A LOCAL-CLUSTERED EVOLVING NETWORK MODEL
Chao Li,Hai Zhao,Shaoqian Yuan
2008-01-01
Abstract:In this paper, we propose an improved network model with the conception of local cluster to describe the evolving phenomenon. We find the passivation effect during the evolving process, which makes degree distribution of the network, follows a power- law distribution with a cut-off. The formula of the mean cluster coefficient of the model is calculated in some given situations, and the result is validated through a numerical simulation. The performance comparison of the improved model with BA and LW model under attacks and errors suggests that networks with high cluster coefficient is more fragile than we supposed. Keywords: Local-world, Local-cluster, Passivation effect, Scale-free 1. Introduction. Recently, we have witnessed a substantial new movement in network research, with the focus shifting away from the analysis of regular and random graphs to complex network. The new approach has been driven largely by two innovative researches (1,2), which depicts the network dynamic mechanism of the small world effect and scale- free property respectively. To depict the evolving growth mechanism of network comprehensively, people have pre- sented modified models based on scale-free network (3). Dorogovtsev and Mendes (4) have considered the aging effect on nodes. Li and Chen (5) have found that preferential attach- ment rule limited in certain local area of the whole network, and they proposed a local world evolving model (LW model), which constructs network whose degree distribution ranges between the power law and exponent. Although the local evolving conception was introduced by LW model, the way that selected the local world randomly adopted by the LW model was out of accord with the real case. For instance, regional groups in world trade network divided by areas have already been determined before the entry of new countries, and the member within the groups have established close relationship. The probability that selecting the local world by connecting edges between the nodes in random choice is quite small, therefore, the way that selects the local world should be improved. In this paper, an improved local clustered model (LC model) is present, which selects ag roup of affinitive nodes as the local world. The degree distribution of the LC model suggests that there exist critical values, nodes with degrees higher than the critical value are more preferred to be selected, and we denote it as phenomenon passivation. With the passivation effect during the evolving process, the generated network follows a power-law degree distribution with a cut-off, and heterogeneity of the network decrease gradually as the scale of local world increases. Moreover, LC model has steady network mean cluster coefficient compared with BA and LW model as the scale of network grows. For its high mean cluster coefficient, LC model is more fragile than BA and LW model under malicious attack and random errors.