Expand this Topic clickable element to expand a topic
Skip to content
Optica Publishing Group

Transmission of 112 Gb/s PM-QPSK signals over up to 635 km of multimode optical fiber

Open Access Open Access

Abstract

We investigate transmission of 112 Gb/s PM-QPSK signals over 50 μm core diameter OM3 multimode fiber using the center launch approach. We demonstrate successful transmission of 16 DWDM channels over a distance of 635 km for a capacity-distance product of 1016 Tb/s-km. The limiting impairment appears due to mode coupling and multipath interference effects.

©2011 Optical Society of America

1. Introduction

It is well known that the theoretical informational capacity of a multi-mode optical fiber (MMF) is much larger than that of a single-mode fiber. However, most of the practical systems using MMF are limited to a transmission distance of a few hundred meters and channel data rates of 10 Gb/s and less, due to detrimental effects of inter-modal dispersion. With bandwidth demands ever increasing, it seems important to explore the ultimate limits of the transmission speed and reach for MMF links. Several techniques were proposed to overcome the inter-modal dispersion. One such technique, requiring feedback from the receiver to the transmitter, is to use adaptive optics to launch signal into one of the principal modes of the MMF, and 10 Gb/s transmission over 11 km was successfully demonstrated [1]. Another and much simpler technique is known as “center launch”, where a butt-joint or splice of MMF to a standard single-mode fiber is used at the link input to preferentially launch the signal into the fundamental mode of the MMF and at the output to filter out unwanted higher-order modes (HOM). A great advantage of this technique is that standard single-mode transceiver equipment can be used without any additional components. With center launch and direct detection, 10 Gb/s transmission over 12.2 km [2], 20 Gb/s over 5 km [3], 40 Gb/s over 3.7 km [4], and 86 Gb/s over 2 km [5] of MMF were demonstrated. 10 Gb/s performance resistance to connector offset, fiber bending and shaking was studied in [2] and found acceptable. Recently, world record performance was demonstrated using coherent optical orthogonal frequency division multiplexing (CO-OFDM) – 21.4 Gb/s over 200 km [6] and 107 Gb/s over 100 km [7] of MMF. Improvement of performance versus direct detection was attributed to the extreme resistance of this format to linear impairments such as chromatic (CD) and polarization mode (PMD) dispersion. Another transmission format rapidly gaining popularity which is using coherent detection and digital signal processing (DSP) to compensate large amounts of CD and PMD is polarization-multiplexed quadrature phase-shift keying (PM-QPSK) [8]. In this work, we report on the study of 112 Gb/s PM-QPSK data transmission performance over OM3 MMF and demonstrate 1.6 Tb/s transmission over up to 635 km.

2. Experimental set-up

The fiber used in our experiments was Corning ClearCurve® 50 μm core diameter OM3 fiber with specified nominal multimode bandwidth of 2000 MHz-km at 850 nm. The theoretically estimated effective area of the fundamental mode at 1550 nm is 172 μm2. Since this fiber is not optimized for operation in the 1550 nm wavelength range, we first performed a differential mode delay (DMD) measurement to assess both the magnitude of modal dispersion and the feasibility of the selective fundamental mode excitation by center-launch. In the experiment, a standard single-mode fiber carrying 6.2 ps long light pulses at 1550 nm was scanned across the input face of the 17.66 km long segment of MMF, and oscilloscope traces of the output recorded as function of the centerline offset. The result is shown in Fig. 1 .

 figure: Fig. 1

Fig. 1 DMD measurement results for OM3 MMF fiber using 6.2 ps long pulses at 1550 nm.

Download Full Size | PDF

As can be seen from Fig. 1, up to 8 different mode groups can be excited by offset launch, and the largest modal delay is ~35 ns. For zero offset or perfectly centered launch, only two mode groups are excited but the relative portion of power in a higher order mode is still significant. For our experiments, standard single-mode fiber patch cords were spliced directly to the MMF (no bridge fiber) at the input and output of the span. As was shown in [2], the selectivity of exciting the fundamental mode of the MMF can be greatly improved by achieving the best possible mode field diameter match between it and the fundamental mode of the single-mode fiber at the splice point. Therefore, we optimized the splice recipe using arc sweep function to slightly expand and taper the single-mode fiber core. The effectiveness of the center-launch technique was monitored by measuring the excess loss (due to filtering out higher order modes) at the output splice. With a short length of MMF (~2 m) between two patch cords, we measured excess loss as low as 0.2 dB. At the end of the 70.6 km long MMF fiber span, the excess loss increased to ~0.5-0.7 dB.

Transmission studies were performed in two basic experimental configurations. We first studied the OSNR sensitivity performance of a single 112 Gb/s PM-QPSK channel in a straight line configuration over various lengths of MMF. This experimental configuration is shown in Fig. 2 .

 figure: Fig. 2

Fig. 2 Experimental set-up of straight-line configuration for basic OSNR sensitivity measurements.

Download Full Size | PDF

A distributed feedback (DFB) laser or external cavity laser (ECL) at wavelength 1550.92 nm was modulated with a QPSK modulator driven by two independent 215-1 PRBS patterns with a symbol rate of 28 Gbaud. The output from the QPSK modulator was optically polarization multiplexed to produce the 112 Gb/s PM-QPSK signal. The signal was launched into a span of OM3 fiber and then noise-loaded at the receiver to control the OSNR. At the receiver end, the channel passed through an optical filter with 0.4 nm bandwidth and was detected in a polarization- and phase-diverse digital coherent receiver with a free-running local oscillator with 100 kHz nominal linewidth. The four signals from the balanced photodetectors were digitized by analog-to-digital converters operating at 50 Gs/s using a real-time sampling oscilloscope with 20 GHz electrical bandwidth. The sampled waveforms were processed off-line in a computer, with standard digital signal processing steps [9] including (i) quadrature imbalance compensation, (ii) up-sampling to 56 Gsamples/s and dispersion compensation using a frequency-domain equalizer, (iii) digital square and filter clock recovery, (iv) polarization recovery and equalization using an adaptive butterfly structure with filter coefficients determined using the constant modulus algorithm, (v) carrier frequency offset using a spectral domain algorithm, (vi) phase recovery using a pre-decision algorithm, and (vii) bit decisions. The bit error rate (BER) was measured for each 28 Gb/s tributary by direct error counting.

The second part of the study consisted of 8- and 16-channel DWDM transmission experiments with a re-circulating loop. The configuration for the WDM loop experiments is shown in Fig. 3 . DFB lasers with 50 GHz channel spacing were multiplexed and modulated together by the 112 Gb/s PM-QPSK transmitter. The channels were launched into the re-circulating loop with a flat spectrum. The loop was comprised of a single 70.6 km long span of MMF assembled by splicing together 4 x 17.66 km spools. Single-mode fiber jumpers were again spliced directly to the MMF in a center-launch condition at the span input and output. The measured total span loss including all splices was 18.4 dB. A single-stage EDFA followed the fiber span to compensate for the fiber loss. The channel launch power into the fiber span was controlled by a VOA on the transmitter side. A loop synchronous polarization scrambler (LSPS) was used to mitigate possible loop polarization artifacts and a bandpass filter implemented with a dynamic gain equalizer (Finisar WaveShaper) was also deployed in the loop to filter out ASE noise outside the signal spectrum. As in the straight-line system, CD was compensated electronically in the digital coherent receiver.

 figure: Fig. 3

Fig. 3 Experimental set-up of DWDM transmission system with re-circulating loop.

Download Full Size | PDF

3. Experimental results

The first measurements made were that of the OSNR sensitivity for single channel transmission over several different lengths of the MMF. The different span lengths were created by splicing together individual spools of MMF. The transmission laser was a DFB laser and the channel launch power was about −4 dBm. The results are shown in Fig. 4a for 3 different span lengths ranging from 28.7 km to 70.6 km, as well as the back-to-back (B2B) data. We observe a penalty from transmission over the multimode fiber, increasing with span length. The penalty is likely due to multipath interference (MPI) from mode coupling between the fundamental mode and higher order modes during propagation and is relatively small for lengths up to 70.6 km, with < 0.5 dB penalty at the BER = 1x10−3 level and approximately 1 dB at the BER = 1x10−5 level.

 figure: Fig. 4

Fig. 4 (a) BER vs. OSNR for different span lengths of OM3 multimode fiber with DFB laser. (b) BER vs. OSNR for a continuous 70.6 km span and a 70.6 km span comprised of two 35.3 km pieces of multimode fiber with 5 m of single-mode fiber in between. (c) Comparison of performance for DFB and ECL for 70.6 km transmission.

Download Full Size | PDF

To help understand the nature of the mode coupling-induced penalty, we next divided the 70.6 km of multimode fiber into two 35.3 km lengths with 5 m of standard single-mode fiber spliced between them. This configuration provided for additional higher order mode (HOM) filtering, but as shown in Fig. 4b, did not affect the performance compared to the continuous 70.6 km MMF span, suggesting that the transmission penalty is dependent mainly on the total length of the MMF.

Finally, we also compared the performance of the DFB transmission laser with a few MHz linewidth to a narrow linewidth (100 kHz nominal linewidth) ECL We observed no significant difference in performance between the two lasers, as shown in the results given in Fig. 4c.

In the re-circulating loop DWDM experiments, we first investigated the performance dependence on channel launch power in an 8 channel 50 GHz system. The channel plan ranged from 1549.32 nm to 1552.12 nm. The power per channel was varied over a wide range and the BER was measured for each power level for the center 1550.92 channel at two different distances. The results obtained at 353 km and 565 km are given in Fig. 5a . At the longer distance, we also removed the 4 channels closest to the measurement channel and repeated the measurement. In this configuration the closest channels to the measurement channel were 150 GHz away. For both distances, and both channel spacings at 565 km, we observed relatively flat performance between 2 dBm and 6 dBm per channel. Given these results, we then used a launch power level of 4 dBm/channel to perform transmission experiments with the 8-channel 50 GHz system for a range of distances from 141 km to 706 km (2 to 10 loop repetitions). The results for the measurement channel at 1550.92 nm are given in Fig. 5b. The Q value (as derived from the BER data) approaches the FEC threshold at 706 km, suggesting that error-free transmission after FEC is possible out to at least 635 km.

 figure: Fig. 5

Fig. 5 (a) BER vs. channel launch power. (b) OSNR and 20log(Q) vs. distance for the 8 channel 50 GHz system with 4 dBm/channel launch power.

Download Full Size | PDF

We then increased the channel count to 16, transmitted with the same launch power of 4 dBm/channel, and measured the BER of all 16 channels at 635 km. The results are given in Fig. 6a . All channels had Q values over the FEC threshold at that distance. Representative QPSK constellation diagrams for the X and Y polarizations for the 1550.92 nm channel are shown in Fig. 6b. The transmission of 16 x 100 Gb/s over this distance represents a capacity-distance product of 1016 Tb/s-km, which we believe is the largest demonstrated over MMF to date.

 figure: Fig. 6

Fig. 6 (a) Q values for 16 channels after 635 km of MMF. (b) X and Y constellations for 1550.92 nm channel.

Download Full Size | PDF

While the transmission distance of 635 km may be the longest reported for a system built with multimode fiber, the Q and OSNR data in Fig. 5b can be viewed in a different way that illustrates the magnitude of the penalty experienced by transmission over the MMF. In Fig. 7a , the measured BER values at each distance are given as a function of the measured OSNR values. These results are compared to simple back-to-back data with no MMF transmission. The results are interesting in that there appears to be a fairly uniform OSNR penalty of about 11-12 dB for all of the distances measured from 141 to 706 km. The transmission OSNR data are the values that were measured for each distance at the receiver, while the back-to-back OSNR data was produced by noise-loading the signal at the receiver.

 figure: Fig. 7

Fig. 7 (a) BER vs. OSNR for MMF transmission at different distances (4 dBm/channel) and back-to-back. (b) BER vs. OSNR for MMF transmission at 565 km and back-to-back.

Download Full Size | PDF

There are two main potential sources of the observed penalty - nonlinearity and mode coupling-induced multipath interference (MPI). It is somewhat difficult to easily separate out these two effects since the transmission data was taken with an optimized channel power of 4 dBm, and therefore some of the penalty is due to nonlinear impairments. However, it is likely that a large fraction of the penalty is due to mode coupling and MPI effects. This can be reasonably inferred from the data in Fig. 7b, in which the same back-to-back BER vs. OSNR data is compared to the transmission data at 565 km for which the channel power was varied from −4 dBm to + 7 dBm as shown in Fig. 5a. The first point in the transmission data corresponds to the −4 dBm channel power level, which is in the linear transmission regime. However, even at this low power, the OSNR penalty is almost 6 dB, which must have its origins in the mode coupling and MPI effects since nonlinearity should not be an issue, given the estimated fiber effective area. Therefore it appears likely that the OSNR penalty at most of the distances is probably due in large part to the effects from multimode transmission within the transmission fiber. We believe that the origin of the penalty is from mode coupling in which either light initially launched into the first higher order mode couples back into the fundamental mode during propagation, or part of the fundamental mode is coupled out to higher order modes, and then coupled back into the fundamental mode before the end of the span, thus causing multipath interference between delayed copies of the signal.

4. Summary and conclusions

We have experimentally demonstrated transmission of 1.6 Tb/s with 16 channels of 112 Gb/s PM-QPSK over up to 635 km of 50 μm core diameter OM3 MMF. To our knowledge, this represents the longest transmission at this data rate over multimode fiber to date and a record capacity-distance product of 1016 Tb/s-km. The penalty imputed to mode coupling and MPI grows with transmission distance and is the limiting impairment in the transmission system.

References and links

1. X. Shen, J. M. Kahn, and M. A. Horowitz, “Compensation for multimode fiber dispersion by adaptive optics,” Opt. Lett. 30(22), 2985–2987 (2005). [CrossRef]   [PubMed]  

2. D. H. Sim, Y. Takushima, and Y. C. Chung, “High speed multimode fiber transmission by using mode-field matched center-launching technique,” J. Lightwave Technol. 27(8), 1018–1026 (2009). [CrossRef]  

3. I. Gasulla and J. Capmany, “1 Tb/s x km multimode fiber link combining WDM transmission and low-linewidth lasers,” Opt. Express 16(11), 8033–8038 (2008). [CrossRef]   [PubMed]  

4. S. S.-H. Yam and F. Achten, “Single wavelength 40 Gbit/s transmission over 3.4km broad wavelength window multimode fibre,” Electron. Lett. 42(10), 592–594 (2006). [CrossRef]  

5. H. S. Chung, S. H. Chang, and K. Kim, “6 x 86 Gb/s WDM transmission over 2 km multimode fiber using center launching technique and multi-level modulation,” Opt. Express 17(10), 8098–8102 (2009). [CrossRef]   [PubMed]  

6. Z. Tong, Q. Yang, Y. Ma, and W. Shieh, “21.4 Gbit/s transmission over 200 km multimode fibre using coherent optical OFDM,” Electron. Lett. 44(23), 1373–1374 (2008). [CrossRef]  

7. Y. Ma, Y. Tang, and W. Shieh, “107 Gbit/s transmission over multimode fibre with coherent optical OFDM using centre launching technique,” Electron. Lett. 45(16), 848–849 (2009). [CrossRef]  

8. K. Roberts, M. O’Sullivan, K.-T. Wu, H. Sun, A. Awadalla, D. J. Krause, and C. Laperle, “Performance of Dual-Polarization QPSK for Optical Transport Systems,” J. Lightwave Technol. 27(16), 3546–3559 (2009). [CrossRef]  

9. J. D. Downie, J. E. Hurley, J. Cartledge, S. R. Bickham, and S. Mishra, “Transmission of 112 Gb/s PM-QPSK signals over 7200 km of Optical Fiber with Very Large Effective Area and Ultra-Low Loss in 100 km Spans with EDFAs Only,” in Optical Fiber Communication Conference and Exposition (OFC) and National Fiber Optic Engineers Conference (NFOEC) (Optical Society of America, Washington, DC, 2011), paper OMI6.

Cited By

Optica participates in Crossref's Cited-By Linking service. Citing articles from Optica Publishing Group journals and other participating publishers are listed here.

Alert me when this article is cited.


Figures (7)

Fig. 1
Fig. 1 DMD measurement results for OM3 MMF fiber using 6.2 ps long pulses at 1550 nm.
Fig. 2
Fig. 2 Experimental set-up of straight-line configuration for basic OSNR sensitivity measurements.
Fig. 3
Fig. 3 Experimental set-up of DWDM transmission system with re-circulating loop.
Fig. 4
Fig. 4 (a) BER vs. OSNR for different span lengths of OM3 multimode fiber with DFB laser. (b) BER vs. OSNR for a continuous 70.6 km span and a 70.6 km span comprised of two 35.3 km pieces of multimode fiber with 5 m of single-mode fiber in between. (c) Comparison of performance for DFB and ECL for 70.6 km transmission.
Fig. 5
Fig. 5 (a) BER vs. channel launch power. (b) OSNR and 20log(Q) vs. distance for the 8 channel 50 GHz system with 4 dBm/channel launch power.
Fig. 6
Fig. 6 (a) Q values for 16 channels after 635 km of MMF. (b) X and Y constellations for 1550.92 nm channel.
Fig. 7
Fig. 7 (a) BER vs. OSNR for MMF transmission at different distances (4 dBm/channel) and back-to-back. (b) BER vs. OSNR for MMF transmission at 565 km and back-to-back.
Select as filters


Select Topics Cancel
© Copyright 2024 | Optica Publishing Group. All rights reserved, including rights for text and data mining and training of artificial technologies or similar technologies.