9.7. Further Reading
Ellison explores methods of binding an identity to a public key without using certificates [297].
The Internet Security Association and Key Management Protocol [599] deals with key exchange and authentication on the Internet. Several key exchange protocols are based on classical cryptosystems [146, 686]. Protocols based on public key methods abound (see, for example, [682, 705, 895, 951]).
Key generation is based on random numbers generated from physical phenomena [12, 234, 289, 307, 539, 740]. Generating keys pseudorandomly is tricky [711]; the most common method, using polynomial congruential generators, is not safe [128, 129, 532, 746]. Rabin [738] and Adleman, Pomerance, and Rumley [10] discuss generating large prime numbers for use in RSA; their method relies upon pseudorandom number generation.
Several papers discuss issues in public key infrastructure, including interoperation [451, 452, 761], organization [558, 579], requirements [37, 762], and models [207, 714]. Park and Sandhu [710] have proposed extensions for X.509v3 certificates. Adams and Lloyd [7] discuss many aspects of public key infrastructures.
Merkle [621] notes that certificates can be kept as data in a file. Changing any certificate changes the file. This reduces the problem of substituting faked certificates to a data integrity problem.
Key escrowing allows the recovery of data if the cryptographic key is not accessible. The best known such system is the U.S. government's Escrowed Encryption Standard (EES) [116, 251, 665, 667, 678] Beth, Knobloch, Otten, Simmons, and Wichmann [86] identify five desirable properties of such a system; Ganesan [346] developed Yaksha, which meets all of these. Denning and Branstad [246] discuss the architecture of key escrow systems.
Several key escrow schemes explore different ways to control access. Burmester et al. [145] present a protocol with a limited time span. Several authors discuss the nontechnical aspects of the proposed U.S. key escrow system (for example, see [628, 794, 866]). Clark [176] and Walker et al. [929] discuss the relationship between key recovery and key escrow. Others have proposed enhancements and extensions of various Internet protocols for key recovery [53, 593, 798].
Translucent cryptography [66, 67] allows some fraction of the messages to be read. This is not a key escrow system, because the keys are not available, but it does serve the ends of such a system in that the messages can be read with a specified probability.
Digital signature protocols abound. One standard, the DSS [666], uses a variant of El Gamal [294]; Rivest and others have criticized some of its features [755]. Others, especially those associated with the ITU's X.500 series of recommendations, recommend (but do not require) RSA. Grant's book [372] discusses digital signatures in general and presents many case studies.
The electronic commerce protocol SET [812, 813, 814] uses dual digital signatures to tie components of messages together in such a way that neither the messages nor their association can be repudiated. Ford and Baum [330] discuss SET and the supporting infrastructure. Ghosh [353] provides a balanced view of the dangers of Internet commerce using the Web.
|