Optimizing elastic optical networks quality of transmission with an empirical optical amplifier gain saturation model

Abstract

In this paper, we revisit the problem of adjusting transceiver power, amplifier gain, and nodal gain in optical-bypass-enabled networks, taking into account an empirical model of erbium-doped fiber amplifier (EDFA) gain saturation. We assume that network demands have been routed via the shortest possible paths. A heuristic algorithm, which estimates the degradation of the signal-to-noise ratio (SNR) for each demand, is used to assign the modulation level and spectrum slots to the routed demands. Our optimization process aims to maximize the minimum SNR margins of demands, by adjusting EDFA small-signal gain, variable optical attenuator (VOA) attenuation, and transceiver power. Constraints and impairments imposed by the physical-layer include EDFA and VOA gain limits, transceiver power limits, receiver sensitivity, insertion loss of optical switch input ports, EDFA gain saturation, EDFA noise, and fiber non-linear interference. These are modeled using the Gaussian noise model. The proposed optimization problem is formulated as a geometric programming problem and then converted into an equivalent convex problem.Numerical results suggest that in the optimal solution of our optimization, all demands are served with typically low bit error rates (BER) (<10−4). We calculate the operational gains of the EDFA to set the spans to full compensation whenever possible. Pre-amplifiers and line amplifiers do not experience gain saturation due to relatively small input powers, while approximately 10% of boosters undergo more than a 1 dB gain reduction due to high input powers. We observe that overlooking the effect of EDFA gain saturation and physical-layer-imposed constraints such as nodal gain limits or transceiver power limits during network optimization leads to a considerable number of blocked demands due to SNR degradation or high BER (>10−3).