Figure 5

An overview of NIST randomness test results for the raw and post-processed two-bit sequences generated at a rate of 1 Mbps recorded from the arrangement as shown in Fig. 3. The raw data is of 3 Mbits long and passes 13 out of 15 tests. Toeplitz post-processing brings the bit length down by only 0.5% (post-processed bit length ∼ 2.985 Mbits) while passing all the tests. XOR post-processing also passes all statistical tests but results in a loss of 50% bits. The dashed line (\(p = 0.01\)) represents the critical p-value above which the sequence is considered random. These results indicate that no statistical patterns were found in the tested samples of random bits. The same analysis done on two-bit generated at 50% visibility in the experimental setup for noise revealed that raw data passes only 5 tests while XOR post-processed data passes 11 tests. Similarly, the raw data generated at 60% visibility passes only 4 tests, but is able to pass 12 tests after XOR post-processing