Modulation format identification based on constellation diagrams in adaptive optical OFDM systems

Adaptive orthogonal frequency division multiplexing (OFDM) systems are promising for high-bit-rate short-reach communication by optimizing the allocation of modulation format and power to each subcarrier. Because adaptive OFDM systems involve multiple modulation formats, automatic modulation format identification (AMFI) is significant for subsequent signal processing and symbol decision of OFDM receivers. In this work, we proposed and experimentally demonstrated a blind modulation format identification technique based on constellation diagrams achieved from channel estimation of OFDM system. In the proposed AMFI scheme, the identification feature is the number of signals' constellation clusters. First, we use the peak-density clustering algorithm to track the centers of signals' constellation clusters by plotting the density-distance graph, where the clusters' centers have much larger density and distance values than that of the other clusters' points. Then, we use the K nearest neighbor regression algorithm to automatically calculate the number of signals' constellation clusters. Finally, we evaluated and measured the identification accuracies of QPSK, 8-QAM, 8-PSK, 16-QAM, 64-QAM and 128 QAM signals in simulation and experiment.