Research On Cascading Failure Of Mission-Oriented Complex Communication Networks

As a critical national information infrastructure,communication networks provide services for various applications and missions at the upper level.However,the actual network environment is very complicated,and at the same time,cascading disasters or failures frequently occur in the communication network,resulting in large-scale service interruptions in the network.It is challenging to construct a simple and accurate model to describe the communication network and its missions.This thesis conducts model-ing by extracting critical features in the network,analyzes the impact of these features on cascading failures in specific mission scenarios,and further studies the network vul-nerability based on the cascading failure model,and designs the recognition network in consideration of cost heterogeneity Algorithm of critical nodes.The main research contents of this thesis are:(1)Construct a cascading failure model of a complex communication network.In the actual network environment,the communication network is very complicated.This thesis extracts the communication network parameters from the engineering perspective and combines the network science domain knowledge to model the complex communica-tion network.By extracting the key features of the complex communication network,the constructed model is divided into two parts: the network communication model and the cascading failure random model.In the network communication model part,the layered architecture is mainly considered.The network is divided into an edge layer and a com-munication layer.The communication between edge layer hosts is mapped to the commu-nication layer,and the communication layer is responsible for the storage and forwarding of data packets.In the cascading failure random model,according to the congestion of the router,it is assumed that the router has three states: normal,congested,and overloaded,and the three states switch to each other.The probability of the router being overloaded is related to the current packet loss rate.(2)Vulnerability analysis of complex communication network based on cascad-ing failure model.This thesis proposes a network vulnerability analysis framework based on the cascading failure model and analyzes the vulnerability of complex commu-nication networks when the network faces different attack strategies from router perfor-mance,routing algorithm,and network structure.Through a large number of experiments,it has been found that the enhancement of router performance can effectively deal with random network failures,but it has no effect when the network is maliciously attacked?routing algorithms with congestion avoidance can effectively alleviate cascading failures,but routing shocks will occur under certain conditions? The real network and the scale-free network still have the characteristics of ”robust and fragile” under the cascade failure model.(3)Mission-oriented complex communication network cascade failure model and vulnerability analysis.There are missions with different characteristics in com-munication networks,and different missions will affect the distribution and dynamics of traffic in the network.This thesis proposes a network mission model to model missions based on communication interaction behaviors and traffic generation.It mainly considers P2 P and C/S missions and uses the ON/OFF model to simulate traffic generation.Af-ter many simulation experiments,it is concluded that due to the continuity and volatility of traffic under P2 P missions,the network is more prone to cascading failures and more fragile.(4)Identification of critical nodes in mission--oriented complex communication networks based on heterogeneous costs.In view of the problem that the more impor-tant nodes in the network have higher attack costs,a heterogeneous cost model based on node value is proposed,assuming that the node attack cost is exponentially related to the node value.The heterogeneity of attack costs increases the difficulty of identifying criti-cal nodes.This thesis designs a directed simulated annealing algorithm.Adding an initial solution generation strategy and a directed perturbation strategy based on average cost to the traditional simulated annealing algorithm dramatically increases the convergence speed and effectiveness of the algorithm.Finally,through a large number of experiments on simulated networks and real networks,it is found that with the increase of cost het-erogeneity,low-value nodes play an increasingly important role.Compared with P2 P or hybrid missions,servers under C/S missions The corresponding edge router has a greater impact on network performance.