We propose signal-carrier interleaved (SCI) optical OFDM for direct detected transmission systems. Such a scheme can be considered as a variation of self-coherent detection where the carrier and signal are supplied at the transmitter and extracted at the receiver for coherent-like detection. This provides high OSNR sensitivity while maintaining very low carrier-to-signal power ratio (CSR). Our experiment results show that with 0 dB CSR, 43.2 Gb/s 16 QAM OFDM signal can be successfully delivered over 80 km standard single mode fiber (SSMF) with 24 dB OSNR requirements at 7% FEC limit.
©2013 Optical Society of America
Direct-modulation (DM) and direct detection (DD) based optical communication systems have continued to attract attentions because of its low cost and simple implementation. Despite the fact that coherent detection has achieved dramatic success in the past few years due to adoption of powerful electronic digital signal processing [1–4], the optical gears for coherent detection are rather sophisticated, requiring polarization multiplexing/de-multiplexing, double IQ modulators and balanced receivers. Subsequently, for cost-sensitive short-reach high-speed communication, such as inter-cabinet communication, DM/DD is a more desirable option and has been actively explored [5–15]. One of the common DD schemes studied in the last a few years is the offset single sideband (SSB) . By leaving a gap between signal and the main carrier, the signal can be free from signal-to-signal beat noise (SSBN). Another technique to avoid interference from SSBN is to utilize subcarrier interleaved configuration where only odd number subcarriers are filled . Furthermore, by adopting SSB modulation, the signals are immune to dispersion induced fading. Although the performance of offset SSB or subcarrier interleaved direct detected signals is relatively superior, such configuration sacrifices electrical bandwidth and thus has relatively low electrical spectral efficiency (SE). To improve the SE of DD systems, Peng et al. have proposed a virtual SSB DD (VSSB-DD) scheme in which advanced digital signal processing (DSP) is employed to cancel SSBN . In this way, the guard band between the main carrier and the signal is not required, and electrical SE is doubled. Recently, block-wise phase switching (BPS) DD is proposed to improve electrical SE by using double sideband (DSB) modulation . However, there are two drawbacks for either VSSB-DD or BPS-DD: (i) in order to remove SSBN, high carrier-to-signal power ratio (CSR) is required, and (ii) complicated DSP is involved for iterative SSBN cancellation.
In this paper, we propose signal-carrier interleaved optical OFDM (SCI-OOFDM) for direct detected optical communication systems. The SCI-OOFDM signals assign the main carrier and the signals to different time slots, for instance, to different OFDM symbols. At the receiver, optical hybrid and differential detection are utilized to recover the double sideband signals. In this way, as long as balanced receiver provides high enough common mode rejection ratio (CMRR, defined as the ratio between the differential-mode gain and the common-mode gain inside a balanced receiver), SSBN cancellation in DSP is not needed.
SCI-OOFDM possesses the following advantages: (i) it gains significant sensitivity improvement over BPS-OFDM or VSSB-OFDM, (ii) the DSP procedure for SCI-OOFDM can be as simple as offset OFDM, (iii) the system does not require any filtering at the receiver side to pick up the main carrier, thus the frequency tracking of the main carrier is not needed as well. Furthermore, since SCI-OOFDM signals can be DSB modulated, the SE of such system is twice as that of offset SSB OFDM. Note that the SCI-OOFDM approach we proposed is essentially one variation of self-coherent scheme , but with much longer optical delay at receiver, our signals are robust against chromatic dispersion. Furthermore, comparing with conventional self-coherent detection, in SCI direct detection (SCI-DD) scheme the received electrical signal bandwidth will not be doubled as the CW main carrier (rather than signal itself) is mixing with the signals.
2. Principle of signal-carrier interleaved direct detection
The idea of SCI-DD is to interleave signals with the main carrier on block basis. The ‘block’ here refers to one (or a few) OFDM symbol(s). In single-carrier modulation format, the ‘block’ can be a certain length of signal waveform. At the receiver, the signal is split into two paths, with one of the paths delayed by one block against the other. These two paths are then fed into a coherent receiver consisting of an optical hybrid and two balanced receivers to recover the DSB complex signal.
The structure of SCI signals and structure of SCI-DD receiver are shown Fig. 1. The delayed path at receiver is also depicted for comparison. There can be a few different variations when complementing the SCI-DD. Figure 1(a) illustrates the most straightforward implementation of SCI-DD, where every signal block is followed by a main carrier block . Therefore, in Fig. 1(a), the same signal either in the pink shaded block (shaded area with dashed outer line) or the blue shaded block (shaded area with solid outer line) can be recovered. This means the overall SE is reduced by a factor of 2, since half of the waveform is allocated for the main carrier in the time domain. To achieve better SE, an optimized scheme is illustrated in Fig. 1(b), where every two signal blocks are followed by one main carrier block. Hence, at the receiver, as shown in the figure, 2/3 of the waveform (pink shaded areas) can be recovered. This version of ‘2/3 SE’ is the optimal SE we can achieve for our proposed SCI-DD scheme. This is because using either shorter signal blocks to interleave with carrier block would reduce the overall SE, and using longer blocks would result in increased computational complexity, the very problem we try to avoid in this report.
After the original received signal and the delayed signal pass through an optical hybrid, the output of the hybrid are shown in Fig. 1(c). Since the delay time is designed to be exactly one block, in each given time frame, there is a carrier beating with the signal. Namely, from the first pair of the hybrid outputs, we can recover the real part of the signal, and from the second pair we recover the imaginary part. The photocurrent and at the output of two balanced receivers would respectively be the following
Since the signals are recovered without SSBN compensation, the DSP of SCI-OOFDM can be as computationally efficient as offset OFDM. The procedure of the receiver DSP is the same of single-polarization coherent optical OFDM : (i) IQ imbalance compensation, (ii) FFT window synchronization for identifying the start of the OFDM symbol, (iii) CP removal and Fast Fourier transform (iv) phase noise compensation, (v) channel estimation in terms of Jones Matrix , and (vi) constellation construction for each carrier and bit error rate (BER) computation. Note that the above DSP procedure are applicable to both the ‘1/2 SE’ version shown in Fig. 1 (a) and the ‘2/3 SE’ version shown in Fig. 1 (b), and as such the DSP complexity for different variations with different SE is similar.
The SCI-DD scheme is essentially a variation of self-coherent detection where the carrier and signal are supplied at the transmitter, extracted at the receiver for coherent-like detection [16,18]. Comparing with previously proposed digital self-coherent detection (DSCD) scheme , the advantage of our SCI-DD approach is twofold. First, since long optical delay (10 ns or longer) is utilized for SCI-DD, the signals is resilient to fiber dispersion. Second, the mixing products generated at receiver are between the main carrier and the signals, signifying that the signal bandwidth after DD will not be doubled as for conventional DSCD. Meanwhile, comparing with the other self-coherent approach in , SCI-DD does not require a sharp optical filter to extract the main carrier and thus does not need to acquire and active tracking the main carrier at the receiver.
Comparing with conventional DD systems using one single photodiode, our SCI-DD system requires extra components such as an optical delay line, an optical hybrid, and a pair of balanced receivers. Hence we would like to touch upon some unique implementation requirements related to SCI-DD. First, we consider the optical delay line the most critical part of receiver that affects the performance of the SCI-DD system. The main feature about the delay line is the polarization insensitivity. Namely, the delay line should be able to maintain the polarization state so that the polarization of original signal and delayed signal can always be aligned. This is true since the delay line is typically a few meters (for instance, 3 meters in our demonstration), and the change of polarization state along such a short fiber can be ignored. SCI-DD uses almost half of the coherent detection including one optical hybrid and two balanced receiver. It is shown that 90° optical hybrid can be achieved by judicious configuring two optical couplers , and therefore the cost is moderate. For the pair of balanced receivers, since the two photodiodes or even the two pairs of the balanced receivers can be packed into one single module, with mass production, we expect the cost of balanced receivers would not be much higher than a single photodiode.
3. Experimental setup
In this paper, we demonstrate SCI-DD with OFDM (multicarrier) modulation, although the SCI scheme can be applied to either single-carrier or multicarrier signals. The experimental setup is illustrated in Fig. 2. As shown in the figure, an external cavity laser (ECL) is split into two branches, one fed into a multiband OFDM generator, the other to an intensity modulator for supplying the main carrier. The OFDM generator consists of an intensity modulator for tone generations, and an optical IQ modulator driven by an arbitrary waveform generator (AWG) at 10 GSa/s. Three optical tones are spaced at 8.75 GHz, which is multiple of subcarrier spacing and therefore satisfies orthogonal-band multiplexing (OBM) condition . The FFT size is 128 points. The cyclic prefix (CP) is 22 points (2.2 ns). The middle 108 subcarriers out of 128 are filled for each OFDM band. 40 training symbols are attached to the beginning of the OFDM frame. 16-QAM modulation is employed, thus the raw data rate of our demonstrated system is 43.2 Gb/s for 3-band OFDM signal (with total optical bandwidth of 26.25 GHz). The main carrier is switched on/off aligned with OFDM frames by driving the intensity modulator with (1, 0, −1, 0) supplied by an AWG on corresponding OFDM frames. Note that the carrier on/off window is synchronized with the OFDM symbol, and therefore the bandwidth requirement for such intensity modulator is very low. For instance, in this demonstration the carrier is switched at 33 MHz. The fiber lengths of the two paths (carrier path and signal path) are carefully matched. It is worth noting that we split the main carrier and signal into two paths because of the bandwidth limitation of our digital-to-analog converters (DACs). The 3-dB bandwidth of the DACs we use is 3 GHz. In order to achieve a data rate of 40 Gb/s, we have to do optical band multiplexing. In practice, both carrier and signal will be provided by a single DAC, and the additional path for carrier is thus not needed. It is also worth noting that because 2 AWGs are used, as mentioned, our CP length relatively long to combat time jitter of two AWGs. In practice, the length of CP can be greatly reduced because only one AWG is needed. After the two paths are combined, the signals are amplified and launched into an 80-km standard single-mode fiber (SSMF). At the receiving end, the signal is re-amplified and filtered with a 100-GHz WDM filter. The signals are then split into two paths, with one branch delayed by one OFDM symbol (15 ns) with the other, and then both fed into an optical hybrid for ‘self-coherent’ detection. The signals from optical hybrid are then detected by balanced receivers and sampled by a real time oscilloscope at 50 GSa/s.
To eliminate interference from any residual SSBN in balanced receivers, only odd subcarriers of training symbols are filled for channel estimation . The training symbols in practice will be only used at acquisition stage and the subsequent channel estimation can be recovered via the data, and therefore training symbols are not considered in overhead computation for this demonstration. 2.16x106 bits of data are collected for BER computation.
4. Experimental results and discussion
Assuming the total power fed into receiver is fixed, it is easy to show that when signals are mixed with the main carrier, the maximum mixing product is achieved when the main carrier has the same power as signals. Namely, the optimal CSR for DD is expected to be 0 dB. To verify whether this is the case in our experimental setup, we measure BER as a function of CSR at back-to-back condition. The measurements are shown in Fig. 3(a). As expected, the optimal CSR of 0 dB is observed for the 16-QAM signal. We also note that in the regime of low CSR (for instance, −8 dB), the signal quality quickly degrades. This is mainly because the CMRR of the balanced receivers is finite (typically 25 dB), leading to residual SSBN. If the CSR is too low, the power of SSBN products would be significant compare with signal itself, and thus the performance is degraded. We then identify the optimal launch power for the 80-km SSMF transmission. We measure BER as a function of launch power at CSR of 0 dB. From the results in Fig. 3(b) we find that 0 dBm is the optimal launch power for the 43.2 Gb/s signal. When launch power is lower than 0 dBm, the transmission performance is limited by amplified spontaneous emission (ASE) noise. While signal launch power goes beyond 0 dBm, signals would suffer from fiber nonlinear noise.
At last, we measure OSNR sensitivity of the system at the back to back and after 80-km transmission. For all following measurements, the CSR is set to 0 dB, and launch power is 0 dBm. To obtain a reference for evaluating our proposed SCI-DD system, we first measure OSNR sensitivity using coherent detection. By coherent detection, we mean the condition that the main carrier is turned off at transmitter side, and at the receiver, instead of delaying the signal and applying differential detection, we have a local oscillator for coherent detection. As the green diamond dotted curve depicted in Fig. 4(a), the 14.4 Gb/s (single band) coherently detected OFDM (CO-OFDM) 16-QAM signal requires 11 dB and 14 dB OSNR for 20% FEC (BER = 1.9x10−2) and 7% FEC (BER = 3.8x10−3), respectively. We then measure the 14.4 Gb/s SCI-OOFDM signals. It is found that the single band 16-QAM 14.4 Gb/s SCI-OOFDM requires 15 dB and 18 dB OSNR for 20% FEC and 7% FEC. Therefore, the OSNR sensitivity of SCI-OOFDM is about 4 dB worse than CO-OFDM. The reason for the 4-dB difference is as follows: while in coherent detection systems the local oscillator is free from ASE noise, in this SCI-DD system the main carrier, propagating through the transmission fiber, has the same signal-to-noise ratio as signal itself, and thus ideally the OSNR sensitivity should be sacrificed by 3 dB. Additional 1 dB penalty is attributed to be some residual SSBN in SCI-DD system. For reference, we also perform numerical simulation of back-to-back transmission. In the simulation, the CMRR of balanced receivers are set to infinity. The simulation results are shown in Fig. 4 (dashed curves). We can see that for 14.4 Gb/s SCI-OOFDM signals, our experimental results agree well with the simulation results at FEC threshold of 20%, and have 1 dB difference at 7% FEC threshold.
We then measure 3-band signal (with raw data rate of 43.2 Gb/s, or 40.4 Gb/s if excluding 7% FEC). From the results shown in Fig. 4(b), it can be seen that the 43.2 Gb/s signal requires 20 dB OSNR for 20% FEC, and 24 dB OSNR for 7% FEC. In comparison with the numerical simulation, at 20% FEC limit, our experimental results are 2 dB away from simulation, and 3.5 dB away from simulation at 7% FEC threshold. Comparing with the BPS-DD scheme that proposed earlier which is also DSB modulated, we find the SCI-DD has much better OSNR sensitivity. For instance, to achieve 49.4-Gb/s date rate using 8-QAM OFDM modulation, the BPS-DD system requires 32-dB OSNR at a BER of 1.9x10−2 . This is 12 dB higher than a 43.20-Gb/s SCI-DD system with 16-QAM OFDM modulation. Note that the balanced receiver we are using is off-the-shelf designed for 10-Gb/s data rate, and the CMRR can be higher than 20 dB only within about 5 GHz bandwidth. For the 43.2 Gb/s signal, the electrical bandwidth is about 13 GHz, which is much wider than 5 GHz. There could be some residual SSBN at frequency range beyond 5 GHz when we performed SSBN cancellation for the 3-band signal . Since majority of SSBN is removed by balanced receivers, the cancellation needs only one iteration. We find that though SSBN cancellation does reduce the noise floor of the 16-QAM 43.2 Gb/s signal, the difference between the two cases are very small as long as OSNR becomes lower than 26 dB. This indicates that balanced receivers can effectively reject majority of SSBN.
We have proposed signal-carrier interleaved (SCI) optical OFDM for direct detected optical communication. With such a scheme, we have successfully delivered 43.2-Gb/s OFDM signals over 80-km SSMF without polarization multiplexing and with high electrical SE. Our experimental results show that the 43.2-Gb/s 16-QAM OFDM signal requires 24-dB OSNR when 7% FEC is employed.
References and links
1. S. J. Savory, G. Gavioli, R. I. Killey, and P. Bayvel, “Electronic compensation of chromatic dispersion using a digital coherent receiver,” Opt. Express 15(5), 2120–2126 (2007). [CrossRef] [PubMed]
2. C. R. S. Fludger, T. Duthel, D. van den Borne, C. Schulien, E.-D. Schmidt, T. Wuth, J. Geyer, E. de Man, G.-D. Khoe, and H. de Waardt, “Coherent equalization and POLMUX-RZ-DQPSK for robust 100-GE transmission,” J. Lightwave Technol. 26(1), 64–72 (2008). [CrossRef]
3. W. Shieh and C. Athaudage, “Coherent optical orthogonal frequency division multiplexing,” Electron. Lett. 42(10), 587–589 (2006). [CrossRef]
4. X. Liu, F. Buchali, and R. W. Tkach, “Improving the nonlinear tolerance of polarization-division-multiplexed CO-OFDM in long-haul fiber transmission,” J. Lightwave Technol. 27(16), 3632–3640 (2009). [CrossRef]
5. S. C. J. Lee, F. Breyer, S. Randel, M. Schuster, J. Zeng, F. Huiskens, H. P. A. van den Boom, A. M. J. Koonen, and N. Hanik, “24-Gb/s transmission over 730 m of multimode ﬁber by direct modulation of 850-nm VCSEL using discrete multi-tone modulation,” in Optical Fiber Communication Conference/National Fiber Optic Engineers Conference (OFC/NFOEC), 2007, PDP6.
6. J. L. Wei, D. G. Cunningham, R. V. Penty, and I. H. White, “Feasibility of 100G ethernet enabled by carrierless amplitude/phase modulation and optical OFDM,” in European Conference and Exhibition on Optical Communication (ECOC), 2012, P6.05.
7. D. Qian, N. Cvijetic, J. Hu, and T. Wang, “108 Gb/s OFDMA-PON with polarization multiplexing and direct detection,” J. Lightwave Technol. 28(4), 484–493 (2010). [CrossRef]
8. B. Schmidt, A. J. Lowery, and J. Armstrong, “Experimental demonstrations of electronic dispersion compensation for long-haul transmission using direct-detection optical OFDM,” J. Lightwave Technol. 26(1), 196–203 (2008). [CrossRef]
9. M. Schuster, S. Randel, C. A. Bunge, S. C. J. Lee, F. Breyer, B. Spinnler, and K. Petermann, “Spectrally efficient compatible single-sideband modulation for OFDM transmission with direct detection,” IEEE Photonics Technol. Lett. 20(9), 670–672 (2008). [CrossRef]
10. N. Cvijetic, M. Cvijetic, M. Huang, E. Ip, Y. Huang, and T. Wang, “Terabit optical access networks based on WDM-OFDMA-PON,” J. Lightwave Technol. 30(4), 493–503 (2012). [CrossRef]
11. R. P. Giddings, X. Q. Jin, E. Hugues-Salas, E. Giacoumidis, J. L. Wei, and J. M. Tang, “Experimental demonstration of a record high 11.25Gb/s real-time optical OFDM transceiver supporting 25km SMF end-to-end transmission in simple IMDD systems,” Opt. Express 18(6), 5541–5555 (2010). [CrossRef] [PubMed]
12. L. A. Neto, G. B. de Farias, N. Genay, S. Menezo, B. Charbonnier, P. Chanclou, and C. Aupetit-Berthelemot, “On the limitations of IM/DD WDM-FDMA-OFDM PON with single photodiode for upstream transmission,” in Optical Fiber Communication Conference/National Fiber Optic Engineers Conference (OFC/NFOEC),2013, OW4B.1.
13. W. R. Peng, X. Wu, V. Arbab, B. Shamee, L. Christen, J. Yang, K. Feng, A. Willner, and S. Chi, “Experimental demonstration of a coherently modulated and directly detected optical OFDM system using an RF-tone insertion,” in Optical Fiber Communication Conference/National Fiber Optic Engineers Conference (OFC/NFOEC), 2008, paper OMU2. [CrossRef]
14. W. R. Peng, B. Zhang, K. M. Feng, X. X. Wu, A. E. Willner, and S. Chi, “Spectrally efficient direct-detected OFDM transmission incorporating a tunable frequency gap and an iterative detection techniques,” J. Lightwave Technol. 27(24), 5723–5735 (2009). [CrossRef]
15. X. Chen, A. Li, D. Che, Q. Hu, Y. Wang, J. He, and W. Shieh, “High-speed fading-free direct detection for double-sideband OFDM signal via block-wise phase switching,” in Optical Fiber Communication Conference/National Fiber Optic Engineers Conference (OFC/NFOEC), 2013, paper PDP5B.7.
17. Q. Yang, Y. Tang, Y. R. Ma, and W. Shieh, “Experimental demonstration and numerical simulation of 107-Gb/s high spectral efficiency coherent optical OFDM,” J. Lightwave Technol. 27(3), 168–176 (2009). [CrossRef]
18. L. Xu, J. Hu, D. Qian, and T. Wang, “Coherent optical OFDM Systems using self optical carrier extraction,” in Optical Fiber Communication Conference/National Fiber Optic Engineers Conference (OFC/NFOEC), 2008, paper OMU4. [CrossRef]