Two-Lane DNN Equalizer Using Balanced Random-Oversampling for W-Band PS-16QAM RoF Delivery over 4.6 km

Abstract

For W-band long-range mm-wave wireless transmission systems, nonlinearity issues resulting from photoelectric devices, optical fibers, and wireless power amplifiers can be handled by deep learning equalization algorithms. In addition, the PS technique is considered an effective measure to further increase the capacity of the modulation-constraint channel. However, since the probabilistic distribution of m-QAM varies with the amplitude, there have been difficulties in learning valuable information from the minority class. This limits the benefit of nonlinear equalization. To overcome the imbalanced machine learning problem, we propose a novel two-lane DNN (TLD) equalizer using the random oversampling (ROS) technique in this paper. The combination of PS at the transmitter and ROS at the receiver improved the overall performance of the W-band wireless transmission system, and our 4.6-km ROF delivery experiment verified its effectiveness for the W-band mm-wave PS-16QAM system. Based on our proposed equalization scheme, we achieved single-channel 10-Gbaud W-band PS-16QAM wireless transmission over a 100 m optical fiber link and a 4.6 km wireless air-free distance. The results show that compared with the typical TLD without ROS, the TLD-ROS can improve the receiver‘s sensitivity by 1 dB. Furthermore, a reduction of 45.6% in complexity was achieved, and we were able to reduce training samples by 15.5%. Considering the actual wireless physical layer and its requirements, there is much to be gained from the joint use of deep learning and balanced data pre-processing techniques.