Broadband Wavelength Conversion for Hybrid Multiplexing Signals Based on a Parallel Dispersion-Engineered Silicon Waveguide

Abstract

Broadband all-optical wavelength conversion (AOWC) for hybrid wavelength-and mode-division multiplexing (WDM-MDM) signals is experimentally demonstrated based on degenerate four-wave mixing in a silicon chip with a parallel dispersion-optimized multimode nonlinear waveguide and mode (de)multiplexers. By simultaneously coupling two modes into the waveguide using an optical fiber array, the intermodal crosstalk is measured to be as low as −25.9 dB for the fundamental mode of the transverse electric mode (TE 0 ) (or −23.6 dB for the first-order mode TE 1 ), and the conversion efficiency is −25.4 dB for TE 0 (or −26.3 dB for TE 1 ) mode. A wide conversion bandwidth of ∼68 nm is measured except for the influence of the crosstalk, which is the first time to experimentally demonstrate the broadband wavelength conversion for MDM signal. Using a 4 × 10 Gbit/s on-off keying (OOK) hybrid WDM-MDM signal, four AOWC channels are obtained on the idlers and the power penalty of each channel is less than 2.7 dB at the bit-error-ratio of 1 × 10 −9 .

On‐chip all‐optical multicasting of mode‐division multiplexing QPSK signals

“…The waveguide is with a height of 220 nm and a width of 770 nm, which is dispersion‐engineered for the TE 0 and TE 1 modes to realize their zero‐dispersion wavelengths are both near 1550 nm. The conversion bandwidths of waveguide are measured to be about 68 nm for the two modes [17]. The non‐linear coefficients of the two modes at 1550 nm are calculated to be 139.2 W −1 m −1 and 134 W −1 m −1 , respectively.…”

“…In the waveguide, a series of FWM processes occur among the incident MDM‐QPSK signal S and the dual pumps P 1 and P 2 , including inter‐mode and intra‐mode processes in principle. The waveguide length is optimized to be 2.9 mm to ensure the intra‐FWM efficiency and to suppress the inter‐modal crosstalk by increasing the phase mismatches [17]. The intra‐mode FWMs can be divided into two series: degenerate FWMs (2 P 1 → S + I 1 and 2 P 2 → S + I 2 ) and non‐degenerate FWMs ( P 1 + S → P 2 + I 3 , P 1 + P 2 → S + I 4 , and P 2 + S → P 1 + I 5 ).…”

“…In the past, on-chip mode-selective wavelength conversion for MDM signals has been carried out, just to satisfy the phase-matching condition for a single mode [15]. Moreover, we have demonstrated the wavelength conversion for MDM signal by carefully designing the silicon waveguide structure to support multiple FWM processes simultaneously [16], as well as realized the wavelength conversion for hybrid WDM-MDM signals [17]. However, AOM for MDM signals has not been reported so far.…”

On-chip all-optical multicasting of mode-division multiplexing QPSK signals

“…In the waveguide, a series of FWM processes occur among the incident MDM-QPSK signal S and the dual pumps P 1 and P 2 , including inter-mode and intramode processes in principle. The waveguide length is optimized to be 2.9 mm to ensure the intra-FWM efficiency and to suppress the intermodal crosstalk by increasing the phase mismatches [17]. The intramode FWMs can be divided into two series: degenerate FWMs (2P 1 →S + I 1 and 2P 2 →S + I 2 ) and non-degenerate FWMs (P 1 + S→P 2 +…”