Is SecureRandom really necessary for generating initialization vectors or is it random enough?

Question

Is SecureRandom really necessary for generating initialization vectors or is it random enough?

In order for the stream cipher to be protected from repeated key attacks, IV should not be repeated. But does SecureRandom have an advantage over simple, unprotected Random in this regard (or is it just to create an unpredictable sequence)?

Assuming I use fixed-size messages with AES CBC mode, and I generate a new Random for each IV (using the current time nano as a seed), does this increase the likelihood of repeating IV compared to SecureRandom?

+5

java random cryptography encryption

user9397757 Sep 03 '12 at 14:03

source share

4 answers

48- , 2 ^ 48 . , long . . , SecureRandom, .

+1

Peter Lawrey 03 . '12 14:06

, IV . , , . , XOR'ed . , IV , , , , ECB. , , .

0

Maarten Bodewes 04 . '12 0:17

IV, , . , , , IV.

The strongest random number generators that are used generate just a few tens of bits of entropy per second. Most algorithms see IV as secondary to the key. Therefore, a slow and strong RNG must be reserved for the key and for sowing a fast, long RNG period for IV.

0

Lunatic experimentalist Sep 05 '12 at 5:13

source share