Th4C.5.pdf
OFC 2016 © OSA 2016
Trellis Coded Modulation Transmission over 40km 6-LP Mode Fiber D. Yu1,2, J.J.A van Weerdenburg2, E. Silekens2, R.G.H. van Uden2, M. Tang1, D. Liu1, A.M. Velazquez-Benitez3, P. Sillard4, D.Molin4, M. Bigot-Astruc4 R. Amezcua-Correa3, H. de Waardt2, A.M.J. Koonen2, S. Fu*1, and C.M. Okonkwo2 1
National Engineering Laboratory for Next Generation Internet Access System, School of Optics and Electronic Information, Huazhong University of Science and Technology, Wuhan, China, 430074 2 COBRA Research Institute, Eindhoven University of Technology, Eindhoven, the Netherlands, 5612 AZ 3 CREOL, the College of Optics and Photonics, University of Central Florida, Orlando 32815, USA 4 Prysmian Group, 644 Boulevard Est. Billy Berclau, 62092 Haisnes Cedex, France Author e-mail address:
[email protected]
Abstract: We demonstrate transmission of trellis-coded-modulated-12-spatial-and-polarization modes over 40km 6-LP-mode fiber with low differential mode group delay. By employing 8-PSK trellis coded modulation, an OSNR gain of 1.8dB is observed in comparison to QPSK. OCIS codes: (060.4510) Optical communications; (060.1660) Coherent communications; (060.4230) Multiplexing; (060.2330) Fiber optics communications.
1. Introduction With the fundamental limit of single mode fiber (SMF) transmission systems being approached, several orders of magnitude capacity increase has been demonstrated using novel fibers employing space division multiplexing [1-3]. N-times capacity increase can be realized if N spatial modes or N cores are utilized as independent channels for transmission within a fiber which supports no less than N spatial channels. Recently, 10GBaud quadrature phase shift keying (QPSK) signal of 20 spatial and polarization modes over 4.45km 6-linearly polarized (LP) mode fiber [4] and 30GBaud QPSK transmission of 30 spatial and polarization modes over 22.8km 9-LP mode fiber [5] have been demonstrated. The transmission distance is constrained when fully exploiting the supported spatial and polarization modes of the fiber. By reducing the number of multiplexed mode channels, potential better results may be obtained due to the improved spatial diversity. In addition, advanced coding schemes have been employed to improve the performance of few-mode fiber (FMF) transmission systems by exploiting the spatial dimension [6]. In [7, 8], trellis coded modulation (TCM) was demonstrated to provide a minimum signal coding optical signal to noise ratio (OSNR) gain of 0.4dB whilst maintaining spectral efficiency in a SMF fiber loop transmission systems. Therefore, similar coding gains in terms of OSNR can be exploited in mode division multiplexing (MDM) systems by employing FMFs. Moreover, it is more meaningful to guarantee that the effective spectral efficiency is maintained when employing such advanced coding schemes. In this work, we experimentally demonstrate 12-spatial-and-polarization modes multiplexed transmission over 22.5km and 40km of 6-LP mode fiber without inline multi-mode amplification. In particular, we employ TCM based 8-phase shift keying (TCM-8-PSK, 3 bits/symbol) at 20Gbaud, employing 1-bit for parity, thereby allowing 2 bits for data. In comparison with 20Gbaud quadrature phase shift keying (QPSK, 2 bits/symbol), we observe an OSNR gain of 1.8dB after 40km 6-LP mode fiber transmission. 2. TCM encoded 12-Spatial-and-Polarization Modes Transmission System As shown in Fig. 1 (a), the TCM encoder consists of three delay elements (boxes marked with “T”) and XOR operator (⊕). The state of the encoder, i.e., the state of the delay elements, is determined by the systematic bit sequence and produces a parity bit, generating 1-bit penalty. The 2 systematic bits are both inputs and outputs of the convolutional encoder. Finally, 3 bits are directly mapped to an 8-PSK constellation. In the decoding process, determining the most probable input bits requires the decoder to know all probabilities of the constellation, i.e. for every symbol, for every constellation point, the squared Euclidean distance between the received and the most probable transmitted symbol must be known. Thus, the soft-input soft-output Bahl Cocke Jelinek Raviv (BCJR) decoder is employed. The output of the BCJR decoder is used to make hard decisions on bits and calculate the BER. In this work, we employ 8-PSK using a parity bit dictated by a convolutional encoder and compare it to conventional QPSK transmission over the FMF transmission system, under the condition of the same spectral efficiency. Accordingly, the data throughput for both formats should be the same, 2 bits/symbol. The experimental setup is shown in Fig. 1 (b). The output of an external cavity laser (ECL) has a central wavelength of 1550.12nm, whose linewidth is less than 100kHz. It is divided into two beams by a 3dB optical coupler. One is guided into the modulator as the signal light, and the other one is used as local oscillator (LO) to
Th4C.5.pdf
OFC 2016 © OSA 2016
T
Parity bit
ECL
TCM real
TCM Encoder
Noise Loading 485ns
VOA
1
BPF
(0.8nm)
PBC
8-PSK Mapper
Systematic bit 1
Modulator
τ1 τ2 τ3
TCM TCM imag Encoder
ASE
OSA
τ4 τ5
Mode_6 X-Pol. Mode_6 Y-Pol.
(a)
BER Demap CPE BER Demap CPE
CD FE Comp. Comp.
TDM-SDM Receiver
T
CD FE Comp. Comp.
Digital Re-alignment
BER Demap CPE BER Demap CPE
Resampling to 40 Gs/s
Mode_1 Y-Pol.
12×12 MIMO TDE
Mode_1 X-Pol.
3 4 5 6
(c)
(d)
(e) 6LP-Mode Fiber
LO
Digital Domain T
2
Photonic Lantern Mode DeMUX
Transmitter
TCM Encoder Systematic bit 2
Photonic Lantern Mode MUX
minimize the complexity of the setup. The signal light is modulated by a 20GBaud fully uncorrelated pseudo random binary sequence (PRBS) of length 215. By transmitting over a sufficient distance larger than the coherence length of the ECL, the LO and transmitter laser can be considered uncorrelated. In addition, polarization decorrelation of 485ns is performed. An amplified spontaneous emission (ASE) source is used to set the OSNR of generated signal at the noise loading stage. Further decorrelation is performed on all the 6 spatial modes before launching into the 6-mode PL based mode MUX thus creating the 12 fully decorrelated spatial and polarization channels.
Mode decorrelation Delays: τ1 =49ns; τ2 =85ns; τ3 =134ns; τ4 =193ns;τ5 =237ns
(g)
(f)
(h)
(b)
Fig. 1 (a)TCM encoder. (b) Experimental setup. (c)-(h) are the output near field patterns of the 6-LP mode PL when launching from the six input port independently. PBC: polarization beam combiner; LO: local oscillator; CD Comp.: chromatic dispersion compensation; FE Comp.: front-end compensation; CPE: carrier phase estimation; VOA: variable optical attenuator.
For the mode division multiplexer, we employ an all fiber photonic lantern (PL) [9]. Six single mode fibers are positioned asymmetrically and adiabatically tapered to a small size to match the core of a 4-LP mode fiber, in order to excite LP01, LP11a, LP11b, LP21a, LP21b, and LP02, respectively. The output near field patterns are shown in Fig. 1 (c)-(h) when launching from each input single mode fiber independently. Although the output pattern is not an accurate LP mode, strong mode selectivity can be easily found for LP21 and LP02 mode according to their modal patterns. When the output of PL is coupled with the 6-LP mode fiber which supports 6 LP modes (LP01, LP11, LP21, LP02, LP12, and LP31) and has a low differential mode group delay of 85ps/km [10], the insertion loss (IL) of individual mode is 0.88dB, 1.57dB, 1.12dB, 1.6dB, 1.59dB, and 1.16dB, respectively, with the loss of the PL included. Even though FMF was not tapered to match the core size of the PL, the coupling loss from the PL to the 6LP FMF is less than 1dB. In order to manage the splitting loss of the used coupler and equalize the mode dependent loss (MDL) of the PL based MUX, 6 Erbium doped fiber amplifiers (EDFAs) are employed before the mode MUX, as labeled in Fig. 1 (b). Although the mode MUX is mode selective, we get full mixed signals after transmission over the 6-LP mode fiber. After the mode DeMUX, the time-division-multiplexed space division multiplexer (TDMSDM) receiver is employed to reduce the number of required receivers [11]. Consequently, only 2 sets of coherent receivers and corresponding real-time oscilloscopes are needed to measure the 12 mode channels. After digital resampling and re-alignment, front-end impairments and chromatic dispersion (CD) are compensated. To unravel the 12 transmitted spatial channels, we utilize a 12×12 minimum mean squared error (MMSE) time domain equalizer (TDE) algorithm. The weight matrix of the TDE is heuristically updated using the least mean squares (LMS) algorithm during convergence and decision directed least mean squares (DD-LMS) during data transmission, with adaptive step size [3]. The frequency offset between the signal laser and LO is compensated using one carrier phase estimation (CPE) block per spatial channel in the form of a digital phase locked loop (DPLL). After demapping process, the bit error ratio (BER) and MDL of the system transmission matrix are calculated. 4. Results and Discussions The input power at each port of the PL is set to be 4.88dBm, 5.57dBm, 5.12dBm, 5.6dBm, 5.59dBm, and 5.16dBm, respectively. Consequently, the powers of six modes excited by each port at the 6-LP mode fiber are all equal to 4dBm. The overall launch power at the 6-LP mode fiber input is measured to be 11.82 dBm and the output power after 40km FMF transmission is 3.75 dBm. The power at each output port of mode DeMUX is around −13 dBm when only one port is launched. The power difference between the output ports is lower than 2dB, confirming the almost full mode coupling within the 6-LP mode fiber. We first demonstrate the 12-spatial-and-polarization modes multiplexing transmission over 22.5km and 40km 6-LP mode fiber with 20GBaud dual-polarization (DP)-QPSK
Th4C.5.pdf
OFC 2016 © OSA 2016
signal. As depicted in Fig. 2(a), for the 7% hard-decision forward error correction (HD-FEC), an OSNR of 18.1dB can guarantee such transmission over 22.5km, while 2dB penalty is experimentally found when the distance is extended to 40km. When it comes to the 20% soft-decision (SD)-FEC, the OSNR penalty is reduced to 0.6dB, demonstrating a relatively small OSNR penalty when extending the transmission distance. Next, we explore the performance of advanced coding scheme TCM-8-PSK in this system. From Fig. 2 (a) we observe that the BER curve crossing point for conventional QPSK and TCM-8-PSK is between the HD-FEC and SD-FEC. For 22.5km and 40km transmission, the OSNR required by the 20% SD-FEC for TCM is 14.68dB and 15.53dB, respectively, showing a penalty of 0.98dB and 1.18dB, compared with conventional QPSK. However, at the 7% HD-FEC, the advantage of TCM becomes obvious. Compared with conventional QPSK, the OSNR gain of TCM is 0.95dB, 1.3dB, and 1.8dB for back-to-back, 22.5km, and 40km transmission, respectively. The OSNR gain for the MDM system is larger than that of the SSMF systems. Compared with the back-to-back transmission, about 4 dB of OSNR penalty is observed for both the conventional QPSK transmission and the TCM-8PSK transmission when the distance is extended to 22.5 and 40km.
(a) (b) Fig. 2. (a) System BER for various OSNRs. (b) System MDL variation with OSNR.
According to the singular value decomposition of system transmission matrix, we can obtain the MDL of transmission system. For the 40km transmission of QPSK and TCM-8PSK, the variation of MDL with launching OSNR is shown in Fig. 2 (b). There is no definite relationship between MDL and system OSNR. The MDL is around 4.5dB when sweeping the system OSNR, showing a small variation less than 1.2dB. 4. Conclusions In conclusion, we have experimentally demonstrated trellis coded modulation over 12-spatial-and-polarization modes multiplexing transmission of both 22.5km and 40km 6-LP mode fiber using low loss 6-mode photonic lantern based mode MUX. By employing 12×12 MIMO DSP, 20GBaud DP-QPSK and DP-TCM-8PSK signal after the transmission link are recovered successfully. The advanced coding scheme TCM shows favorable OSNR gain of 1.8dB when compared to conventional QPSK signal, without reducing the effective spectral efficiency. This work was partially supported by the 863 High Technology Plan (2015AA015502), National Natural Science Foundation of China (61275068, 61331010), and the NWO Dutch photonics PhD program.
References [1] A. D. Ellis et al., “Approaching the Non-Linear Shannon Limit,” J. Lightwave Technol. 28(4), 423-433 (2010). [2] R. G. H. van Uden et al., “Ultra-high-density spatial division multiplexing with a few-mode multicore fibre,” Nature Photonics 8, 865 (2014). [3] R. G. H. van Uden et al., “MIMO equalization with adaptive step size for few-mode fiber transmission systems,” Opt. Express 22(1), 119126 (2014). [4] J. J. A. van Weerdenburg et al., “10 Spatial mode transmission using low differential mode delay 6-LP fiber using all-fiber photonic lanterns,” Opt. Express 23, 24759-24769 (2015) [5] N. K. Fontaine et al., “30x30 MIMO Transmission over 15 Spatial Modes,” in Proc. OFC, Th5C.1 (2015). [6] E. Sillekens et al., “Experimental Demonstration of 8 state Turbo Trellis coded Modulation Employing 8 Phase Shift Keying,” in Proc. ECOC, Tu3.4.1 (2015). [7] C.M.Okonkwo et al., “Advanced coding techniques for few mode transmission systems,” Opt. Express 23(2), 1411-1420 (2015). [8] S. Ishimura et al., “Eight-state trellis-coded optical modulation with signal constellations of four-dimensional M-ary quadrature-amplitude modulation,” Opt Express 23(5), 6692-6704 (2015). [9] A. Velazquez-Benitez et al., “Six mode selective fiber optic spatial multiplexer,”Opt Letters 40(8), 1663-1666 (2015). [10] P. Sillard et al., “Low-DMGD 6-LP-mode fiber,” in Proc. OFC, M3F.2 (2014). [11] R. G. H. van Uden et al., “Time domain multiplexed spatial division multiplexing receiver,” Opt. Express 22(10), 12668-12677 (2014).
