Journal of Networks - Academy Publisher

More documents

Recommendations

Info

158 JOURNAL OF NETWORKS, VOL. 5, NO. 2, FEBRUARY 2010 Signal Processing Algorithms in 100Gbit/s Optical Coherent and Non-coherent Receivers with PSK Modulation Christina Hebebrand and Werner Rosenkranz Chair for Communications, University of Kiel, Kaiserstr. 2, D-24143 Kiel, Germany Email: {ch, wr}@tf.uni-kiel.de Abstract— This paper addresses multi-level PSK modulation formats with coherent and non-coherent detection and reviews signal processing algorithms like mitigation of phase noise, the clock and carrier recovery algorithms as well as the equalizer structures and their performance. Index Terms— optical communications, modulation, phase shift keying, phase noise, equalizer I. INTRODUCTION Electronic signal processing is currently introduced in various subsystems of modern optical high speed transceivers. Traditionally electronic processing was basically limited to tasks like MUX/DEMUX, clock and data recovery and various synchronization loops as well as analog circuitry like modulator drivers or limiting amplifiers. In order to deal with advanced systems and with the need for increased performance at long reach and higher speed, additional signal processing requirements have appeared. Examples are FEC and electronic dispersion equalizers (EDC), electronic precompensation, differential encoders, clock and carrier recovery, and polarization control. Recent challenges employ pure digital processing, where ADCs and DACs are required, like OFDM modulation (orthogonal frequency division multiplexing) or MLSE (maximum likelihood sequence estimation) equalizers. In this paper some of these signal processing algorithms, which may be used in PSK modulated systems are considered in detail and their impact on performance and cost in terms of implementation effort is reviewed. II. MITIGATION OF PHASE NOISE IN NON- COHERENT PSK-RECEIVERS In a first part, phase noise is investigated. Phase noise is an important limiting effect in high-speed optical communication systems that make use of phase-shift keying modulation formats [1]. One method to mitigate nonlinear phase noise distortions is the compensation of the mean nonlinear phase shift (MEAN) [2]. Alternatively, multi-symbol phase estimation (MSPE) [3, 4] can be used to reduce the loss from direct detection compared to coherent detection by estimating a reference © 2010 ACADEMY PUBLISHER doi:10.4304/jnw.5.2.158-164 phase over several consecutive received symbols. Both phase noise compensation methods, MEAN (in a post detection version for DQPSK) and MSPE require signal processing at the receiver side after optical-to-electrical conversion. We investigate the performance of different phase noise compensation options for a high speed RZ-DQPSK transmission system with various dispersion maps and direct detection. The efficiency of MEAN, MSPE and the combination of both is examined by Monte-Carlo simulations for varying pre- and post-compensation. Furthermore the performance of these three strategies for different average fiber input powers is investigated. Not only the optimum phase noise compensation strategy is identified, also the sensitivity of all three compensation variants to the chosen dispersion map is compared to the conventional receiver. For the best noise tolerant receiver option a maximum Q-gain of approximately 2 dB can be achieved for a wide range of fiber input power values [13]. A. Simulation Setup The simulation setup for the 43 Gbit/s RZ-DQPSK multi-span transmission system is shown in Fig. 1. (Here 43 Gbit/s is used, but the methods are also applicable for 100 Gbit/s). At the transmitter side an optical I/Qmodulator based on a parallel Mach-Zehnder-Modulator (MZM) structure is used to generate the DQPSK signal. After RZ-pulse carving the optical signal is filtered (Gaussian filter 1 st order, fFWHM = 50 GHz), taking into account an optical multiplexer, and then amplified to achieve the desired average input power. The optical channel is characterized by a variable chromatic dispersion (CD) pre-compensation, eight fiber spans and a variable CD post-compensation. The pre- and postcompensation values are varied in steps of 50 ps/nm. Each span consists of 80 km SMF (D = 17 ps/nm/km, γ = 1.37 W -1 km -1 ), a DCF and an EDFA, which fully compensates for the attenuation of the whole span. The OSNR of the system is adjusted to 14 dB. In each span the DCF compensates for 90% of the CD of the SMF (10% inline under-compensation). In the simulation the transmission fibres are assumed to be nonlinear. At the receiver side the signal is filtered (Gaussian filter 1.5 th order, fFWHM = 50 GHz) and then received by two Mach- Zehnder delay interferometers (MZDI) followed by a balanced detector. The inphase- and quadrature
JOURNAL OF NETWORKS, VOL. 5, NO. 2, FEBRUARY 2010 159 components are then filtered by an electrical lowpass filter (Bessel 5 th order, f3dB = 15.05GHz). Both phase noise compensation methods, MEAN and MSPE, are implemented. Either one of them or both can be used for these investigations. The MEAN method [2], which exploits the fact, that the mean nonlinear phase shift is proportional to the received power, is implemented in the electrical domain using an additional intensity detection branch. The intensity difference of two consecutive symbols has to be determined to get a value that is proportional to the received nonlinear phase noise. This is the difference of the nonlinear phase noise of two consecutive symbols due to the reception with MZDIs. The intensity difference is multiplied by a scaling factor α and then used to rotate the phase of the inphase and quadrature component of the received signal. The MSPE method as described in [3, 4] calculates recursively a new decision variable x(n) using the received signal u(n) and the previous variable x(n-1). A forgetting factor w slowly fades out the contribution of the previous symbols. This method averages out the phase noise of the received symbols, leading to a better phase reference and thus eliminating the loss of direct detection compared to coherent reception. CW laser data real' data imag' Post- 8 spans compensation BP EDFA DCF SMF c(n) T S MZM push-pull data real MZM push-pull data imag D-FF � j 4 e �/2 MSPE B. Simulation Results x(n) T S T S T S x(n-1) 10% under compensation �/4 �/4 + w RZ-pulse carving clock u(n) BP 80km Boosteramplifier Precompensation + x LP LP LP - T S Phase Rotator MEAN Figure 1. Simulation setup for 43Gb/s; MZM: Mach-Zehnder Modulator; BP: Bandpass; LP: Lowpass; D-FF: D-flip-flop; MEAN: Compensation of the mean nonlinear phase shift; MSPE: Multi-symbol phase estimation. To investigate the efficiency of the examined methods with respect to the conventional receiver, Fig. 2 depicts © 2010 ACADEMY PUBLISHER � the Q gain versus the residual dispersion of the MEAN and the MSPE for three different pre-compensation values and the three fibre input powers. Figure 2. Q gain [dB] with respect to the conventional receiver vs. residual dispersion [ps/nm] for MEAN (left column) and MSPE (right column) for different pre-compensations and 4 dBm (upper row), 6 dBm (centre row) and 8 dBm (lower row) average fibre input power. It can be seen that for 4 dBm the Q gain is almost the same for the three pre-compensations as well for MEAN as for MSPE. For MEAN there is almost no improvement, whereas the MSPE achieves a Q gain of 1- 1.5 dB. For 6 dBm and 8 dBm the efficiency of MEAN increases for those values of pre-compensation where the performance of the conventional receiver is worse. Moreover, it can be seen that the performance of MEAN increases with increasing average fibre input power. For 6 dBm a maximum Q gain of 0.9 dB and for 8 dBm of 2.3 dB can be achieved. This leads to the conclusion that MEAN is most efficient for systems with performance loss due to the influence of nonlinearities, as in the case of non-optimised dispersion maps. The improvement of MSPE, however, is almost the same for all investigated pre-compensation values for each of the input powers. Furthermore, the efficiency of MSPE decreases for increasing input power due to the higher influence of the nonlinearities. The efficiency with respect to the conventional receiver of the combination of MEAN and MSPE is shown in Fig. 3 for different pre-compensations and the three fibre input powers. The Q gain of the combined method is approximately equal to the sum of the Q gains of each method alone. The combined method is hence most efficient if the improvement of MEAN and MSPE is relative high, as e.g. for 6 dBm input power and -900
Page 1 and 2: Journal of Networks ISSN 1796-2056
Page 3 and 4: JOURNAL OF NETWORKS, VOL. 5, NO. 2,
Page 23 and 24: APPENDIX EVALUATION OF THE OPTICAL
Page 31: JOURNAL OF NETWORKS, VOL. 5, NO. 2,
Page 83 and 84:
JOURNAL OF NETWORKS, VOL. 5, NO. 2,
Page 85 and 86:
Page 87 and 88:
Page 89 and 90:
Page 91 and 92:
Page 93 and 94:
Page 95 and 96:
Page 97 and 98:
Page 99 and 100:
Page 101 and 102:
Page 103 and 104:
Page 105 and 106:
Page 107 and 108:
Page 109 and 110:
Page 111 and 112:
Page 113 and 114:
Page 115 and 116:
Page 117 and 118:
Page 119 and 120:
Page 121 and 122:
Page 123 and 124:
Page 125 and 126:
Page 127 and 128:
Page 129 and 130:
Page 131 and 132:
Page 133 and 134:
Page 135 and 136:
Page 137 and 138:
Page 139 and 140:
Aims and Scope. Call for Papers and
show all

Journal of Networks - Academy Publisher

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?