Network correlation plays a key role in the research of Internet topology. Current researches focus on the clustering, mixing and rich-club characteristics separately. This paper makes a deep study on the three network correlation characteristics. First, it points out the possible inconsistency between mean clustering and clustering coefficient, which are two main metrics for measuring the clustering characteristic. Then it shows that the local clustering coefficient is highly correlated with the nodes’ degree. After a deep study of the PFP (positive-feedback preference) model, the intrinsic mechanism governing the rise of rich-club phenomenon is discovered and verified. The work is extended by exploring the relationships between these network correlation characteristics.