Controlling Security of Software Development with Multi-agent System

Software systems become distributed and complex. Distributed systems are crucial for organizations since they provide possibility to share data and information, resources and services. Nowadays, many software systems are not developed from scratch: system development involves reuse of already developed components. However, with the intrusion in the computer systems, it has become important that systems must fulfill security goals and requirements. Moreover, interdependencies of components create problems during integration phase. Therefore, security properties of components should be considered and evaluated earlier in the lifecycle. In this paper, we propose an agent-oriented process that supports verification of fulfillment of security goals and validation of security requirements during different phases of development lifecycle. Moreover, the system needs to support mapping of security requirements to threat list to determine if any of the attacks in the list is applicable to the system to be developed. This is performed by the meta-agents. These meta-agents automatically create a security checklist, as well as, provide control of actions taken by human agent.