Open Access
Add to BibSonomyAbstract
Utilizing low-cost, 2.2GHz modulation bandwidth, uncooled and standalone directly modulated VCSEL (DM-VCSEL)-based real-time dual-band optical OFDM (OOFDM) transmitters, aggregated 16.375Gb/s transmissions of OOFDM signals having bandwidths approximately 3.8 times higher than the VCSEL manufacturer-specified modulation bandwidths, are experimentally demonstrated, for the first time, over 200m OM2 MMF links based on intensity modulation and direct detection. The aggregated signal transmission capacities of the aforementioned links vary by just 8% for various OM2 MMFs ranging from 100m to 500m, and by just 10% over a 1GHz passband carrier frequency detuning range. Such dual-band OOFDM adaptability-induced excellent performance robustness and large passband frequency tunability can significantly relax the requirements on VCSEL modulation bandwidth for achieving specific transmission performances for cost-sensitive application scenarios such as data centers.
© 2015 Optical Society of America
1. Introduction
In data centers, multi-mode fiber (MMF) links incorporating directly modulated vertical cavity surface emitting lasers (DM-VCSELs) are widely adopted as a cost-effective, high-capacity, scalable and low-power consumption technical strategy [1]. To satisfy the exponential data traffic growth driven by a large number of emerging services such as streaming video, social networking and higher performance computing, those existing DM-VCSEL-based MMF links are facing great challenges to be upgraded, in a cost-effective manner, to 10Gb/s/λ and far beyond [1]. In addressing such a technical challenge, significant cost-savings can be made if already installed MMF transmission links can be utilized with minimum modifications rather than pulling new plants. However, narrow optical bandwidths of legacy MMFs and low-modulation bandwidths of commercially available VCSELs become the most significant obstacles to practically achieving the desired transmission performance of the aforementioned transmission links based on conventional transmission techniques.
To considerably broaden the optical bandwidths of existing MMF links, several MMF bandwidth enhancement techniques have been proposed including, for example, mode-field center launch [2,3], spatial light modulation [4,5] and mode-group division multiplexing [6]. These techniques are, however, strongly system-dependent, implying that massive technical barriers still remain to be solved before they are viable for future practical deployment in data centers. On the other hand, optical injection locking (OIL) has also been shown to be effective in considerably broadening VCSEL modulation bandwidths [7,8]. However, apart from the additional requirement of CW light sources with very high output optical powers, the 3dB modulation bandwidth and corresponding frequency response profile of a directly modulated OIL-VCSEL are very sensitive to OIL conditions applied. For extremely cost-sensitive application scenarios such as data centers, it is prohibitive to maintain all those OIL conditions sufficiently stable over a long time period. Therefore, it is greatly advantageous if a standalone, uncooled, low-modulation bandwidth DM-VCSEL can still be applicable in upgrading existing MMF links to >10Gb/s/λ via fully exploiting the link’s available system frequency response over a spectral region far beyond its’ 3-dB optical bandwidth. In addition, it is also highly desirable if the transmission performance of the above-mentioned transmission link has sufficient robustness against variations in both optical signal spectral width and transmission distance.
Due to its’ unique features namely excellent performance adaptability to system frequency response variation, highly spectral efficiency as well as low-power consumption potential, optical orthogonal frequency division multiplexing (OOFDM) has been considered as one of the strongest “future-proof” techniques for practical implementation in DM-VCSEL-based MMF links [9,10]. Recently, end-to-end real-time serial transmissions of 11.25Gb/s single-band OOFDM (SB-OOFDM) signals over 2000m legacy OM1/OM2 MMFs have been experimentally achieved using standalone low-bandwidth DM-VCSELs and simple intensity modulation and direct detection (IMDD) [11,12]. To further improve the OOFDM signal transmission capacity, and to still remain those low-cost key components associated with the SB-OOFDM transceivers, dual-band OOFDM has been reported to successfully transmit end-to-end real-time 15.125Gb/s signals over directly modulated OIL-VCSEL-based 100m OM2 MMF links incorporating digital-to-analogue converters (DACs) and analogue-to-digital converters (ADCs) operating at sampling speeds as low as 4GS/s [13]. In the real-time dual-band OOFDM transceiver, adaptively encoded baseband and passband sub-bands are independently generated, and then electrically frequency-division-multiplexed (FDM) for intensity modulation of the OIL-VCSEL. It has also been experimentally shown [13] that the dual-band transceiver frequency tunability in both the electrical and optical domains enables the full use of entire OIL-enhanced dynamic frequency responses. Considering the aforementioned disadvantages associated with directly modulated OIL-VCSELs, the use of standalone DM-VCSELs to realize dual-band OOFDM transmissions over MMF links with improved capacity versus reach performances is of great importance for practical implementation in data centers.
In this paper, by making use of a low-cost, low modulation bandwidth, uncooled, standalone DM-VCSEL, we report, for the first time, aggregated 16.375Gb/s dual-band OOFDM transmissions over 200m OM2 MMF links based on simple IMDD. In comparison with the directly modulated OIL-VCSEL-based MMF links [13], the considerably improved transmission performance and the simplified transmitter configuration are achieved by: a) appropriate adjustments of the relative sub-band signal power levels, and b) further optimizations of DM-VCSEL operating conditions. Our experimental results also show that the standalone VCSELs can intensity-modulate adaptive electrical OFDM signals having bandwidths approximately 3.8 times wider than its’ manufacturer-specified 3dB modulation bandwidth. In addition, a 1GHz passband RF frequency detuning range is also experimentally demonstrated, over which aggregated signal transmission capacity variations of <10% are measured. Furthermore, excellent robustness of the standalone DM-VCSEL-based dual-band OOFDM transmission capacity is also observed over a wide OM2 MMF length range varying from 100m to 500m.
2. Experimental setup of DM-VCSEL-based dual-band OOFDM MMF systems
Figure 1 shows the DM-VCSEL-based IMDD MMF link employing the real-time dual-band OOFDM transmitters, whose physical design architectures and core field programmable gate array (FPGA)-based OFDM digital signal processing (DSP) functions can be found in [14,15]. The adopted parameters for the key transceiver components and corresponding transmission link are presented in Table 1 and Table 2. The real-time OFDM transmitter of each sub-band consists of a 4GS/s@8-bit DAC and a FPGA for performing high-speed transmitter DSP functions including, for example, pseudo-random data generation, pilot-tone insertion, online adaptive bit and power loading for 15 data-carrying subcarriers with signal modulation formats selected from 16-quaternatry amplitude modulation (QAM), 32-QAM and 64-QAM, a 32-point inverse fast Fourier transform (IFFT) with input complex data being arranged to satisfy the Hermitian symmetry, online signal clipping level adjustment, 8-bit sample quantization and cyclic prefix (CP) insertion. The above real-time OFDM transmitter design offers live optimizations of both subcarrier bit/power allocation and digital signal clipping level using a FPGA embedded memory editor via a joint test action group (JTAG) connection to a personal computer. Such online parameter manipulation capability enables not only the rapid identification of optimum transceiver/system parameters, but also easy explorations of the highest achievable system performances for different conditions.
Fig. 1 Experimental system setup of DM-VCSEL-based dual-band OOFDM OM2 MMF link utilizing real-time OOFDM transmitters. VEA: variable electrical attenuator; AMP: RF amplifier; BPF1: band pass filter (5.6-7.0GHz); BPF2: band pass filter (3.9-9.8GHz); LPF: low pass filter; VOA: variable optical attenuator; LO: local oscillator; MCPC: mode conditioning patch-cord; PIN + TIA: photodetector with integrated transimpedance amplifier.
Table 1. OOFDM Transceiver and Link Parameters
Table 2. Dual-band OFDM Signal Power Levels
As shown in Fig. 1, to simultaneously generate two separate electrical OFDM sub-band signals, independent digital and RF electronics are adopted in each individual sub-band: one electrical sub-band, referred to as the baseband, occupies a spectral region from 0 to 2GHz, whilst the second electrical sub-band, termed the passband, is produced by amplitude modulating a 6.125GHz RF carrier with another 0-2GHz OFDM baseband signal. Such up-conversion is performed via a double-balanced mixer followed by a bandpass filter to attenuate the unwanted out-of-band spectral components. This leads to the generation of a double sideband passband OFDM signal occupying a spectral region from 4.125 to 8.125GHz. After appropriately adjusting both the relative sub-band signal power levels and the total dual-band signal power levels, these two OFDM sub-bands are combined in a low-loss RF multiplexer and then added to an optimum 6.4mA DC bias current in a 26.5GHz bias tee. The resulting dual-band OFDM signal is finally employed to drive a standalone, uncooled, single-mode VCSEL to generate an intensity-modulated OOFDM signal. The power of the OOFDM signal is adjusted by a variable optical attenuator (VOA), from which the optical signal is launched into an OM2 MMF link via a commercially available mode conditioning patch-cord (MCPC). The MCPC enables stable conventional offset launch, thus different launch condition-induced variations in system frequency response can be eliminated.
At the receiver, a −19dBm sensitivity MMF-pigtailed 12GHz P type-intrinsic-N type (PIN) with integrated transimpedance amplifier (MMF-PIN + TIA) is adopted to perform the optical to electrical conversion of the dual-band OOFDM signal via direct detection. The received analog dual-band electrical OFDM signal is then captured and digitized by a real-time sampling oscilloscope and consequently processed using MatLab. To recover the baseband (passband) OFDM signal, a digital down-conversion section whose functions are identical to the analog RF down-conversion circuit reported in [15,16] is omitted (included), as illustrated in Fig. 1. The main functionalities of the digital down-conversion process include estimations of the RF carrier frequency and phase offsets, and the corresponding offset compensations. After passing through digital low pass filters (LPFs), both the received baseband and down-converted passband signals are subject to conventional OFDM receiver DSP procedures including symbol synchronization, pilot-subcarrier detection, channel estimation/equalization, as well as all other receiver DSP functions that are just inverse to their transmitter DSP counterparts [17]. The total sub-band bit error rate (BER) and individual subcarrier BER of each individual sub-band are continuously and simultaneously calculated and displayed in the oscilloscope. This enables the rapid optimization of the overall transmission systems via careful component and link parameter adjustments.
3. Experimental results
To experimentally demonstrate the feasibility of utilizing the dual-band OOFDM technique to upgrade existing DM-VCSEL-based MMF links to 10Gb/s/λ and beyond, in Subsection 3.1, the optimized dual-band OFDM signal characteristics are first presented in terms of their adaptive bit/power loading profiles and relative electrical sub-band power levels. The online optimizations of these parameters pave a solid path leading to the successful experimental demonstrations of aggregated 16.375Gb/s dual-band OOFDM transmissions over standalone DM-VCSEL-based 200m OM2 MMF IMDD links in Subsection 3.2. In addition, the excellent signal transmission capacity robustness of the demonstrated transmission link is also experimentally explored in Subsection 3.3 to variations in both passband RF frequencies and OM2 MMF transmission distances of up to 500m.
3.1 Optimized dual-band OFDM signal characteristics
To effectively compensate for the overall system frequency response roll-off effect [16] and simultaneously maximize the system transmission capacity, parameter optimizations are first conducted via adaptive subcarrier bit and subcarrier/sub-band power loading. For the considered 200m OM2 MMF IMDD link incorporating the DM-VCSEL under the operating conditions listed in Table 1 and Table 2, for each OFDM sub-band, the adaptively loaded and received subcarrier power profiles are given in Fig. 2(a), where the system frequency response of each individual sub-band is also plotted which is measured from the IFFT input in the transmitter to the FFT output in the receiver and subsequently normalized to the first subcarrier power of the corresponding sub-band.
Fig. 2 (a) Adaptively loaded and received subcarrier powers, and normalized overall system frequency response for each sub-band; (b) Adaptively loaded optimum subcarrier bit profiles for two sub-bands offering aggregated 16.375Gb/s OOFDM transmissions over 200m OM2 MMFs.
It can be seen in Fig. 2(a) that, compared to the received baseband signal, much smaller subcarrier power variations between low and high frequency subcarriers are observed for the received passband signal. This is due to the relatively flat passband system frequency response, as shown in Fig. 2(a). On the other hand, the resulting optimum subcarrier bit allocation profiles for both sub-bands are also presented in Fig. 2(b). The co-existence of the finite FPGA/DAC power dynamic range-induced imperfect compensation for the passband system frequency response roll-off as well as the limited relative sub-band power allocation leads to the completely dropping of the last four high-frequency subcarriers in the passband signal, as seen in Fig. 2(b). Based on Fig. 2(b) and the transceiver parameters listed in Table 1, it can be easily worked out that the achieved signal bit rates for the baseband and passband are 10.875Gb/s and 5.5Gb/s, respectively, thus giving rise to an aggregated signal transmission capacity of 16.375Gb/s.
Various 16.375Gb/s dual-band OOFDM signal spectra are shown in Fig. 3 at different points of the considered transmission link: Fig. 3(a) for the output signal of the RF multiplexer in the transmitter; Fig. 3(b) for the received signal at the output of the PIN/TIA after optical back-to-back (BTB) transmissions, and Fig. 3(c) for the output signal of the PIN/TIA after signal transmissions over the DM-VCSEL-based 200m OM2 MMF link. For both the optical BTB and MMF transmission cases, the received optical powers (ROPs) are fixed at −3dBm, and the measured sub-band power levels are listed in Table 2.
Fig. 3 Measured spectra of 16.375Gb/s dual-band OOFDM signals. (a) Output signal of the RF multiplexer in the transmitter; (b) received signal at the output of the PIN for the optical BTB case (ROP = −3dBm); (c) received signal at the output of the PIN after the 200m OM2 MMF transmission (ROP = −3dBm).
In order to achieve similar BER developing curves for both sub-bands, it is necessary to employ an optimum sub-band signal power ratio of 5.7dB between the passband and the baseband at the transmitter, as listed in Table 2. After the optical BTB transmission, such a signal power ratio is reduced to −4.3dB, indicating that the passband OFDM signal suffers from much more attenuation than the baseband OFDM signal. This agrees very well with experimental measurements reported in [15]. Although the passband to baseband power ratio is further reduced to −4.7dB after the 200m OM2 MMF transmission, very similar BER developing curves for these two sub-bands can still be achieved as discussed in Subsection 3.2. This is due to adaptive modulation-associated linear trade-off between the sub-band signal transmission capacity and the ROP [18]: for a specific transmission system, a high transmission capacity and/or modulation format requires a large ROP for achieving the similar BER performance. Given the fact that the system frequency response of a MMF link depends on the MMF length, the optimum sub-band signal power ratio between the passband and the baseband at the transmitter is also a function of MMF length.
3.2 16.375Gb/s dual-band OOFDM transmissions over 200m OM2 MMFs
In this Subsection, experimental measurements are undertaken of the 16.375Gb/s dual-band OOFDM transmission performance over the DM-VCSEL-based 200m OM2 MMF link based on IMDD. Here the optimum bit and power loading profiles presented in Fig. 2 are employed, and the optimum DM-VCSEL operating conditions presented in Table 1 are also adopted together with the optimized sub-band power levels listed in Table 2.
For both the optical BTB and the entire 200m OM2 MMF IMDD link, Fig. 4(a) presents the measured BER performances of both sub-bands as a function of ROP. It can be seen in Fig. 4(a) that the BER developing trends for both sub-bands under different link configurations are very similar, indicating that the standalone DM-VCSEL can support the successful transmissions of adaptively modulated 16.375Gb/s dual-band OOFDM signals having their bandwidth far beyond its’ 3dB modulation bandwidth. Compared to the passband, for various link configurations considered here, relatively high BER values for the baseband occur at ROPs of < −7dBm. This is because unwanted inter- and intra-sub-band intermixing frequency products generated upon square-law photon detection in the receiver are predominantly located in the baseband spectral region. On the other hand, BER values for the baseband are slightly better than those for the passband at ROPs of >-5dBm, mainly because the adopted sub-band power ratio cannot fully compensate for the large absolute system frequency response roll-off experienced by the passband. In addition, Fig. 4(a) also shows that, in comparison with their corresponding optical BTB configurations, the optical power penalties at a forward error correction (FEC) BER limit of 2.3 × 10−3 [17] are 1.1dB and 0.5dB for the baseband and passband, respectively. The relatively large baseband power penalty is mainly contributed by its’ high signal transmission capacity.
Fig. 4 (a) Baseband and passband BER performances of 16.375Gb/s dual-band OOFDM signals for both the optical back-to-back and 200m OM2 MMF transmission link; (b) subcarrier BER distribution across all subcarriers for the baseband and passband OFDM signals after transmission through the 200m OM2 MMF link (ROP = −3dBm). BB: baseband; PB: passband; OBTB: optical back-to-back.
The corresponding subcarrier error distributions for both the baseband and passband after 200m OM2 MMF transmissions are plotted in Fig. 4(b), in obtaining which experimental measurements are undertaken at ROPs of −3dBm. In Fig. 4(b), it is clearly observed that, by making use of adaptive bit and power loading, both the averaged sub-band and individual subcarrier BER performances of each sub-band are below the adopted FEC limit. Compared to the passband case, the relatively high BER distribution fluctuation for the baseband is mainly due to the large residual frequency response roll-off effect shown in Fig. 2(a).
Before and after performing channel equalization in the receiver, the corresponding constellations of representative subcarriers are exemplified in Fig. 5 and Fig. 6 for the baseband and passband signals, respectively. The constellations are measured at ROPs of −3dBm after 200m OM2 MMF transmissions. As expected from the results presented in Fig. 2(a), in comparison with the passband, large variations in subcarrier amplitude levels prior to channel equalization for the baseband are clearly observed, which can, however, be effectively rectified by channel estimation and equalization.
Fig. 5 Representative received subcarrier constellations before/after channel equalization for the baseband signal.
Fig. 6 Representative received subcarrier constellations before/after channel equalization for the passband signal.
To explore the impact of inter-sub-band interference on the performance of 16.375Gb/s over DM-VCSEL-based 200m OM2 MMF transmissions, the BER performance of each sub-band is measured with the other sub-band switched off in the DSP but all corresponding components still connected and powered. Experimental measurement results in Fig. 7 show that the inter-sub-band interference induced power penalty to the passband transmission at the adopted FEC limits is approximately 0.7dB, whilst the corresponding power penalty to the baseband transmission is approximately 1.5dB. This indicates that, for the present system, the inter-sub-band interference has relatively strong impacts on the baseband OOFDM signal. Such observed power penalties are in close agreement with those reported in [15]. The physical reasons underpinning these behaviors are: a) the above-mentioned square-law photon detection-induced intermixing frequency products which are mainly located in the baseband region, and b) the utilization of a large RF power ratio between the passband and baseband.
Fig. 7 BER performance of the baseband (passband) with passband (baseband) being switched off (stayed on) after 16.375Gb/s over 200 OM2 MMF dual-band OOFDM transmissions.
3.3 Transmission performance robustness
From discussions in Subsection 3.2, it is clear that adaptive bit and subcarrier/sub-band power loading offers a simple and effective means of compensating for the system frequency response roll-off effect. Apart from the maximization of the aggregated transmission capacity, the adaptive loading technique also considerably enhances the transceiver flexibility and the performance robustness against variations in passband RF carrier frequency and transmission distance.
The excellent passband RF carrier frequency tunability of the DM-VCSEL-based dual-band OOFDM transceiver is shown in Fig. 8(a), where the sub-band signal transmission capacities are measured after 200m OM2 MMF transmissions. In obtaining Fig. 8(a), for a specific RF carrier frequency, extensive optimizations of all other parameters are undertaken until the maximum total signal transmission capacity is achieved under a condition that each sub-band BER is below the adopted FEC limit at ROPs of −4dBm. The optimized parameters include subcarrier bit/power, electrical sub-band power and sub-band signal power ratio between the passband and the baseband, as well as VCSEL bias/driving current. It is very interesting to note that, in Fig. 8(a), an almost flat baseband signal capacity is obtainable for the entire passband frequency variation range of 1GHz, over which the passband signal capacity, however, varies by approximately 24%. The passband signal transmission capacity reduction is mainly due to the limited VCSEL modulation bandwidth, the limited dynamic range of the applied sub-band power ratio and the wide signal spectrum-enhanced modal noise effect. It is also shown in Fig. 8(a) that the 1GHz passband frequency variation range corresponds to a relative aggregated signal transmission capacity variation of <10%.
Fig. 8 Measured robustness of the aggregated signal transmission capacity. (a) Aggregated signal transmission capacities versus passband frequency; (b) aggregated signal transmission capacities versus MMF transmission distance.
The measured robustness of aggregated signal transmission capacity against variations in OM2 MMF transmission distance is presented in Fig. 8(b), where, for a specific transmission distance, the system optimization procedure adopted in measuring Fig. 8(a) is considered, together with the passband carrier frequency optimization. It can be seen in Fig. 8(b) that, when the transmission distance increases from 100m to 500m, the maximum signal transmission capacities for both sub-bands are observed around the 200m-300m MMF length region, and that the aggregated signal transmission capacity only varies by <8%. Such behaviors are mainly attributed to the combined effects of differential mode delay (DMD) and modal noise [11]: for a long (short) MMF link, the DMD effect is increased (decreased) while the modal noise effect is reduced (increased). As a direct result, the maximum sub-band/aggregated signal transmission capacity occurs in the 200m-300m MMF transmission distance region.
4. Conclusions
Making use of standalone DM-VCSEL-based real-time adaptive OOFDM transmitters, aggregated 16.375Gb/s transmissions of dual-band OOFDM signals having bandwidths approximately 3.8 times higher than the VCSEL modulation bandwidths, have been experimentally demonstrated, for the first time, over 200m OM2 MMF IMDD links. Experimental results have also shown that the aggregated signal transmission capacities of the transmission links vary by just 8% for various OM2 MMFs ranging from 100m to 500m, and by just 10% over a 1GHz passband carrier frequency detuning range. The strong OOFDM adaptability mainly attributes to the aforementioned performance robustness and passband frequency tunability, which result in significantly relaxed requirements on VCSEL modulation bandwidth for achieving specific transmission performances for cost-sensitive application scenarios such as data centers.
Acknowledgments
This work was supported in part by the PIANO + under the European Commission’s ERA-NET Plus Scheme within the project OCEAN under Grant Agreement 620029, in part by Sino-UK Higher Education Research Partnership for PhD Studies, National High Technology Research and Development Program of China (863 Program) (2012AA011302, 2012AA011304, 2013AA010503), NSFC (No. 61071097, No. 61107060, No. 61101095, No. 61301156 and No. 61471087).
References and links
1. L. Paraschis, “Advancements in data-center networking and the importance of optical interconnections,” in European Conference and Exhibition on Optical Communication (ECOC), (Institution of Engineering and Technology, London, 2013), paper Th.2.F.3. [CrossRef]
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. R. E. Freund, C. A. Bunge, N. N. Ledentsov, D. Molin, and C. Caspar, “High-speed transmission in multimode fibers,” J. Lightwave Technol. 28(4), 569–586 (2010). [CrossRef]
4. A. Amphawan, “Holographic mode-selective launch for bandwidth enhancement in multimode fiber,” Opt. Express 19(10), 9056–9065 (2011). [CrossRef] [PubMed]
5. A. Okamoto, K. Aoki, Y. Wakayama, D. Soma, and T. Oda, “Multi-excitation of spatial modes using single spatial light modulator for mode division multiplexing,” in Optical Fiber Communication /National Fiber Optic Engineers Conference (OFC/NFOEC), (Optical Society of America, Los Angeles, California, 2012), paper JW2A.38.
6. C. P. Yu, J. H. Liou, Y. J. Chiu, and H. Taga, “Mode multiplexer for multimode transmission in multimode fibers,” Opt. Express 19(13), 12673–12678 (2011). [CrossRef] [PubMed]
7. C.-H. Hang, L. Chrostowski, and C. J. Chang-Hasnain, “Injection locking of VCSELs,” IEEE J. Sel. Top. Quantum Electron. 9(5), 1386–1393 (2003). [CrossRef]
8. L. Chrostowski, B. Faraji, W. Hofmann, M.-C. Amann, S. Wieczorek, and W. W. Chow, “40 GHz bandwidth and 64 GHz resonance frequency in injection-locked 1.55 μm VCSELs,” IEEE J. Sel. Top. Quantum Electron. 13(5), 1200–1208 (2007). [CrossRef]
9. C. Kachris and I. Tomkos, “Energy-efficient bandwidth allocation in optical OFDM-based data center networks,” in Optical Fiber Communication /National Fiber Optic Engineers Conference (OFC/NFOEC), (Optical Society of America, Los Angeles, California, 2012), paper JTh2A.34.
10. Y. Benlachtar, R. Bouziane, R. I. Killey, C. R. Berger, P. Milder, R. Koutsoyannis, J. C. Hoe, M. Püschel, and M. Glick, “ Optical OFDM for the data center,” in International Conference on Transparent Optical Networks (ICTON), (Munich, Germany, 2010), paper We.A4.3.
11. E. Hugues-Salas, X. Q. Jin, R. P. Giddings, Y. Hong, S. Mansoor, A. Villafranca, and J. M. Tang, “Directly modulated VCSEL-based real-time 11.25-Gb/s optical OFDM transmission over 2000-m legacy MMFs,” IEEE Photon. J. 4(1), 143–154 (2012). [CrossRef]
12. E. Hugues-Salas, N. Courjault, X. Q. Jin, R. P. Giddings, C. Aupetit-Berthelemot, and J. M. Tang, “Real-time 11.25Gb/s optical OFDM transmission over 2000m legacy MMFs utilizing directly modulated VCSELs,” in Optical Fiber Communication /National Fiber Optic Engineers Conference (OFC/NFOEC), (Optical Society of America, Los Angeles, California, 2012), paper OW3B.1.
13. H. B. Zhang, X. W. Yi, Q. W. Zhang, Y. Ling, M. L. Deng, E. Hugues-Salas, R. P. Giddings, Y. Hong, M. Wang, and J. M. Tang, “Robust real-time 15.125 Gb/s adaptive optical OFDM transmissions over 100 m OM2 MMFs utilizing directly modulated VCSELs subject to optical injection locking,” Opt. Express 22(1), 1163–1171 (2014). [CrossRef] [PubMed]
14. X. Jin, J. Wei, R. Giddings, T. Quinlan, S. Walker, and J. Tang, “Experimental demonstrations and extensive comparisons of end-to-end real-time optical OFDM transceivers with adaptive bit and/or power loading,” IEEE Photon. J. 3(3), 500–511 (2011). [CrossRef]
15. R. P. Giddings, E. Hugues-Salas, and J. M. Tang, “Experimental demonstration of record high 19.125 Gb/s real-time end-to-end dual-band optical OFDM transmission over 25 km SMF in a simple EML-based IMDD system,” Opt. Express 20(18), 20666–20679 (2012). [CrossRef] [PubMed]
16. Q. W. Zhang, E. Hugues-Salas, Y. Ling, H. B. Zhang, R. P. Giddings, J. J. Zhang, M. Wang, and J. M. Tang, “Record-high and robust 17.125 Gb/s gross-rate over 25 km SSMF transmissions of real-time dual-band optical OFDM signals directly modulated by 1 GHz RSOAs,” Opt. Express 22(6), 6339–6348 (2014). [CrossRef] [PubMed]
17. M. L. Deng, Y. Ling, X. F. Chen, R. P. Giddings, Y. H. Hong, X. W. Yi, K. Qiu, and J. M. Tang, “Self-seeding-based 10Gb/s over 25km optical OFDM transmissions utilizing face-to-face dual-RSOAs at gain saturation,” Opt. Express 22(10), 11954–11965 (2014). [CrossRef] [PubMed]
18. E. Hugues-Salas, R. P. Giddings, X. Q. Jin, Y. Hong, T. Quinlan, S. Walker, and J. M. Tang, “REAM intensity modulator-enabled 10Gb/s colorless upstream transmission of real-time optical OFDM signals in a single-fiber-based bidirectional PON architecture,” Opt. Express 20(19), 21089–21100 (2012). [CrossRef] [PubMed]
