Design of Novel Optical Fiber Communication Electronic System and Big Data Prediction Method of Its Loss

Abstract

For the shortcomings of traditional data communication electronic mode and bus interface, optical fiber communication electronic technology and compact peripheral component interconnect (CPCI) bus interface mode with industry standard are adopted to design a kind of optical fiber communication electronic data transmission mode. The optical fiber communication electronic board based on CPCI bus constitutes a novel optical fiber communication electronic system. In the process of hardware design, the small form-factor pluggable (SFP) transceiver module is used to convert the photoelectric signal and complete the transmission of optical fiber data. Regarding the existence of differential signal in high-speed serial circuit of electronic board, the wiring is reasonably arranged. The hardware test platform of optical fiber communication electronic system is built based on PCI9054 CPCI bus and industrial personal computer (IPC) communication electronics. FPGA with embedded high-speed serial transceiver module is adopted to design the hardware logic function. Binary gray code is used to count. FIFO protocol and Aurora protocol are supported. Protocol IP core is customized to realize the full duplex electronic communication of high-speed data. Considering the loss problem of the system, the grey model GM(1, 1), ARIMA model and Elman neural network are employed as the big data level research scheme, which are combined into the loss prediction model according to a certain proportion. In the experiment, the test of DMA data reading and writing function is designed, and the embedded highspeed transceiver module and optical fiber communication electronic logic function are simulated. The results show that the symbol error rate (SER) is always 0, and the low-pass filtering characteristics of the channel are effectively compensated, which makes the signal quality stabler. Compared with the single big data level research scheme, the combined model has smaller prediction error and higher prediction accuracy.