Passive optical networks (PON) have become one of the main solutions for optical access networks because of their low cost and large bandwidth capacity. With the continuous increase in bandwidth demand, such as high-definition videos and online conferences, in recent years, existing PON technology has faced capacity bottlenecks. Recently, few-mode fiber (FMF)-based mode-division multiplexing (MDM) technology has been proposed, which is expected to further improve the capacity of PON by introducing a new multiplexing dimension to support higher transmission rates and more users. However, for the mode-division multiplexing passive optical network (MDM-PON), which divides users by mode, the mode crosstalk in the FMF causes the user signals to be loaded on different modes to interfere with each other, which affects the transmission performance. To further improve the transmission performance of the MDM-PON, we propose a MIMO pre-equalization based mode crosstalk mitigation method for the MDM-PON and build a simulation system with VPI Transmission Maker for verification.
Owing to the point-to-multipoint structure of the MDM-PON downlink, it is impossible to simultaneously receive and eliminate mode crosstalk for all modes at the receiver. Therefore, an MIMO pre-equalization based mode crosstalk mitigation method is proposed in this study. To estimate the downlink channel impulse response, we design a time-division training sequence that is inserted into the frame header. The time-division training sequence consists of the same number of time slots as the modes, and each time slot corresponds to only one mode and contains the corresponding training symbol sequence. The receiver of each mode uses coherent detection and adopts a training sequence-based least mean square (LMS) adaptive algorithm for channel estimation. The channel estimates are fed back to the transmitter for pre-equalization. The transmitter-side MIMO equalizer used in this study has a linear structure and uses the feedback channel impulse response to calculate the tap coefficients based on the zero-forcing (ZF) criterion. Considering the inter symbol interference (ISI) caused by fiber chromatic dispersion, we use cascaded FIR filters after the MIMO equalization for pre-compensation at the transmitter. Because the impulse response and dispersion coefficient of the chromatic dispersion are known, the tap coefficients of the FIR filter used for dispersion pre-compensation can be directly calculated.
By adjusting the transmitted optical power in the range of -20 dBm to -44 dBm, we analyze the performance of each mode under OBTB and 5 km FMF transmission, during which the bit error rate (BER) threshold of 7% hard decision forward error correction (HD-FEC) is adopted. The curves of the BER with respect to the transmitted optical power with and without crosstalk under the OBTB and 5 km FMF are shown in Fig. 6. When the BER meets the 7% HD-FEC threshold, compared with tranmission without crosstalk, the transmitted optical power of each mode is increased by 1 dB, 2.8 dB, 2.5 dB, and 2 dB in tranmission with crosstalk. After connecting the 5 km FMF, the transmitted optical power of the LP01 mode is increased by 2.8 dB compared with the transmission without crosstalk. The BER curves of the other three modes can only be close to or at the threshold. The above results show that the influence of the mode crosstalk from the FMF is greater than that of the mode multiplexer/demultiplexer, which is the main factor of performance degradation in the current system. We further study the BER performance before and after using MIMO pre-equalization and compare it with the ZF-pre-coding based crosstalk mitigation. According to the results in Fig. 7, after using MIMO pre-equalization and ZF pre-coding, the BER performance of each mode in transmission with and without crosstalk is improved. When the BER meets the 7% HD-FEC threshold, compared with the ZF pre-coding, the transmitted optical power of each mode using MIMO pre-equalization is reduced by 1.6 dB, 1.2 dB, 1.5 dB, and 1.4 dB under OBTB and is reduced by 3.0 dB, 4.1 dB, 1.2 dB, and 9.2 dB under 5 km FMF. The above results show that MIMO pre-equalization performs better than ZF pre-coding in both cases because it can eliminate the ISI caused by modal dispersion and the differential mode group delay after connecting the FMF. Additionally, the adaptive channel estimation used in MIMO pre-equalization is more accurate than the least squares (LS) estimation used in ZF pre-coding. Considering the influence of fiber chromatic dispersion, we analyze the BER performance before and after using pre-dispersion compensation based on MIMO pre-equalization. As shown in Fig. 8, when the BER meets the 7% HD-FEC threshold, the transmitted optical power of each mode using pre-dispersion compensation is 0.9 dB, 2.0 dB, 1.2 dB, and 2.4 dB less than without pre-dispersion compensation.
We propose an MIMO pre-equalization based mode crosstalk mitigation method for MDM-PONs and build a simulated MDM transmission system for verification. In both cases of OBTB and 5 km FMF transmission, we compare the performance of the proposed method with that of ZF pre-coding based crosstalk mitigation method. The results show that when using four LP modes (LP01, LP11, LP21, LP31) for 4×25 Gbaud QPSK transmission and the BER meets a 7% HD-FEC threshold, compared with the ZF pre-coding, the transmitted optical power of each mode using MIMO pre-equalization is reduced by 1.6 dB, 1.2 dB, 1.5 dB, and 1.4 dB under OBTB, whereas it is reduced by 3.0 dB, 4.1 dB, 1.2 dB, and 9.2 dB under 5 km FMF. The above results demonstrate that the proposed method can effectively mitigate the mode crosstalk in MDM-PONs and exhibits better performance for ISI caused by modal dispersion and differential modal group delay compared with linear pre-coding. Moreover, considering the influence of fiber chromatic dispersion, the performance of MIMO pre-equalization is further improved after cascading dispersion pre-compensation.
