Center-launching technique appears as a promising method to allow single-mode propagation in multi-mode fibers, guaranteeing full transparency to the transmitted optical signal also for applications in board-to-board and data server interconnects. In this paper we show that this technique is robust to mechanical perturbations up to about 1 kHz, demonstrating that the vibrations do not affect the transmission performances. Different experimental configurations are tested in order to exclude multimode propagation and to confirm the only fundamental mode propagation. Finally, a theoretical discussion comments the experimental results.
© 2013 OSA
The exponential growing of demand for bandwidth capacity and the need to develop fast, energy-efficient and low cost solutions in board-to-board and data interconnects inside massive data centers lead to find alternative electrical and optical solutions [1–4]. Recently, multi-mode fibers (MMFs) have assumed a growing interest in datacom networks  Usually, the transmission distance in a backplane system is less than 1.5 meters. It has already been demonstrated that MMF is a good choice thanks to its easy handling, flexibility, and high tolerance to alignment offset in fiber-to-transmitter/receiver and fiber-to-fiber connections. Moreover, MMF is currently employed also in short-reach data center interconnections and local-area networks (LAN) in conjunction to directly-modulated vertical-cavity surface emitting lasers (VCSELs), which permit to reach bit rates as high as 10 Gb/s with low costs, low network complexity and low power consumption up to hundreds of meters. When the covered distances are short the impairment induced during propagation by MMF intermodal dispersion does not generally affect data transmission, also in case of high transmission bit rate . Since all these advantages are required also in backplane and datacom applications, the exploitation of MMF with fast VCSELs for shorter distances should be mandatory. Moreover, the new optical backplane would be completely compatible with local area networks. To increase the whole transported capacity, in MMFs it is possible even to exploit mode division multiplexing  with standard LP modes  or alternative optical vortices [9–11].
Optical data links should guarantee also a complete transmission transparency in the interconnection with the external network, i.e. a travelling signal must be allowed to pass through the backplane network on a determined internal link without any need of opto-electronic and electro-optic conversion and processing. This hypothesis should allow transferring very high bit rate optical signals also with complex modulation formats and multiplexing (for example in wavelength  or in polarization ). In this configuration, the connected input/output signal is presumably travelling into a single-mode fiber (SMF) coming from and going to the external network. The employment of MMF in the interconnect link between two SMFs might be critical because of the introduction at the output interface of power losses and modal noise, which cause undesired link loss fluctuations and can have a strong detrimental impact on the quality of the received signal . Moreover, simplicity and scalability are two main requirements of the new backplanes, which introduce another constrain to network architecture. Hence, the optical backplane should be composed by MMfibers for their advantages but, in the meantime, allowing the transit of complex signals from the external network without introducing any impairment or exploiting optoelectronic conversions.
Recently, for MMF propagation the mode-field matched (MFM) center-launching technique has been proposed, in which the beam profile of the incident light is precisely matched to the fundamental mode (LP01 mode) of the MMF . Exploiting this technique, since only the fundamental mode of MMF is excited, it is possible to directly connect the input SMF to the MMF ensuring the transparency to the transmitted signal and SMF-like propagation [16,17].
On the other hand, ETSI standards  recommend the robustness to mechanical perturbations required to the telecommunications equipment. If MFM center-launching condition is exploited, in case of external perturbations the system transmission performances could be deteriorated due to higher order modes excitation in MMF. Hence, it is mandatory to investigate the impact of mechanical perturbations on the performances of these board-to-board MMF systems in order to guarantee error-free transmission .
In this paper we present the analysis of the robustness of the center-launching technique in MMF in presence of mechanical vibrations up to about 1 kHz applied to the system. Polarization analysis and transmission performances at 1-Gb/s bit rate are evaluated in case of propagation along 2 m of a standard graded-index legacy MMF, when the mechanical perturbation is applied. To reproduce different operating conditions in backplane applications, we have analyzed two experimental configurations: at first, the transmitted signal is received directly on a receiver, set on the board in the internal network; later, the transmitted signal is transferred into another external network in a transparent way passing through the MMF backplane. In the former configuration a simple free-space photodiode is adopted at the receiver side, while in the latter one a SMF pigtailed photodiode is used.
Finally, we considered a 1-Gb/s transmission on a long MMF span (4.5-km length), e.g. when a large amount of modal dispersion could affect the performances if multimode propagation occurs in the link. This configuration emulates a direct connection between the backplane and a data center interconnection or a local area network in which the maximum transmission distance is not limited to 300 meters. As explained in the theoretical discussion, all these different experimental tests are necessary to exclude the generation of higher order modes in the MMF and to demonstrate the robustness of the center-launching technique to the applied mechanical vibrations.
2. Experimental validation of the robustness to mechanical vibrations of the center-launching technique in 2-m MMF link
The transparent connection achieved in a MMF by exploiting the center-launching technique was subjected to mechanical tests. The experimental set up used for the mechanical characterization is shown in Fig. 1 . The transmitter employed in our experimentation was a single-mode VCSEL source [20,21] suitable for high-speed applications. The VCSEL was pigtailed by a standard step-index SMF; the output power was about 1 mW at 1335 nm with a bias current about 11 mA in uncooled conditions. Direct modulation at 1 Gb/s with NRZ data (231-1 PRBS pattern length) was performed.
In order to achieve the optimal center-launching condition into the MMF, the SMF pigtail was fusion-spliced directly to the MMF spoil of the length of 2m. The mode field pattern coming from the SMF matched very well with the fundamental mode of the standard graded-index MMF, which is a OM2 fiber with 50-μm core diameter, characterized by overfilled launch (OFL) bandwidth of 2000 MHz*km at 1310 nm. If inaccurate SMF-MMF coupling is performed mode-field mismatch occurs and the higher-order modes are excited into the MMF with a power comparable with the one of the fundamental mode. The center-launching technique effectiveness was estimated without any applied vibration in presence of a free-space photodiode (PD) at the output. The measured power losses of the optical signal between the output of the pigtailed input SMF and the output of the MMF was about 0.4 dB. The intensity profile at the MMF output was monitored thanks to a CCD camera (in the inset of Fig. 1 the detected light spot is shown). The very low power loss and the Gaussian intensity pattern visible after MMF propagation sustain the effectiveness of the realized center-launching technique.
Then, the MMF was rigidly fixed on the shaker, therefore the fiber doesn’t move with respect to the shaker, when the vibration is applied, as shown in Fig. 1. A sinusoidal signal at different frequencies (between 10 Hz and 800 Hz) was applied on a shaker to generate the vibration on the optical fiber. To verify the real mechanical transfer of the vibration to the MMF link, we chose to analyze the impact of the vibration on the state of polarization of the optical signal by means of a linear polarizer placed at the fiber output, before the free-space PD. In Fig. 2 some examples of the signal at the output of the MMF are reported. Due to frequency response of the shaker checked by a suitable accelerometer the signal transferred to the MMF is very distorted at very low vibration frequencies. As the vibration frequency increases, it becomes more regular up to a good sine shape.
To check if the vibration affects the performances of the MMF interconnect link, BER measurements were performed. The experimental setup was in this case without the linear polarizer used for the polarization analysis. At the receiver side, for the first round of measurement, the MMF optical output was directly detected by a free-space coupled PIN PD connected to the BER tester for the transmission performances characterization. The eye diagram of the received signal after 2-m MMF propagation is shown in Fig. 3 in two different conditions: when no vibration and when a vibration of 97 Hz is applied (an extinction ratio of 6.2 dB was achievable optimizing the amplitude modulation). No visible distortions are present. The BER curves as a function of the received power are reported in Fig. 4 : Theperformances after 2-m MMF link at different vibration frequencies are shown. It is interesting to note that no significant penalties are visible, in spite of polarization changes previously measured.
In order to verify the stability of the center-launching technique, a temporal BER analysis is carried out. The receiver power is fixed to be about −10.3dBm. Every 10 minutes the condition of the system was changed alternatively with no vibration and with vibration at different frequency. No significant BER variation around the value of 10−6 induced by the vibration is noticed (Fig. 5 ). BER fluctuations are limited in the normal error measurement range. These results demonstrate that the vibration does not modify the initial condition of the system and the polarization changes do not impair the system performances.
As we will better explain in the theoretical discussion presented in Section 3, the experimented absence of impairment in the propagation performances when mechanical vibrations affect the system does not entirely exclude the rise of higher modes in the under-test MMF. Hence, BER measures were performed considering a second configuration at the receiver side: the end of the 2-m MMF was coupled with a SMF. In this configuration, the final SMF works as a spatial filter. If just the fundamental mode is propagating in the MMF, the mode matches with the single mode of SMF, showing fixed power losses for not-perfectly centered coupling. However, if higher order modes are excited in the MMF, they can couple with the fundamental mode of SMF, causing modal noise and affecting the system performances. Figure 6 shows BER measurements at different vibration frequencies in the case of SMF-coupling at the output of the experimental system before the receiver. A PD with a sensitivity of −17 dBm has been used. No visible degradation is present. The system performances do not change when the vibration is applied if the center launching technique is optimized.
Finally, to check if the effects of the vibration change for different positions of the applied vibration on the system, we placed the shaker on the first SMF span of the VCSEL pigtail. The measures were repeated in the same condition explained previously. Also in this case the vibrations do not affect the performances system.
3. Theoretical discussions
In the previous Section we have shown that an external vibration in the range 10-800 Hz does not introduce any significant impairment in 2-m MMF propagation when exploiting center-launching technique to excite just the fundamental mode in the MMF link. This result has been verified with and without the filtering effect achieved by coupling the MMF with a final SMF.
It is important to know what value of frequency can induce the rise of high-order modes in the fiber. Generally, the external vibration produces an acoustic grating with a determined βeq, which can permit the energy flow from one mode to the other . To make possible this transfer the βeq of the grating must be similar to the difference of the propagation coefficient Δβ between the two modes. More a mode is close to another one in terms of propagation coefficient, more the energy transfer between them is possible. Furthermore, in order to have a significant coupling efficiency, the spatial pattern of the perturbation must show the symmetry for matching one mode to another.Eq. (1) it is possible to estimate the frequency needed for matching the difference in the propagation constants of the LP01 and LP11 modes. Entering the geometrical parameter of the OM2 fiber, the difference Δβ01/11 has been evaluated by BeamPROP Tool, obtaining , which corresponds to a beat length between LP01 and LP11 modes of 5.3 mm. Therefore, in order to obtain the coupling between the two considered LP modes it’s required a spatially-periodic perturbation  with quite small pitch. Using Eq. (1) we obtain an acoustic frequency f of about 1 MHz. Due to this high frequency value, a realistic external perturbation can difficulty lead to a power transfer between the fundamental and a higher order mode.
Moreover, it is important to notice that generally no penalties can be observed for very short propagation distance and free space detection after MMF propagating (without SMF filter) if the whole optical field is received by the PD. Also when multimode propagation occurs, thanks to the orthogonality of the different modes of a multimode fiber, no beating will be seen at the receiver if the whole spatial distribution of the output optical field is collected in the PD area. In fact, considering two generic modes in cylindrical coordinates, the electrical signal after the detector can be expressed asEq. (2) will vanish thanks to the orthogonality between the modes. In this situation, the performances of the system may be limited only by modal dispersion. Equation (2) can be generalized for n incident modes leading to the same results. When a spatial filter is introduced (for example by coupling the MMF with a SMF), instead, only a part of the spatial distribution at the output of the MMF can couple to the fundamental mode propagating in the SMF. Since the MMF to SMF alignment cannot be perfectly controlled, the presence of higher order modes is traduced in a reduction of the coupled power and in the generation of modal noise. Moreover, the arising of modal noise normally leads to an increasing of polarization dependence in the performances of the optical system.
Hence, while short-reach free-space detection is robust also in presence of higher order modes generated in input or by an external perturbation, a SMF-MMF-SMF system demands strict requirements in terms of modal purity in the multimode section.
4. Experimental validation of the robustness to mechanical vibrations of the center-launching technique in presence of large modal dispersion
In Section 2 also the SMF-MMF-SMF link exploiting center-launching in 2-m MMF has been demonstrated very robust to external mechanical perturbations. But, as explained in Section 3, this configuration employing SMF filtering remains extremely sensitive to the presence of higher order modes in fiber. The experimentally measured absence of modal noise in our measurement in Fig. 6 could be due to a particular splicing between MMF and SMF. A large amount of splicing with different offset should be tested to guarantee that no higher order modes are generated and so no modal noise can occur in case of spatial filtering due to fiber connections.
To demonstrate that actually no higher-order modes are excited when vibrations affect the link exploiting center-launching, we have analyzed the propagation on a longer OM2 MMF span (4.5 km). For this reach, operating at 1-Gb/s bit rate at 1335 nm, modal dispersion should be evident in case of multimode propagation (OFL bandwidth is 2000 MHz*km at 1310 nm).
We considered both detection configurations analyzed for 2-m MMF propagation. By exploiting free-space PD configuration, we are able to eventually check the impact of pure modal dispersion in the BER measures; in case of SMF-filter configuration at the receiver side, also the penalties due to modal noise could be observed. The BER measurement at different vibration frequencies for both configurations are shown in Fig. 7 . For free-space configuration, no significant penalty is shown for the different vibration frequency. The slight spread of the curves at very low BER (10−9 ÷ 10−10) is due to the absence of clock and data recovery and it is in the range of the measurement error. When the SMF filter is inserted, a slight spread can be observed for BER lower than 10−6. However, all the curves are enclosed in a range of about 0.5 dB and no error floor is visible, showing that the introduced modal dispersion has a negligible effect on the performances and hence experimentally demonstrating that no higher-order modes are excited when vibrations affect the link exploiting center-launching.
Also in case of long MMF propagation, we monitored the output intensity profile with a CCD camera, also both in free space and SMF-filter configuration, before and during the vibration, without detecting any significant changing.
Moreover, the experimented robustness also in this long MMF propagation suggests the capability of exploiting the center launching technique to transmit directly from the backplane to the data center interconnections or to a LAN longer than hundreds of meters. Thanks to the center launching, in fact, the reach limitation due to modal dispersion will be greatly reduced. A complete integration between optical backplanes, the data center interconnections and the local area networks could be mandatory in future short-reach networks for limiting costs, power consumption, guaranteeing greater scalability and simplicity.
We have shown that the exploitation of center-launching technique guarantees an almost SMF-like propagation in MMF. The transmission of a 1-Gb/s NRZ OOK signal remains robust to mechanical perturbations as required by ETSI recommendations, demonstrating that no higher order modes are generated owing to the applied vibrations.
No BER penalties have been obtained for vibrations with frequency up to about 1 kHz, both by directly detecting the optical signal at the MMF output by a PD and by coupling the MMF with a SMF in order to achieve a spatial filtering of the field propagating in the MMF. To confirm SMF-like propagation and to exclude multimode propagation in the MMF subjected to vibrations, also BER tests in case of a long MMF span in presence of a large amount of modal dispersion have been performed.
The actual account of the mechanical perturbation frequency that can induce the generation of the first higher order mode has been achieved by suitable simulations, obtaining a frequency of about 1 MHz, much larger than the frequencies characteristic of the realistic environmental perturbations.
Therefore, from the mechanical point of view, center-launching technique appears as a promising method to ensure full transparency to the transmitted signals for applications in board-to-board and data server interconnects.
The authors wish to thank VERTILAS company for the VCSEL supply.
References and links
1. S. Nishimura, K. Shinoda, Y. Lee, G. Ono, K. Fukuda, K. Fukuda, T. Takemoto, H. Toyoda, M. Yamada, S. Tsuji, and N. Ikeda, “Components and interconnection technologies for photonic-assisted routers toward green networks,” IEEE J. Sel. Top. Quantum Electron. 17(2), 347–356 (2011). [CrossRef]
2. N. Fehratović and S. Aleksić, “Power consumption and scalability of optically switched interconnects for high-capacity network elements,” in Proceedings OFC/NFOEC 2011, paper JWA84, Los Angeles, CA, USA (2011).
3. M. A. Taubenblatt, “Optical interconnects for high-performance computing,” J. Lightwave Technol. 30(4), 448–457 (2012). [CrossRef]
4. N. Y. Li, C. L. Schow, D. M. Kuchta, F. E. Doany, B. G. Lee, W. Luo, C. Xie, X. Sun, K. P. Jackson, and C. Lei, “High-performance 850 nm VCSEL and photodetector arrays for 25 Gb/s parallel optical interconnects,” in Proceedings OFC/NFOEC 2010, paper OTuP2, San Diego, CA, USA (2010).
5. D. Molin, G. Kuyt, M. Bigot-Astruc, and P. Sillard, “Recent advances in MMF technology for data networks,” in Proceedings OFC/NFOEC 2011, paper OWJ6, Los Angeles, CA, USA (2011).
6. T. Irujo and J. Kamino, “Multimode or single-mode fiber?” www.ofsoptics.com (2012).
8. J. Carpenter, B. C. Thomsen, and T. D. Wilkinson, “2x56-Gb/s Mode-division multiplexed transmission over 2km of OM2 multimode fiber without MIMO equalization,” in Proceeding ECOC 2012, paper Th.2.D.3, Amsterdam, NL (2012).
9. S. Ramachandran, N. Bozinovic, P. Gregg, S. E. Golowich, and P. Kristensen, “Optical vortices in fibres: a new degree of freedom for mode multiplexing,” Proceedings ECOC2012, paper Tu.3.F.3, Amsterdam, NL (2012).
10. A. Gatto, M. Tacca, P. Martelli, P. Boffi, and M. Martinelli, “Free-space orbital angular momentum division multiplexing with Bessel beams,” J. Opt. 13(064018), 1–5 (2011).
11. P. Martelli, A. Gatto, P. Boffi, and M. Martinelli, “Free-space optical transmission with orbital angular momentum division multiplexing,” Electron. Lett. 47(17), 972–973 (2011). [CrossRef]
12. G. Charlet, E. Corbel, J. Lazaro, A. Klekamp, R. Dischler, P. Tran, W. Idler, H. Mardovan, A. Konczykowska, F. Jorge, and S. Bigo, “WDM transmission at 6-Tbit/s capacity over transatlantic distance, using 42.7-Gb/s differential phase-shift keying without pulse carver,” J. Lightwave Technol. 23(1), 104–107 (2005). [CrossRef]
13. P. Boffi, M. Ferrario, L. Marazzi, P. Martelli, P. Parolari, A. Righetti, R. Siano, and M. Martinelli, “Stable 100-Gb/s POLMUX-DQPSK transmission with automatic polarization stabilization,” IEEE Photon. Technol. Lett. 21(11), 745–747 (2009). [CrossRef]
14. P. Pepeljugoski, D. Kuchta, and A. Risteski, “Modal noise BER calculations in 10-Gb/s multimode fiber LAN links,” IEEE Photon. Technol. Lett. 17(12), 2586–2588 (2005). [CrossRef]
15. 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]
16. 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]
17. P. Boffi, A. Gatto, A. Boletti, P. Martelli, and M. Martinelli, “12.5 Gbit/s VCSEL-based transmission over legacy MMFs by centre-launching technique,” Electron. Lett. 48(20), 1289–1290 (2012). [CrossRef]
18. European Standards: ETSI EN 300 019-2-2 V2.2.1 (2012-01); ETSI EN 300 019-2-3 V2.2.2 (2003-04); ETSI EN 300 019-2-4 V2.2.2 (2003-04).
19. D. H. Sim, Y. Takushima, and Y. C. Chung, “Robustness evaluation of MMF transmission link using mode-field matched center-launching technique,” in Proceedings OFC/NFOEC 2008, paper OWR3, San Diego, CA, USA (2008) [CrossRef]
20. A. Gatto, A. Boletti, P. Boffi, E. Ronneberg, C. Neumeyr, M. Ortsiefer, and M. Martinelli, “1.3 μm VCSEL transmission performance up to 12.5 Gbit/s for metro access networks,” IEEE Photon. Technol. Lett. 21(12), 778–780 (2009). [CrossRef]
21. A. Gatto, A. Boletti, P. Boffi, and M. Martinelli, “Adjustable-chirp VCSEL-to-VCSEL injection locking for 10-Gb/s transmission at 1.55 microm,” Opt. Express 17(24), 21748–21753 (2009). [CrossRef] [PubMed]
23. A. Li, A. Al Amin, X. Chen, and W. Shieh, “Reception of mode and polarization multiplexed 107-Gb/s COOFDM signal over a two-mode fiber,” in Proceedings OFC/NFOEC 2011, paper PDPB8, Los Angeles, CA, USA (2011).