Feature Analysis On Opencore Application Network Based On Complex Network

SoC technology has become the main design method of large-scale digital integrated circuit today. With the core of embedded system and basis of reusable IP cores, SoC technology is an integrated chip which integrates both software and hardware and pursuits a short time to hit the market. IP cores are widely used in the design of FPGA and ASIC. The use of IP cores could improve the stability of the system, shorten the development cycle and reduce the cost of projects. Opencores website provides engineers with a large number of free and open-source IP cores projects. With the society becoming more and more net-oriented, people tend to demand more reliability of the network. The topology of different networks shows different capacities to network faults. Therefore, it is of great significance to study the topology and stability of the application network formed by IP cores.72 stable IP cores projects described by Verilog language, with source files can being compiled and can being downloaded to the FPGA verification are selected from opencores website and then decomposed into many cores and sub-cores using top-down method. A binary and undirected opencore application network is constructed according to the basic principle of complex network and both the cores and sub-cores are abstracted into nodes, the connections between cores and cores, cores and sub-cores, sub-cores and sub-cores are abstracted into edges. Node degree distribution, clustering coefficient, average path length, modularity, betweenness and other characteristics are used to indicate the topology of the network and to draw that the opencore application network has characteristics of both scale-free network and small world, and in the network only a few nodes are involved in external connections indicating that the network has poor connectivity. The relative size of largest connected subgraphs and global efficiency are introduced to evaluate the stability of the complex network. By comparing the variations of the relative size of largest connected subgraphs, global efficiency, average path length and the maximum path length after different strength of random attack and targeted attacks, we conclude that the network shows strong robustness against random attack but is vulnerable to targeted attacks.