Load balancing method for low-orbit constellation network

Abstract

The invention belongs to the technical field of networks, and particularly relates to a load balancing method for a low-orbit constellation network, which comprises the following steps: constructing the low-orbit constellation network, and taking each satellite of the low-orbit constellation network as a node; acquiring a source node and a destination node in the low-orbit constellation network; dividing a minimum hop count set area according to the source node and the destination node; acquiring link state information among the nodes in the minimum hop count set area, and processing the link state information among the nodes by adopting a hybrid ant colony-genetic algorithm to obtain a selection probability of each information transmission path; performing priority updating on the path selection probability; data transmission is carried out according to the updated path selection probability, and load balancing of the nodes is achieved; according to the method, the satellite network load is reasonably distributed, the utilization efficiency of a communication link is effectively improved, network delay and congestion are remarkably reduced, and meanwhile high flexibility and expansibility are achieved.

Description

Technical Field

[0001] The present invention belongs to the field of network technology, and in particular relates to a load balancing method for a low-orbit constellation network. Background Art

[0002] With the large-scale deployment of low earth orbit (LEO) satellite constellations, satellite communication networks face challenges such as high-speed mobility, frequent topology changes, and limited link resources, which may lead to overload, congestion, and unstable performance of communication links, affecting communication quality and system capacity. To solve these problems, an improved inter-satellite link load balancing method is proposed to improve system efficiency and reliability.

[0003] Traditional routing and load balancing methods have limitations in the dynamic environment of LEO satellite networks. Static path planning makes it difficult to adapt to satellite position changes and link status fluctuations in a timely manner, which can easily lead to congestion or overload of some links. As the size of the constellation increases, the number of links increases sharply, and fixed rules and historical data are difficult to meet the needs of real-time and flexibility. Therefore, an optimization method that can adapt to the dynamic network environment and achieve load balancing among multiple intersatellite links is needed. Summary of the invention

[0004] In order to solve the problems existing in the above-mentioned prior art, the present invention proposes a load balancing method for low-orbit constellation networks, which includes: constructing a low-orbit constellation network, taking each satellite of the low-orbit constellation network as a node; obtaining source nodes and destination nodes in the low-orbit constellation network; dividing a minimum hop count set area according to the source nodes and the destination nodes; obtaining link state information between each node in the minimum hop count set area, using a hybrid ant colony-genetic algorithm to process the link state information between each node to obtain the selection probability of each information transmission path; performing priority update on the path selection probability; performing data transmission according to the updated path selection probability to achieve node load balancing.

[0005] Beneficial effects of the present invention:

[0006] The present invention fully considers the high dynamics and periodic changes of the low-orbit satellite constellation network topology, and realizes efficient load balancing of inter-satellite links through an accurate path selection algorithm. Specifically, the invention first calculates the optimal transmission node for the next hop based on the relative positions of the source node and the destination node, and defines the minimum hop set area to narrow the path search range and improve the efficiency of path calculation. During the path exploration process, the forward detection mechanism simulates the behavior of ants. The forward detection signal explores along the potential path and leaves pheromones as a reference for path selection. Subsequently, the backward feedback mechanism simulates the backward ant behavior, returns from the target node to the source node, and releases pheromones based on the path quality to accurately reflect the quality of the path.

[0007] The present invention combines genetic algorithms to perform selection, crossover and mutation operations on the path population, thereby improving path diversity and global optimization capabilities. The genetic algorithm uses high-fitness paths as parents, generates new paths through crossover and mutation, and avoids local optimality. The pheromone concentration, heuristic information and genetic fitness are combined to calculate the selection probability of each path and obtain the optimal path combination. This method not only effectively avoids high load and network congestion, but also continuously adjusts the path priority in a dynamic environment to achieve improved communication link utilization, reduced latency and enhanced system reliability. The solution is highly flexible and scalable, adapting to satellite network layouts of different scales and complexities, and significantly improving user experience and overall communication efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 It is the overall flow chart of the present invention;

[0009] Figure 2 It is a scene schematic diagram of the present invention. DETAILED DESCRIPTION

[0010] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0011] A load balancing method for a low-orbit constellation network, the method comprising: constructing a low-orbit constellation network, taking each satellite of the low-orbit constellation network as a node; obtaining a source node and a destination node in the low-orbit constellation network; dividing a minimum hop count set area according to the source node and the destination node; obtaining link state information between each node in the minimum hop count set area, using a hybrid ant colony-genetic algorithm to process the link state information between each node, and obtaining a selection probability of each information transmission path; performing priority update on the path selection probability; performing data transmission according to the updated path selection probability, and realizing node load balancing.

[0012] The present invention proposes a load balancing method for a low-orbit constellation network, which is applied to a low-orbit constellation network, such as Figure 1 , Figure 2 As shown, the method includes the following contents:

[0013] S1. Divide the minimum hop count set area according to the relative positions of the satellites and initialize the parameters of the hybrid ant colony-genetic algorithm.

[0014] This application uses the Iridium constellation as a scenario to illustrate the implementation plan. The Iridium satellite network consists of 66 low- and medium-Earth orbit (LEO) satellites. Each satellite is located at an altitude of approximately 781 kilometers and orbits the Earth with an orbital inclination of 86.4 degrees, forming 6 orbital planes, each of which contains 11 satellites.

[0015] In a low-orbit constellation, each satellite has multiple adjacent satellite nodes and can usually transmit data in four directions (up, down, left, and right); assuming that the distance between the source satellite and the destination satellite is:

[0016] ΔX=|X d -X s |

[0017] ΔY=|Y d -Y s |

[0018] Where ΔX and ΔY represent the distance between the source satellite and the destination satellite on the X-axis and Y-axis respectively. The minimum number of hops is:

[0019]

[0020] in, Indicates rounding up to ensure that the number of hops is an integer.

[0021] The purpose of dividing the rectangular area is to give priority to those satellite nodes that can transmit along the shortest path within the minimum hop count range, according to the source satellite position (X s ,Y s ) and the destination satellite position (X d ,Y d ), the boundary of the rectangular area can be defined as the X-direction boundary: min(X s ,X d ) to max(X s ,X d ), Y direction boundary: min(Y s ,Y d ) to max(Y s ,Y d ), the set of candidate satellite nodes in the rectangular area can be expressed as:

[0022] R={(X,Y)|min(X s ,X d )≤X≤max(X s ,X d ),min(Y s ,Y d )≤Y≤max(Y s ,T d )}

[0023] The set R includes all satellite nodes that can transmit between the X-axis and the Y-axis.

[0024] The non-rectangular area is used to supplement nodes that are not in the rectangular area but can still reach the destination in multiple hops. Assume that the rectangular area already contains n hops. min All paths of the non-rectangular area are mainly considered to have a hop number of n min +1 node, the candidate node set of the non-rectangular area can be defined as:

[0025] N={(X,Y)|X∈[X s -1,X s +1]∪[X d -1,X d +1],Y∈[Y s -1,Y s +1]∪[Y d -1,Y d +1]}

[0026] The minimum hop count set area can be determined based on the rectangular area and the non-rectangular area. The total set M of the minimum hop count candidate area can be defined as the union of the rectangular area R and the non-rectangular area N:

[0027] M=R∪N

[0028] Wherein, M is a candidate set of all possible paths during the transmission process from the source satellite to the destination satellite. For each satellite node in the candidate node M, heuristic information (such as bandwidth, delay, etc.) can be further used to perform path optimization.

[0029] The initialization of the ant colony algorithm includes the number of ants, pheromone importance factor, heuristic information importance factor, pheromone volatility rate, etc.; the initialization of the genetic algorithm includes population size, crossover probability, mutation probability, etc.

[0030] Specifically, according to the network scale and complexity, the number of ants is linearly related to the number of network nodes N. The number of ants is set to k*N, and the value of k is based on the network complexity and computing resources. In this scenario, k=5, m is 330, and all links have the same pheromone concentration at the initial moment, and the population size and number of ants are set to be the same.

[0031] S2. In the minimum hop count set area, a group of forward detection packets are sent from the source node to collect link state information. The collected path information is encoded into chromosomes to form the initial population.

[0032] Specifically include: virtual ants at the source node (X s ,Y s) is released, each ant has a unique ID to identify itself and its path; the source node sends a group of forward ants to the destination node and sets the initial window size W init Each forward ant records the current hop count H and distance D during the exploration process. If H ≥ W or a loop is formed or the forward ant enters the path to the destination node, further diffusion is stopped. If the forward ant approaches the destination node and D decreases, the window size W is updated according to the window reduction formula. When the window is reduced to W min When the ant is near the target node, the exploration range of the forward ant is greatly limited, so that it converges to the destination node more quickly.

[0033] During the path selection process, the forward ants collect the following link status information along the way for subsequent path quality evaluation and pheromone update, including: Path delay: the data transmission delay d from node i to the next hop node j ij Bandwidth: The bandwidth B of path i→j ij , which represents the transmission capacity of the link. Packet loss rate: the data packet loss rate P of path i→j ij , reflecting the stability of the path. Hop count: records the number of hops taken by the ant from the source node to the destination node, used to evaluate the length of the path. The current ant reaches the destination node (X d ,Y d ), the path exploration ends.

[0034] After each forward detection packet reaches the destination node, it records the node sequence and its corresponding link status information. A complete path is encoded as a chromosome of the genetic algorithm. The gene sequence of the chromosome is the ID of each relay node on the path. A path chromosome can be expressed as:

[0035] Chromosome=[Node 0 ,Node 1 ,Node 2 ,…,Node d ]

[0036] Among them, Node 0 Indicates the starting relay node ID, Node 1 Indicates the first relay point ID.

[0037] By introducing a multi-objective optimization fitness function, the delay, bandwidth and packet loss rate indicators are combined into one goal. The fitness formula of the genetic algorithm can be expressed as:

[0038]

[0039] Where Path is the path, i is the link source node, j is the link destination node, and w d is the bandwidth impact weight, Delay ijis the delay, w b Bandwidth is the bandwidth weight. ij is the bandwidth, w l is the packet loss rate weight, Loss ij is the packet loss rate.

[0040] S3. The destination node brings back the collected information through the backward detection packet, releases pheromones on the path passed, updates the local priority of the path, and performs genetic operations on the path population, including selection, crossover and mutation.

[0041] Specifically, on the source node side, these paths and their performance data are used as the initial population of the genetic algorithm, and high-fitness paths are retained preferentially; new path combinations are generated through methods such as partial mapping or sequential crossover, and operations such as node exchange or insertion are used to increase path diversity.

[0042] Specifically, genetic operations are performed based on chromosomes and fitness values. High-quality paths are selected as parents based on fitness to increase the proportion of good genes in the population. Random sampling is performed based on total fitness and selection probability to select several paths as parents:

[0043]

[0044] By combining the genes of the parents to generate new offspring paths, explore better path combinations, select two parent chromosomes, and randomly select two crossover points; select a offspring chromosome to mutate, randomly select the positions of two relay nodes to exchange the two nodes, verify the feasibility of the mutated path, and ensure that there are valid links between adjacent nodes. If not, reselect the mutation point or abandon the mutation.

[0045] Furthermore, the updated pheromones are stored in the system. Specifically, when the forward ant passes through the node, the collected link state information is stored in the node system; when the backward ant passes through the node, the LRU cache mechanism is used to retain only the pheromones of the hotspot path. Two LUR linked lists are maintained in the node storage system, the active linked list active_list and the inactive linked list inactive_list. The active linked list stores the most recently released pheromones, and the inactive linked list stores the pheromones released recently.

[0046] When the backward ant passes a node, it will determine whether the pheromone on the path is recently released. If so, the pheromone on the path is added to the head of the active list and the pheromone is updated. The head of the active list always stores the most recently updated pheromone path. For those pheromone paths that have not been updated for a long time, they will be gradually transferred to the inactive list; the end of the inactive list is the pheromone with the longest storage time. The pheromones on these paths are no longer preferred and will be eliminated; in the active list and the inactive list, if the pheromone has not been accessed or updated for a certain period of time, it will be "evaporated" or eliminated.

[0047] S4, calculating the selection probability of each path according to the pheromone concentration, the path heuristic value and the fitness function of the genetic algorithm;

[0048] The pheromone released by the backward ant on the path (i, j) is updated and its calculation formula is:

[0049]

[0050] The heuristic value formula of link quality can be expressed as:

[0051]

[0052] The comprehensive scoring formula is:

[0053]

[0054] Normalize the S values ​​of all paths to get the selection probability of the i-th path

[0055]

[0056] Among them, Δτ ij is the updated pheromone, σ is the pheromone attenuation coefficient, Q is the constant of the total amount of pheromone, L ij is the length from node i to node j, L e is the current shortest path length, T e is the node set of the optimal path, is the average value of pheromone, α is the weight of pheromone, is the average value of the heuristic function, β is the weight of the heuristic value, Fitness is the fitness function, γ is the path adaptation weight, and n is the number of paths.

[0057] According to the probability interval that the random number r falls into, the corresponding next hop node is selected.

[0058] In order to introduce randomness in path selection, ants can use the roulette wheel selection method when selecting the next hop node. The roulette wheel selection method allows ants to have a certain degree of randomness in path selection, thus avoiding falling into the local optimal solution.

[0059] Use the roulette wheel selection method to select the next hop node and calculate the selection probability p of all neighbor nodes ij , and accumulate these probabilities to construct a cumulative probability interval ranging from [0,1]; randomly generate a number r between [0,1]; and select the corresponding next hop node according to the probability interval that the random number r falls into.

[0060] S5. Data packets are transmitted according to the optimized path weights to achieve load balancing of nodes. The priority of the path is dynamically adjusted by using the volatilization mechanism of multi-dimensional pheromones and the generational update of the genetic algorithm.

[0061] Specifically: Transmit data packets along the optimal path to achieve link load balancing; transmit data packets along the current optimal path to improve link utilization efficiency; and dynamically adjust path priorities through independent volatilization and incremental updates of multi-dimensional pheromones.

[0062] After each path selection, the multidimensional pheromone will decay according to the volatility coefficient, the formula is:

[0063] τ ij (t+1)=(1-ρ·Ω ij )·τ ij (t)+Δτ ij

[0064] Among them, ρ represents the pheromone volatility coefficient, Δτ ij represents the incremental change in the pheromone released by the backward ant on the path (i, j), Ω ij It is the load factor (bandwidth occupancy, delay, and packet loss rate are maintained together).

[0065] The above embodiments further illustrate the purpose, technical solutions and advantages of the present invention in detail. It should be understood that the above embodiments are only preferred implementation modes of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made to the present invention within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A load balancing method for a low-orbit constellation network, characterized in that: include: Build a low-orbit constellation network, and use each satellite in the low-orbit constellation network as a node; Obtain source nodes and destination nodes in the low-orbit constellation network; The minimum hop count set area is divided according to the source node and the destination node; the link state information between each node in the minimum hop count set area is obtained, and the link state information between each node is processed by a hybrid ant colony-genetic algorithm to obtain the selection probability of each path for information transmission; the path selection probability is updated with priority; data is transmitted according to the updated path selection probability to achieve node load balancing.

2. A load balancing method for a low-orbit constellation network according to claim 1, characterized in that: Dividing the minimum hop count set area includes: obtaining the location of the source node and the location of the destination node, determining the rectangular area and the non-rectangular area according to the locations of the two nodes; obtaining the candidate satellite node set in the rectangular area and the candidate node set in the non-rectangular area.

3. A load balancing method for a low-orbit constellation network according to claim 1, characterized in that: The hybrid ant colony-genetic algorithm is used to process the link state information between each node, including: initializing the parameters of the hybrid ant colony-genetic algorithm; encoding the link state information into chromosomes to form an initial population; the destination node brings back the collected information through backward detection, releases pheromones on the passed path, and updates the local priority of the path; performs genetic operations on the population in the path; uses a fitness function to calculate the fitness value of the population after the genetic operation, and optimizes the fitness function through multiple objectives, and calculates the selection probability of each path according to the fitness value.

4. A load balancing method for a low-orbit constellation network according to claim 3, characterized in that: Initializing the parameters of the hybrid ant colony-genetic algorithm includes initializing the parameters of the ant colony algorithm and initializing the parameters of the genetic algorithm; the parameters of the ant colony algorithm include the number of ants, the multidimensional pheromone importance factor, the heuristic information importance factor and the multidimensional pheromone volatility rate; the genetic algorithm parameters include the population size, the crossover probability and the mutation probability.

5. The load balancing method for a low-orbit constellation network according to claim 3, characterized in that: Encoding link state information into chromosomes includes: virtual ants at the source node (X s ,Y s ) is released at each ant, and each ant has a unique ID to record the path it explores; after each forward detection packet reaches the destination node, it records the path node sequence and its corresponding link state information; a complete path is encoded as a chromosome of the genetic algorithm, and the gene sequence of the chromosome is the ID of each relay node on the path. A path chromosome can be expressed as: Chromosome=[Node0,Node1,Node2,…,Node d ] Among them, Node0 represents the starting relay node ID, and Node1 represents the first relay node ID.

6. A load balancing method for a low-orbit constellation network according to claim 3, characterized in that: Genetic operations on the path population include selection, crossover and mutation operations on the population.

7. A load balancing method for a low-orbit constellation network according to claim 3, characterized in that A multi-objective optimization fitness function is introduced to combine the delay, bandwidth and packet loss rate indicators into one objective. The fitness function expression is: Where Path is the path, i is the link source node, j is the link destination node, and w d is the bandwidth impact weight, Delay ij is the delay, w b Bandwidth is the bandwidth weight. ij is the bandwidth, w l is the packet loss rate weight, Loss ij is the packet loss rate.

8. The load balancing method for a low-orbit constellation network according to claim 3, characterized in that: Calculating the selection probability of each path includes: The pheromone released by the backward ant on the path (i, j) is updated and its calculation formula is: The comprehensive scoring formula is: Normalize the S values ​​of all paths to get the selection probability of the i-th path Among them, Δτ ij is the updated pheromone, σ is the pheromone attenuation coefficient, Q is the constant of the total amount of pheromone, L ij is the length from node i to node j, L e is the current shortest path length, T e is the node set of the optimal path, is the average value of pheromone, α is the weight of pheromone, is the average value of the heuristic function, β is the weight of the heuristic value, Fitness is the fitness function, γ is the path adaptation weight, and n is the number of paths.

9. The load balancing method for a low-orbit constellation network according to claim 1, characterized in that: Prioritizing the path selection probability includes: maintaining multiple pheromone dimensions for each link during the path search process, the pheromone dimensions include bandwidth, delay, and packet loss rate; updating the pheromone dimensions using an attenuation strategy; fusing multiple pheromone dimensions using a multi-objective weighted method, and calculating the next-hop selection probability of the ant based on the fused information dimensions.

10. A load balancing method for a low-orbit constellation network according to claim 9, characterized in that: The attenuation of pheromone dimensions includes attenuation of bandwidth pheromone and delay pheromone, and adding pheromone to the paths with better performance according to the corresponding indicators. The attenuation of pheromone includes: t ij (t+1)=(1-ρ·Ω ij )·t ij (t)+Δτ ij Among them, ρ represents the pheromone volatility coefficient, Δτ ij represents the incremental change in the pheromone released by the backward ant on the path (i, j), Ω ij is the load factor.

Images