A Generation Method of Network Security Hardening Strategy Based on Attack Graphs

2.1. Generation of Attack Graph

Attack graph generation goes through a period from manual to automatic. In the study of attack graph generation, researchers propose two models respectively: one is based on system state transformation (Swiler et al., 2001) and the other is based on the dependencies of vulnerabilities and system states (Ammann et al., 2002). The former has a problem of state explosion with problem scale growth. The latter gets widespread attention due to its good scalability. Ammann et al. (2002) first propose the monotonic hypothesis of ability of attackers, providing a speedup for generating attack graph from exponential to polynomial time. A range of attack graph generation algorithms is proposed (Ingols et al, 2006; Ou et al., 2006; Carvalho & Teng, 2010; Jajodia & Noel, 2010), and generation tools are developed, such as MulVAL (Ou et al., 2005), CAULDRON (Jajodia & Noel, 2007) and NetSPA (Ingols et al., 2009). The time complexity of attack graph generation algorithms is further reduced.