ZTE Communications

More documents

Recommendations

Info

Multi-Carrier Source • • • • • • TX1 Data 1 TX2 Data 2 TX3 Data 3 Optical Filter Optical Filter Optical Filter S pecial Topic Super-Receiver Design for Superchannel Coherent Optical Systems Cheng Liu, Jie Pan, Thomas Detwiler, Andrew Stark, Yu-Ting Hsueh, Gee-Kung Chang, and Stephen E. Ralph Multiplexer Channel 123 DSP: digital signal processor LO: local oscillator as at the transmitter side. The synchronized information across the channels is captured for future joint signal processing. For this reason, the optical path and electrical path of each demultiplexed channel need to be equal in length, and digital sampling must be synchronized across the channels. However, these requirements can be relaxed in a joint DSP block if time-domain memory size is increased (Fig. 2). In a joint DSP block, information is available from multiple channels, and this enables joint signal processing to compensate for both linear and nonlinear impairments between channels. This also enables joint carrier phase recovery in carrier-locked Nyquist WDM systems. 3 Joint DSP Based on Adaptive LMS Algorithm for Cancelling Linear ICI Fig. 2 shows the proposed joint DSP for cancelling linear Ch1 Ch2 Ch3 λ1 λ2 λ3 XI XQ YI YQ XI XQ YI YQ XI XQ YI YQ ADC ADC ADC ADC ADC ADC ADC ADC ADC ADC ADC ADC Synchronized Sampling Chromatic Dispersion Equalizer Chromatic Dispersion Equalizer Chromatic Dispersion Equalizer CMA Polarization DEMUX CMA Polarization DEMUX CMA Polarization DEMUX Demultiplexer • • • • • • Optical Filter Optical Filter Optical Filter Multi-Carrier LO Source Timing & Carrier Phase Recovery Timing & Carrier Phase Recovery Timing & Carrier Phase Recovery λ1 λ2 λ3 Coherent Receiver 1 Coherent Receiver 2 Coherent Receiver 3 Synchronized Sampling Channel 1 Electrical Filter Channel 2 Electrical Filter Channel 3 Electrical Filter Joint DSP • • • Data 1 Data 2 Data 3 • • • ◀Figure 1. Proposed super-receiver architecture. ICI. Three channels are considered, and ICI equalization for the center channel (channel two) is shown. After the three-channel signal is sampled with synchronized ADCs, each channel undergoes conventional DSP, that is, chromatic dispersion compensation, polarization demultiplexing, timing recovery, and carrier phase estimation. After timing recovery and carrier phase recovery, an adaptive ICI equalizer based on adaptive least mean square (LMS) algorithm is used across all three channels for both X and Y polarizations. For each polarization, side channels (channels one and three) are shifted in the frequency domain by the amount of channel spacing. By shifting the side channels into the original spectral location, the overlapped spectra are aligned for subsequent ICI equalizer. After then the signals from the three channels are fed into the ICI filter, where the filter coefficients W12, W22, and W32 are jointly and adaptively updated according to the slicing error. Each filter coefficient Wij represents the ADC: analog-to-digital converter CMA: constant modulus algorithm ICI: interchannel interference ISI: intersymbol interference LMS: least mean square ▲Figure 2. Joint DSP based on adaptive LMS algorithm for cancelling linear ICI. XPol XPol XPol YPol YPol YPol Timing & Carrier Phase Recovery Timing & Carrier Phase Recovery Timing & Carrier Phase Recovery Ch1 Ch2 Ch3 Ch1 Ch2 Ch3 Freq. Shift Freq. Shift Freq. Shift Freq. Shift W12 W22 W32 Adaptive Update Adaptive LMS ICI Equalizer W12 W22 W32 Adaptive Update Ch2 XPol + - + - e Ch2 YPol e ISI Equalizer ISI Equalizer Symbol Detection Symbol Detection • • • Ch2 XPol Ch2 YPol • • • March 2012 Vol.10 No.1 <strong>ZTE</strong> COMMUNICATIONS 31
S pecial Topic Super-Receiver Design for Superchannel Coherent Optical Systems Cheng Liu, Jie Pan, Thomas Detwiler, Andrew Stark, Yu-Ting Hsueh, Gee-Kung Chang, and Stephen E. Ralph Req. OSNR (dB) at BER=10 -3 Required OSNR (dB) 32 28 26 24 22 20 18 16 ❋ ● ❋ ● ❋ ❋ ● ● <strong>ZTE</strong> COMMUNICATIONS ❋ ● ●❋ ❋ ● ●❋ ● ❋ 14 24 27 30 33 36 39 42 45 48 Channel Spacing (GHz) (a) Back-to-back lab and simulation results of required OSNR at BER = 10 -3 vs. channel spacing for two-channel 28 Gbaud cases. 28 24 22 20 18 16 14 12 31 ❋ >5 dB Gain ● ❋ ● 32 33 ❋ ● ❋ ● ❋ ● ❋ ● ●❋ ●❋ ●❋ BER: bit error rate DP-QPSK: dual polarization quadrature phase shift keying weighted crosstalk from channel i to channel j. For each Wij, there are a few taps in time domain so that timing offset between channels, caused by imperfect synchronized sampling or path mismatch, can be compensated for. After the ICI equalizer, ISI equalizers are used for each polarization to compensate for the residual ISI induced by optical or electrical filtering in the link. Conventional DSP in Nyquist WDM systems uses an ISI equalizer to compensate for the penalty introduced by strong filtering. However, a joint ICI equalizer relaxes the narrow filter bandwidth and adaptively cancels a significant part of the linear crosstalk. Channel one and three are processed in the same way using information from neighboring channels. 4 Experiment and Simulation Results We conducted a proof-of-concept experiment on the super-receiver architecture and joint ICI algorithm and compared the results with those using simulated data. Two independent lasers each carrying 28 Gbaud dual-polarization quadrature phase-shift keying (DP-QPSK) ● ❋ ● ❋ BER: 10 -2 BER: 10 -3 34 35 36 37 38 Channel Spacing (GHz) { { : Exp. LMS ICI Eq. : Exp. Conventional : Sim. LMS ICI Eq. : Sim. Conventional ● ❋ ● ❋ ●❋ ● March 2012 Vol.10 No.1 : LMS ICI Eq. : Conventional : LMS ICI Eq. : Conventional (c) Required OSNR at BER = 10 -2 and 10 -3 vs. channel spacing after 1280 km SSMF transmission for three-channel 32 Gbaud cases. 39 40 51 41 OSNR Gain (dB) 5◆▲ ❋ ❋▲ ▲❋ ❋▲ ❋ ❋ ❋ ◆ BER: 10 -3 2 ❋▲◆ 0 0 ▲❋ ◆ ICI: interchannel interference LMS: least mean square ▲Figure 3. Comparison of proposed joint LMS ICI equalizer with conventional method. Req. OSNR (dB) at BER=10 -3 20 19 18 17 16 6 4 3 1 4 ❋ ● ❋ ● ◆ ◆ BER: 10 ❋▲ ▲❋ ❋ ❋ ❋ -2 ◆ 8 : 5 Taps ▲ ▲ : 30 Taps ▲ ▲ ◆ 12 ◆ 16 ▲ ◆ 20 ▲ ◆ ▲ 24 Number of Offset Symbols OSNR: optical signal-to-noise ratio SSMF: standard single-mode fiber ▲◆ 28 ❋ : 50 Taps ❋ ▲◆ ▲ ❋ 32 36 (d) OSNR gain at BER = 10 -2 and 10 -3 vs. number of offset symbols under different filter length. (3 channel 32 Gbaud at 32 GHz channel spacing). data were MUXed together without optical filters. At the receiver side, the signals were split into two paths and were received by two coherent receivers with synchronized 40 GSa/s Agilent sampling oscilloscopes. The electrical filters at the sampling head were 16 GHz. Fig. 3(a) and (b) shows the OSNR required to reach a BER of 10 -3 at different channel spacings. The results are shown for experimental and simulated data. When channel spacing decreases, absolute performance degrades in all cases. However, with a joint ICI equalizer, performance degradation is reduced. For this simple two-channel case, there is a 2 dB required OSNR penalty at 30 GHz channel spacing when conventional ISI equalization is used. Joint ICI equalization reduces this penalty by 1 dB. To give a more complete picture of the proposed system with three-channel conditions, we simulated a 3 × 32 Gbaud (128 Gb/s) DP-QPSK Nyquist WDM optical coherent system by using RSOFT OptSim. The bandwidths of optical filters at optical multiplexing were set to 43.75 GHz in order to shape the signal spectrum. After optical multiplexing, the three channel signals were transmitted through 16 spans of 80 km ● ❋ ● ❋ ❋ ❋ : Exp. LMS ICI Eq. : Exp. Conventional : Sim. LMS ICI Eq. : Sim. Conventional 15 29 30 31 32 33 34 35 36 (b) Back-to-back lab and simulation results of required OSNR at BER = 10 -3 Channel Spacing (GHz) vs. channel spacing for two-channel 28 Gbaud cases. ❋ ❋ ● ●❋ ❋ 40
Page 1 and 2: www.zte.com.cn/magazine/English ISS
Page 3 and 4: 34 ..... Design of a Silicon-Based
Page 5 and 6: S pecial Topic Guest Editorial of 1
Page 7 and 8: S pecial Topic High Spectral Effici
Page 9 and 10: S pecial Topic High Spectral Effici
Page 11 and 12: λ1 S pecial Topic High Spectral Ef
Page 13 and 14: S pecial Topic Greater than 200 Gb/
Page 21 and 22: S pecial Topic Spatial Mode-Divisio
Page 27 and 28: S pecial Topic Exploiting the Faste
Page 33: S pecial Topic Super-Receiver Desig
Page 37 and 38: S pecial Topic Design of a Silicon-
Page 43 and 44: S pecial Topic The Key Technology i
Page 49 and 50: S pecial Topic Compensating for Non
Page 51 and 52: S pecial Topic Compensating for Non
Page 53 and 54: S pecial Topic 1 Tb/s Nyquist-WDM P
Page 55 and 56: S pecial Topic 1 Tb/s Nyquist-WDM P
Page 57 and 58: R esearch Papers Hardware Architect
Page 67 and 68: R esearch Papers A Histogram-Based
Page 73: R esearch Papers A Histogram-Based

ZTE Communications

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?