科研之友
微信
新浪微博
Facebook
分享链接
摘要
Based on pre-allocated relative sampling delays among optical network users (ONUs) in an orthogonal frequency division multiplexing (OFDM) passive optical network (PON), the delay-division-multiplexing (DDM) technique employs signal pre-processing to enable demand data to be received by sub-Nyquist analog-to-digital sampling. This work employed an avalanche photodiode (APD) in a DDM-OFDM-PON to achieve higher loss budget, compared to a p-i-n photodiode (PIN). To successfully enable the pre-processing, however, the knowledge of channel response is required; thus, it is critical to estimate complete channel response in DDM-OFDM systems under the constraint of sub-Nyquist sampling. To avoid spectral overlapping in channel estimation, the subcarriers of training symbols in different frequency zones are allocated to different time slots. Unfortunately, this scheme may sustain inaccurate channel estimation due to different degrees of APD saturation for training subcarriers at different frequency zones, leading to significant penalty in performance. Thus, careful control of electrical driving power and optical received power for the training symbols will be important to reduce the impact of saturation on channel estimation, thereby improving transmission performance. We experimentally examined the impact of the accuracy of channel estimation in a 25-Gbps APD-based OFDM-PON. Compared to a PIN, we successfully demonstrated the DDM scheme based on an APD and 1/32 of Nyquist rate with 5-dB improvement in sensitivity after 25-km fiber transmission.
- 出版日期2018
- 单位中山大学
