Title : AI techniques for broadband photonics MMW communication towards 6G
Abstract:
5G defines below 100 GHz as the millimeter-wave (MMW) bands, whereas 100 GHz - 3 THz is categorized as THz band in 6G. Photonics-aided MMW generation is a key technique in the fiber-wireless network, which overcomes the bottleneck of the deployed electrical devices. Moreover, deep learning (DL) is expected to enable a significant paradigm shift in 6G wireless networks. For W-band (75-110 GHz), we proposed a dual gated recurrent unit neural network based nonlinear equalizer (dual-GRU NLE) for radio-over-fiber (ROF) communication systems. The dual- equalization scheme is mainly based upon GRU algorithm, which can be trained via two steps including I-GRU and Q-GRU. By using the dual-GRU equalizer, 60-Gbps 64-QAM signal generation and transmission over 10-km SMF and 1.2-m wireless link at 81-GHz can be achieved. For D-band (110-170 GHz), we proposed a novel scheme to effectively mitigate the nonlinear impairments in a PAM-8 radio-over-fiber (ROF) delivery by a joint deep neuron network (J-DNN) equalizer, which has more superiority in terms of good training accuracy, satisfactory tracking speed, and over-fitting suppression compared with a typical deep neuron network (DNN) equalizer. By using the proposed J-DNN equalizer, 60-Gbps PAM-8 signal generation and transmission over 10-km SMF and 3-m wireless link at 135-GHz can be achieved. Moreover, a novel complex-valued neural network (CVNN) equalizer using ‘CReLU’ activation function to directly recover PAM-4 signals from received noised signals is demonstrated. D-band 90-Gbps single channel PAM-4 signal generation and transmission over 10-km SMF and 3-m wireless link at 140-GHz can be achieved. Thanks to the aid of traditional mathematical-oriented models including FOE and CPR, the computation burden of CVNN is released significantly. Furthermore, we compare the performance of CVNN and real-valued neural network (RVNN) in terms of BER decision accuracy, time complexity and receiver sensitivity. Followed by the same DSP with the same complexity, the comparison result between DNN and CVNN in the same structure shows that CVNN performs better due to its reservation of phase information. Therefore, we believe that the joint use of modelbased, e.g., FOE, CPR steps and complex DL-based techniques has a potential for the future 6G wireless physical layer algorithms.
What will audience learn from your presentation?
- Figure out what role fiber wireless communications (FWC) could play in Beyond 5G and 6G Wireless Networks.
- Present our proposed high speed FWC achievements beyond 100 GHz.
- Cover the key challenges and perspectives on machine learning techniques for the future 6G wireless physical layer algorithms.
- Provide a complex-valued deep machine learning scheme for the high speed coherent photonics MMW communication.
