115.2 Gb/s Optical OFDM Transmission with 4 bit/s/Hz Spectral Efficiency Using IEEE 802.11a OFDM PHY Lenin Mehedy, Masuduzzaman Bakaul and Ampalavanapillai Nirmalathas NICTA Victoria Research Laboratory, Dept. of Electrical and Electronic Engineering, University of Melbourne, VIC 3010, Australia. Phone: +61430225424, E-mail:
[email protected] Abstract - We investigate the possibility of spectrally efficient 100 Gb/s transmission using IEEE 802.11a OFDM PHY based coherent optical OFDM and confirm that such systems may operate at 115.2 Gb/s with 26.5 GHz signal bandwidth. Introduction Optical Orthogonal Frequency Division Multiplexing (O-OFDM) has been demonstrated to be a promising technology for the delivery of ultra high speed data over longer distances without significant effects from fiber impairments [1, 2]. Recent coherent O-OFDM demonstrations have confirmed its effectiveness over hundreds of kilometers of fiber at an overall data-rate of 100 Gb/s with an approximate spectral efficiency of 2 bit/s/Hz [2-4]. Other variants of O-OFDM such as direct detection O-OFDM, adaptively modulated O-OFDM, and hybrid of coherent and incoherent approaches are also found to be equally capable of demonstrating similar performances [5-8]. However, most of these OOFDM approaches use different configurations in modulation formats, baud-rate/subcarrier, FFT sizes, number of pilot subcarriers and training symbols to achieve the desired goals, and contain minimum connections among them. This ‘choose your own’ approach makes the whole process complicated to deploy O-OFDM technology practically. This paper upholds the existing IEEE 802.11a OFDM PHY [9] standard as a benchmark to generate O-OFDM signals and measures the performance boundaries. The investigation found the potentials of the existing wireless standard to use in optical domain that may lead the development of a modified standard for O-OFDM technologies in the near future. Simulation Setup Fig. 1 shows a coherent optical OFDM (CO-OFDM) system modelled in VPItransmissionMakerTM platform [10]. The baseband OFDM signals are generated using the IEEE 802.11a OFDM PHY specifications: a Fast Fourier Transform (FFT) size of 64 with the center frequency subcarrier filled with zero, 48 data subcarriers, 4 binary phase shift keyed (BPSK) pilot subcarriers, 2 BPSK modulated training sequences and a cyclic prefix (CP) of 16 samples. A pseudo random bit sequence length of 215-1 is used to generate the data. The generated complex baseband OFDM signal is then converted to in-phase (I) and quadrature (Q) components to drive two Mach-Zehnder modulators (MZM) biased at the minimum transmission point and with MZM
Fig. 1. Simulation Setup
insertion loss of 4 dB. A variable line-width light source (LD1) with average output power of 3 mW is used as the optical source. The composite optical signal was then transported over a block of single mode fiber (SSMF) with dispersion of 16 ps/nm/km, nonlinear index of 2.6 ×10−20 m2/W, PMD of 0.1 ps/√km, the length of which can be varied when required. An optical gain block with noise figure of 6 dB was used to maintain the launched optical power at -8 dBm. At the receiving end, the I and Q components of the signal are recovered by down conversion using another laser diode (LD2) with characteristic parameters same as LD1. During demodulation, training sequences are used to compensate for the effect of chromatic dispersion induced phase noise and pilot carriers are used to compensate for laser phase noise. Different modulation schemes are adopted in IEEE 802.11a OFDM PHY for various functionalities. For example, BPSK modulation is used for preamble (training sequences) and pilot subcarriers whereas quadrature amplitude modulation (QAM) is used for data subcarriers. As different modulation schemes are used across subcarriers in IEEE 802.11a OFDM PHY, it is necessary to ensure a constant average power across all subcarriers and normalization is therefore performed [9]. Normalized data symbols ( Cnorm ) are generated by Cnorm = Ci × K mod ,
normalization K mod =
factor
modulation K mod
is
dependent
calculated
by
, M is the highest number of constellation
M M
∑C n =1
where
2
n
points, |Cn| is the amplitude of the nth ideal constellation symbol in the complex plain (i.e. Cn = In + jQn) and Ci is the ith data symbol in complex plain [9, 11]. The value of Kmod, for example, for BPSK, 64-QAM, and 128-QAM are 1, 1/√42, and 1/√82 respectively that ensures average constellation power of 1 (one) after normalization. Received data symbols are de-normalized before demodulation at the receiver.
Simulation Results and Discussion Error vector magnitude (EVM) [9, 11] and baud rate (or symbol rate) of subcarriers are used as the performance measures in this paper. For simplicity the system is first characterized in optical back-to-back (B2B) configuration without any fiber in the link and the respective results are shown in Fig. 2 and 3. Shown in Fig. 2, since EVM is calculated with respect to normalized ideal and normalized received symbols, different modulation schemes follow the same curve, which confirms that the normalized EVMs of O-OFDM signals are independent of the order of modulation. However for a certain EVM, bit error ratio (BER) varies with modulation orders as expected. Fig. 3 shows such a relationship between BER and EVM in B2B configuration with a laser linewidth (LLW) of 0 Hz. At a
Fig. 2. EVM follows the same curve independent of modulation order and EVM performance improves with lower LLW as well as for higher baud rates in B2B configuration. Table 1: EVM Thresholds for BER 10-3
Fig. 3. EVM thresholds of different modulation order for different BER
Modulation
EVM (dB)
BPSK 4-QAM 8-QAM 16-QAM 32-QAM 64-QAM 128-QAM
-5 -10 -15 -17 -20 -23 -26
BER 10-3, EVM thresholds for different modulation schemes are shown in Table 1. Fig. 2 also confirms that the EVM performance of the system improves with the increase of baud rates and the decrease of laser-linewidth which can be attributed to the effects of reduced intercarrier-interference (ICI) caused by the loss of orthogonality of the sub carriers due to laser frequency offset at the receiver. Now, to see the overall impact of fiber impairments, the signal is transmitted over a variable length of fiber from 0 to 80 km with varying baud-rate/subcarrier from 20 to 400 MBd and the results are shown in Fig. 4. These plots are generated using 64-QAM and 50 kHz LLW, which will be similar for other orders of QAM with the same LLW as explained in Fig. 2. Fig. 4 shows that the performance of the system remains steady over 80 km
Fig. 4. EVM vs. fiber length with different baud-rate/carrier Table 2: Design parameters for short reach optical link Distance (km)
QAM (M)
Baud Rate (MBd /subcarrier)
Signal BW (GHz)
Bit Rate (Gb/s)
80
64
100
6.625
18.8
40
64
200
13.25
57.6
10
64
400
26.50
115.2
SSMF for baud rates 20 to 50 MBd. As baud rate increases beyond 50 MBd, which is indispensable to achieve a 100 Gb/s O-OFDM system using IEEE 802.11a OFDM PHY, EVM performance degrades slowly with the increase of fiber length. This is because, at higher baud rates the effects of fiber impairments dominate over the effects of ICI. Table 2 shows the different achievable bit rates with various baud rates and signal bandwidth over different lengths of fiber. These are calculated by comparing the plots in Fig. 4 with the EVM thresholds shown in Table 1. It shows that by using IEEE 802.11a OFDM PHY, a 115.2 Gb/s (48 data subcarriers × 400 MBd × 6 bit/symbol) O-OFDM system can be designed to operate over 10 km SSMF with an opto-electronic bandwidth of 26.50 GHz (53 subcarriers including data, pilot and zero filled center subcarrier × 400 MBd × (1+ 25% CP overhead)). This reach, however, can be extended to 40 km by applying suitable dispersion compensation as shown in Fig. 4. Conclusion Optical OFDM signal is generated based on IEEE 802.11a OFDM PHY and its performance boundaries have been investigated. The results confirm that IEEE 802.11a OFDM PHY can be used as a benchmark to design a 100 Gb/s O-OFDM system for last mile optical network with a spectral efficiency of 4 bit/s/Hz. References 1. W. Shieh et al., Opt. Express 16, 841-859 (2008). 2. S. L. Jansen et al., OFC 2008. Paper: PDP2 3. W. Shieh et al., Opt. Express, vol. 16, pp. 6378-6386, 2008. 4. Y. Tang et al., Electron. Lett., vol. 44, pp. 588-589, 2008. 5. B. J. C. Schmidt et al. J. of Light. Tech., vol. 26, pp. 196-203, 2008. 6. J. M. Tang et al., J. of Light. Tech., vol. 25, no 3, pp.787-798, 2007. 7. P. Wei-Ren et al., OFC 2008. Paper: OMU2 8. I. Djordjevic et al., IEEE Phot. Tech. Lett., 18, 2006, pp. 1576-1578 9. IEEE Standard 802.11™-2007, Part 11, Clause 17 10. VPItransmissionMakerTM V 7.6 (www.VPIsystems.com) 11. M. D. McKinley et al., ARFTG Conf., pp. 45-52, 2004.