Multi-Eigenvalue Demodulation Using Complex Moment-Based Eigensolver and Neural Network

Optical eigenvalues originating in optical solitons are the potential for becoming information carriers not affected by chromatic dispersion and nonlinear effects fibers. They obtained attributing associated eigenvalue equations deduced solving Schrödinger equation with inverse scattering transform (IST) to matrix problem, maintaining constant values regardless of transmission distance. The communication systems require solve soliton-by-soliton. While effective solution methods have been studied well telecommunication systems, one most well-known is QZ decomposition-based method. However, algorithm requires a large complexity. To reduce complexity, method demodulate using complex-moment solver (CME) was investigated. CME parallelizable that can extract any eigenvalue. This paper proposes novel demodulation combines an artificial neural network (ANN) based on employing on-off encoded discrete modulation scheme. ANN sensitive input order units; therefore, must be sorted. A lightweight sorting hence required. In addition proposed scheme, this study partial ANN. Here, 2000 km fiber-transmission experiments on-off-encoded four-discrete- were conducted. experimental results indicated bit error rate (BER) characteristics comparable conventional devising extraction range CME.