For investigating the proposed scheme, the encoder and decoder proof-of-principle experiment for QNSC system is set up, which converts the 4-bit parallel digital streams into QNSC signal and realizes the decryption of ciphertext based on the optical DAC. To demonstrate 4-bit operation, one plaintext signal and three running keys with the bit rate of 10Gbps are fed into the DMLA consisting of 4 directly modulated lasers with different wavelengths. The center wavelengths of 4 directly modulated lasers cover from 1549.32 nm to 1551.72 nm with a separation of 0.8 nm. By adjusting the driving current of DMLA, the output optical power of each directly modulated laser is about − 8 dBm. The outputs of 4 directly modulated lasers are combined in ODCA. And a FODA is used to synchronize the binary optical signals as accurately as possible. After the multi-wavelength mixed signal is fed into OEOC with a bandwidth about 18 GHz. Consequently, the QNSC signal with single-wavelength is generated.
The measured evolution of the output waveforms with the average optical power for encoder is shown in Fig. 4. The red areas represent the amplitude signal level distribution of the data 1, and the red areas correspond to represent the data 0. It is shown that the signal amplitude level of data 1 decrease exponentially with the decrease of the average optical power, meanwhile, the signal amplitude level of data 0 increases logarithmically with the decrease of average optical power. However, as average optical power is small enough, both the two levels will tend to a same level. It should be noted that when the average optical power is small enough, the level spacing of 0 and 1 is too small to be distinguished under the fluctuation of quantum noise.
The relationship between the average optical power of encoder and signal-to-noise ratio (SNR) is shown in Fig. 5. It is noted that the SNR of the signal increases with the increase of the optical power. When the average optical power of least-significant bit for optical DAC is less than − 19dBm, the SNR is less than 0dB. It should be noted that the sufficient quantum noise masking was achieved and the gap between two adjacent level is so small that the attacker cannot discriminate the adjacent levels. While the average optical power of most-significant bit for optical DAC is more than − 15dBm, the SNR is more than 6dB, so it can keep the signal qualities adequately, the gap between the two levels is large enough to allow for proper discrimination of data 0 and 1.
The experimentally observed signals of the optical encoder are shown in Fig. 6. The logic value of the running key with 3-bit from PRGN are set to (11010101001100011101), (10100110001110101010), (10011000111010101001), and the amplitude of each signal is 7.05mV, 15.90mV, 31.99mV, the average optical powers are corresponding to -19.98 dBm, -16.97 dBm, -13.98 dBm, as shown in Fig. 6 (a) ~ 6 (c) respectively. Moreover, the logic value of input binary data of plaintext is (10101001100011101010), and the amplitude of the signal is 65.30mV, and the average optical power is -11.07dBm, as shown in Fig. 6 (d). It should be noted that the parallel signals with different wavelengths are weighted by the factors of about 2–4, 2–3, 2–2, 2− 1, and then combined incoherently by ODCA. The QNSC waveform with single wavelength is measured in Fig. 6(e) by OEOC corresponding to Eq. (2), and the binary characteristics of plaintext in Fig. 6 (d) have been disturbed. The eye-diagram is measured in Fig. 6(f), it seems that the eye-diagram of QNSC signal is closed, because the plaintext with binary signal is encrypted with 24-level and masked by noise.
The structure of optical DAC in decoder is the same as that of encoder, and the observed waveforms of the optical DAC for the decoder are shown in Fig. 7, experimentally. The waveforms of the DMLA which are controlled by the output of PRNG2 are shown from Fig. 7(a) to Fig. 7(d). Obviously, the signals in the Fig. 7(a) and Fig. 7(b) are the same as those of Fig. 6(a) and Fig. 6(b), respectively. To generate threshold signal, the logic value of input binary data is set to (01100111000101010110) in Fig. 7(c), which is the reverse of the signal logic value in Fig. 6 (c). In addition, the signal logic value in Fig. 7 (d) is the same as that in Fig. 6 (c), but the average optical power corresponds to that of Fig. 6 (d). So, the threshold of the decision waveform is achieved, as shown in Fig. 7(e).
The decoded signals of the QNSC are shown in Fig. 8. Where the waveform of bits as received by legitimate user is shown in Fig. 8(a), it should be noted that the waveforms in Fig. 6(d) and Fig. 8(a) are similar, corresponding to the correct plaintext. In addition, the clear separation of the 0 and 1 histogram allows him to make bit decisions with 2.19×10–11, as shown in Fig. 8(b). At the same time, the waveform from the eavesdropper in wiretap channel as shown in Fig. 8(c) in which it can take all the optical signals from main channel that would have gone to legitimate users, except for the correct key K, it should be noted that the received signal is still a M-ary signal masked by noise. Moreover, the corresponding histogram in the Fig. 8(d) shows that distinct bits are not distinguishable, which it is giving a very high probability of error with 0.425 in bit decisions because of lack of the key.