Update README.md

Added SP800-90B results.
This commit is contained in:
Pol Henarejos
2022-08-22 14:08:58 +02:00
committed by GitHub
parent a79296b423
commit 4137861a21

View File

@@ -89,8 +89,9 @@ Fig. 1 depicts the distribution of the bytes, from `0` to `255`. Ideally, it mus
Fig. 2 depicts the distribution of chi square tests and excess percentages. Ideally, the distribution of chi square tests must follow a chi square distribution with mean at `255`. Excess percentage distribution must follow a uniform distribution.
Pico-rng is also tested with `ent`, `rngtest` and `dieharder` tests. These are the results obtained with pico-rng:
Pico-rng is also tested with `ent`, `rngtest`, `sp800-90b` and `dieharder` tests. These are the results obtained with pico-rng:
### Ent
```
$ ent sample.rng
Entropy = 7.999999 bits per byte.
@@ -114,6 +115,7 @@ Serial correlation coefficient is 0.000006 (totally uncorrelated = 0.0).
These results show that pico-rng is pretty random.
### FIPS 140-2
```bash
$ cat sample.rng | rngtest
@@ -140,6 +142,27 @@ In these results, we have a total number of trials `429496 (429141+355)`.
- The acceptable result of Monobit test is `1` failed trial for every `9662` trials. For `429496` trials, the number of failed trials should be less than `429496/9662=44.45`.
- The acceptable result of Poker test is `1` failed trial for every `10078` trials. For `429496` trials, the number of failed trials should be less than `42.62`.
### SP800-90B
SP800-90b is the superseded version of FIPS 140-2 (`rngtest`). Can be obtained from here [https://github.com/usnistgov/SP800-90B_EntropyAssessment](https://github.com/usnistgov/SP800-90B_EntropyAssessment).
```bash
$ ./ea-iid sample.rng
Calculating baseline statistics...
H_original: 7.995587
H_bitstring: 0.999863
min(H_original, 8 X H_bitstring): 7.995587
** Passed chi square tests
** Passed length of longest repeated substring test
** Passed IID permutation tests
```
### Diehard
It is a set of different tests.
```bash
$ dieharder -a -g 201 -k 2 -Y 1 -m 2 -f sample.rng
#=============================================================================#