Multiple intelligent agent supported Internet security system: issues, current solutions, and a proposed approach

The Internet has become a common target to attack because of security concerns. Reports of incidents, such as attempted and successful intrusions, have grown dramatically. Several studies have shown that many individuals and companies are abstaining from joining the Internet simply because of such security concerns. The article presents an innovative approach to Internet security which is based on a multiple intelligent agent architecture. This multi agent based security system is focused on one network domain (subnet) only. Four kinds of agents are deployed in the system. They are User Agent, Domain Security Agent (DS Agent), Access Control Agent (AC Agent), and Audit Agent. A Security Information Repository is used to store security information for the network domain, including access information, user profiles, application profiles, authentication keys, etc. These agents communicate with each other through the Information Broker which runs as a co-ordinator not only for the security agents but also for all of the accessible applications in the subnet. A firewall is used as the interface between the subnet and the Internet. Its main function is to filter out the requests which intend to bypass the Information Broker. Additionally, in order to handle the requests from users in other subnets, agents may contact their peer agents in the remote network through the Internet.