Low earth orbit satellite networking resource allocation method based on improved PSO (Particle Swarm Optimization)

Abstract

The invention discloses a low earth orbit satellite networking resource allocation method based on improved PSO (Particle Swarm Optimization), and relates to the field of satellite internet. The method comprises the following steps: firstly, initializing a particle swarm according to a low earth orbit satellite networking resource allocation demand, then calculating a fitness value of each particle, selecting an optimal particle individual and a suboptimal particle individual, then carrying out local exploration on the optimal particle individual, and if the maximum number of iterations is reached, carrying out local exploration on the suboptimal particle individual; if yes, according to the optimal particle individual after local exploration in the current iteration process, a final allocation scheme of the low earth orbit satellite networking resource allocation requirement is obtained, and otherwise, the particle speed is updated, the particle swarm is updated according to the particle speed, and the next iteration process is carried out until the maximum number of iterations is achieved. According to the low-orbit satellite transponder resource allocation method, the requirements of the low-orbit satellite transponder resource allocation method on the orbit characteristics of the low-orbit satellite and the solving performance of a larger network scale are met, and the satellite transponder resource allocation requirements of thousands of low-orbit satellites are supported.

Description

Technical Field

[0001] The present invention relates to the field of satellite Internet, and in particular to a low-orbit satellite networking resource allocation method based on improved PSO. Background Art

[0002] The low-orbit satellite transponder resource allocation method refers to a method that, driven by the networking needs of low-orbit satellite networks, eliminates the conflict of satellite network transponder bandwidth allocation under multiple networking needs by modeling and optimizing satellite network needs and satellite transponder bandwidth, while improving the efficiency of satellite transponder resource utilization. As a typical NP-Hard problem, the low-orbit satellite transponder resource allocation method cannot calculate the optimal solution in a limited time through an explicit solution method. At the same time, due to the characteristics of the low-orbit orbit, the transponder bandwidth allocation must also meet the transit time constraints and the resource discrete constraints brought by the frequent switching of satellite switching links. Therefore, compared with the transponder resource allocation of high-orbit satellites, the low-orbit satellite transponder resource allocation problem is more complicated.

[0003] As one of the combinatorial optimization problems, the low-orbit satellite transponder resource allocation problem, if solved by the full permutation method, will cause the group resource allocation permutation to increase exponentially, especially in the large-scale transponder resource allocation process of low-orbit satellite networks, if the number of users applying for networking resources increases sharply, the time complexity and space complexity required by the algorithm will not be able to meet the actual needs; if the reinforcement learning method is adopted, it is necessary to adopt multi-agent game modeling to resolve conflicts. At present, the relevant research on decentralized reinforcement learning still cannot actually meet the high reliability requirements of satellite transponder resource allocation, and is limited by the limited action space of reinforcement learning. The application effect of transponder resource allocation for large-scale low-orbit satellite networks is limited, and the actual application effect is not good. Summary of the invention

[0004] In view of this, the present invention proposes a low-orbit satellite networking resource allocation method based on improved PSO. This method meets the low-orbit transit time constraints and the resource discrete constraints brought by frequent switching of satellites and switching links. At the same time, the improved PSO achieves better global exploration and easier convergence characteristics, meets the low-orbit satellite transponder resource allocation method for low-orbit satellite orbit characteristics and the performance requirements for solving larger network scales, and supports the satellite transponder resource allocation requirements of thousands of low-orbit satellites.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0006] A low-orbit satellite networking resource allocation method based on improved PSO comprises the following steps:

[0007] Step 1: Set the particle swarm size i for low-orbit satellite network resource allocation maxAfter receiving the resource allocation requirements of d low-orbit satellite networks requested by the user, the satellite control system randomly sorts the resource allocation requirements of d low-orbit satellite networks and randomly selects i max Different sorting results to construct low-orbit satellite network resource allocation particle swarm

[0008] Among them, X i =(x1,x2,…,x d ), i=1,2,3,......,i max ; x1,x2,…,x d They correspond to the resource allocation requirements of a low-orbit satellite network, x1, x2, ..., x d The value of corresponds to an integer from 1 to d. The smaller the integer, the higher the priority of the corresponding low-orbit satellite network resource allocation demand;

[0009] Step 2: For the i-th particle X i , set the corresponding particle speed V i =(v1,v2,…,v d ), where V i Normal distribution: V i ~N(0,1);

[0010] Step 3, set the maximum number of iterations T and the iteration number variable t, and set t = 1;

[0011] Step 4: For each particle in the current iteration process, calculate its corresponding fitness value, and select the optimal particle individual and the suboptimal particle individual in the current iteration process according to the criterion of the maximum fitness value;

[0012] Step 5: Perform local exploration of the optimal particle individuals in the current iteration process according to the optimal reordering formula and greedy strategy;

[0013] Step 6: Determine whether the maximum number of iterations T is reached. If so, obtain the final allocation plan for the resource allocation requirements of d low-orbit satellite networks based on the optimal particle individual after local exploration in the current iteration process, and complete the low-orbit satellite network resource allocation based on the improved PSO; if the maximum number of iterations T is not reached, execute step 7;

[0014] Step 7, iteratively update the particle velocity corresponding to each particle in the current iteration process according to the exploration supervision weight and the self-supervision weight;

[0015] Step 8, iteratively update the low-orbit satellite networking resource allocation particle swarm according to the iteratively updated particle speed, set t=t+1, and continue to execute step 4.

[0016] Furthermore, step 4 specifically includes the following steps:

[0017] Step 401, for the particle in the tth iteration process According to the high total resource occupancy rate of the satellite resource pool, the Bottom-Left packing method is used to obtain the corresponding optimal low-orbit satellite allocation plan, and the particles are The total resource occupancy rate of the satellite resource pool corresponding to the optimal low-orbit satellite allocation solution is recorded as particle The fitness value of

[0018] in,

[0019] Step 402: record the particle with the largest fitness value as the optimal particle individual in the current iteration process. The particle with the second largest fitness value is recorded as the suboptimal particle individual in the current iteration process.

[0020] Furthermore, step 5 specifically includes the following steps:

[0021]

[0022] in, is the optimal reordering result, γ t is the random disturbance in the current iteration process, and its value range is: t =rand(-1,1),O t is the exploration range weight in the current iteration, and O t The value of is: O is the initial exploration range weight, O>0.

[0023] Furthermore, in step 6, according to the optimal particle individual after local exploration in the current iteration process, the final allocation plan for the resource allocation requirements of the d low-orbit satellite network is obtained, and the specific method is:

[0024] The best individual particle after local exploration Among the d values ​​contained, the smaller the value, the higher the priority of the low-orbit satellite network resource allocation demand, so as to obtain the corresponding low-orbit satellite network resource allocation demand ranking, and then according to the total resource occupancy rate of the satellite resource pool as the standard, the bottom-left packing method is used to obtain the optimal particle individual after local exploration The corresponding optimal low-orbit satellite allocation plan completes the low-orbit satellite networking resource allocation based on improved PSO.

[0025] Furthermore, the specific method of step 7 is:

[0026]

[0027] in, Indicates the particle size in the current iteration The corresponding particle speed is express Iterative update value of ω t represents the momentum weight in the current iteration process, ω t =rand(ω min ,ω max ),ω min and ω max Respectively represent the minimum and maximum values ​​of momentum weight; and Respectively represent the exploration supervision weight and self-supervision weight in the current iteration process, and s1 and s2 represent the initial exploration supervision weight and initial self-supervision weight, respectively, and both s1 and s2 are greater than 0; and They represent the first random Gaussian distribution sampling and the second random Gaussian distribution sampling in the current iteration process respectively. and The value range is rand(0,1).

[0028] Furthermore, the specific method of step 8 is:

[0029] Step 801,

[0030] in, express The iterative update value of ; Express Reorder the results: The values ​​of each element in the array are sorted in ascending order, and each element is reallocated to an integer value corresponding to 1 to d according to the sorting order of the value of each element to obtain the iterative update value.

[0031] Step 802, set t=t+1, and continue to execute step 4.

[0032] Due to the adoption of the above technical solution, the beneficial effects of the present invention compared with the prior art are:

[0033] The present invention proposes a low-orbit satellite networking resource allocation method based on improved PSO. Aiming at the characteristics that the transponder bandwidth allocation in the low-orbit satellite Internet needs to meet the resource discrete constraints brought by the transit time and the frequent switching of satellite switching links, the non-deterministic problem of quickly solving the requirements of multiple transponders and multiple networking in the low-orbit satellite network is realized, and the resource discrete constraints brought by the low-orbit transit time constraints and the frequent switching of satellite switching links are met. At the same time, the improved PSO realizes better global exploration and easier convergence characteristics. The present invention meets the requirements of the low-orbit satellite transponder resource allocation method for the low-orbit satellite orbit characteristics and the solution performance for a larger network scale, and supports the satellite transponder resource allocation requirements of thousands of low-orbit satellites. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 The figure is an overall flow chart of a low-orbit satellite networking resource allocation method based on improved PSO in an embodiment of the present invention.

[0035] Figure 2 It is a schematic diagram of resource allocation in an embodiment of the present invention. DETAILED DESCRIPTION

[0036] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0037] A low-orbit satellite network resource allocation method based on improved PSO, such as Figure 1 As shown, the following steps are included;

[0038] Step 1: Set the particle swarm size i for low-orbit satellite network resource allocation max After receiving the resource allocation requirements of d low-orbit satellite networks requested by the user, the satellite control system randomly sorts the resource allocation requirements of d low-orbit satellite networks and randomly selects i max Different sorting results to construct low-orbit satellite network resource allocation particle swarm

[0039] Among them, X i =(x1,x2,…,x d ), i=1,2,3,......,i max ; x1,x2,…,x d They correspond to the resource allocation requirements of a low-orbit satellite network, x1, x2, ..., x d The value of corresponds to an integer from 1 to d. The smaller the integer, the higher the priority of the corresponding low-orbit satellite network resource allocation demand;

[0040] Step 2: For the i-th particle X i , set the corresponding particle speed V i =(v1,v2,…,v d), where V i Normal distribution: V i ~N(0,1);

[0041] Step 3, set the maximum number of iterations T and the iteration number variable t, and set t = 1;

[0042] Step 4: For each particle in the current iteration process, calculate its corresponding fitness value, and select the optimal particle individual and the suboptimal particle individual in the current iteration process according to the criterion of the maximum fitness value;

[0043] Step 5: Perform local exploration of the optimal particle individuals in the current iteration process according to the optimal reordering formula and greedy strategy;

[0044] Step 6: Determine whether the maximum number of iterations T is reached. If so, obtain the final allocation plan for the resource allocation requirements of d low-orbit satellite networks based on the optimal particle individual after local exploration in the current iteration process, and complete the low-orbit satellite network resource allocation based on the improved PSO; if the maximum number of iterations T is not reached, execute step 7;

[0045] Step 7, iteratively update the particle velocity corresponding to each particle in the current iteration process according to the exploration supervision weight and the self-supervision weight;

[0046] Step 8, iteratively update the low-orbit satellite networking resource allocation particle swarm according to the iteratively updated particle speed, set t=t+1, and continue to execute step 4.

[0047] Furthermore, step 4 specifically includes the following steps:

[0048] Step 401, for the particle in the tth iteration process According to the high total resource occupancy rate of the satellite resource pool, the Bottom-Left packing method is used to obtain the corresponding optimal low-orbit satellite allocation plan, and the particles are The total resource occupancy rate of the satellite resource pool corresponding to the optimal low-orbit satellite allocation solution is recorded as particle The fitness value of

[0049] in,

[0050] Step 402: record the particle with the largest fitness value as the optimal particle individual in the current iteration process. The particle with the second largest fitness value is recorded as the suboptimal particle individual in the current iteration process.

[0051] Furthermore, step 5 specifically includes the following steps:

[0052] Calculate according to the optimal reordering formula:

[0053]

[0054] This strategy aims to enhance R g1 The exploration efficiency in the neighborhood can greatly improve the iteration speed of the algorithm, and at the same time has stronger exploration ability, which helps the algorithm avoid falling into suboptimal solutions during the convergence process. Therefore, it is suitable for applications with discrete resources and a large number of satellite transponder nodes;

[0055] According to the greedy strategy Conduct local exploration:

[0056]

[0057] in, is the optimal reordering result, γ t is the random disturbance in the current iteration process, and its value range is: t =rand(-1,1),O t is the exploration range weight in the current iteration, and O t The value of is: O is the initial exploration range weight, O>0;

[0058] In this embodiment, O = d, in the initial stage of the iteration process R g1 The Euclidean distance from the global optimal resource allocation sorting is large, and the larger O t The value can promote local search in a larger range, thereby accelerating the convergence of the algorithm; when it is close to convergence in the late stage of iteration, R g1 The distance between the global optimal resource allocation sorting gradually decreases, and a smaller weight O is used at this time. t Ability to accurately locate the optimal resource allocation order in a smaller exploration area,

[0059] Furthermore, in step 6, according to the optimal particle individual after local exploration in the current iteration process, the final allocation plan for the resource allocation requirements of the d low-orbit satellite network is obtained, and the specific method is:

[0060] like Figure 2 FIG. 1 is a schematic diagram showing the principle of resource allocation. After the user terminal submits the networking application requirement, the low-orbit satellite networking resource allocation method based on the improved PSO is used to obtain the final allocation plan for the low-orbit satellite networking resource allocation requirement. Figure 2 The upper right coordinate diagram is the carrier principle diagram, and the upper left coordinate diagram represents the available resources of the corresponding satellite resource pool in each period, where black represents occupied resources and white represents unoccupied resources; the optimal particle individual after local exploration Among the d values ​​contained, the smaller the value, the higher the priority of the low-orbit satellite network resource allocation demand, so as to obtain the corresponding low-orbit satellite network resource allocation demand ranking, and then according to the total resource occupancy rate of the satellite resource pool as the standard, the bottom-left packing method is used to obtain the optimal particle individual after local exploration The corresponding optimal low-orbit satellite allocation plan completes the low-orbit satellite networking resource allocation based on improved PSO.

[0061] Furthermore, the specific method of step 7 is:

[0062]

[0063] in, Indicates the particle size in the current iteration The corresponding particle speed is express Iterative update value of ω t represents the momentum weight in the current iteration process, ω t =rand(ω min ,ω max ), which can prevent the general particle swarm algorithm from falling into the problem of local suboptimality, ω min and ω max Respectively represent the minimum and maximum values ​​of the momentum weight. In this embodiment, ω min =-1,ω max =1; and Respectively represent the exploration supervision weight and self-supervision weight in the current iteration process, and s1 and s2 represent the initial exploration supervision weight and initial self-supervision weight, respectively, and both s1 and s2 are greater than 0; and They represent the first random Gaussian distribution sampling and the second random Gaussian distribution sampling in the current iteration process respectively. and The value range is rand(0,1).

[0064] Specifically, in this embodiment, s1=0.5, s2=1; when s1 decreases with the iteration rounds, the particles will reduce the independent exploration behavior, promote the group to converge faster, and a larger exploration supervision weight can obtain the optimal resource allocation sorting with a higher probability. When s2 increases with the iteration rounds, the iteration rounds of the particle swarm algorithm can be greatly reduced, guiding the particles to approach a better resource allocation result;

[0065] Furthermore, the specific method of step 8 is:

[0066] Step 801,

[0067] in, express The iterative update value of ; Express Reorder the results: The values ​​of each element in the array are sorted in ascending order, and each element is reallocated to an integer value corresponding to 1 to d according to the sorting order of the value of each element to obtain the iterative update value.

[0068] Step 802, set t=t+1, and continue to execute step 4.

[0069] In summary, the present invention meets the resource discreteness constraints brought by the low-orbit transit time constraints and the frequent switching of satellite switching links. At the same time, the improved PSO achieves better global exploration and easier convergence characteristics, meets the low-orbit satellite transponder resource allocation method for the low-orbit satellite orbit characteristics and the solution performance requirements for a larger network scale, and supports the satellite transponder resource allocation requirements for thousands of low-orbit satellites.

[0070] Those skilled in the art will appreciate that the embodiments described are intended to help readers understand the principles of the present invention, and should be understood that the scope of protection of the present invention is not limited to the embodiments described. For those skilled in the art, the present invention may have various changes and variations. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A low-orbit satellite networking resource allocation method based on improved PSO, characterized in that: The steps include: Step 1: Set the particle swarm size i for low-orbit satellite network resource allocation max After receiving the d low-orbit satellite network resource allocation requirements applied by the user, the satellite control system randomly sorts the d low-orbit satellite network resource allocation requirements and randomly selects i max Different sorting results to construct low-orbit satellite network resource allocation particle swarm Among them, X i =(x1, x2, ..., x d ),i=1,2,3,...,i max ; x1, x2, …, x d They correspond to the resource allocation requirements of a low-orbit satellite network, x1, x2, ..., x d The value of corresponds to an integer from 1 to d. The smaller the integer, the higher the priority of the corresponding low-orbit satellite network resource allocation demand; Step 2: For the i-th particle X i , set the corresponding particle speed V i =(v1, v2, ..., v d ), where V i Normal distribution: V i ~N(0,1); Step 3, set the maximum number of iterations T and the iteration number variable t, and set t = 1; Step 4: For each particle in the current iteration process, calculate its corresponding fitness value, and select the optimal particle individual and the suboptimal particle individual in the current iteration process according to the criterion of the maximum fitness value; Step 5: Perform local exploration of the optimal particle individuals in the current iteration process according to the optimal reordering formula and greedy strategy; Step 6: Determine whether the maximum number of iterations T is reached. If so, obtain the final allocation plan for the resource allocation requirements of d low-orbit satellite networks based on the optimal particle individual after local exploration in the current iteration process, and complete the low-orbit satellite network resource allocation based on the improved PSO; if the maximum number of iterations T is not reached, execute step 7; Step 7, iteratively update the particle velocity corresponding to each particle in the current iteration process according to the exploration supervision weight and the white supervision weight; Step 8, iteratively update the low-orbit satellite networking resource allocation particle swarm according to the iteratively updated particle speed, set t=t+1, and continue to execute step 4.

2. The low-orbit satellite network resource allocation method based on improved PSO according to claim 1, characterized in that: Step 4 specifically includes the following steps: Step 401, for the particle in the tth iteration process According to the high total resource occupancy rate of the satellite resource pool, the Bottom-Left packing method is used to obtain the corresponding optimal low-orbit satellite allocation plan, and the particles are The total resource occupancy rate of the satellite resource pool corresponding to the optimal low-orbit satellite allocation solution is recorded as particle The fitness value of in, Step 402: record the particle with the largest fitness value as the optimal particle individual in the current iteration process. The particle with the second largest fitness value is recorded as the suboptimal particle individual in the current iteration process.

3. The low-orbit satellite network resource allocation method based on improved PSO according to claim 2, characterized in that: Step 5 specifically includes the following steps: in, is the optimal reordering result, γ t is the random disturbance in the current iteration process, and its value range is: t =rand(-1,1),O t is the exploration range weight in the current iteration, and O t The value of is: O is the initial exploration range weight, O>

0.

4. The low-orbit satellite network resource allocation method based on improved PSO according to claim 3 is characterized in that: In step 6, according to the optimal particle individual after local exploration in the current iteration process, the final allocation plan for the resource allocation requirements of d low-orbit satellite networks is obtained, and the specific method is as follows: The best individual particle after local exploration Among the d values ​​contained, the smaller the value, the higher the priority of the low-orbit satellite network resource allocation demand, so as to obtain the corresponding low-orbit satellite network resource allocation demand ranking, and then according to the total resource occupancy rate of the satellite resource pool as the standard, the bottom-left packing method is used to obtain the optimal particle individual after local exploration The corresponding optimal low-orbit satellite allocation plan completes the low-orbit satellite networking resource allocation based on improved PSO.

5. The method for allocating resources of low-orbit satellite networks based on improved PSO according to claim 4, characterized in that: The specific method of step 7 is: in, Indicates the particle size in the current iteration The corresponding particle speed is express Iterative update value of ω t represents the momentum weight in the current iteration process, ω t =rand(ω min ,ω max ),ω min and ω max Respectively represent the minimum and maximum values ​​of momentum weight; and Respectively represent the exploration supervision weight and white supervision weight in the current iteration process, and S1 and s2 represent the initial exploration supervision weight and the initial white supervision weight, respectively, and both s1 and s2 are greater than 0; and They represent the first random Gaussian distribution sampling and the second random Gaussian distribution sampling in the current iteration process respectively. and The value range is rand(0, 1).

6. The method for allocating resources of low-orbit satellite networks based on improved PSO according to claim 5, characterized in that: The specific method of step 8 is: Step 801, in, express Iterative update value of ; Express Reorder the results: The values ​​of each element in the array are sorted in ascending order, and each element is reallocated to an integer value corresponding to 1 to d according to the sorting order of the value of each element to obtain the iterative update value. Step 802, set t=t+1, and continue to execute step 4.

Images