Method for searching optimal network topology applying distributed multi-short-path routing

Abstract

The invention provides a method for searching an optimal network topology applying distributed multi-short-path routing. According to the method, the optimal network topology applying the distributed multi-short-path routing in an application scene given by a user is searched based on a framework of a genetic algorithm. According to the method, an initial population of chromosomes is constructed in a mode of generating an ER random connected network, and each chromosome corresponds to one network topology; the method comprises the following steps of: performing multiple iterations, generating offspring chromosomes by taking chromosomes in a previous generation of population as parent chromosomes in three modes of cloning, variation and crossover during each iteration, forming a new generation of population, performing a distributed multi-short-path routing network transmission test on the chromosomes in the new generation of population, and eliminating chromosomes with low fitness according to a test result; and after multiple iterations, the chromosome with the highest fitness in the latest generation of population corresponds to the optimal network topology. According to the method, the required optimal network topology can be efficiently found out, and the method has relatively good compatible expansibility and a good application prospect.

Description

Technical Field

[0001] The present invention relates to the fields of network optimization technology, artificial intelligence and genetic optimization algorithm, and in particular to a method for finding an optimal network topology for distributed multi-short path routing. Background Art

[0002] With the continuous development of information technology, the network has become an infrastructure to support data transmission, resource sharing and information exchange. In artificial network systems such as communication and transportation networks, in order to complete transmission tasks more efficiently and quickly, the collaborative work of routing algorithms is usually required. The shortest path routing and its variants are one of the most widely used algorithms.

[0003] The basic principle of shortest path routing is to guide the transmission object along the shortest path from the starting node to the target node in the network, where the "short" path means low transmission cost, such as short transmission distance and short transmission time. In order to complete the transmission task with the best performance, shortest path routing needs to grasp or accurately predict the transmission cost information of each node and edge in the network in real time, so as to calculate the shortest path suitable for the current network transmission conditions. Shortest path routing is simple to implement, but in actual application, it requires a large amount of computation to calculate the shortest path for the transmission object in real time, and the shortest path using historical records may not be able to adapt to the dynamic changes in network conditions. Therefore, in application scenarios with high requirements for computing power and real-time transmission response, shortest path routing often cannot meet network transmission needs and its scope of use is limited.

[0004] In order to overcome the limitations of the shortest path routing, people have proposed a variety of improvement methods in recent years. Among them, distributed multi-short path routing has attracted attention because it can effectively improve network transmission performance with lower complexity. The basic idea of ​​distributed multi-short path routing is to pre-calculate and store multiple short paths with less overlap between nodes or edges between the transmission start node and the destination node as candidate transmission paths. When the actual transmission occurs, the best path is selected from the candidate paths according to the real-time transmission status of the network as the transmission path of the transmission object. Compared with the traditional shortest path routing, distributed multi-short path routing effectively balances the contradiction between path calculation complexity and transmission efficiency.

[0005] The performance of distributed multi-short path routing depends to a large extent on the design of network topology. Factors such as the spatial distribution of homogeneous nodes in the network and the community structure will significantly affect the network transmission performance of distributed multi-short path routing. Therefore, when actually using it, optimizing the network topology according to the specific application scenario is of great significance to improving the performance of distributed multi-short path routing and achieving efficient network transmission.

[0006] The optimization of network topology often requires comprehensive consideration of multiple factors, such as the scale of the network, the connectivity between nodes, the calculation of short paths, etc. The search space for the optimal topology is huge and the objective function is complex. Traditional optimization methods often find it difficult to effectively solve and give a satisfactory optimal solution in a short time.

[0007] Genetic algorithm is a general optimization algorithm inspired by biological evolution. Its core idea is to simulate the inheritance, mutation, and selection evolution process of biological populations in nature, and gradually approach the optimal solution to the problem through the individual survival mechanism. Genetic algorithm performs well in global optimization problems, especially suitable for solving problems with large search space, complex structure, and diverse or nonlinear optimization objectives. Introducing genetic algorithm in finding the best network topology for distributed multi-short path routing can make full use of the powerful global search ability and adaptability of genetic algorithm to complex problems, and efficiently find the best network topology for distributed multi-short path routing in the application scenario given by the user. Summary of the invention

[0008] In the above background, the present invention proposes a method for finding the best network topology for applying distributed multi-short path routing. The method simulates the evolution process of the network topology and uses a genetic algorithm to complete the search for the best network topology for applying distributed multi-short path routing in an application scenario given by a user, thereby improving the transmission throughput and efficiency of the network when applying distributed multi-short path routing.

[0009] The specific content of the present invention is as follows:

[0010] The invention relates to a method for finding an optimal network topology for applying distributed multi-short path routing. In the initialization stage of the method, a population including μ chromosomes is constructed by generating an ER random connected network, wherein each chromosome represents a network topology, and the network topology includes N nodes and E edges. Based on the framework of a genetic algorithm, T iterations are performed in the process of finding the optimal topology for applying distributed multi-short path routing. In each iteration, μ chromosomes in the previous generation population are used as parent chromosomes, and λ child chromosomes are generated by cloning, mutation and crossover. A new generation population is formed by all parent chromosomes and child chromosomes. A network transmission test using distributed multi-short path routing is performed on the network topology corresponding to each chromosome in the new generation population in an application scenario given by a user. The average transmission completion time of the transmission task is calculated according to the test result, and the negative value of the average transmission completion time is used as the fitness of the chromosome, and the λ chromosomes with the lowest fitness are eliminated. After completing T iterations, the network topology corresponding to the chromosome with the highest fitness in the latest generation population is the optimal network topology suitable for applying distributed multi-short path routing in the application scenario given by the user.

[0011] The cloning operation is to randomly select μ*P from the parent generation cl chromosomes are cloned to generate μ*P cl The network topology corresponding to the daughter chromosomes of each clone is consistent with the network topology corresponding to its parent chromosome.

[0012] The mutation operation is to randomly select μ*P from the parent generation mu chromosomes mutate to generate μ*P mu Daughter chromosomes: For each selected parent chromosome, first generate a daughter chromosome by cloning, randomly select an edge in the network topology corresponding to the daughter chromosome, change one of the endpoint nodes of the edge to connect it to other nodes in the network. If this change destroys the connectivity of the network, the edge adjustment action is undone and the edge adjustment is performed again according to the same rules until the connectivity of the network is not destroyed.

[0013] The crossover operation refers to randomly selecting μ*P from the parent generation cr Crossover the chromosomes to generate μ*P cr Daughter chromosomes: For each pair of selected parent chromosomes, first use the two parent chromosomes to generate a daughter chromosome by superimposing edges. The network topology corresponding to the daughter chromosome contains all the edges corresponding to the network topologies of the two parent chromosomes. If the network topology contains more than E edges, some of the edges in the network topology are randomly deleted so that the number of edges in the network topology after deletion is E. If this change destroys the connectivity of the network, the edge deletion action is undone, and the edge deletion is performed again according to the same rules until the connectivity of the network is not destroyed.

[0014] The beneficial effects of the present invention are as follows: the present invention, combined with a genetic algorithm, can efficiently find the optimal topology structure suitable for applying distributed multi-short path routing in a given application scenario by a user, thereby improving network transmission efficiency. The network topology optimization method provided by the present invention has good performance and compatibility and scalability, can cope with different network scales, application scenarios and transmission tasks, and has good application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is a flow chart of the present invention;

[0016] Figure 2 This is a schematic diagram of the cloning operation of the present invention;

[0017] Figure 3 It is a schematic diagram of the variation operation of the present invention;

[0018] Figure 4 This is a schematic diagram of the crossover operation of the present invention. DETAILED DESCRIPTION

[0019] The following are specific embodiments of the present invention and the accompanying drawings to further describe the technical solution of the present invention, but the present invention is not limited to the following embodiments.

[0020] like Figure 1 As shown, the present invention relates to a method for finding the best network topology for applying distributed multi-short path routing. In the initialization stage of the method, a population including μ chromosomes is constructed by generating an ER random connected network, wherein each chromosome represents a network topology, and the network topology includes N nodes and E edges. Based on the framework of a genetic algorithm, T iterations are performed in the process of finding the best topology for applying distributed multi-short path routing. In each iteration, μ chromosomes in the previous generation population are used as parent chromosomes, and λ child chromosomes are generated by cloning, mutation and crossover. A new generation population is composed of all parent chromosomes and child chromosomes. The network topology corresponding to each chromosome in the new generation population is tested for network transmission using distributed multi-short path routing in an application scenario given by a user. The average transmission completion time of the transmission task is calculated according to the test results, and the negative value of the average transmission completion time is used as the fitness of the chromosome, and the λ chromosomes with the lowest fitness are eliminated. After completing T iterations, the network topology corresponding to the chromosome with the highest fitness in the latest generation population is the best network topology suitable for applying distributed multi-short path routing in the application scenario given by the user.

[0021] Refer to Figure 2 In the example given, the cloning operation refers to randomly selecting μ*P from the parent generation cl chromosomes are cloned to generate μ*P cl The network topology corresponding to the daughter chromosomes of each clone is consistent with the network topology corresponding to its parent chromosome.

[0022] Refer to Figure 3 In the example given, the mutation operation refers to randomly selecting μ*P from the parent generation mu chromosomes mutate to generate μ*P mu Daughter chromosomes: For each selected parent chromosome, first generate a daughter chromosome by cloning, randomly select an edge in the network topology corresponding to the daughter chromosome, change one of the endpoint nodes of the edge to connect it to other nodes in the network. If this change destroys the connectivity of the network, the edge adjustment action is undone and the edge adjustment is performed again according to the same rules until the connectivity of the network is not destroyed.

[0023] Refer to Figure 4 In the example given, the crossover operation refers to randomly selecting μ*P from the parent generation cr Crossover the chromosomes to generate μ*P crDaughter chromosomes: For each pair of selected parent chromosomes, first use the two parent chromosomes to generate a daughter chromosome by superimposing edges. The network topology corresponding to the daughter chromosome contains all the edges corresponding to the network topologies of the two parent chromosomes. If the network topology contains more than E edges, some of the edges in the network topology are randomly deleted so that the number of edges in the network topology after deletion is E. If this change destroys the connectivity of the network, the edge deletion action is undone, and the edge deletion is performed again according to the same rules until the connectivity of the network is not destroyed.

[0024] The present invention can efficiently find the optimal topology structure suitable for applying distributed multi-short path routing in the application scenario given by the user, thereby improving the network transmission efficiency. The network topology optimization method of the present invention has good performance and expansion compatibility, can cope with different network scales, application scenarios and transmission tasks, and has good application prospects.

[0025] The above is a specific implementation case of the present invention, which is used to make the present invention clearer. Any modification or equivalent replacement of the present invention within the spirit of the present invention and the protection scope of the claims shall fall within the protection scope of the present invention.

Claims

1. A method for finding an optimal network topology using distributed multi-short path routing, characterized in that: In the initialization stage of the method, a population containing μ chromosomes is constructed by generating an ER random connected network, where each chromosome represents a network topology, which contains N nodes and E edges. Based on the framework of the genetic algorithm, T iterations are performed in the process of finding the best topology for applying distributed multi-short path routing. In each iteration, μ chromosomes in the previous generation population are used as parent chromosomes, and λ child chromosomes are generated by cloning, mutation and crossover. A new generation population is composed of all parent chromosomes and child chromosomes. The network topology corresponding to each chromosome in the new generation population is tested for network transmission using distributed multi-short path routing in the application scenario given by the user. The average transmission completion time of the transmission task is calculated based on the test results, and the negative value of the average transmission completion time is used as the fitness of the chromosome, and the λ chromosomes with the lowest fitness are eliminated. After completing T iterations, the network topology corresponding to the chromosome with the highest fitness in the latest generation population is the best network topology suitable for applying distributed multi-short path routing in the application scenario given by the user.

2. The method for finding the best network topology using distributed multi-short path routing according to claim 1, characterized in that: The cloning operation is to randomly select μ*P from the parent generation cl chromosomes are cloned to generate μ*P cl The network topology corresponding to the daughter chromosomes of each clone is consistent with the network topology corresponding to its parent chromosome.

3. The method for finding the best network topology using distributed multi-short path routing according to claim 1, characterized in that: The mutation operation is to randomly select μ*P from the parent generation mu chromosomes mutate to generate μ*P mu Daughter chromosomes: For each selected parent chromosome, first generate a daughter chromosome by cloning, randomly select an edge in the network topology corresponding to the daughter chromosome, change one of the endpoint nodes of the edge to connect it to other nodes in the network. If this change destroys the connectivity of the network, the edge adjustment action is undone and the edge adjustment is performed again according to the same rules until the connectivity of the network is not destroyed.

4. The method for finding the best network topology using distributed multi-short path routing according to claim 1, characterized in that: The crossover operation refers to randomly selecting μ*P from the parent generation cr Crossover the chromosomes to generate μ*P cr Daughter chromosomes: For each pair of selected parent chromosomes, first use the two parent chromosomes to generate a daughter chromosome by superimposing edges. The network topology corresponding to the daughter chromosome contains all the edges corresponding to the network topologies of the two parent chromosomes. If the network topology contains more than E edges, some of the edges in the network topology are randomly deleted so that the number of edges in the network topology after deletion is E. If this change destroys the connectivity of the network, the edge deletion action is undone, and the edge deletion is performed again according to the same rules until the connectivity of the network is not destroyed.

Images