|
DARPA ITO Sponsored Research |
Security for Distributed Computer Systems MIT Laboratory for Computer Science |
| Project Website: | http://theory.lcs.mit.edu/~cis/darpa/home.html -- Additional project information provided by the performing organization |
| Quad Chart: | Quad Chart provided by the performing organization |
| Objective: | Our objective is to address critical issues in the design and implementation of secure distributed computing systems. We focus on some selected applications as a means of identifying and working on general issues. The design of secure communications protocols and public-key certificate infrastructures are central themes, but we also investigate fundamental theoretical constructs required to define and analyze our protocols and higher-level issues such as portability and ease-of-use for new users. |
| Approach: | Our approach is to identify issues and problems whose resolution is likely to yield the greatest benefit in terms of providing technology for the implementation of secure distributed computing systems, and to provide both theoretical foundations and solutions for the problems identified and practical (implementation) studies to illustrate our proposed solutions. We are also interested in participating in relevant standards activities, when appropriate. |
| Recent FY-97 Accomplishments: | We have designed and implemented version 1.0 of SDSI (a Simple Distributed
Security Infrastructure), which promises to provide a more powerful and
elegant way to organize public-key certificates for secure distributed
computing systems. An associated effort designed and implemented a
user-interface for SDSI 1.0.
We have designed and implemented in Java an electronic voting protocol. Based on an earlier proposal by Fujioka, Okamoto, and Ohta, the scheme allows voters to use their Web browers to vote securely in an election. We finished a first round of testing of our newly proposed public-key cryptosystem (PKC) whose security is based on the difficulty of finding the closest vector to a point in a lattice. Based on our experimentation, we propose a set of parameters for an initial use of this PKC. In particular we suggest that working with a lattice in dimension 250-300 is not vulnerable to current techniques. For our experiments we used an implementation of the LLL lattice reduction algorithm due to the LiDIA group. We have initiated a study of the security of the Digital Signature Standard (DSS) when used in conjunction with various pseduo random number generators (PSRGs). In particular, we totally break DSS, retrieving the secret key of the signer, if it uses the linear congruential PSRG or any if its famous variante such as the `truncated' linear congruetial generator. More generally, we show that using any pseudo random generator whose sequeneces can be expressed as solutions of a system of modular linear equations, is insecure within DSS, . We propose yet another public-key cryptosystem and for which breaking the system can be proved at least as hard in the average case as is solving the worst case instance of the problem of finding the non-zero `unique' shortest vector in a lattice. Our system is free of decryption error, whereas previously proposed system with the same security guarantee by Ajtai-Dwork of IBM suffers a 1 in n chance of decryption error for every bit, where n is roughly the size of the public key. |
| FY-98 Plans: | We plan to complete the design of SDSI 2.0 and its merger with SPKI
(so it will become SPKI/SDSI 2.0), and to implement SPKI/SDSI 2.0 in both
C and Java.
We plan to improve the portability of the electronic voting scheme. We plan to finish the implementation of an incremental cryptographic editor, which will support cryptographic functions such as siging, hashing, and encrypting documents in incremental steps in the bacground, while the documents are being modified in the forground. |
| Technology Transition: | There has been tremendous interest in SPKI/SDSI 2.0, and we hope to be able to provide C and Java implementation to interested organizations and users during the upcoming year. The current implementation was for SDSI 1.0, and has not been a subject of technology transition at the implementation level. (At the conceptual level, it has had considerable influence.) |
| Principal Investigator: |
Shafi Goldwasser Lab for Computer Science, Room NE43-332 545 Technology Square Cambridge, Massachusetts 02139 617-253-5914 617-258-8682 fax shafi@theory.lcs.mit.edu |
| Co-Principal Investigator: |
Ronald Rivest Lab for Computer Science, Room NE43-322 545 Technology Square Cambridge, Massachusetts 02139 617-253-5880 617-258-9738 fax rivest@theory.lcs.mit.edu |