Effect of key generators on the automatic search for flaws in key management schemes

Abstract

The protection of financial data in major networks depends upon cryptography, and the security of such systems is heavily dependent upon the schemes employed to distribute cryptographic keys over the network. The complexity of key management schemes increases with the size and diversity of the total system under protection; previous papers have described automatic means of searching for security flaws in such systems. Experience in this field has indicated that key generators, employed to produce random cryptographic keys for use in the network, may be significant features of attack scenarios. The nature of the key generator differs from other key management functions, inasmuch as it provides for a multiple output from a single input. This type of function conflicts with the tree search strategy developed for the automatic search routine. This paper describes the means by which key generators may be incorporated in such a routine, and gives an attack strategy that could be so detected.