In this section, the overall process of mining log file togenerate policy rules and its architecture will be presented asshown in Fig. 1.The process consists of the following five iterativecomponents in sequence: (1) to analyze and generate an initialset of firewall policy rules, to collect firewall log raw data, andto extract the attributes for data mining, (2) to perform datamining of Association Rule Mining and Mining firewall Logusing Frequency) along generalization to discover a new set ofthe updates to the initial firewall policy rules, (3) to identifydecaying rules and dominant rules, and (4) to edit and generatethe updated firewall policy rules, generalized and anomaly-free.12 1+%/)8*8#($#:*-";%//#<(/*&)#=./"8#Initially the firewall contains a set of firewall policy rulesspecified by network security administrator in knowledge of itsorganization’s security policy. Each rule specifies to accept ordrop (deny) a packet of certain attributes. For example, a rulecould specify a packet for Hyper Text Transport Protocol(HTTP) used to transfer HTML documents of the World Wideon port 80 with a specific Source IP Address as shown in Fig. 2.Each rule could consist of up to seven attributes. Theseattributes are in the following format of “ ”. The first attribute of direction isthe packet direction whether it is in-coming or out-going. Thesecond attribute is the protocol of packet such as TCP or UDP.The next attributes are two pairs of IP address and its port forsource and destination of packet. Finally the last attribute is theaction upon packet being accepted or dropped by firewall.Further an attribute may not be present in a firewall rule toimply all or any. An attribute may be a range such as the rangeof TCP port from 1024 to 65535 or the range of IP address in alocal area network by mask.
การแปล กรุณารอสักครู่..
