Efficient network topology detection method based on gradient guidance

Abstract

The invention discloses an efficient network topology detection method based on gradient guidance, which is applied to the technical field of network topology detection, and aims at solving the problem that the coverage rate of nodes in a network detected in the prior art is not ideal. The method comprises the following steps: firstly, randomly selecting some IPs from a network segment to be detected as initial IPs, and carrying out Traceroute detection on the initial IPs; training a neural network by utilizing the mapping relationship between a detected destination IP and an approach coverageIP; then calculating the gradient of the neural network to the input IP; guiding the IP to carrying out mutation based on the gradient to generate a new IP address, and iterating the process. The destination IP measured by the method is used for Traceroute, the final point coverage rate is calculated, more IP addresses can be detected, a larger network structure can be covered, and a more complete network topology can be estimated.

Description

technical field [0001] The invention belongs to the field of the Internet, and in particular relates to a network topology detection technology. Background technique [0002] In the 21st century of the information age, with the rapid development of Internet technology and the rapid expansion of network scale, the current Internet is indispensable to affect all aspects of human life. The data of the "Statistical Report on China's Internet Development Status" released in August 2018 shows that the Internet has profoundly changed human life. In order to better monitor, manage and build the Internet, understand the network topology and explore the structure of the network It is also becoming more and more important to characterize and analyze the behavior of the network. [0003] The scale of the current Internet continues to expand, and the structure of the network has become extremely complex, making it difficult to understand the network. In order to ensure the correct and ...

AI Technical Summary

Problems solved by technology

However, the efficiency of this method is very low, and its limitations are: due to the large network address space, the distribution of network IPs is sparse in some areas, but dense in other areas. By randomly selecting IP addresses for measurement, the The covered area may be very small. The ratio of the nodes in the network detected in the detection to all the nodes is defined as the point coverage rate. The result of the point coverage rate is not very ideal.

Using completely randomly generated input to detect the program, it is difficult to completely discover the loopholes in the program, because many boundary value conditions in the program are difficult to trigger

Images