ρ r})<λ(χ r+1 ,q r ,{p r , ρ r})

[0126] <λ(χ r+1 ,q r+1 ,{p r , ρ r})

[0127] <λ(χ r+1 ,q r+1 ,{p r+1 , ρ r+1})

[0128] Therefore, the problem-solving process is convergent. Therefore, the resource optimization method of the drone-assisted de-cellular large-scale MIMO system proposed in the present invention achieves the improvement of system coverage performance and user communication performance through user scheduling, drone deployment, power allocation and other schemes.

Claims

1. A resource optimization method for a drone-assisted decellularized massive MIMO system, characterized in that: Deploy drones as aerial APs in the target area, build a joint service communication model by performing the following steps, and optimize the configuration of communication resources based on the model; Step 1: Deploy drones as aerial APs in the target area to provide services for users in signal coverage blind areas, and design a user scheduling solution for joint services of ground APs and aerial APs; The aerial AP provides services to users within the wireless coverage area of ​​the target area, which is expressed as a binary constraint: in, represents the horizontal coordinate of the kth user, w a Indicates the horizontal coordinate of the ground AP, represents the horizontal position of the UAV u, R a , R u Respectively represent the service range of ground AP and aerial AP, R a Affected by the cable length, R u Determined by the drone's flight altitude and battery level; Step 2: Based on the user scheduling scheme of the joint service of ground AP and aerial AP, a joint service communication model is constructed; In the scenario where the channel conditions are all known, the downlink uses conjugate beamforming to send signals to users; combined with user scheduling, the received signal of user k is where x ik Indicates the signal sent by the AP; if the user is served by a ground AP, then: Among them, p mk s is the power allocation parameter for the terrestrial AP in the downlink of the cellular massive MIMO. k The signal transmitted to the user satisfies E{s k 2 }=1,ω k is additive Gaussian white noise; In the scenario where the aerial AP only provides business services to users within its fixed range, when the aerial AP is hovering at a fixed position, the signal received by users within its communication range is: Among them, ρ uk Allocate parameters for power; The entire received signal of the user is expressed as Step 3: Design a resource allocation scheme including user scheduling, UAV location deployment, and user power allocation to maximize the user minimum downlink rate of the above joint service communication model; Maximize the minimum downlink communication rate of users under the joint service communication model, combine user scheduling constraints, drone deployment constraints, and power allocation constraints, and build a resource allocation optimization problem for the traditional decellularized MIMO system assisted by drones to achieve blind spot-free coverage of users in the system. The optimization problem is as follows: Among them, (1) and (2) in formula (8) are constraints on user scheduling, constraint (3) is the restriction on the deployment location of the UAV, (4) and (5) are constraints on power allocation, S af is the maximum safe distance between multiple drones; Step 4: Verify the feasibility of the joint service communication model through optimization solution method.

2. The resource optimization method for a drone-assisted decellularized massive MIMO system according to claim 1, characterized in that: The step 1 specifically includes: The AP selection method is that the user connects to multiple APs within the service range each time, using a binary variable {χ mk , χ uk } respectively represent the scheduling of ground AP and aerial AP of user k, and define the user scheduling matrix If user K is served by a ground AP point, then the binary variable χ mk The coefficient is 1, otherwise it is 0. If the user is served by an aerial AP point, the binary variable χ uk =1, otherwise 0; according to the resource heterogeneity and the limitations of UAV deployment, the collaborative access rules of aerial and ground AP points are considered: during each resource allocation period, each user communicates with a single service point in the air or on the ground; the ground AP provides services for all users in the target area; the aerial AP provides services for users within the wireless coverage of the target area, with binary constraints.

3. The resource optimization method for a drone-assisted decellularized massive MIMO system according to claim 1, characterized in that: The step 2 specifically includes: Step 2-1: Design a channel model that combines the effects of small-scale fading and large-scale fading. represents the channel gain between the mth AP and the kth user, and the channel transmission model is Among them, β mk 、h mk They represent the large-scale fading coefficient and small-scale fading factor between the m-th AP and the k-th user, respectively, where the small-scale fading is Rayleigh fading, i.e., h mk ~CN(0,1), large-scale fading coefficient β of ground AP point mk Modeled as the product of path loss and shadow fading, β mk =pl mk ·s mk , where s mk represents the lognormal shading, pl mk The three-stage path loss is expressed as: Where L is a constant that depends on the carrier frequency, user and AP heights, and d mk is the horizontal distance between APm and user k, d0 and d1 are the reference value distances; Step 2-2: The communication link between the aerial AP and the user is expressed as: Where a1 and a2 are the path loss indices of line-of-sight LoS and non-line-of-sight NLoS, respectively. are the probabilities of LoS links and NLoS links, respectively. is the Euclidean distance between the u-th UAV and the k-th user, and H is the fixed flight altitude of the UAV. When designing a communication system based on aerial AP points, considering the randomness of LoS and NLoS links, the LoS transmission probability between the u-th aerial AP and the k-th user is: Where a and b are environmental parameters, and U is the set of aerial APs. is the elevation angle between the kth user and the uth aerial AP. The LoS probability model will increase with the increase of AP height and elevation angle. The average large-scale fading between the kth user and the uth aerial AP is expressed as The channel model is derived in large-scale form as E{g uk }=β uk ; Step 2-3: When the channel conditions are known, the downlink uses conjugate beamforming to send signals to the user.

4. The resource optimization method for a drone-assisted decellularized massive MIMO system according to claim 1, characterized in that: The step 3 specifically includes: Step 3-1: User scheduling scheme based on joint service of ground AP and aerial AP. In the scenario where ground AP shares channel and aerial AP shares channel, ground users are served by one AP at a time, and the user scheduling constraint satisfies Step 3-2. Constrain the AP's transmit power as follows: Among them, P max , max are the maximum transmit powers of AP and UAV respectively; Step 3-3: Maximize the minimum downlink communication rate of users under the joint service communication model, combine user scheduling constraints, drone deployment constraints, and power allocation constraints, and construct the resource allocation optimization problem of the traditional decellularized MIMO system assisted by drones to achieve blind spot-free coverage of users in the system.

5. The resource optimization method for a drone-assisted decellularized massive MIMO system according to claim 4, characterized in that: The step 4 specifically includes: Step 4-1, the above optimization problem (8) is a mixed integer non-convex problem, and the block coordinate descent and continuous convex approximation technology are used to transform the non-convex problem into a convex problem for iterative solution; Step 4-2: First, adjust the binary variable of user scheduling to a continuous variable of 0 to 1. According to the optimization variables χ, q, P(p,ρ) contained in the problem, use the block coordinate descent method to decompose the new non-convex problem into three sub-problems: user scheduling (χ), aerial AP deployment (q), and power allocation P(p,ρ). Based on the continuous convex optimization technology, optimize the aerial AP deployment for a given aerial AP deployment and power allocation by optimizing the user scheduling, and optimize the complex power allocation for a given user scheduling and aerial AP deployment by combining variable fast decomposition. Step 4-3: Let the problem of maximizing the minimum user rate be a function Given the UAV deployment location and power allocation, the user scheduling convex optimization problem is: in, for The first-order Taylor expansion upper bound of the expanded interference term is that in each iteration, the original function is approximated by a more tractable function at a given local point, λ r is the target result in the rth iteration, χ r is defined as the user result obtained in the rth iteration, then When q, p, ρ are given, the problem is a convex problem and is solved by the solver; Step 4-4: For given χ, p, ρ and any given local point q r , the convex optimization problem of aerial AP deployment is: in, YesR u The scaled value of They are The first-order coefficients and constants of is the aerial AP deployment result obtained in the rth iteration; Step 4-5: Introduce a variable block p for representing the transmission power at AP point m m ,m=1...M, with p -m represents the set of other variable blocks except m variable blocks; for given χ, q, the power allocation convex optimization problem is: in, They are The first-order Taylor expansion upper bound; Step 4-6: The feasible set of each subproblem is a subset of the feasible set, then λ(x r ,q r ,{p r ,r r })<λ(x r+1 ,q r+1 ,{p r+1 ,r r+1 }) The solution process is convergent, and the system coverage performance and user communication performance are improved through user scheduling, drone deployment, and power allocation solutions.

Images