Network Topology Overlapping Group Discovery Based On K-plex And Application In IP Geolocation

The network group discovery has a large number of applied researches and has achieved certain results in network virus prevention,IP entity positioning and network load balancing.A comprehensive analysis of the structural characteristics of the network topology,improvement of the operating efficiency of the existing network topology overlapping group discovery algorithm,and research on a group discovery method that is more suitable for network topology has important practical significance.This paper focuses on the Internet field,and conducts research on network topology overlapping group discovery technology and geolocation application.The main work and innovations are as follows:(1)Aiming at the application limitations of the overlapping group discovery algorithm in the existing research on the network topology,h-core decomposition is applied to the overlapping group discovery algorithm of the network topology graph.This paper proposes a network topology overlapping group discovery algorithm based on h-core pruning.The algorithm introduces the h-core idea to filter out a large number of noisy nodes that affect the efficiency of the algorithm,decompose the network into smaller and tighter connected components,and retain the locally densely connected target subgraphs on the target network.The overlapping group detection of the network topology is realized by enumerating the k-plex groups in the network.Experiments on real AS network data sets of different scales show that this algorithm can not only effectively reduce the number of extremely small groups,but also can find almost all k-plex groups larger than the controllable parameters within the effective time.Under the premise of a higher degree of restoration of the full enumeration algorithm,the operating efficiency of the network topology overlapping group discovery algorithm is improved,and the network size that can be handled by the discovery of extremely large k-plex groups is increased by two orders of magnitude.(2)Aiming at the problem that the positioning accuracy of network entities in the existing research is limited by the separation strategy of the backbone network and the positioning accuracy of the IP nodes in the overlapping area is limited,an IP positioning algorithm based on target group discovery is proposed.Put forward the concept of rsubgroups and set the central point to increase the priority of the target node,reserve the nodes closer to the target node in the topology,and guide the group to discover the direction of the target IP.The algorithm applies overlapping group discovery to the location of network entities,and combines voting strategies to determine the geographic location of the network group,thereby obtaining the geographic location information of the target node,and further assessing the credibility of the location results of the IP nodes in the overlapping group.A large number of experiments have proved that the algorithm has better performance in multiple scenarios.Compared with the NNC positioning algorithm,this algorithm uses fewer voting nodes and uses more voting nodes than NNC,and the positioning effect is better.In the urban topology in Henan Province,using the same voting nodes,the positioning accuracy of this algorithm is 29.1% higher than that of the NNC algorithm.