Protecting Autonomous Agents By Using Ccf

Sy Huang,Cl Huang,Q Chen,Cs Gao
2004-01-01
Abstract:How to use cryptographic methods to protect agent from malicious host is the key issue in the research of mobile agent security. Recent approaches based on computing with "encrypted functions" are limited to the case where only the agent originator learns the result of the computation and the host running the agent must not notice anything at all, furthermore, the agent can not give the real time response to the executing host. In this paper, a novel approach based on CCF (crypto-computing with feedback) technique is introduced for supporting autonomous agent to accomplish secure computation on the remote hosts. CCF is such a method to accomplish iterative computation of encrypted function, in which the cipher-text output will be feedback and reused as a new input of next round secure computation. For the garbled circuit has the ability to compute the cipher-text form input directly without decrypting them first, so it is adopted as the basic tool in CCF. Our approach makes it possible that agent can roam around remote nodes and accomplish secure computation without interaction with the agent originator and need not the help of the third trust-party, in addition, the agent can give real time response to the executing host according to the current state of agent and current inputs of host.
What problem does this paper attempt to address?