Linearity Compensation Method for Silicon-Based Modulator Based on Enhanced Maximum Ratio Combined Receiver
Qin Jun,Tao Yuansheng,Jin Ming,Han Changhao,Gangwar Rahul Kumar,Li Yueqin,Sun Jian,Miao Min
DOI: https://doi.org/10.3788/AOS231033
2023-01-01
Acta Optica Sinica
Abstract:Objective Silicon-based modulators feature small size, low power consumption, and easy integration. However, compared with lithium niobate modulators, they suffer poor linearity, which limits their performance in analog communication systems such as radio over fiber access networks. Various improvement methods have been proposed to improve the linearity of silicon-based modulators, including optimizing the p-n junction design, modifying the doping concentration of the p-n junction, and adopting novel waveguide structures, electrode structures, and driving methods. However, these methods generally require altering the physical characteristics or structures of the devices or adding additional driving circuits. The modulator linearity is typically fixed once the device fabrication or packaging is completed, which makes it difficult to change afterward. Currently, there is a lack of compensation schemes for Si modulator linearity after device fabrication or packaging. Therefore, we want to propose a way from the system perspective to conduct the performance compensation caused by the poor linearity of Si modulators. Methods In our paper, a novel enhanced maximum-ratio combined receiver (EMRC-Rx) is proposed and demonstrated through proof-of-concept experiments, and it is conducted to mitigate the system performance degradation caused by the low linearity of Si modulators when the modulators are deployed in passive optical network (PON)-based access networks. The EMRC-Rx leverages the advantages of both direct detection receiver (DD-Rx) and lite coherent detection receiver (Lite CO-Rx) by utilizing the maximum signal-to-noise ratio contribution from both the receiver types to significantly improve receiver sensitivity and mitigate the system performance degradation. The proposed EMRC algorithm considers the contribution of multiple Lite CO-Rx components to the output signal-to-noise ratio, thereby increasing the proportion of signal-to-noise ratio in the lite coherent receiver and further enhancing the receiver sensitivity. As a result, the EMRCRx in the Si modulator system could achieve similar performance compared with the MRC-Rx in the lithium niobate modulator system. The EMRC-Rx consists of three components including DD-Rx, Lite CO-Rx #1, and Lite CO-Rx #2 (Fig. 3). The results of the three components are aggregated and calculated by the EMRC algorithm from Equation 1 to obtain the final output of the EMRC-Rx. The corresponding digital signal processing flow for DD-Rx, Lite CO-Rx # 1, and Lite CO-Rx # 2 is illustrated in Fig. 3. Results and Discussions The experimental results show that when the bit error rate (BER) exceeds the KP4-FEC threshold at 1. 0 x 10(-4), the receiver sensitivity of EMRC-Rx is improved by 5. 5 dB and 8. 8 dB compared with standalone DD-Rx and Lite CO-Rx respectively, with corresponding improvements in error vector magnitude (EVM) of 32. 5% and 41. 1% (Figs. 8 and 9). Finally, the system performance is significantly improved. Through further comparative experiments with lithium niobate modulators, the EMRC-Rx based on Si modulators can improve the receiver sensitivity by 3. 5 dB and 7. 9 dB respectively compared with the Lite CO-Rx and DD-Rx employing lithium niobate modulators (Fig. 11). A comparable system performance with the MRC-Rx in the lithium niobate modulator is realized. The results indicate that the EMRC-Rx can compensate for the performance degradation caused by the low linearity of Si modulators. For the entire experimental system, the optimal range for the frequency spacing between the downlink and uplink optical carrier is 10 GHz to 18 GHz. Beyond this range, the system performance starts to degrade ( Fig. 9). Considering that DD-Rx, Lite CO-Rx #1, and Lite CO-Rx #2 all occupy a certain bandwidth, the total bandwidth utilization within the photodetector (PD) bandwidth is calculated as 70.45%. At different fiber transmission distances (040 km), the EMRC-Rx performance is significantly superior to other receivers (Fig. 10). The bandwidths of the Si modulator and PD employed in our paper are 33 GHz and 22 GHz respectively. By employing higher-order signal modulation schemes and larger bandwidth PDs, further improvements in transmission rates can be achieved and frequency overlap is avoided. On the other hand, the signal beating in the PD indicates that signal-signal beating interference (SSBI) occurs when the two sidebands of the downlink signal beat each other, which can distort across the entire baseband range. However, when the spacing between the carrier and sidebands is sufficiently large, the influence of SSBI is significantly reduced. In the proposed system, Lite CO-Rx #1 and Lite CO-Rx #2 have a significant guard band, allowing them to remain unaffected by SSBI (Fig. 12). As for the DD-Rx component, the left half of the signal may be influenced by SSBI. However, due to the high carrier-to-sideband power suppression ratio (CSPR) in the system, it is sufficient to minimize the influence of SSBI. Therefore, the effect of SSBI in the system in our study can be generally considered negligible. In terms of system cost, compared with other reported representative lite coherent systems, the proposed EMRC-Rx does not introduce additional hardware but mainly differs in the digital signal processing part where the EMRC algorithm is employed. The additional digital signal processing can optimize receiver sensitivity and mitigate the performance degradation caused by the low linearity of Si modulators. The higher receiver sensitivity not only reduces the correction costs of system error but also allows for higher split ratios in the optical distribution network (ODN), further decreasing the deployment costs of PON. From a system perspective, leveraging the advantages of silicon-based devices in CMOS compatibility and large-scale production can further reduce equipment costs when Si modulators are extensively deployed in PONs. Additionally, for distributed units (DUs) and remote radio units (RRUs), integrating more chip-level devices such as lasers, detectors, passive components, and amplifiers can reduce costs and power consumption, which is beneficial for both operators and end-users. Conclusions We propose an EMRC-Rx that leverages the advantages of both direct detection and coherent detection to significantly improve receiver sensitivity and mitigate the system performance degradation caused by the low linearity of Si modulators. By employing EMRC-Rx, the system can ensure consistent transmission performance both under low-received power and high-received power scenarios. During the experimental validation, EMRC-Rx demonstrates superior performance compared with other receivers, making it a promising solution to the challenges associated with Si modulator linearity in optical communication systems. The proposed EMRC-Rx is an algorithm-based linearization compensation scheme specifically designed for Si modulators. It serves as a system-level performance optimization solution for devices after fabrication or packaging to fill a current gap in the industry. Our study provides a valuable guidance for the construction of high-reliable and low-cost photonic integrated access networks based on silicon modulators in the 5G era.