Structure Analysis of Function Call Network Based on Percolation

In software security analysis, selecting more important vertices can effectively improve the efficiency. Based on the theory of complex network, we form the complex function call network according to the call relation in binary software. The percolation of this network is carried out based on the index of vertex degree, betweenness centrality, the PageRank value and the number of neighbor connections. A comparison of vertex deletion was conducted to test the change of network structure by different strategies. The quantitative experiment shows that the two percolation processes of random deletions and preference deletions have great difference on the network structure, which indicates that the different index can be used to judge the importance of vertex.