Research On Complex Information Network Modeling And Performance Evaluation Method

Complex information networks are everywhere in real life,and play an indispensable role.Wired local area network(Wired LAN)and complex networks are widely used.For Wired LAN and complex networks,exploring their performance becomes an important part of network development.Effectively evaluating and predicting network performance is essential to improve the topology of the network and the dynamic transmission capabilities of the network.With the development of internet technology,how to evaluate and predict the effectiveness and invulnerability of complex information networks has become an important research topic.The difficulty of determining the weight of each evaluation index and the complexity of the relationships between the evaluation indexes,so it is a challenge for effectively evaluating network performance.How to make the evaluation result more realistic and objective is the top priority of the network performance evaluation.In this paper,the relationship between network performance and its evaluation indexes is analyzed,and the relationship between the effectiveness and the rate of sending packet is also analyzed.Under two attack modes,node attack and link attack,the invulnerability of Wired LAN and complex networks is analyzed and compared in depth.The cost-effectiveness ratio is critical to the network performance research.Therefore,the cost-effectiveness ratio of Wired LAN and complex networks is analyzed.The main results are as follows:Firstly,an evaluation model is proposed to evaluate the effectiveness and invulnerability of Wired LAN.This model specifically evaluates network performance by structural equation model(SEM).After the evaluation model is obtained,the contribution of the forwarding nodes to the network in terms of effectiveness and invulnerability can be analyzed.The SEM is a statistical method that analyzes the relationship among variables based on the large sample data.Compared to other statistical methods,it can contain several latent variables and consider measurement errors of observed variables.The SEM can also adjust and modify the hypothetical model continuously according to various fitting indexes of the model until all fitting indexes of the model meet the requirements.Network performance can be evaluated by the several wired LAN models,such as star networks,ring networks,and tree networks.At last,simulation results show that the SEM is suitable for the network performance evaluation.Secondly,an evaluation model is proposed to evaluate the effectiveness and invulnerability of complex networks.This model specifically evaluates network performance by principal component analysis(PCA).Different evaluation models are selected according to the testing values of different complex networks.The PCA is a statistical method,which can reduce the dimension of the variables by analyzing sample data.It can improve one-sidedness of the single evaluation index,and performance of complex networks can be predicted according to comprehensive evaluation model.In order to verify that the proposed model is realistic and suitable for a variety of complex networks.Several common complex networks,such as scale-free network and small-world networks with different probability values are chosen to evaluate network performance in this paper.Finally,simulation results show that the PCA is suitable for network performance evaluation as well.