Wireless Resource Allocation Method Based on Downlink Delay in GEO Satellite Communication System

Abstract

The present invention discloses a method for wireless resource allocation based on downlink delay in a GEO satellite communication system. In the method, based on the OFDMA model of downlink transmission of a multi-beam GEO satellite, the satellite antenna gain and channel coefficient are calculated, and a problem model of minimizing the download time based on user data requirements is constructed with the total data download time of all users as the objective function. For the problem model of minimizing the download time based on user data requirements, it is solved by a resource allocation algorithm based on the convex difference relaxation SCA algorithm. During the solution process of the resource allocation algorithm based on the convex difference relaxation SCA algorithm, the initial value of variables is set by an initial value allocation algorithm based on demand weights. The present invention uses an algorithm based on demand weights to allocate initial subcarriers, and proposes a low-complexity SCA algorithm based on convex difference relaxation to optimize the carriers and power, which can better meet the requirement of minimizing the system downlink data download under low complexity.

Description

Technical Field

[0001] The present invention relates to the field of communication technologies, and particularly to a method for allocating radio resources based on downlink delay in a GEO satellite communication system. Background Art

[0002] With the continuous development of communication technologies, the number of users and base stations in satellite communication systems is increasing, and the data demand is also growing. Users in satellite communication systems can be divided into real-time users and non-real-time users. Among them, real-time users usually require high-reliability and low-latency communication services, such as voice calls and video calls, and the system needs to ensure the downlink rate index of real-time users. The service requirements of non-real-time users are often file downloads, large data transmissions, etc., and the requirements for real-time performance are low. Some users need to obtain data through long-term downloads. At this time, the satellite radio resource allocation method will affect the time for users to download data from the satellite, and delay has also become an important indicator for measuring the performance of communication systems. Due to different service scenarios such as the channel conditions, antenna conditions, and required data volumes of users, various radio resources on the satellite need to be flexibly allocated to meet the diverse needs of different users.

[0003] In traditional problems with the following line rate as the objective function, resources such as power and frequency are often used as optimization variables. However, the impact of the total data volume of user requirements on resource allocation is not considered. Many problems with the data download delay as the objective function also do not consider the impact of user demand data volume on optimization. In the scenario of the combination of satellite communication, edge computing, and the Internet of Things, the delay minimization problem can better improve user satisfaction, reduce network congestion, and improve system efficiency. In the existing technology, there is a problem of computing offloading based on queuing theory in the satellite edge computing scenario, and a game framework for computing offloading is designed, with the corresponding time and energy consumption as the objectives, but the resource allocation at the beam level is not considered. In other existing technologies, the constructed models and objective functions are different. For example: (1) A GEO-LEO two-layer satellite communication structure is constructed to minimize the download delay. The communication requirements of devices can be collaboratively computed by LEO satellites or forwarded by GEO satellites to ground gateways, making full use of the characteristics of different satellites for communication, and using deep reinforcement learning and convex optimization methods to solve two sub-problems. (2) For a delay-oriented space-air-ground communication network architecture, with the minimization of the computing offloading time of tasks as the objective function, the online scheduling problem is expressed as a Markov decision process with energy constraints, and a reinforcement learning algorithm is used to solve it. (3) In the satellite hopping beam scenario, three objectives of delay minimization, throughput maximization, and fairness between beams are optimized, and they are weighted and added to form a single objective function for solution. (4) The resource allocation problem in the scenario of multiple satellites and multiple gateways is considered, and the task offloading decision variables, computing, and communication resources are optimized simultaneously. Finally, a single variable composed of delay and energy consumption is solved. (5) Assuming that data streams can be separated during system transmission, a joint routing selection and time slot allocation algorithm is proposed to minimize the total power consumption of the system. (6) Considering the demand situation of each user in a multi-beam satellite, user scheduling and resource allocation between beams are carried out, with the data download delay as the optimization objective. However, the resource allocation in the frequency domain is not considered, and it is required that the power can only be selected between two discrete values.

[0004] The following problems have not been solved in the existing research: When there are multiple users in the satellite communication scenario who need to download data, the data download time will greatly affect the system operation efficiency. However, due to the different data volume requirements and channel conditions of each user, a more intelligent resource allocation algorithm is needed to flexibly schedule satellite resources. Summary of the Invention

[0005] Technical Objective: Aiming at the defects in the prior art, the present invention discloses a method for wireless resource allocation based on downlink delay in a GEO satellite communication system. In the GEO satellite downlink OFDMA communication system, based on the joint allocation of downlink power and carriers, an algorithm based on demand weights is used to allocate initial subcarriers, and a low-complexity Successive Convex Approximation (SCA) algorithm based on convex difference relaxation is proposed to optimize the carriers and power. Compared with baseline algorithms such as fixed access and branch and bound method, the method described in this patent can better meet the requirement of minimizing the system downlink data download at low complexity.

[0006] Technical Solution: To achieve the above technical objective, the present invention adopts the following technical solutions.

[0007] A method for wireless resource allocation based on downlink delay in a GEO satellite communication system, comprising the following steps:

[0008] S1. Based on the OFDMA model of multi-beam GEO satellite downlink transmission, calculate the satellite antenna gain and channel coefficients, initialize the resource allocation mode and estimate the data download time, and construct a model for the shortest download time problem based on user data requirements with the total data download time of all users as the objective function;

[0009] S2. For the model of the shortest download time problem based on user data requirements, solve it through a resource allocation algorithm based on the Successive Convex Approximation (SCA) algorithm with convex difference relaxation;

[0010] S3. During the solution process of the resource allocation algorithm based on the Successive Convex Approximation (SCA) algorithm with convex difference relaxation, set the initial values of variables through an initial value allocation algorithm based on demand weights.

[0011] Beneficial Effects: The present invention establishes a multi-beam GEO satellite downlink OFDMA communication system, allocates resources in combination with the data download volume requirements of different users, so that the data download time is the shortest, and at the same time optimizes the carrier allocation variables and power allocation variables. To handle the mixed optimization problem, the big M method and the CCCP algorithm are used to transform the mixed optimization problem into a continuous problem for solution, a low-computation-complexity Successive Convex Approximation (SCA) algorithm based on convex difference relaxation is introduced by introducing slack variables, and an algorithm based on demand weights is used to allocate initial subcarriers. The simulation results show that compared with the baseline algorithm, the algorithm of this patent can better meet the requirement of minimizing the system downlink data download at low complexity, improve the system efficiency, and the carrier initial allocation method proposed in this patent can also find better initial values and improve the optimization convergence speed. Description of the Drawings

[0012] Figure 1 It is the flowchart of the method of the present invention;

[0013] Figure 2 It is the schematic diagram of the topological structure of the OFDMA model of multi-beam GEO satellite downlink transmission of the present invention;

[0014] Figure 3 It is the beam distribution diagram of the satellite during the simulation process;

[0015] Figure 4 It is the convergence diagram of the objective function and the LEO satellite data download time during the simulation process;

[0016] Figure 5 It is the diagram of the change in data download time when the rated power and the number of ground users change during the simulation process;

[0017] Figure 6 It is the diagram of the change in data download time when the average data volume requirement and the rated power change during the simulation process;

[0018] Figure 7 It is the diagram of the change in data download time with different sub - carrier numbers under different algorithm scenarios during the simulation process;

[0019] Figure 8 It is the diagram of the change in data download time with different numbers of users under different algorithm scenarios during the simulation process. Specific implementation manner

[0020] In order to enable those skilled in the art of this technology to better understand the solution of this application, the following will clearly and completely describe the technical solutions in the embodiments of this application with reference to the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of this application.

[0021] As shown in the attached Figure 1 A wireless resource allocation method based on downlink delay for a GEO satellite communication system in this embodiment includes the following steps:

[0022] S1. Based on the OFDMA model of the downlink transmission of a multi - beam GEO satellite, calculate the satellite antenna gain and the channel coefficient, initialize the resource allocation method and the data download time estimation, and construct a problem model with the shortest download time based on the user data requirements, taking the total data download time of all users as the objective function;

[0023] There are a total of M users within the coverage of N beams of the GEO satellite. The distance from the projection center of the nth beam on the ground to the mth user on the ground is d nm , 1 ≤ n ≤ N, 1 ≤ m ≤ M;

[0024] The coverage areas of the N beams of the satellite are dispersed on the ground and jointly use the full frequency band resources of the satellite. One beam serves multiple users within its coverage area and uses the OFDMA method for resource allocation. The frequency band with a total bandwidth of B is divided into K subcarriers, and the bandwidth of each subcarrier B0 = B / K. Each user within the beam uses one subcarrier, and there is no interference among the users within the beam. Assume that the data download requirement of the m-th user is D m , and D m > 0, with the unit of bit. The model topology structure diagram is as shown in Figure 2 . The GEO satellite is the GEO satellite, Beam is the beam, and user is the user. Among them, f1, f2, etc. represent different subcarrier frequencies. Frequency division multiplexing resources are used within the same beam, and there is frequency aliasing between different beams.

[0025] Assume that there are a total of M users within the coverage areas of the N beams of the GEO satellite. The users can be distributed on land or in the ocean and have different download requirements. Each user is equipped with an isotropic omnidirectional transmitting antenna pointing in the direction of the satellite, and each user accesses the beam closest to it. Assume that the distance from the projection center of the n-th beam of the GEO satellite on the ground to the m-th user on the ground is d nm , 1 ≤ n ≤ N; then the beam selected by this user is the n-th beam that minimizes d nm . Assume that the set of users served by the n-th beam of the GEO satellite is denoted as Then

[0026]

[0027] Among them, represents the number of elements in

[0028] Let the beam gain from the n-th beam of the GEO satellite to the m-th user on the ground be denoted as G(θ nm ), where θ nm is the off-axis angle between the m-th user and the n-th beam, that is, the angle between the main axis direction of the antenna of the n-th beam of the GEO satellite and the line connecting the m-th user and the satellite. The calculation formula for G(θ nm ) is:

[0029] G(θ nm ) = G max (J1(s nm ) / 2s nm + 36J3(s nm ) / (s nm ) 3 ) 2 (2)

[0030] Among them, s nm= 2.07123·sinθ nm / sinθ 3dB ; G max is the maximum antenna gain, θ 3dB is the 3dB power angle of the GEO satellite, and J1(·) and J3(·) are the first-order and third-order Bessel functions of the first kind, respectively.

[0031] Due to the adoption of OFDMA technology, each subcarrier is orthogonal to each other. The channel from the nth beam to the mth user on the kth subcarrier on the ground is expressed as follows

[0032]

[0033] where G R is the receiving gain of the user, l mk is the path loss from the kth subcarrier of the satellite to the mth user, and the calculation formula is: l mk = (4πd m / λ k ) 2 , d m is the distance from the GEO satellite to the mth user, and λ k is the wavelength of the kth subcarrier.

[0034] Initialization of resource allocation method and estimation of data download time:

[0035] In the beam of the GEO satellite, the bandwidth and frequency-domain resources are allocated by OFDMA. Therefore, within each scheduling time slot, each user within a beam can access multiple subcarriers, but a subcarrier can only be used by one user within the beam. When the mth user connects to the kth subcarrier on the nth beam of the GEO satellite, let the indicator variable otherwise Define the following notations

[0036]

[0037] where, 1 ≤ k ≤ K;

[0038] Due to the characteristics of OFDMA, a beam can only be connected to one user, and the following constraints can be obtained:

[0039]

[0040] In addition, only when the mth user accesses the nth beam can it use the subcarriers therein, can it be non-zero, that is, there are the following constraint conditions:

[0041]

[0042] On each sub - carrier, the beam can allocate corresponding power resources. The power obtained by the \(m\) - th user from the \(k\) - th sub - carrier of the \(n\) - th beam of the satellite is denoted as Define the following notations:

[0043]

[0044] where \(p\) (k) is the power of the \(k\) - th sub - carrier, \(P\) is the power matrix of all sub - carriers, with size \(N\times KM\);

[0045] To ensure the upper limit of system resources, the power should also satisfy the following constraint conditions:

[0046]

[0047] where \(P\) 0 is the rated transmission power of the satellite beam, and \(P\) T is the total rated transmission power of the satellite.

[0048] Only when the \(k\) - th sub - carrier of the \(n\) - th beam is allocated power can this sub - carrier be selected for access. It can be represented by the sign function \(sgn(\cdot)\). The relationship between and

[0049]

[0050] Define the following notations:

[0051]

[0052] where \(c\) (k) is the access indication for the \(m\) - th user to access the \(k\) - th sub - carrier of the \(n\) - th beam, and \(C\) is the access indication matrix for all users;

[0053] The signal - to - interference - plus - noise ratio (SINR) received by the \(m\) - th user from the \(k\) - th sub - carrier of its \(n\) - th beam can be expressed as The SINR of the \(m\) - th user served by the GEO satellite beam on the \(k\) - th sub - carrier of the \(n\) - th beam can be expressed by formula (9).

[0054]

[0055] where \(n'\) is the beam index from 1 to \(N\) (not equal to \(n\)), \(m'\) refers to the \(m'\) - th user, and \(m'\neq m\), \(U\) n′ is the set of all users corresponding to beam \(n'\); \(\sigma\) is the noise energy;

[0056] Then the instantaneous achievable downlink rate \(r\) obtained by the \(m\)th user from the \(n\)th beam of the GEO satellite nm is as follows:

[0057]

[0058] where is the instantaneous downlink rate of the \(m\)th user from the \(n\)th beam and the \(k\)th subcarrier of the GEO satellite;

[0059] Assume that for the \(m\)th user, it is within the coverage of the \(n\)th beam, that is Then at this time for this determined \(m\), only For the remaining \(n'\neq n\), According to formula (9), it can be known that: According to formula (10), it can be known that \(r\) n′m = 0, only \(r\) nm \(\neq 0\), then the instantaneous achievable downlink rate \(r\) obtained by the \(m\)th user from the GEO satellite m can be expressed as:

[0060]

[0061] The data download time \(t\) of the \(m\)th user m can be divided into two parts. One part is the time for the electromagnetic wave signal to be sent from the satellite to the user, and the other part is the time to connect to the corresponding beam and download the required data through the instantaneous achievable downlink rate \(r\) m to download the required data.

[0062]

[0063] where \(v\) light is the speed of light;

[0064] At this time, the total data download time of all users is \(t\), and the calculation formula is:

[0065]

[0066] Construct a problem model for the shortest download time based on user data requirements:

[0067] In order to minimize the total data download time \(t\) of all users on the ground, let \(t\) be the objective function. When optimizing the optimization variables \(P\) and \(C\), make \(t\) reach the minimum value. Since the objective function in this optimization problem has considered the fairness factor among different users, the minimum signal-to-interference-plus-noise ratio of users is no longer restricted. The problem model for the shortest download time based on user data requirements can be expressed as:

[0068]

[0069] The meaning of the first constraint condition C1 is that the transmission power of the GEO satellite must be greater than or equal to zero. The meaning of the second constraint condition C2 is that the total transmission power of each beam of the satellite cannot exceed the rated power of the beam. The meaning of the third constraint condition C3 is that the total power of all beams of the satellite does not exceed the maximum rated total transmission power of this GEO satellite. The meaning of the fourth constraint condition C4 is that each subcarrier in a beam can be connected to at most one user. The meaning of the fifth constraint condition C5 is that the sign function of the power on the subcarriers of each beam is the subcarrier allocation variable. The meaning of the sixth constraint condition C6 is that if a user is within the coverage of a certain beam, the subcarrier allocation variable can be 0 or 1 at this time. The meaning of the seventh constraint condition C7 is that if a user is not within the coverage of a certain beam, the subcarrier allocation variable can only be 0.

[0070] S2. For the problem model of the shortest download time based on user data requirements, it is solved by a resource allocation algorithm based on the convex difference relaxation SCA algorithm.

[0071] The optimization problem corresponding to formula (14) is a non-convex problem and contains an integer optimization variable C, so this is a mixed integer programming problem, where the objective function is a non-convex function, and constraint C5 contains an integer variable and is also non-convex. After organizing the objective function, we can get:

[0072]

[0073] Among them, the latter term is the total time delay of the electromagnetic wave in the air during the process of each user receiving information transmission. When the user's position is determined, this term is a constant and can be ignored in the objective function. Therefore, at this time, the optimization problem corresponding to formula (14) can be expressed as

[0074]

[0075] Analyzing the above problem, it can be obtained that since there is a signal-to-interference-plus-noise ratio in the denominator of the objective function, the objective function is non-convex. C5 in the constraint conditions is an integer non-convex constraint. To solve this problem, it is necessary to convert the non-convex constraints and non-convex problems into convex constraints and convex problems, and then use the method of convex optimization to process this.

[0076] The resource allocation algorithm based on the convex difference relaxation SCA algorithm includes the subcarrier and power allocation algorithms based on the convex difference relaxation SCA algorithm. The algorithm includes the following content:

[0077] S2.1. Introduce an upper bound relaxation variable to convert the non-convex objective function into a convex function.

[0078] To solve the non-convex problem of the objective function in the optimization problem corresponding to formula (16), an upper bound relaxation variable φ m , φm It can be expressed as Φ = (φ1,..., φ m ) 1×M , where Φ is the upper bound relaxation variable matrix. When the optimization problem converges, the physical meaning of φ m is the data download time of the m-th user. The upper bound relaxation condition can be expressed as:

[0079]

[0080] At this time, the optimization problem corresponding to formula (16) can be transformed into:

[0081]

[0082] Among them, the eighth constraint condition C8 is added. At this time, the objective function in the optimization problem corresponding to formula (18) is a convex function, meeting the conditions of convex optimization.

[0083] S2.2. On the basis that the objective function is a convex function, convert the non-convex constraint conditions into convex functions, including the following steps:

[0084] S2.2.1. Transformation of constraint condition C5:

[0085] Use the big M method and the CCCP algorithm to process the non-convex constraints in constraint condition C5. When using the CCCP algorithm, iterative solution is required. Assume is the value of calculated in the t-th iteration. The transformation using the big M method is as follows:

[0086]

[0087]

[0088]

[0089]

[0090]

[0091]

[0092] At this time, constraint conditions C5.1 and C5.2 are both linear convex functions. Transform constraint condition C5.3 into:

[0093]

[0094] ​​​S2.2.2, Transformation of Constraint Condition C8

[0095] Observing the constraint condition C8 in the optimization problem corresponding to formula (18), it can be seen that both the numerator and denominator of the fraction on the left side of the inequality sign are positive, so φ on the right side of the inequality sign m is also positive. At this time, by transposing, we can get

[0096]

[0097] Introduce the lower bound constraint variable of the rate function such that Define the symbol:

[0098]

[0099] where is the lower bound constraint variable, and α is the subset of the lower bound constraint variables;

[0100] At this time, the constraint condition C8 can be transformed into the constraint condition C8.1 and the constraint condition C8.2, and the calculation formulas are as follows:

[0101]

[0102]

[0103] Among them, the constraint condition C8.1 is a convex constraint. Due to the existence of the SINR term in the constraint condition C8.2, it is not a convex function. After performing common denominator arrangement on it, and since both the noise term σ 2 and the order of magnitude of the channel coefficients are very small, after normalizing them, we can get:

[0104]

[0105] The problem corresponding to the constraint condition C8.2 is a difference of convex (DC) problem, and the CCCP algorithm can be used to solve it. However, the result when taking the derivative of the logarithmic function is very complex, so an SCA algorithm based on convex difference relaxation is proposed to solve it. To avoid taking the derivative of the logarithmic function, introduce two relaxation variables of the logarithmic function The constraint condition C8.2 is transformed into the following formula:

[0106]

[0107] where is the lower bound constraint, is the upper bound constraint, and at the same time, define the symbol:

[0108]

[0109] where is the lower bound constraint, is the upper bound constraint, η is the subset of lower bound constraints, and ξ is the subset of upper bound constraints;

[0110] By transforming the constraint conditions C8.3 and C8.4, we can obtain:

[0111]

[0112] At this time, the constraint condition C8.1 in formula (23), the constraint condition C8.2.1 in formula (26), and the constraint condition C8.3 in formula (29) are all convex constraints. The constraint condition C8.4 in formula (30) is a concave constraint and needs to be processed by the SCA algorithm. Since the problem is arranged into the relevant expression of the exponential function e x , the calculation will be simpler. The exponential function can be obtained through processing:

[0113]

[0114] where x (t) is the previous value of the variable x in the i-th iteration. Substituting it into the constraint condition C8.4 in formula (30), we can get

[0115]

[0116] At this time, C8.4.1 in formula (32) is a linear constraint and also a convex constraint. At this time, C8 in formula (18) is arranged into a convex constraint.

[0117] The constraint condition C8 is transformed into the following formula:

[0118]

[0119] In S2.2, based on the convex objective function, the non-convex constraint conditions are converted into convex functions, and the calculation formula is:

[0120]

[0121] At this time, the optimization problem corresponding to formula (33) is a convex problem, and most of the constraints are linear constraints. The CVX toolbox in MATLAB can be directly used to solve this convex problem. When performing iteration, T (t) is the objective function value calculated in each iteration. Thus, the subcarrier allocation indication for the corresponding user is obtained. The solution results include P, C, Φ, α, η, ξ, that is, all subcarrier power matrices, all user access indication matrices, upper bound relaxation variable matrices, lower bound constraint variable subsets, lower bound constraint subsets, and upper bound constraint subsets.

[0122] S3. In the solution process of the resource allocation algorithm based on the convex difference relaxation SCA algorithm, the initial value setting of variables is performed by the initial value allocation algorithm based on demand weights;

[0123] In this application, the initial values of variables C and ξ need to be iteratively set. Since the method of this application optimizes the data download time of all users, and the objective function is the data demand of each user divided by the information transfer rate, it is possible to first assume that the power is evenly distributed to obtain the initial value of the orthogonal subcarrier allocation variable. Then, according to the initial allocation value of the orthogonal subcarrier, the power of each beam is evenly divided to obtain the initial value of the relaxation variable ξ.

[0124] The initial value setting of variables includes the following steps:

[0125] S3.1. Perform the initial subcarrier allocation based on demand weights;

[0126] Assume that in the optimization problem corresponding to formula (33), the objective function is the total time for all users to download data, expressed as:

[0127]

[0128] where W m is the download time of the m-th user, and the calculation formula of W m is:

[0129]

[0130] Since the download data time of each user needs to be as small as possible, each user must access more than one orthogonal subcarrier to obtain resources. When studying the users inside the n-th beam, the expression of W m for the users inside the beam can be expressed as:

[0131]

[0132] From the above formula, it can be seen that if the data download demand D m of the user is larger, the numerator should be larger to make W m larger, so more subcarriers need to be allocated to this user. Similarly, if the channel condition of the user inside the beam is worse, the SINR will be lower, and more subcarriers also need to be allocated to prevent W m from being too large.

[0133] When initially allocating orthogonal subcarriers, it is assumed that the power on each beam is the rated power, and the power on the sub-bands of each beam is P0 / K. Since the difference in channel coefficients on different subcarriers comes from fading, and within the frequency band, the influence of frequency on fading is small. Therefore, the channel coefficient on the first subcarrier can be calculated for all subcarriers when setting the iterative process, that is, assuming that the rate value on the first subcarrier is the same as that on other carriers, for user comparison. At this time, W m The expression is

[0134]

[0135] Among them, is the channel capacity from the nth beam to the mth user on the ground on the first subcarrier;

[0136] Therefore, the W value of each user within the beam can be calculated according to formula (37) m value, and then the subcarriers are divided according to the size of the W m value. The larger the W m value, the more subcarriers are allocated.

[0137] Assume that there are a total of users connected within the nth beam, which are respectively At this time, the W m values calculated for each user are respectively The subcarriers are divided according to their sizes. Assume that each user is allocated K m subcarriers. At this time, the calculation formula is:

[0138]

[0139] Among them, is the symbol for rounding down. After each user within the beam determines the number of subcarriers according to formula (38), due to the use of rounding down calculation, there will still be several subcarriers not allocated. At this time, the extra subcarriers are allocated one by one to the user with the largest W m value in the order of the size of the W m value. Assume that if:

[0140]

[0141] At this time, the calculation formula for the number K0 of extra subcarriers is:

[0142]

[0143] Then there are K0 users who are allocated one more subcarrier:

[0144]

[0145] After calculating the number of subcarriers allocated to each user, starting from the first subcarrier \(k = 1\), the subcarriers are allocated to the first user until all subcarriers are allocated.

[0146] The algorithm for allocating orthogonal subcarriers described above is shown in Table 1.

[0147] Table 1 Subcarrier Allocation Algorithm Based on Demand Weight

[0148]

[0149]

[0150] S3.2. Initialize the slack variables;

[0151] After allocating the subcarriers, the allocation of the \(P\) variable can be carried out. Assume that the satellite rated power is evenly distributed to each beam, and then the beam is evenly distributed to each subcarrier according to the usage of the subcarriers. According to the subcarrier initialization algorithm, each subcarrier is used initially.

[0152]

[0153] Among them, is the initial value of , is the initial value of ;

[0154] After obtaining the initial value of the \(P\) variable, the initial value \(\xi\) (0) needs to be obtained. The initial value \(\xi\) of \(\xi\) can be directly set according to the constraint conditions in formula (28). (0) The calculation formula is as follows:

[0155]

[0156] Inducing the above algorithm steps, the solution algorithm for the optimization problem corresponding to formula (33) is shown in Table 2.

[0157] Table 2 Carrier and Power Allocation Algorithm Based on Convex Difference Relaxation SCA Algorithm

[0158]

[0159] Simulation verification:

[0160] In this application, the complexity of the scheme is analyzed first. The optimization problem corresponding to formula (33) has \(5NMK + M\) decision variables and \(8NMK+NK + MK+N + 2M + 1\) constraint conditions. Assume that the optimization problem needs to be iterated \(E\) times to achieve convergence. At this time, the algorithm complexity of the problem is as follows. It is a polynomial function and can be solved by a computer in polynomial time.

[0161] ○(E(5NMK + M) 3 (8NMK + NK + MK + N + 2M + 1)) (44)

[0162] It is assumed that the user's location is randomly generated within the coverage of the GEO satellite beam, and the user's data volume requirement is also randomly generated. Without considering the influence of factors such as altitude and climate on the user's coordinates, distance from the satellite, and data volume requirement, it is assumed that the user's antenna always points to the corresponding satellite beam. Each user has a data download volume requirement D m and it is assumed that the user's data download volume requirement D m follows a normal distribution with the unit of Mbit. The antenna parameters and other user parameters are given in Table 3 below.

[0163] Table 3 GEO Satellite Antenna Parameters and User Parameters

[0164]

[0165] The beam position distribution of the GEO satellite antenna is as Figure 3 shown. The GEO satellite is located directly above the middle beam, that is, the satellite sub - point coincides with the beam center of the middle beam. GEO Beam Center represents the GEO satellite beam center.

[0166] The optimization problem corresponding to formula (33) needs to be solved by an iterative optimization method. When solving the optimization problem, the optimization result of the previous step is continuously used as a constant and substituted into the optimization problem, and multiple iterations are required to achieve the final convergence. Here, the number of users M is changed, and the users are randomly and uniformly distributed within the satellite coverage, and the convergence characteristics of the algorithm for different numbers of users are obtained as Figure 4 shown. Figure 4 The solid line in it is the value of the objective function ∑φ in formula (33) obtained by each iterative optimization m , and the dashed line is the downlink data download time of the LEO satellite calculated after each iteration. It can be seen from Figure 4 that during the iterative process, the value of the objective function ∑φ in formula (33) m can well upper - bound the user data download time, indicating that the introduction of the slack variable φ does not change the convergence of the optimization problem itself. When the number of users increases, the iterative convergence speed slows down, and at the same time, the data download time also increases. Since two slack variables are introduced to relax the logarithmic function and then the CCCP algorithm is used, the algorithm convergence speed is relatively fast. Figure 4 It shows that the wireless resource allocation method in this application can achieve convergence relatively quickly within a limited time.

[0167] An experimental simulation was conducted on the relationship between the satellite downlink data download time and different rated powers of GEO satellite beams. When changing the rated power of the GEO satellite beam, the rated power of the GEO satellite was also changed to 7 times the rated power of the beam. Under different numbers of users and different beam rated powers, 100 Monte Carlo simulation experiments were carried out for each set of parameters, and the average value was calculated. The experimental results are as Figure 5 shown. It can be seen from Figure 5 that as the number of users increases, the data download time becomes longer and longer. This is because the overall system demand is increasing while the total satellite resources remain unchanged, resulting in an increase in the data download time. When the beam rated power increases, the data download time continuously decreases. When P 0 increases from 10W to 30W, the data download time decreases most significantly. This is because the rated power of 10W is too low, and the satellite downlink rate is low, unable to meet the user's needs. When the rated power increases from 30W to 100W, the rate of decrease in the data download time becomes slower and slower. When the rated power increases from 100W to 300W, the power increases by 200W, but the system data download time decreases very little. This is because as the rated power increases, the interference also increases continuously, thus affecting the signal-to-interference-plus-noise ratio value and restricting the rapid increase of the satellite downlink rate. In order to better save resources, a trade-off can be made between the data download time and the system energy consumption to achieve better system performance with lower power consumption.

[0168] The user's data download volume requirement directly affects the final data download time. Assuming the number of users remains unchanged, the average values of the user data volume requirements are 6000, 8000, 10000, 12000, and 14000 Mbit respectively. At the same time, the rated power of the beam is changed, and the rated power of the satellite is changed to 7 times the rated power of the beam. Under different demand average values and different beam rated powers, 100 Monte Carlo simulation experiments were carried out for each set of parameters, and the average value was calculated. The experimental results are as Figure 6 shown. It can be seen from Figure 6 that when the average value of the downloaded data volume requirement continuously increases, the data download time also increases continuously. This is because the downlink rate remains unchanged, but the demand increases. When the rated power of the satellite increases, the downlink rate of the GEO satellite will increase, and the data download time will be shortened. However, the continuous increase in power will not make the download time infinitely shortened. This is because when the power increases, the interference will also increase, resulting in a slower and slower increase in the downlink rate.

[0169] For the case of different numbers of orthogonal subcarriers in the OFDMA system, the downlink data download time of the GEO satellite under three algorithm scenarios was compared and analyzed, and the rated power of the satellite beam was taken as 100W.

[0170] Table 4 Comparison of the algorithms in this patent

[0171]

[0172] The optimization situations of the three algorithm scenarios are shown in Table 4. Among them, Proposed refers to the algorithm proposed in this patent, O-Power refers to the algorithm that only optimizes power, Fixed Orthogonal Subcarrier Allocation is the calculation result of the subcarrier allocation initial value algorithm, O-Carrier refers to the algorithm that only optimizes orthogonal subcarriers, and Fixed Power is the power value obtained by equally dividing the rated power according to the subcarrier initial value. Under different numbers of orthogonal subcarriers, 100 Monte Carlo simulation experiments are carried out under each group of parameters, and the average value is calculated. The experimental results are as Figure 7 shown.

[0173] It can be Figure 7 seen that compared with the algorithms that only optimize power and only optimize orthogonal subcarriers, under the algorithm scenario proposed in this patent, the user data download time is significantly lower. As the number of orthogonal subcarriers K increases, the data download time becomes smaller and smaller. This is because after the number of carriers increases, the power resources allocated to each subcarrier are also smaller. When calculating the signal-to-interference-plus-noise ratio, the interference term will be reduced more, resulting in an increase in the overall rate, and thus the data download time is reduced. Under the algorithm scenario proposed in this patent, as the number of subcarriers continues to increase, the reduction rate of the data download time gradually slows down. This is because the power on the subcarriers will also decrease, resulting in a slower increase in the signal-to-interference-plus-noise ratio and a slower increase in the rate. Therefore, when the total bandwidth is fixed, too large a number of orthogonal subcarriers will not bring a rapid increase in the rate. In the actual scenario, it is still necessary to make a trade-off between the algorithm complexity and the system performance.

[0174] When the number of users is different, compare the GEO satellite downlink data download time under different algorithm scenarios. Under different numbers of users, 100 Monte Carlo simulation experiments are carried out for each group of parameters, and the average value is calculated. Among them, Fixed is the calculation result of the algorithm with fixed power and carrier allocation as the initial value, and Branch and bound is the algorithm for carrier power allocation using the branch and bound method.

[0175] When using the branch and bound method to solve the optimization problem, the orthogonal subcarrier allocation variable C, as a 0-1 variable, can be directly solved without continuous approximation through the CCCP algorithm. The power variable optimization is solved using the SCA algorithm based on convex difference relaxation. At this time, there are 4NMK + M decision variables in the optimization problem. Assuming that the branch and bound method (Branch and bound) is used to divide the optimization problem into several sub-problems, it needs to be iterated E times to achieve convergence. The complexity of the branch and bound algorithm is exponential, and in the worst case, all possible solutions need to be enumerated. At this time, the algorithm complexity of the branch and bound method for solving is

[0176] ○(E12 4NMK+M ) (45)

[0177] According to formula (44), the complexity of the carrier power allocation algorithm based on the convex difference relaxation SCA algorithm proposed in this patent is ○(E(5NMK + M) 3 (8NMK + NK + MK + N + 2M + 1)) (46)

[0178] Compared with the complexity of the algorithm proposed in this patent, the branch and bound method needs to divide the optimization problem into multiple sub - problems for solution, so the complexity is very high. The experimental results under relevant parameters and algorithm scenarios are as Figure 8 shown.

[0179] As Figure 8 can be seen, as the number of users continues to increase, the data download time will also continue to increase. This is because as the number of users increases, the number of sub - carriers allocated to each user will decrease, resulting in a decrease in the downlink rate of users and an increase in the overall data download time. Compared with the fixed allocation algorithm, the algorithm proposed in this patent can reduce the user data download time. When the orthogonal sub - carriers are fixed at the initial value, the data download time is greater than that of the proposed algorithm, indicating that the optimization of orthogonal sub - carriers can improve the system efficiency. When the power is evenly distributed and only the orthogonal sub - carriers are optimized, the data download time is much greater than that of the proposed algorithm, indicating that the optimization of power can greatly help users download data quickly.

[0180] The result of the algorithm proposed in this patent is very close to that of the branch and bound method, but the computational complexity of the algorithm in this patent is much lower than that of the branch and bound method, indicating that the resource allocation of the SCA algorithm based on convex difference relaxation proposed in this patent can complete the resource allocation task with low complexity. Figure 8 The O - Carrier algorithm in [reference] is only slightly better than the Fixed algorithm with fixed power and carrier allocation as the initial value, indicating that the initial value allocation method based on demand weight proposed in this patent has a good effect, and the initial value allocation result is relatively close to the result of carrier optimization, which can improve the optimization convergence speed.

[0181] The above are only the preferred embodiments of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A method for wireless resource allocation based on downlink delay in a GEO satellite communication system, characterized in that, It includes the following steps: S1. Based on the OFDMA model for downlink transmission of multi-beam GEO satellites, calculate the satellite antenna gain and channel coefficients, initialize the resource allocation method and data download time estimation, and construct a model for the shortest download time problem based on user data requirements with the total data download time of all users as the objective function. S2. For the model of the shortest download time problem based on user data requirements, solve it through a resource allocation algorithm based on the successive convex approximation (SCA) algorithm with convex difference relaxation. The resource allocation algorithm based on the SCA algorithm with convex difference relaxation includes subcarrier and power allocation algorithms based on the SCA algorithm with convex difference relaxation, which includes the following: S2.

1. Introduce upper bound relaxation variables to convert the non-convex objective function into a convex function. S2.

2. On the basis of the convex objective function, use the big-M method and the CCCP algorithm to convert the non-convex constraint conditions into convex functions. S3. In the process of solving the resource allocation algorithm based on the SCA algorithm with convex difference relaxation, set the initial values of variables through an initial value allocation algorithm based on demand weights. The setting of the initial values of variables includes the following steps: S3.

1. Based on demand weights, perform initial subcarrier allocation to obtain the C variable, that is, the initial value of the access indication matrix of all users. S3.

2. Perform initial settings on the relaxation variables to obtain the P variable, that is, the initial value of the power matrix of all subcarriers.

2. A method for wireless resource allocation based on downlink delay in a GEO satellite communication system according to claim 1, characterized in that: The model for the shortest download time problem based on user data requirements includes: Among them, is the power obtained by the m-th user from the k-th subcarrier of the n-th beam of the satellite, where 1 ≤ k ≤ K, 1 ≤ n ≤ N, 1 ≤ m ≤ M. There are M users in total within the coverage of the N beams of the GEO satellite, and the frequency band with a total bandwidth of B is divided into K subcarriers; P 0 is the rated transmit power of the satellite beam, and P T is the total rated transmit power of the satellite. sgn(·) is the sign function. is the access indication for the m-th user to access the k-th subcarrier of the n-th beam, and v light is the speed of light, D m is the data volume download requirement of the m-th user, and d m is the distance from the GEO satellite to the m-th user, and r m is the instantaneous achievable downlink rate obtained by the m-th user from the GEO satellite. is the set of users served by the n-th beam of the GEO satellite. P and C are the power matrix of all subcarriers and the access indication matrix of all users, respectively.

3. A method for wireless resource allocation based on downlink delay in a GEO satellite communication system according to claim 1, characterized in that: On the basis of the convex objective function, use the big-M method and the CCCP algorithm to convert the non-convex constraint conditions into convex functions, and the calculation formula is: Among them, φ m is the upper bound relaxation variable, is the power obtained by the m-th user from the k-th subcarrier of the n-th beam of the satellite, 1 ≤ k ≤ K, 1 ≤ n ≤ N, 1 ≤ m ≤ M. There are M users in total within the coverage of the N beams of the GEO satellite, and the frequency band with a total bandwidth of B is divided into K subcarriers; P 0 is the rated transmit power of the satellite beam, P T is the total rated transmit power of the satellite, sgn(·) is the sign function, is the access indication for the m-th user to access the k-th subcarrier of the n-th beam, is the value calculated in the t-th iteration, D m is the data volume download requirement of the m-th user, is the set of users served by the n-th beam of the GEO satellite, is the lower bound constraint, is the upper bound constraint, is the lower bound constraint variable. P, C, Φ, ɑ, η, ξ are the power matrix of all subcarriers, the access indication matrix of all users, the upper bound relaxation variable matrix, the subset of lower bound constraint variables, the subset of lower bound constraints, and the subset of upper bound constraints, respectively.

4. A method for wireless resource allocation based on downlink delay in a GEO satellite communication system according to claim 1, characterized in that: The formula for initial subcarrier allocation includes: Among them, W m is the download time of the m-th user, σ is the noise energy, is the channel capacity from the n'-th beam to the m-th user on the ground on the first subcarrier, is the channel capacity from the n-th beam to the m-th user on the ground on the first subcarrier, P 0 is the rated transmission power of the satellite beam, D m is the data volume download requirement of the m-th user, 1 ≤ n ≤ N, there are M users in total within the coverage of N beams of the GEO satellite, the frequency band with a total bandwidth of B is divided into K subcarriers, and B0 is the bandwidth of each subcarrier; is the set of users served by the n-th beam of the GEO satellite; Allocate K m subcarriers to each user, and the calculation formula is as follows: Among them, W m′ is the symbol of rounding down; K0 users are allocated one more subcarrier: After calculating the number of subcarriers allocated to each user, starting from the first subcarrier k = 1, allocate the subcarriers to the first user until all subcarriers are allocated.

5. A method for wireless resource allocation based on downlink delay in a GEO satellite communication system according to claim 1, characterized in that: The formula for the initial value of the P variable includes: The initial value of ξ, ξ (0) The calculation formula includes: Among them, is the upper bound constraint, is the power obtained by the m-th user from the k-th subcarrier of the n-th beam of the satellite, where 1 ≤ k ≤ K, 1 ≤ n ≤ N, 1 ≤ m ≤ M. There are M users in total within the coverage of the N beams of the GEO satellite, and the frequency band with a total bandwidth of B is divided into K subcarriers; P 0 is the rated transmit power of the satellite beam, is the channel from the n beams to the m-th user on the ground at the k-th subcarrier.

Images