Abstract
This paper proposes an improved itinerary recording protocol for securing distributed architectures based on mobile agents. The behavior of each of the cooperating agents is described, as well as the decision process establishing the identities of offenders when an attack is detected. Our protocol is tested on a set of potential attacks and the results confirm our assumption regarding offender designations and moments of detection. More precisely, the performance evaluation shows that our protocol detects the attack where there is collaboration between a platform on the cooperating agents' itinerary and another on the mobile agent's itinerary. As a result, this protocol constitutes a suitable option for electronic commerce applications where security concerns prevail over cost factors.