Low-Complexity Adaptive Background Compensation of Coherent Optical Transmitters

Next generation optical communications will demand coherent transceivers with data rates in the 1.6 Terabit-per-second range and beyond, requiring both high symbol rate and large constellations. In this scenario coherent systems become increasingly sensitive to transmitter imperfections such as In-phase/Quadrature time skew, amplitude and phase errors, bandwidth limitations and mismatches, etc. These impairments must be compensated to achieve satisfactory performance. In this work, we propose a novel and efficient technique for the precompensation of such impairments, requiring minimum additional hardware in current transmitter designs. This scheme is adaptive, as it readjusts itself to cope with time and environment dependent impairments, and background, as interruptions of normal operation are not required for the continuous recalibration of the system. Naturally, such in-field adaptation demands the real-time system identification of the coherent optical transmitter, which is performed from the sole information provided by a low bandwidth photodiode/analog-to-digital converter chain at the transmitter output. This fact represents a meaningful advance with respect to the usual channel estimation methods, which require the full bandwidth coherent demodulation of the optical output. A theoretical foundation for the proposed scheme is presented. Numerical simulations of a realistic coherent optical transmitter show excellent results for high symbol rate modulation formats.

Automated summary

Coherent transceivers with high data rates in the Terabit-per-second range and beyond are required for next generation optical communications. Precompensation of transmitter imperfections such as time skew, amplitude and phase errors is proposed in this work using a novel and efficient technique. The scheme is adaptive and can adjust to time and environment dependent impairments in real-time without interrupting normal operation. The proposed scheme only requires low bandwidth information from the transmitter output, providing a meaningful advance compared to usual channel estimation methods.