Inferring sequences produced by pseudo-random number generators
Journal of the ACM (JACM)
Random number generators: good ones are hard to find
Communications of the ACM
Numerical recipes in C (2nd ed.): the art of scientific computing
Numerical recipes in C (2nd ed.): the art of scientific computing
The art of computer programming, volume 2 (3rd ed.): seminumerical algorithms
The art of computer programming, volume 2 (3rd ed.): seminumerical algorithms
Cryptography: Theory and Practice,Second Edition
Cryptography: Theory and Practice,Second Edition
PGP: Pretty Good Privacy
| Hi-index | 0.00 |
Theoretically, using any Linear Congruence Generator (LCG) to generate pseudo-random numbers for cryptographic purposes is problematic because of its predictableness. On the other hand, due to its simplicity and efficiency, we think that the LCG should not be completely ignored. Since the random numbers generated by the LCG are predictable, it is clear that we cannot use them directly. However, we shall not introduce too much complication in the implementation which will compromise the reasons, simplicity and efficiency, of choosing the LCG. Thus, we propose an easy encryption method using an LCG for email encryption. To see how practical in predicting random numbers produced by an LCG, we implement Plumstead's inference algorithm [2] and run it on some numbers generated by the easiest congruence: Xn+1 = aXn+ b mod m. Based on the result, we confirm the theoretical fault of the LCG, that is, simply increasing the size of the modulus does not significantly increase the difficulty of breaking the sequence. Our remedy is to break a whole random number into pieces and use them separately (with interference from another source, in our case, English text). We use 16-bytes random numbers and embed each byte of the random number as noise in one text character. In such a way, we can avoid revealing enough numbers for the attacker to predict.