The Design And Implementation Of Security Strategy For Protecting Mobile Agent From Malicious Host

Mobile agent is a program that can migrate autonomously among hosts in the heterogeneous network. It determines when and where to migrate, clone itself and create sub-agent that cooperate to finish the complicated task. The characteristic calculate can be migrated to the server where resources exist. It has following advantages: open, flexibility, robustness, scalability, efficiency, and supporting mobile customer. Because mobile agent migrates independently, and network is open and untrusts, mobile agent is facing serious security threat. If this problem can't be solved, it will be impossible to apply this technology in those areas that need security. Therefore, the security of mobile agent has become a research hot spot of scientific research workers. Designing of a safe and reliable mobile agent system has important theoretical significance and practical value.The issue about malicious hosts is the focus and difficult of research in the mobile agent system now. The mobile agent will migrate to the target host and run according to the will of source host. It is obvious that mobile agent is completely exposed to the host. The host which runs the mobile agent can modify code, data and status. Agent may also be attacked maliciously in the network transmission .So on the basis of analyzing effective policies which have been put forward , we proposes an effective security policy based on trust, which can prevent attacking from malicious host. The policy applies the trust to the mobile agent system. Before the mobile agent migrating, we got the host which in the pass of the agent to terminable host, then figured out the trust of these hosts, choose host which had a higher trust level made static routing path protected migration path. This will minimize the number of malicious hosts in migration path, thereby reduced the possibility of mobile agent attacked by malicious hosts. In the migration process, using SSL guarantee the safety of communication, agent and host should verify identity firstly, then host detected code and state ,agent authorize host access control under the trust. Thus configure the safety management system and strategy documents of Java, as well as sand-box technology to protect the implementation of the code, eventually logging. When agent returned to the source host, based on the use of the reference state after inspection methods, source host detected the state information and log records collected by agent to accomplish the task, finally, update the trust level of the host.This dissertation introduces some basic knowledge about mobile agent and cryptography firstly. On the basis of analyzing security issues and introducing the existing security strategy which can prevent hosts from maliciously attacking,the mobile agent security strategy based on trust has been proposed. Through analyzing the strategy show that the security strategy in this dissertation can solve flexibly multiple security attacks in mobile agent. Finally, the simulation experiment verified the effectiveness and feasibility of this strategy.