LAUSR

Abstract In this paper, we propose a successive interference cancellation (SIC) technique, which uses K-means clustering algorithm for a standard single mode fiber (SSMF) based multi-user on-off keying (OOK) modulated transmission system. The transmission system employs power division non-orthogonal multiple access (PD-NOMA) scheme to enhance the capacity. In this work, we have experimentally shown the efficacy of our proposed direct detection (DD) technique for 20 Gbps PD-NOMA signal transmissions at 1.55 μ m wavelength. The results show that the unsupervised machine learning (ML) based DD receiver can produce BER, which is well below the pre-FEC limit, for the data rate of 5 Gbps, 5.5 Gbps, and 10 Gbps after 50 km of standard single mode fiber (SSMF) for far user. Using the ML algorithm, the maximum measured power penalty for aggregated 20 Gbps per channel data rate is measured as 2.7 dB at BER = 1 0 − 3 for far user. The proposed system does not include any compensation technique for mitigating fiber induced impairments and still can detect the data for near and far user, which is 50 km away from the transmitter.