Resource Allocation Method for Multiple UAVs Assisting Macrocell Base Stations in Downlink NOMA Networks

Abstract

A resource allocation method for multiple unmanned aerial vehicle (UAV) - assisted macro - cell base stations in a downlink non - orthogonal multiple access (NOMA) network, comprising the following steps: Step 1, multiple UAV base stations randomly select a location according to their location strategy with probability weights; Step 2, according to the location where the UAV base station is located, perform channel allocation for several users of the UAV base station; Step 3, perform power allocation for two users already allocated under the same channel according to the optimal power allocation strategy; Step 4, based on the power allocation results obtained in Step 3, calculate the utility function values of all UAV base stations at this location, and calculate the selection probabilities of all UAV base stations using the general utility function value; Step 5, repeat Steps 1 to 4. When the selection probability converges to a certain location and no longer changes, the resource allocation is completed. This method can be used in complex scenarios of air - ground heterogeneous wireless NOMA access, with multiple UAV base stations providing dynamic services, wide coverage, reduced co - channel interference, dynamic resource allocation, and improved overall user QoS.

Description

Technical Field

[0001] The present invention relates to the field of communication technologies, and specifically to a method for resource allocation in a downlink NOMA network with multiple unmanned aerial vehicles (UAVs) assisting a macrocell base station. Background Art

[0002] Driven by the sharp increase in the demand for high data rate and low latency applications in the fifth-generation (5G) network, power-domain non-orthogonal multiple access (NOMA) has emerged as a promising multiple access technology. NOMA enables multiple users to transmit simultaneously within the same orthogonal resource, and the receiver uses successive interference cancellation (SIC) technology to decode the received signals. Therefore, compared with traditional orthogonal multiple access (OMA) technology, NOMA can effectively improve the spectral efficiency.

[0003] In recent years, unmanned aerial vehicles (UAVs), or drones, have attracted much attention due to their high mobility and cost-effectiveness. They are widely used to perform complex applications such as cargo delivery, surveillance, traffic monitoring, communication enhancement and restoration. In the field of cellular communication, UAVs are mostly deployed as aerial base stations, or flying base stations, whose positions and trajectories can be optimized to provide a better experience for ground users in a NOMA manner. UAV-assisted macrocell base station communication makes the resource allocation problem including user pairing, power allocation and user scheduling more challenging in the establishment of a radio resource optimization system in an air-ground three-dimensional communication network. It is worth noting that as an auxiliary aerial base station, the change of the UAV's position is related to the achievable data rate of relevant users, and different horizontal positions will bring changes in the interference of co-frequency adjacent base stations. By appropriately controlling the UAV's trajectory and utilizing the potential line-of-sight (LoS) communication between the UAV and ground users, the performance of UAV-based wireless networks in terms of coverage, throughput and energy efficiency can be significantly improved. Therefore, the optimal deployment of UAV positions is becoming a key issue.

[0004] In [1], a distributed matching algorithm for solving the user pairing problem in cognitive radio NOMA networks was proposed. In [2], Zhu et al. studied user pairing in downlink NOMA networks, considered the minimum rate constraint, and proposed a closed-form optimal user pairing solution to maximize the sum rate. In [3], Kang et al. studied the optimal user grouping in downlink NOMA systems to maximize the sum rate, and proposed a low-complexity suboptimal scheme based on swapping. In [4], a distributed user pairing scheme based on group handover was proposed under the premise of considering demand diversity. Most of the existing studies only consider the single BS scenario. Sun et al. studied the joint optimization of multi-user communication scheduling using periodic NOMA and UAV trajectory in [5] to maximize the minimum throughput of ground users. However, the above studies mainly focus on the optimization of the trajectory, altitude, etc. in the single UAV deployment and single UAV scenario.

[0005] Most of the studies in references [6]-[9] focus on the resource allocation problem in a single UAV scenario, paying attention to the energy efficiency of a single UAV and the interference between a small number of base stations. There is a lack of research on multi-dimensional joint optimization of power, coordinate position, and pairing in the complex scenario of multi-UAV networking, and it is impossible to obtain an improvement in the user service quality QoS from a system and global perspective. Therefore, the research on the multi-dimensional resource optimization system of NOMA and multi-UAV assisted communication still lacks exploration.

[0006] [1] W.Liang, Z.Ding, Y.Li, et al., “User pairing for downlink non-orthogonal multiple access networks using matching algorithm,” IEEE Trans. Veh. Technol., vol.65, no.12, pp.5319-5332, Dec. 2017.

[0007] [2] L.Zhu, J.Zhang, Z.Xiao, et al., “Optimal user pairing for downlink nonorthogonal multiple access (NOMA),” IEEE Wireless Commu. Lett., vol.8, no.2, pp.328-331, Apr. 2019.

[0008] [3] J.Kang, I.Kim, “Optimal user grouping for downlink NOMA,” IEEE Wireless Commu. Lett., vol.7, no.5, pp.724 - 727, Oct. 2018。

[0009] [4] Y.Sun, H.Shao, Z.Du, “QoE - oriented resource allocation for downlink non - orthogonal multiple access,” IEEE Commu. Lett., vol.25, no.7, pp.

[0010] 2362 - 2365, Apr. 2021。

[0011] [5] J.Sun, Z.Wang, and Q.Huang, “Cyclical NOMA based UAV - enabled wireless network,” IEEE Access, vol.7, pp.4248C4259, 2019。

[0012] [6] UAV - Aided Air - to - Ground Cooperative Non - Orthogonal Multiple Access。

[0013] [7] W.Mei and R.Zhang, “Uplink cooperative NOMA for cellular connected UAV,” IEEE J.Sel.Areas Commun., vol.13, no.3, pp.644 - 656, Jun. 2019。

[0014] [8] M.Sohail, C.Leow and S.Won, “Energy - efficient non - orthogonal multiple access for UAV communication system,” IEEE Trans.Veh.Technol., vol.68, no.11, pp.10834 - 10845, Nov. 2019。

[0015] [9]A. Bhowmick, S. Roy, S. Kundu, “Throughput maximization of a UAV-assisted CR network with NOMA-based communication and energy harvesting,” IEEE Trans. Veh. Technol., vol. 71, no. 1, pp. 362-374, Oct. 2021。 Summary of the Invention

[0016] To solve the above technical problems, the present invention provides a framework and method for resource allocation in a downlink NOMA air-ground heterogeneous network with multiple UAVs assisting macrocell base stations, realizing multi-dimensional resource optimization of multiple UAV base stations assisting multiple ground macrocells.

[0017] To achieve the above technical objectives, the technical solution adopted is: A method for resource allocation in a downlink NOMA network with multiple UAVs assisting macrocell base stations, including the following steps:

[0018] Step 1, multiple UAV base stations randomly select a location according to their location strategy with probability weights;

[0019] Step 2, according to the location where the UAV base station is located, perform channel allocation for several users of the UAV base station:

[0020] If the number of channels is greater than or equal to the number of users, access in the OMA manner, and one user accesses one channel; if the number of users is greater than the number of channels and less than or equal to 2 times the number of channels, one channel accesses 2 users in the NOMA manner. As the number of allocated users decreases, when the remaining number of users is equal to the remaining number of channels, the remaining users access in the OMA manner;

[0021] If 2 times the number of channels is less than the number of users, pair them two by two in descending order of effective power to achieve access;

[0022] Step 3, perform power allocation for the 2 users already allocated under the same channel according to the optimal power allocation strategy; Step 4, considering the situation of its own utility and the utility of interfering neighbor base stations based on the power allocation result obtained in Step 3, calculate the utility function values of all UAV base stations at this location, and calculate the selection probabilities of all UAV base stations using the general utility function value;

[0023] Step 5, repeat Step 1 to Step 4. When the selection probability converges to a certain location and no longer changes, the channel allocation strategy and power allocation strategy corresponding to this location are optimal, that is, the resource allocation is completed.

[0024] The specific implementation method of the NOMA method is to arrange the users in descending order of actual received effective power The user with the minimum actual received effective power and the user with the maximum actual received effective power are paired to the same channel, the user with the second minimum actual received effective power and the user with the second maximum actual received effective power are paired to the same channel, and so on for allocation.

[0025] The optimal power allocation strategy method is

[0026]

[0027]

[0028] where is the actual received effective power of the weak user m, represents the power allocation ratio of the strong user n, represents the power allocation ratio of the weak user m, and the actual received effective power of the strong user n is greater than or equal to that of the weak user m.

[0029] The method for selecting interfering neighbor base stations is to calculate the normalized interference level between any two base stations BS i and BS j. If the normalized interference level from base station BS i to base station BS j is greater than the interference threshold, and the normalized interference level from base station BS j to base station BS i is less than or equal to the interference threshold, then base station BS j is the neighbor base station interfered by base station BS i.

[0030] The beneficial effects of the present invention are as follows: This method can be used in complex scenarios of air-ground heterogeneous wireless NOMA access. Multiple UAV base stations can provide dynamic services. Through channel allocation, power optimization, NOMA user pairing, and UAV position adjustment, a wide coverage range can be achieved, co-channel interference can be reduced, resources can be dynamically allocated, and the overall user QoS can be improved. The present invention aims at a multi-cell scenario of UAV-assisted ground base stations, and through joint optimization of local multi-dimensional wireless resources, utilizes the characteristics of NOMA and a game-based optimization framework and algorithm to maximize the network-wide rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 is the system model diagram of the present invention;

[0032] Figure 2 is the non-symmetric interference relationship diagram;

[0033] Figure 3 is the 5×5 grid network topology diagram;

[0034] Figure 4 is the convergence performance comparison diagram of the random pairing and head-to-tail schemes;

[0035] Figure 5 is the convergence performance comparison diagram of the 5×5 grid network topologies with different d;

[0036] Figure 6 Comparison graph of the convergence performance of the 5×5 grid network topology for different active users l. Detailed implementation manner

[0037] Different from the single-dimensional optimization in most existing studies on UAV-assisted NOMA networks, the present invention establishes an air-ground wireless communication system combining multiple UAVs and traditional cellular base stations in a downlink NOMA network, realizing multi-dimensional resource optimization of UAV positioning, user scheduling, NOMA user pairing, and power allocation. First, in a cellular scenario with a fixed base station location, user scheduling, user pairing, and power allocation strategies are given. Then, for a multi-cell scenario, the UAV position optimization problem is transformed into a local altruistic game problem, and the effectiveness of this problem is proved. Finally, specific action steps are given.

[0038] In summary, the main contributions of this paper can be summarized as follows:

[0039] (1) Multiple UAVs are involved in our communication system as flying base stations (FBSs). As Figure 1 shown, the system model includes FBSs and a ground base station (GBS). Each flying base station FBS independently determines its own user pairing, power allocation, and user scheduling strategies. In addition, the FBSs need to optimize their positions to obtain higher benefits.

[0040] (2) Given the positions of the FBSs, we derive the optimal power allocation, user scheduling, and user pairing in closed form. The optimal user pairing scheme in a multi-cell scenario is different from that in a single-cell scenario.

[0041] (3) To find the optimal FBSs deployment to maximize the defined network utility, we describe the UAV position optimization problem as a local altruistic game and prove that it is an exact potential game with at least one pure strategy Nash equilibrium (PNE). A distributed learning algorithm is designed to quickly obtain the optimal PNE.

[0042] A method for resource allocation in a downlink NOMA network assisted by multiple UAVs for a macro cellular base station includes the following steps:

[0043] Step 1: Each of the multiple UAV base stations selects a position randomly from its available movable hovering positions.

[0044] A UAV-assisted NOMA air-ground network composed of a total of M traditional ground base stations and multiple UAV base stations (FBSs), that is, an air-ground three-dimensional wireless communication network (system model) jointly composed of high-altitude base stations with UAVs and traditional macro cellular base stations.

[0045] The wireless communication network consists of M1 terrestrial base stations GBS and M2 drone base stations FBS, i.e., M = M1 + M2. The index set of the base stations is denoted as BS = (1, 2,..., M), where M is the total number of base stations. The base station type vector is T. Let any base station be BS i , where T(i) = 0 indicates that the base station BS i is a GBS; otherwise, T(i) = 1 indicates that BS i is an FBS. It can be determined whether the base station is a terrestrial base station or a drone base station according to T(i) = 0 or 1. Whether it is a terrestrial base station or a drone base station, both have K accessible channels and U i pre - served users. The user set of the base station can be denoted as UE i =(1, 2,..., U i ). It is set that K < U i < 3K, indicating that a certain number of users must adopt the NOMA mode, and there may be a situation where after 2 users are connected to each channel, no more users can be served. For the flying base station i, its movable hovering position can be denoted as l i =(x i , y i , z i ) ∈ L i , where |L i | represents the number of movable hovering positions of the flying base station i. That is, the position of the ground station is fixed, with only one definite three - dimensional coordinate. The aerial drone base station can have multiple movable hovering positions at the same time, as shown in Figure 1 .

[0046] Step 2: According to the positions of the drone base stations, allocate channels to several users of the drone base stations.

[0047] [[ID=3,6]]First, the positions of all drone base stations are determined. For a relatively simple system, each drone base station has only two users and K channels.

[0048] For a drone base station with 2K users, without loss of generality, assume that the users are numbered and arranged according to the relationship (actual received effective power relationship). Using m and n to represent the arranged users, the optimal user pairing strategy is

[0049]

[0050] The optimal NOMA pairing method for multiple users belonging to the same base station is to pair them based on their actual received effective power: the user with the lowest actual received effective power is paired with the user with the highest actual received effective power, the user with the second lowest actual received effective power is paired with the user with the second highest actual received effective power, and so on. For example, when K = 2, there are 4 users. After numbering and arranging them according to their actual received effective power, user 4 ≤ user 2 ≤ user 1 ≤ user 3, that is, m = 1, n = 4 can be combined, and user 4 and user 3 can be paired to one channel. If m = 1, n = 4 can be combined, and user 2 and user 3 can be paired to another channel.

[0051] For flying base station BS i, if U m >2K, that is, the number of users to be served exceeds 2 times the number of channels. Without loss of generality, users press The relationships are numbered and arranged, and the optimal user pairing strategy is

[0052]

[0053] Use formula (2) to perform user pairing.

[0054] Therefore, since the number of users that each base station needs to serve is variable, K m <3K, assuming each base station can access K channels. Our generalized pairing method is as follows:

[0055] 1. If the number of users served by a base station is U m ≤K, then access is performed in the general OMA manner, with one user accessing one channel.

[0056] 2. If the number of users served by a base station is K m ≤2K, since the number of channels is less than the number of users, a mixed access method is required, with some users using OMA and some using NOMA. In NOMA, one channel accesses two users, so The actual received effective power is sorted by size. The user with the largest actual received effective power is paired with the user with the smallest actual received effective power to access one channel. The user with the second smallest actual received effective power is paired with the user with the second largest actual received effective power to access one channel. As the number of users decreases, if the number of remaining users equals the number of remaining channels, the remaining users are accessed using OMA, with one user accessing one channel. If the number of users is exactly equal to twice the number of channels, U m =2K, then directly use the method of formula (1) to pair the head and tail.

[0057] 3. If the number of users served by a base station is U m >2K, since the number of channels is much less than the number of users, the effective power​​ The larger 2K users are paired in pairs according to the method of formula (1) to achieve the NOMA mode. According to formula (2), there are still 2K - U m users who cannot be connected.

[0058] Step 3: Perform power allocation on the two users already assigned to the same channel according to the optimal power allocation strategy;

[0059] For a given UAV base station, if the optimal decoding order of the two users m and n that jointly form the NOMA group at this time is Φ i ={m, n}, assuming α n +α m =1, then the optimal power allocation strategy is:

[0060]

[0061] Among them, is the actual received effective power of the weak user m, represents the power allocation ratio for the strong user n, represents the power allocation ratio of the weak user m. The actual received effective power of the strong user n is greater than or equal to the actual received effective power of the weak user m.

[0062] Note that the proposed rule is different from the single - base - station scenario. In the single - base - station NOMA scenario, the decoding order is only based on the channel gain of the users. On the other hand, in the 2 - user NOMA, we can find only related to and the smaller user m.

[0063] Step 4: Considering the situation of its own utility and the combined utility of interfering neighbor base stations, calculate the utility function value of all UAV base stations at this position according to the power allocation result obtained in Step 3. The general utility function value calculates the selection probability of all UAV base stations.

[0064] Next, we use game theory to design a distributed location optimization scheme with local cooperation. Formally, we model the dynamic location deployment game G = [B, {A i} i∈B , {U i} i∈B , where Ui is the utility function of base station i as a game participant; A i represents the set of strategic actions of base station i as a game participant. For the FBS UAV base station, A i = L i and the optional hovering point positions of the UAV, while for the GBS ground base station, since it is fixedly installed, there is only one optional position A i= L i Let a i ∈ A i represent the selection position of participant i, and a -i be the position selection strategy of all participants other than participant i.

[0065] To ensure that a globally optimal FBS position solution can be given, for each FBS, it not only needs to consider its own revenue and find the position strategy that maximizes the sum of the service user rates of the FBS in its position set, but also needs to consider its own position strategy. We define the utility function of participant i in the form of local altruism as:

[0066]

[0067] We can use an iterative method. In each round, first fix the positions of other FBSs. As a game participant, base station i selects the position strategy that maximizes its U i (a i , a -i ). Through several number-theoretic iterations, the optimal strategies for the positions of each FBS base station can be obtained. The above method can be proven to conform to the corresponding rules of an exact potential game and can reach a pure strategy Nash equilibrium.

[0068] Considering its own utility sum, which is the sum of the rates of the users allocated to itself, represents the set of the utility sums of interfering neighbor base stations. The utility sum of an interfering neighbor base station is the sum of the rates of the users accessing the interfering neighbor base station.

[0069] The rate calculation method for users is as follows. Users m and n are a pair of NOMA users in the same base station and on the same channel. The maximum achievable rates of users m and n in base station i can be expressed as:

[0070]

[0071]

[0072] Among them, N represents the noise power. Among them, the power allocation coefficients of the two NOMA users satisfy α n + α m ≤ 1.

[0073] The method for determining whether a neighboring base station is self-interfering is to establish an interference graph model for judgment. We use an asymmetric interference graph to capture the complex interference relationships between various types of base stations. Specifically, we apply a widely used interference metric (IM) to calculate the normalized interference level between any two base stations BS i and BS j, as shown in Equation 5:

[0074]

[0075] |UE j | = U j , representing the number of users served by base station BS j , P i , P j are the transmit powers of the base stations, is the channel gain from base station i at position (l i , l j ) to user m of base station j, m ∈ UE j indicates that user m belongs to the user set of base station j, represents the channel gain from base station j to user m (the user it needs to serve) of base station j, is the statistical average of the interference intensity from base station i to base station j, representing the macroscopic interference level between adjacent base stations. If the two base stations select the same channel or frequency band, it will cause interference. What this patent calculates is whether the average level of the sum of the interferences of neighboring base station j to all users of base station i / the effective power sum of all users' interferences of base station i (excluding the total number of users) exceeds a certain threshold. If it exceeds, it means that base station j interferes with base station i, and there will be an arrow on the interference graph, base station j → base station i. It should be noted that this is not necessarily bilateral interference. We use the interference graph G IM = (BS, ε) to describe the interference situation between various types of base stations, where BS represents the base station nodes of the drone-assisted NOMA air-ground network, and ε represents the set of edges. Each edge represents a pair of nodes with an interference relationship. The construction of the interference graph of the nodes is determined according to the IM value. If is greater than the set interference threshold I th , there is an edge from BS i to BS j, which means there is an interference relationship between BS i and BS j. The topological structure of the interference graph is completely random, and there may be loops. Due to the asymmetric interference, that is, the interference relationship between two adjacent base stations is not symmetric and bidirectional, it is necessary to determine the mutual interference between these two base stations according to the calculation, and use Figure 2 to represent the asymmetric interference graph of network nodes. Each node represents a base station, and each edge represents the corresponding interference relationship. From this, we can see that node 3 interferes with node 2, but conversely, due to its Less than the set interference threshold, Node 2 does not interfere with Node 3.

[0076] Define four interference relationships:

[0077] 1. Calculate whether it is greater than the set interference threshold I th , and find all neighbor sets that interfere with SBS i, indicating the set of neighbor base stations that will interfere with base station i.

[0078] 2. Calculate whether it is greater than the set interference threshold I th , and find all neighbor sets interfered by SBS i, indicating the set of neighbor base stations interfered by base station i.

[0079] 3. If mutual interference can be defined between base stations j and i, and there is a two-way interference edge between the two nodes of the interference graph;

[0080] 4. If it means that there is no co-channel interference between base stations j and i, and there is no interference edge.

[0081] Therefore, to calculate in formula 4, what needs to be found is the neighbor base station that interferes with the UAV base station.

[0082] Finally, we designed a centralized-distributed learning algorithm (CDLA) to achieve the optimal PNE, and the general utility function value calculates the selection probability of all UAV base stations.

[0083] In the k-th iteration, use Ψ Ai (k) to represent the probability distribution of the set of strategic actions A i of FBS i, indicating the selection probability of a certain action a i (k) of the game participant BSi. The probability of the selected BSs updating their behavior follows the Boltzmann-Gibbs rule as follows:

[0084]

[0085] N i represents the neighbor set of i. β is the learning parameter, satisfying β > 0, which reflects the rational level of the player.

[0086] Step 5: Repeat Steps 1 to 4. When the selected probability converges to a certain position and no longer changes, the channel allocation strategy and power allocation strategy corresponding to this position are optimal, that is, the resource allocation is completed.

[0087] In the simulation, we set up a network of 25 base stations, where the ground base stations and the UAV base stations are arranged alternately. Among them, 13 are UAV base stations, and each UAV base station has 9 mobile suspension position coordinates. The simulation scenario is as Figure 3 shown. We evaluate the performance metric of the average network utility by conducting 500 Monte Carlo experiments and comparing different schemes. Figure 4 In, we conduct a comparison of the algorithm convergence performance. It can be seen that compared with the "head-to-tail" scheme and the random scheme, the performance of the proposed scheme has increased by about 5% and 26%.

[0088] In Figure 5 we study the average performance comparison between the proposed scheme and other schemes. d is the distance between base stations, ranging from 200 meters to 400 meters. The base station spacing affects the interference situation in the network (where the number of channels K is 6 and the number of users per base station L is 24). As d increases, the interference between base stations decreases, so a higher network utility value can be obtained. Figure 6 In we show the performance comparison of the proposed scheme by changing the number of active users L in each base station BS (where the number of channels K is 6 and the base station spacing d is 300m). As the number of users L served by each base station increases, through user scheduling in the scheme, the performance gain of our scheme increases from 2% to 7%. In summary, the present invention utilizes the NOMA and UAV-assisted methods, and through multi-dimensional optimizations such as channel allocation, power optimization, NOMA user pairing, and UAV position adjustment, it can achieve the improvement of the user QoS performance in multi-cell wireless communication. The UAV base station wireless resource method can be applied to multiple fields of UAV-assisted communication.

Claims

1. A resource allocation method for multiple UAV-assisted macro base stations in a downlink NOMA network, characterized in that: Step 1: Each of the multiple UAV base stations selects a location. Step 2: According to the locations of the UAV base stations, perform channel allocation for several users of the UAV base stations: If the number of channels is greater than or equal to the number of users, access is carried out in the OMA manner, and one user accesses one channel. If the number of users is greater than the number of channels and less than or equal to twice the number of channels, two users access one channel in the NOMA manner. As the number of allocated users decreases, when the remaining number of users is equal to the remaining number of channels, the remaining users access in the OMA manner. If twice the number of channels is less than the number of users, pairing is achieved from the largest to the smallest in terms of effective power for access. Step 3: Perform power allocation for the two users already allocated under the same channel according to the optimal power allocation strategy. Step 4: Based on the power allocation results obtained in Step 3, considering the situation of its own utility and the combined utility of interfering neighboring base stations, calculate the utility function values of all UAV base stations at this location. The general utility function value calculates the selection probabilities of all UAV base stations. Step 5: Repeat Step 1 to Step 4. When the selection probability converges to a certain location and no longer changes, the channel allocation strategy and power allocation strategy corresponding to this location are optimal, that is, the resource allocation is completed.

2. The method for resource allocation in a downlink NOMA network with multiple UAVs assisting a macro base station as claimed in claim 1, wherein: The specific implementation method of the NOMA manner is to arrange the users in descending order of the actual received effective power. The user with the smallest actual received effective power and the user with the largest actual received effective power are paired to the same channel, the user with the second smallest actual received effective power and the user with the second largest actual received effective power are paired to the same channel, and so on for allocation.

3. The method for resource allocation in a downlink NOMA network with multiple UAVs assisting a macrocell base station according to claim 1, wherein: The optimal power allocation strategy method is Among them, is the actual received effective power of the weak user m, represents the power allocation ratio of the strong user n, represents the power allocation ratio of the weak user m, and the actual received effective power of the strong user n is greater than or equal to the actual received effective power of the weak user m.

4. The method for resource allocation in a downlink NOMA network with multiple UAVs assisting a macro cell base station as claimed in claim 1, wherein: The method for selecting interfering neighboring base stations is to calculate the normalized interference level between any two base stations BSi and BSj. If the normalized interference level from base station BSi to base station BSj is greater than the interference threshold, and the normalized interference level from base station BSj to base station BSi is less than or equal to the interference threshold, then base station BSj is the interfering neighboring base station of base station BSi.

Images