Fig. 1. (a) Schematic setup of the proposed quasi-OSSB modulation approach. (b) Illustration of the operation principle in the frequency domain. (c) Evolution of the polarization state.

Fig. 2. The original CSR (in decibels) versus the modulation index m<1.

Fig. 3. The output CSR (in decibels) versus the polarizing angle θ at different modulation indexes m.

Fig. 4. The required θ (left axis) and corresponding LUR (right axis) versus the modulation index m (the output CSR is fixed at its optimum value, 0 dB).

Fig. 5. Radio frequency power oscillation versus LUR. The insert denotes the simulated radio frequency power variation at different fiber lengths. (The radio frequency is fRF=30  GHz, the fiber dispersion index is D=17  ps/km·nm, and the attenuation factor is α=0.2  dB/km.)

Fig. 6. Spectrum response of the 50/100 GHz OI from B-C (dashed line) and from B-D (dotted line), REW=0.01  nm.

Fig. 7. (a) Optical spectra before (blue dotted line) and after (black real line) the OI (B-C) and the transmission response (red dashed line) of the OI. (b) Optical spectra before (blue dotted line) and after (black real line) the OI (B-D) and the transmission response (red dashed line) of the OI.

Fig. 8. Experimental optical spectra when CSR=22, 18, 9.5, 3.5, and 0 dB.

Fig. 9. Experimental optical spectra with different driving frequencies ƒRF=14, 16, 18, and 20 GHz.

Fig. 10. Simulated radio frequency power versus CSR.

Fig. 11. Simulated BER curves and corresponding eye diagrams at BTB and 100 km SMF transmission.