We present a novel optical transmission system to experimentally demonstrate the possibility of mode division multiplexing. Its key components are mode multiplexer and demultiplexer based on a programmable liquid crystal on silicon panel, a prototype few-mode fiber, and a 4x4 multiple input multiple output algorithm processing the information of two polarization diversity coherent receivers. Using this system, we transmit two 100Gb/s PDM-QPSK data streams modulated on two different modes of the prototype few-mode fiber. After 40km, we obtain Q2-factors about 1dB above the limit for forward error correction.
©2011 Optical Society of America
Increasing the constellation complexity has been the most frequently-reported approach to raise the transmission capacity in the past few years, but it can have a detrimental impact on reach, e.g. a reduction by a factor of five from 100Gb/s Polarization Division Multiplexed (PDM) Quaternary Phase-Shift Keying (QPSK) to 200Gb/s PDM-16-QAM (Quadrature Amplitude Modulation) . A new and disruptive approach for increasing capacity would be Mode Division Multiplexing (MDM). Recently, the possibility to perform MDM transmissions has been demonstrated experimentally, even though the number of exploited spatial modes was limited to a maximum of three and the transmission distances were still far from achievable single-mode distances [2–5]. The new key enablers for these experiments are mode converters, few-mode transmission fibers, and advanced multiple-input multiple-output (MIMO) algorithms for the digital signal processor in coherent optical receivers. For all of these elements, different approaches have been proposed.
In our work, we demonstrate the transmission of two independent modes over a prototype few-mode fiber (FMF). The exploited modes are the two degenerate variants of the LP11 mode, which we shall call ‘LP11a’ and ‘LP11b’. While these two modes have naturally the same propagation constant, our FMF has the advantage of exhibiting very large effective-index differences and very large mode delays between modes of different order. Thus, low linear [6,7] and nonlinear  crosstalk between modes occurs during transmission, in combination with low loss of only 0.22dB/km. For mode multiplexing and demultiplexing, we designed and tested an integrated spatial mode converter incorporating a Liquid Crystal on Silicon (LCOS) spatial light modulator (SLM), which is reprogrammable to transform the incoming LP01 mode to almost any desired mode profile . The LCOS is an alternative to the simpler but fixed mode-selective phase masks fabricated in glass . At the receiver side, we combine the SLM that generates mode diversity with two coherent receivers with joint digital signal processing for the detection of two modes at a time.
The two modes are recovered with excellent performance after 20km, and interestingly, only a limited additional penalty is measured when the distance is increased to 40km.
2. Experimental set-up
In our set-up, depicted in Fig. 1 , the light from a laser at 1533.47nm is passed into an integrated transmitter. The transmitter uses a serializer to produce four 28Gb/s electrical pseudo-random signals of length 215-1, each shifted by 8192 bits, which feed a quad-driver and a polarization-multiplexed nested Mach-Zehnder modulator. It generates a data stream at 112Gb/s PDM-QPSK, including 12% protocol and forward error correction (FEC) overhead. This stream is replicated along two fiber paths decorrelated by several thousands of symbols using an SMF patch-cord, and fed to two optical amplifiers, connected to two of the four inputs of a mode multiplexer where they are converted into the LP11a and LP11b modes. Another independent PDM-QPSK transmitter is used for the generation of the fundamental mode LP01 and connected to another port of the mode multiplexer, but not more than two transmitters are used at once.
The output of the mode multiplexer is realized by using a pigtail fiber supporting up to 3 modes (LP01 as well as the two degenerate LP11 modes ‘a’ and ‘b’). This pigtail is spliced to 40 km of the prototype FMF described in section 3.2. After transmission, the output of the FMF is connected to the mode demultiplexer, again by splicing it to a pigtail of the same type as at the output of the mode multiplexer. In the mode demultiplexer, two arbitrary modes of the FMF that can be extracted and converted back into LP01 before they are injected into two SMFs.
With this setup we investigated several configurations: single mode transmissions (LP01 only) and MDM transmission (LP11a + LP11b or LP01+LP11b). In all configurations, the power into each receiver is varied using dedicated attenuators in order to adjust the optical signal to noise ratio (OSNR) and the performance is measured. The coherent receivers incorporate optical preamplifiers and four balanced photodiodes in a polarization diversity configuration. Sampling is performed with two real-time oscilloscopes (one per receiver), which have an analog bandwidth of 16GHz and are synchronously triggered. When the LP01 is detected a constant phase mask and a single receiver are used. Whereas, when the phase masks are set to select LP11a and LP11b, two receivers are used to perform joined signal processing. It is described more in detail in section 3.3.
3. Detailed description of key elements
3.1 Mode multiplexer and demultiplexer
To perform MDM, we rely on SMF-to-FMF mode conversion based on the principle of a so-called 4f correlator depicted in Fig. 2 . Input fiber and output fiber are collimated by a lens, whose distance of the fiber output corresponds to its focal length ‘f’. As a result, the two-dimensional Fourier transform of the spatial light distribution in the fiber is obtained in the central plane, spaced by another ‘f’ between the lenses. This Fourier transform can be modulated in phase or in amplitude using a multiplicative mask. The resulting field distribution in the output fiber can be easily calculated by applying the inverse Fourier transform to the product of mask and incoming Fourier transform.
We used a spatial light modulator realized in LCOS technology in order to implement the multiplicative masks. As a compromise between practical implementation complexity and performance we chose to use only phase masks, which should nevertheless allow for sufficiently high mode conversion ratios, according to simulation results. The employed LCOS device is a panel of 1920x1080 pixels, which can all be programmed independently to produce a phase shift between 0 and 2π. That offers the possibility to obtain various mode shapes, limited by the resolution inside the modulated light beam. The LCOS device operates in reflective mode and it works only along one polarization axis.
Based on these constraints imposed by the LCOS panel we constructed a 4x1 mode-converter used as multiplexer and a 2x1 mode-converter used as demultiplexer. It should be noted that both devices are symmetric, and thus the roles of multiplexer and demultiplexer could be inverted. Their structure is depicted in Fig. 3 . In the 4x1 multiplexer, all four inputs enter the free-space device through a collimated SMF pigtail. Then, three of them are spatially modulated as it is depicted in the grey box in the upper left of Fig. 3. The light of each of them is split along two optical paths by a polarization beam splitter (PBS), and one of the two resulting beams passes by a half-wave plate in order to rotate it by 90° and to align both beams along the correct polarization axis of the LCOS. After passing a lens to slightly refract them, the beams hit the LCOS panel on two of six possible spots on its upper half (see Fig. 3. in the upper right). The refraction by ~0.2° is necessary in order to allow the beam not to hit the components of the input path after reflection at the LCOS. The spots on the LCOS device count approximately 80x80 pixels and are programmed with one of the three phase masks depicted in the center of Fig. 3., which give the light beam the shape of either LP01, LP11a or LP11b mode from then on. After spatial modulation, each light beam is sent back through a comparable assembly of lens, half-wave plate and PBS in order to put together again the two polarization tributaries. Finally, the three spatially modulated beams are combined with the fourth, unchanged input beam and are collimated into the FMF. At the receiver end, the 2x1 converter is designed similarly, but with only two SMF-fiber pigtails, which can both be spatially modulated, making use of four spots on the lower half of the LCOS panel to program the phase masks.
Clearly the most important characteristic of a mode-converter and -multiplexer/-demultiplexer are the provided mode conversion ratios and the resulting values for intermodal cross-talk. Unfortunately, with our set-up we were not able to determine these values precisely because the mode multiplexer and demultiplexer were developed in collaboration with a project partner and assembled by them, both together in one “black box” as it is shown in Fig. 4 . Thus, we were not able to apply any changes to the set-up in order to do specific measurements. The only points where we could access were the SMF fibers at the input of the mode multiplexer and at the output of the mode demultiplexer. As a consequence, for every measurement, the light had to pass through two phase masks in the mode multiplexer (for the two polarizations), the FMF pigtails, the transmission FMF and two phase masks in the mode demultiplexer (again for the two polarizations). With this configuration, we were not able to decouple the crosstalk occurring at different positions in our system such as mode multiplexer or demultiplexer.
However, some of the experimental results presented in section 4 as well as the results presented in  and  give some information concerning the intermodal crosstalk generated in our mode multiplexer and demultiplexer
3.2 Prototype few-mode fiber
For the prototype FMF, an optimized step-index profile has been chosen, as it is depicted in Fig. 5a . It offers simplicity in terms of fabrication and good compatibility with standard step-index single-mode fibers. The step core has a radius of 7.5µm and an index difference with the cladding of Δn = (ncore - ncladding,) = 9.7∙10−3 at 1550nm, giving a normalized frequency V of 5.1 that ensures the full guidance of the first 4 modes (the fundamental LP01 mode and the higher-order LP11, LP21 and LP02 modes) in the C-band while cutting off the next higher-order modes. This is illustrated in Fig. 5b, where the normalized propagation constant , with neff,lm as the effective index of the LPlm mode, is plotted as a function of V and it can be seen that for the given value of V = 5.1, the first four modes, LP01, LP11, LP21, LP02, have positive B. The 4 guided modes have low macro-bend losses (<10dB/turn at 10mm radius at 1550nm), as well as large effective-index differences (>0.8∙10−3) between each other, which ensures low mode couplings [6,7]. Furthermore they have large differential mode group delays (DMGD) (between 1 and 9ps/m for all modes and 4.35ps/m between LP01 and LP11) and large effective areas Aeff (between 118 and 133µm2 depending on modes), which ensures small inter- and intra-modal non-linear effects. They also exhibit low losses (<0.22dB/km), which should differ by a maximum of 0.01dB/km between the guided modes according to our predictions. More detailed characteristics about the fiber can be found in .
3.3 MIMO digital signal processing
The two receivers provide four complex signals representing the optical field. To discriminate between the degenerate modes LP11 along the two polarization axes, a 4x4 MIMO equalizer is needed, as opposed to the conventional 2x2 MIMO used for polarization demultiplexing over SMF. The structure of the 4x4 equalizer is depicted in Fig. 6a . The four complex signals are split into four, each sent into an FIR filter having up to 15 taps. Each of the four outputs is the combination of the four inputs, filtered by a dedicated FIR filter. In order to estimate the required length of the equalizer we vary the FIR tap count. We use the 100Gb/s PDM-QPSK performance in back-to-back as a reference (MIMO 2x2 over SMF). An almost flat Q2-factor is measured from 9 taps to 15 taps, both for 100Gb/s PDM-QPSK and 2x100Gb/s MDM after 40km, as shown in Fig. 6b. The 4x4 MIMO receiver operates in blind mode, using the traditional constant modulus algorithm (CMA) for FIR update. The complexity per bit of such an architecture is only doubled compared to standard single mode operation, as 16 adaptive filters are required to process 2x100Gb/s (4x4 MIMO) compared to four filters for 1x100Gb/s (2x2 MIMO). It should be noted that the CMA tends to converge to the more powerful signal tributaries. In particular, in presence of multiple inputs it is necessary to correctly initialize the filter taps for allowing the detection of all the polarization/mode tributaries. Failing to do so would cause different outputs converging to the same input. To initialize the equalizer, we employed a blind source separation algorithm , which determines the starting value for the central tap of each FIR filter and sets all the other taps to zero. After detection we performed identification of polarizations and modes by comparing the delays of the different tributaries in order to ensure that the four modes have been correctly recovered.
4. Experimental results
The first experimental results were recorded after 20km of FMF fiber. To ease comparisons, we choose to focus on the same received power in all configurations namely −31dBm. In Fig. 7 , the fundamental mode LP01 shows 1.1dB sensitivity penalty after 20km of FMF compared to the reference back-to-back case where only single mode fibers and no mode multiplexer and demultiplexer are used. We attribute this penalty to intermodal crosstalk in the MDM part of the system. After the same distance, the average Q2-factor of the LP11a + LP11b modes detected with the 4x4 MIMO receiver is reduced by another 0.6dB with respect to the LP01 mode. After 40km of FMF, the modes LP11a + LP11b are further degraded, but by no more than 1.1dB. Their maximum performance is still about 1dB above the FEC-limit. Nonetheless, this suggests that there might be some mode coupling during fiber propagation, which gets stronger with the increased length.
The values represented in Fig. 7 are averaged over all the received mode/polarization tributaries (two for the LP01 mode and four for the LP11 mode). Figure 8a depicts the actual performance of all mode/polarization tributaries of LP11, which are depicted schematically in Fig. 8b. The Q2-factors of one polarization of mode LP11a (‘LP11a-polX’) appear significantly lower (by typically 2.5dB) than the other Q2-factors. This is attributed to degraded characteristics of our mode multiplexer/demultiplexer along one polarization. But as mentioned in section 3.1, it is not possible to identify if the problem is located in the mode multiplexer or in the demultiplexer or in both.
Further insight on modal crosstalk can be obtained when propagating both LP01 and LP11b modes, as depicted in Fig. 9 . We keep the LP11b’s power constant at −32dBm on the receiver and vary the power of the LP01 mode. When the LP01 mode reaches −36dBm, 1dB of Q2-factor penalty can already be observed, as a result of linear crosstalk. The LP01 mode should reach −19dBm power to yield a reasonable 10dB Q2-factor, i.e. a 12dB higher power than without LP11b (see ‘LP01 only’ in Fig. 9). This result shows that it hasn’t been possible with our setup to reach a sufficiently low level of intermodal crosstalk between the modes LP01 and LP11. In [10,11] we showed that an important amount of this crosstalk is generated by the elements for mode conversion and multiplexing / demultiplexing, which leads to the conclusion that improvements in these devices are necessary to enable a transmission over more than two modes. Furthermore, the exact amount of crosstalk in other critical parts of the system, notably during fiber propagation, has to be quantified to allow a deeper analysis of the reasons for performance degradation.
The transmission of two 100Gb/s PDM QPSK signals at the same wavelength has been demonstrated over two modes of a 40km few-mode fiber. The key enablers are a reprogrammable free-space mode multiplexer and demultiplexer, a prototype few-mode fiber with low mode coupling and two polarization- and mode-diversity coherent detectors supported by MIMO processing. This experiment paves the way to future high-capacity multimode fiber systems, even if numerous challenges still have to be solved.
This work has been partly supported by the French government, in the frame of STRADE research project (ANR-09-VERS-010).
References and links
1. G. Bosco, V. Curri, A. Carena, P. Poggiolini, and F. Forghieri, “On the performance of Nyquist-WDM Terabit Superchannels Based on PM-BPSK, PM-QPSK, PM-8QAM or PM-16QAM Subcarriers,” J. Lightwave Technol. 29, 53–61 (2011). [CrossRef]
2. A. Li, A. Al Amin, X. Chen and W. Shieh, “Reception of Mode and Polarization Multiplexed 107-Gb/s CO-OFDM Signal over a Two Mode Fiber,” OFC 2011, Paper PDPB8 (2011).
3. M. Salsi, C. Koebele, D. Sperti, P. Tran, P. Brindel, H. Mardoyan, S. Bigo, A. Boutin, F. Verluise, P. Sillard, M. Astruc, L. Provost, F. Cerou, G. Charlet, “Transmission at 2x100Gb/s, over Two Modes of 40km-long Prototype Few-Mode Fiber, using LCOS based Mode Multiplexer and Demultiplexer,” OFC 2011, Paper PDPB9 (2011).
4. R. Ryf, S. Randel, A. H. Gnauck, C. Bolle, R.-J. Essiambre, P. J. Winzer, D. W. Peckham, A. McCurdy, and R. Lingle, Jr., “Space-division multiplexing over 10 km of three-mode fiber using coherent 6×6 MIMO processing,” OFC 2011, Paper PDPB10 (2011).
5. N. Hanzawa, K. Saitoh, T. Sakamoto, T. Matsui, S. Tomita, and M. Koshiba, “Demonstration of mode-division multiplexing transmission over 10 km two-mode fiber with mode coupler,” OFC 2011, Paper OWA4 (2011).
6. R. Olshansky, “Mode Coupling Effects in Graded-index Optical Fibers,” Appl. Opt. 14(4), 935–945 (1975). [PubMed]
7. F. Yaman, N. Bai, Y. K. Huang, M. F. Huang, B. Zhu, T. Wang, and G. Li, “10 x 112Gb/s PDM-QPSK transmission over 5032 km in few-mode fibers,” Opt. Express 18(20), 21342–21349 (2010). [CrossRef] [PubMed]
8. C. Koebele, M. Salsi, G. Charlet, S. Bigo, “Nonlinear Effects in Long-Haul Transmission over Bimodal Optical Fibre,” ECOC 2010, Paper Mo.2.C.6 (2010).
9. G. Stepniak, L. Maksymiuk, and J. Siuzdak, “Increasing Multimode Fiber transmission Capacity by Mode Selective Spatial Light Phase Modulation,” ECOC 2010, Paper P6.03 (2010)
10. D. Sperti, M. Salsi, C. Koebele, P. Tran, H. Mardoyan, S. Bigo, A. Boutin, P. Sillard, G. Charlet. “Experimental Investigation of Modal Crosstalk using LCOS-based Spatial Light Modulator for Mode Conversion,” ECOC 2011, Paper Th.12.B.2 (2011).
11. C. Koebele, M. Salsi, D. Sperti, P. Tran, P. Brindel, H. Mardoyan, S. Bigo, A. Boutin, P. Sillard, F. Cerou, G. Charlet, “Two-Mode Transmission with Digital Inter-Modal Cross-Talk Mitigation,” ECOC 2011, Paper Tu.5.B.4 (2011).
12. P. Sillard, M. Bigot-Astruc, D. Boivin, H. Maerten, L. Provost, “Few-Mode Fiber for Uncoupled Mode-Division Multiplexing Transmissions,” ECOC 2011, Paper Tu.5.LeCervin.7 (2011).
13. J.-F. Cardoso and B. H. Laheld, “Equivariant adaptive source separation,” Trans. Signal Processing 44(12), 3017–3030 (1996). [CrossRef]