Online optimization of bidirectional amplifiers in long-haul fiber-optic time transfer systems based on improved genetic algorithm

Bidirectional erbium-doped fiber amplifiers (Bi-EDFAs) are widely adopted in long-haul fiber-optic time transfer systems to compensate for the optical loss in the fiber links. The optical gains of the Bi-EDFAs should be elaborately optimized as they will induce unwanted noises at the same time, resulting in significant deterioration of the signal-to-noise ratio (SNR). In the practical long-distance fiber links with multiple Bi-EDFAs involved, the online optimization of the optical gains for the best SNR becomes a hard task. In this paper, we proposed a novel method of optimizing the optical gains of the working Bi-EDFAs in the fiber link by using an improved genetic algorithm, which can achieve an equivalent SNR result to that with the idealized solution based on the enumerative method meanwhile with the calculation time no long increases exponentially with the increased number of Bi-EDFAs. This method was further verified experimentally on 276 and 439 km-long coiled fiber links, in which three and five Bi-EDFAs were respectively incorporated. For both cases, the achieved SNRs showed more than 15 dB improvement compared with the original strategy of fixing the optical gains. Furthermore, compared with the results based on the idealized enumerative method, the SNRs were only 2 dB lower, while the iteration time was shortened by three orders of magnitude.