Mediating Hosts’ Malicious Character

Abstract

Right now there isn’t any published solution to protect mobile agents against hosts’ attack. This is known as the problem of malicious hosts. To solve this problem, Minsky proposed an approach based on replication and vote. Within this approach, vote diverging information hosts are considered to be malicious. We consider such distinctive qualification as simple, straightforward and not flexible. Our paper relates malicious property to a more elaborated process. We claim that our process is cautious and flexible. In this process we: (1) Explore how to attach a trust or suspicious quality with a host. This attachment must be temporal. (2) Perform Vote among agents handing out information from trusted hosts only. Suspicious hosts’ information is dismissed. (3) Delegate the faculty to declare the malicious character to an adjudicator. The designation requires a dialogue and could be parameterized with a threshold. (4) Capture trusted hosts’ bizarre behavior and disclose it.