Intelligent bandwidth-estimation technique for orthogonal frequency division multiplexing-based elastic optical networking

Abstract—In orthogonal frequency division multiplexing (OFDM)-based elastic optical networking (EON), it is imperative to identify the parameters of the unknown OFDM signal intelligently, quickly, and robustly. Among these parameters, the bandwidth should be the first to be identified for the reason that other useful parameters such as number of subcarriers, baud rate, and frequency spacing can be obtained subsequently. In this paper we propose, for the first time to our knowledge, an intelligent bandwidth-identification technique for OFDM-based EON. The proposed technique can be divided into three sub-stages: estimation of power spectrum density, noise filtering based on empirical mode decomposition, and bandwidth identification based on a sliding window. When the number of samples is fixed at 215 and the optical signal-to-noise rate varies from 15 to 25 dB, the simulation results demonstrate that a bandwidth range from 2 to 40 GHz can all be successfully recognized with a minimum resolution of 39.1 MHz, and the average estimation absolute accuracies (EAAs) exceed 95%. Moreover, the technique has a good tolerance to chromatic dispersion and is independent of the number of subcarriers and frequency offsets. Finally, we verify the effectiveness of this technique by an experiment involving a 7.04 Gbps OFDMbased EON with minimum EAA of 94.12%, in which its bandwidth can be successfully recognized.