An improved forward integrity protocol for mobile agents

Abstract

Forward integrity is an important mobile agent security property. Forward integrity in mobile agents guarantees that results contained in a mobile agent from previously visited servers cannot be changed by a malicious server. Many mobile agent forward integrity protocols are based on a protocol family proposed by Karjoth et al. This paper identifies a "colluding servers" attack on these protocols where two (or more) visited servers can conspire to modify the existing collected data from other servers. We propose an improved Karjoth protocol by applying split-knowledge when constructing digital signatures, which can defend against the colluding servers attack. © Springer-Verlag 2004.