A Hardware Implementation Of A Quadrature Amplitude Modulation Communication System With Mimo Applications PDF Download

Coherent detection with digital signal processing (DSP) is currently being deployed in longhaul optical communications. Dual-polarization (DP) quadrature phase shift keying (QPSK) is a modulation format suitable for long-haul transmission (1000 km or above). Another modulation, DP-16-QAM (quadrature amplitude modulation) has been deployed recently in metro regions (between 100 and 1000 km). Extending the reach of DP-16QAM is an active research area. For short-reach transmission (shorter than 100 km), there is still an open question as to when the technology will be mature enough to meet cost pressures for this distance. In this dissertation, we address mainly on phase recovery and polarization demultiplexing in digital coherent receivers for short-reach applications. Implementation of real-time Gbaud (Gsymbol per second) optical coherent systems for singlecarrier higher-level modulation formats such as 64-QAM depends heavily on phase tracking. For offline DSP, decision-directed phase recovery is performed at the symbol rate with the best performance and the least computational effort compared to best-known algorithms. Real-time implementations at Gbaud requires significant parallelizing that greatly degrades performance of this algorithm. Hardware parallelization and pipelining delay on the feedback path impose stringent requirements on the laser linewidth, or the frequency noise spectral level of laser sources. This leads to the paucity of experiments demonstrating real-time phase tracking for 64- or higher QAM. We experimentally investigated the impact of opticallyfiltered lasers on parallel and pipelined phase tracking in a single-carrier 5 Gbaud 64-QAM back-to-back coherent system. For parallelization levels higher than 24, the optically-filtered laser shows more than 2 dB improvement in optical signal-to-noise ratio penalty compared to that of the same laser without optical filtering. In addition to laser phase noise, parallelized phase recovery also creates greater sensitivity to residual frequency offset induced by the presence of sinusoidal tones in the source. Sinusoidal frequency modulation may be intentional for control purposes, or incidental due to electronics and environmental fluctuations. We experimentally investigated the impact of sinusoidal laser phase noise on parallel decision-directed phase recovery in a 5 Gb 64-QAM system, including the effects of frequency offset compensation and equalization. MIMO (multi-input multi-output) FIR (finite-impulse response) filters are conventionally used for polarization demultiplexing in coherent communication systems. However, MIMO FIRs suffer from acquisition problems such as singularity and long convergence for a certain polarization rotations. To reduce the chip power consumption required in short-reach coherent systems where differential group delay is not prominent, we proposed a novel parallelizable DSP architecture. Our approach introduces a polarization pre-rotation before MIMO, based on a very-coarse blind SOP (state of polarization) estimation using only a single Stokes parameter (s1). This method eliminates the convergence and singularity problems of conventional MIMO, and reduces the number of MIMO cross taps responsible for cancelling the polarization crosstalk. We experimentally presented a tradeoff between hardware reduction and performance degradation in the presence of residual chromatic dispersion for short-reach applications. Finally, we extended the previous blind SOP estimation method by using a low-complexity discrete-time extended Kalman filter in order to reduce the memory depth and redundant computations of the previous design. We experimentally verified that our extended Kalman filter-based polarization prerotation at ASIC rates enhances the clock tone of polarization-multiplexed signals as well as the bit-error rate performance of using reduced-complexity MIMO for polarization demultiplexing.