[Cryptech Tech] Statistical Checks on the Random Numbers

[Russ Housley]

(Resending without the cut-and-paste error.)

I have started looking at the random values provided by Joachims.  Here are some initial results using the FIPS 140 tests:
--  The Consecutive Value Test 
--  The Monobit Test
--  The Poker Test
--  The Runs Test
--  The Long Run Test

For each of the data sets, I ran the test 48 times, each one consuming 20,000 bits.

filtered_concat_data_2014-06-07.bin
-- One Constant Value Test failure (Iteration = 3)
  ----  Number of 2-byte Block Repetitions = 1

filtered_fpga_entropy_2014-06-08.bin
-- Two Constant Value Test failures (Iteration = 12 and 39)
  ----  Both times, number of 2-byte Block Repetitions = 1
-- Seven Runs Test failures (Iteration = 1, 2, 12, 24, 33, 34, 41, and 46)

filtered_fpga_entropy_2014-06-10.bin
-- 47 Poker Test failures
-- 48 Run Test Failures

filtered_mix_data_2014-06-06.bin 
-- No failures !

Russ