Performance Analysis of a 28 GHz Wideband Analog Radio-over-Fiber Fronthaul with Channel Nonlinearity Compensation

Analog radio-over-fiber (ARoF) fronthaul emerges as a possible solution to overcome the fronthaul bottleneck for 5G and beyond mobile networks. ARoF technology exhibits better spectral efficiency compared to the traditional common public radio interface (CPRI) that is based on digitized transport of wireless signals. The deployment of ARoF links at millimeter-wave (mmWave) bands has added advantage of being able to support ultra-broadband wireless signal transmission, but the performance is subjected to channel nonlinearity and noise. The transport of mmWave modulated optical signal is also subjected to fiber chromatic dispersion when transported in a double sideband format. In this paper, we investigate a 28 GHz mmWave-based ARoF fronthaul link with 1 GHz single wideband transmission over 10 km fiber using optical single sideband modulation format to combat the dispersion-based power fading. We propose, investigate, and demonstrate three approaches for compensating channel nonlinearity, including novel digital modulation, i.e., modified cut-and-paste-based probabilistic shaping (CAP-based PS), and directly using nonlinear post-equalizers, i.e., Volterra series-based equalizer and neural network-based (NN-based) equalizer. The experimental results show that the CAP-based PS approach using signal transformation at the transmitter and the receiver can improve the bit-error-rate (BER) performance for 64-quadrature amplitude modulation (64-QAM) with low computation complexity, but introduces slight data rate loss. On the other hand, both Volterra and fully connected NN-based equalizers can achieve BER< 3.8 × 10
^-3
for 64-QAM without data rate loss, with a superior BER obtained by NN equalizer under a relatively increased computation complexity. Furthermore, the NN equalizer applied after orthogonal frequency division multiplexing (OFDM) 16-QAM using least-square equalization also achieves an improved BER below 3.8 × 10
^-3
, which highlights the NN equalizer's capability of addressing channel distortions under different modulation formats. These practical channel nonlinearity compensation approaches provide promising solutions for addressing channel nonlinearity in the future deployment of ARoF fronthaul, especially for fronthaul operating at mmWave bands.