Coordinated Multi-Input and Single-Output Photonic Millimeter-Wave Communication in W-Band Using Neural Network-Based Waveform-To-Symbol Converter

• DOI: 10.3390/photonics12030248
• Publication Date: 2025-03-10T10:59:59Z
• Authors: Kexin Liu; Boyu Dong; Zhongya Li; Yinjun Liu; Yaxuan Li; Fangbing Wu; Yongzhu Hu; Junwen Zhang
• Keywords: Symbol (formal); Extremely high frequency

Photonic millimeter-wave communication systems are promising for high-capacity, high-speed wireless networks, and their production is driven by the growing demand from data-intensive applications. However, challenges such as inter-symbol interferences (ISIs), inter-band (IBIs), symbol timing offsets (STOs), nonlinearity impairments exist, especially in non-orthogonal multiband configurations. This paper proposes demonstrates neural network-based waveform-to-symbol converter (NNWSC) a coordinated multi-input single-output (MISO) photonic system with multiplexing. The NNWSC replaces conventional matched filtering, down-sampling, equalization, simplifying receiver enhancing interference resilience. Additionally, it reduces computational complexity, improving operational feasibility. As proof of concept, experiments conducted 16QAM carrierless amplitude phase (NM-CAP) modulation MISO configurations scenario where two base stations have 5 km 10 fiber links, respectively. Data were collected across various roll-off factors, sub-band spacings, received optical power (ROP) levels. Based on proposed method, (mmWave) at 91.9 GHz demonstrated transmission speed 30 Gbps. results show that NNWSC-based achieves significant bit error rate (BER) reductions compared to receivers all tolerances STO also studied. These findings highlight integration solution high-frequency, interference-prone environments, potential improvements low-SNR dynamic scenarios.