Answers are grounded in this paper and cite their sources.
We experimentally demonstrated Intensity-Modulated Direct-Detection (IMDD) single-channel 1,040 km linear transmission and 800 km non-linear transmission at 10 Gb/s over standard single-mode (G.652) fiber, without any optical dispersion compensation or mitigation, using a Maximum-Likelihood Sequence-Estimation (MLSE) receiver employing the square-root (SQRT) branch metric with off-line processing. These experiments were designed as to probe the limits of the MLSE approach. They successfully showed that long-haul uncompensated transmission is in principle possible with MLSE, even in the presence of large uncompensated dispersion and strong intra-channel fiber non-linearities, provided that enough complexity can be built into the receiver. In the linear 1,040 km experiment, a Bit Error Rate (BER) of 10(-3) was achieved with an Optical Signal-to-Noise Ratio (OSNR) penalty with respect to back-to-back of 2.9 dB, using two samples per bit and 16,384 trellis states. Several other set-ups were tested as well, including the use of only one sample per bit and fewer trellis states. In the non-linear 800 km experiment, power was ramped up to 12 dBm, exciting substantial Kerr non-linearity, whose induced spectral-broadening exacerbated the effects of the large uncompensated dispersion of the link. Using an MLSE receiver with 1,024 states, we demonstrated a non-linear threshold of 9 dBm. We benchmarked this experiment towards simulations addressing various electrical and optical dispersion compensation strategies. We also carried out an analysis of error run-lengths, on both experiments, which showed that error burstiness may change considerably depending on the number of processor states, OSNR and the amount of non-linearity in the link.
No takes yet. Share an insight, caveat, or question.
Poggiolini et al. (2008) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: