Demon-stration of a low-complexity memory-polynomial-aided neural network equalizer for CAP visible-light communication with superluminescent diode
The surge of visible-light communication (VLC) derives from the ever-increasing demand of wireless network capacity and the limited spectral resources in the radiofrequency domain. VLC offers high-speed and high-capacity data links while being free from electromagnetic interference (EMI), and has become a suitable technology for complementing the fifth-generation (5G) or beyond-5G network. Furthermore, VLC has the potential to relieve the wireless communications from radiation-related concerns attributed to the millimeter wavelength infrastructure.
In VLC systems, both the light emitters and detectors are key elements for achieving high-speed communication links. InGaN-based laser diodes (LDs) and light-emitting diodes (LEDs), including micro-LEDs, have been used in the vast majority of the high-speed VLC systems up to date, achieving multi-Gbit/s (multi-Giga-bit-per-second) data rates. A third type of device known as the superluminescent diode (SLD) combines the advantageous characteristics of both LDs and LEDs: Low etendue, high optical power density, low temporal coherence, wide frequency bandwidth, and droop-free emission, which make the SLD attractive for VLC and other applications. In previous research works, non-return-to-zero on-off keying (NRZ-OOK) and discrete multi-tone modulation have been applied in SLD-based VLC. However, carrier-less amplitude and phase (CAP) modulation and machine learning have not been studied in the visible-light communication technology of SLD. In particular, the research of SLD is still in the initial stage, so some linear and nonlinear damages are serious. Machine learning techniques, such as neural network, can help the SLD based visible-light communication system mitigate linear and nonlinear noise as much as possible, and improve the transmission rate of the system. However, at present, the complexity of neural networks is very high, and in the perspective of practicality, low-complexity neural networks deserve more researches. Therefore, the group aims at designing a low-complexity neural network based on polynomial expansion to achieve high-speed SLD based visible-light communication.
Figure.1 SLD device and electro-optical characteristics: (a) Light-output-power – current – voltage (L–I–V) and external quantum efficiency (EQE) characteristics of the SLD at CW injection current. (i) Scanning electron microscope image of the SLD showing the metal contact (on p-side) and the 12° tilted front facet. (b) Electroluminescence (EL) spectra of the SLD under various injection currents. (ii) SLD under operation coupled with a collimating lens. (c) Full-width at half maximum (FWHM) and peak position of the SLD as measured from (b).
Figure.2 (a) The schematic of the Memory-Polynomial-aided Neural Network equalization pre-processing for the received symbols. (b) The structure of the applied Memory-Polynomial-aided Neural Network. (c) The structure of the DNN for comparison.
About The Group
The research group of Prof. Nan Chi from Fudan University and the research group of Prof. Boon S. Ooi from King Abdullah University of science and technology propose an innovative Memory-Polynomial-aided Neural Network (MPANN) for the novel high-speed SLD-based VLC system with CAP modulation. In addition, the device characteristics and photoelectric characteristics of the designed SLD are given in detail. In the visible light communication system, the modulation bandwidth of SLD can reach 400 MHz when the injection current is 700 mA.In addition, this paper also found that there are serious linear and nonlinear noise in the SLD-based visible-light communication system, therefore, this paper proposed a low-complexity memory-polynomial-aided neural network. This paper optimizes the parameters of the network and compares it with traditional linear and nonlinear filters and neural networks in terms of equalization performance and complexity. The results show that the neural network is robust, efficient and of low-complexity .Finally, the ultra-high speed SLD visible light communication of 2.9Gbit/s is realized by the neural network. The article is entitled “Demonstration of a low-complexity memory-polynomial-aided neural network equalizer for CAP visible-light communication with superluminescent diode” and published in Opto-Electronic Advances Vol. 8 2020.
The research group of Prof. Nan Chi from Fudan University has been engaged in the research of visible-light communication and optical fiber communication for a long time, and is one of the leading teams in developing high-speed visible light communication. The research interests of the team include visible light communication, digital signal processing, high-capacity high-speed optical fiber communication, underwater communication and other cutting-edge basic and application technologies.
The research group of Prof. Boon S. Ooi from King Abdullah University of science and technology aims at delivering compact and energy saving integrated laser-diode based devices and solutions for applications requiring light spanning from the ultraviolet to the visible and near-infrared regime. The group has conceived plethora of practical laser-based solutions, and built proof-of-concept models for data collection and testing of new classes of innovative multi-function laser devices. The research interests of the team include visible-light communication, nanostructures, device research and underwater wireless optical communication.
Hu F C, Holguin-Lerma J A, Mao Y, Zou P, Shen C et al. Demonstration of a low-complexity memory-polynomial-aided neural network equalizer for CAP visible-light communication with superluminescent diode. Opto-Electron Adv 3, 200009 (2020)