Assessment of laser noise influence on direct-detection transmission system performance

A new semi-analytical method of estimating the influence of laser noise on directly modulated direct-detection transmission system performance is proposed. Its accuracy is assessed for different propagation situations and receiver noise type dominance using rigorous mixed semi-analytical simulation. Two different relative intensity noise (RIN) powers at receiver input associated with the two laser currents of the binary levels are considered, and not just a single one corresponding to the laser average current as in other published works. It is shown that, to perform rigorous simulation, suitable methods of solving stochastic laser rate equations should be used, and a number of samples high enough in order to stabilize the sensitivity estimator should be considered. It has been found that, for power penalties of about 1 dB, more than 150 samples may be necessary. For a power penalty due to laser noise of about 4.5 dB, discrepancies not exceeding 0.4 dB are observed between the results obtained using the new semi-analytical approach and using the rigorous simulation. It is shown that the new semi-analytical approach gives more precise results than other approach currently used. For a power penalty due to laser noise not exceeding 2 dB, the discrepancies between the two approaches can reach roughly 1 dB. It is shown that these discrepancies can increase with the laser linewidth. Explicit expressions for system power penalty due to the RIN at the fiber output are presented, for both signal-independent and -dependent noise dominance. It is shown that these expressions give good upper bounds of the real power penalties.