Open Access
Add to BibSonomyAbstract
We numerically demonstrate colorless reception of dense wavelength division multiplexed channels in the C-band for high-order QAM (16-64 QAM) signals on a 120° monolithically integrated downconverter, based on a 2x3 MMI with calibrated analog IQ recovery. It is shown that the proposed calibrated 120° downconverter can increase up to 80 the number of coincident channels in an efficient way, exhibiting good signal dynamic range and high fabrication yield. As this downconverter makes use of the minimum number of power outputs required for perfect recovery of IQ signals, it becomes an interesting alternative to conventional 90° based downconverters.
© 2013 Optical Society of America
1. Introduction
The deployment of reconfigurable optical add-drop multiplexers (ROADM) in transport optical networks has provided flexibility and configurability capabilities to network operators. The introduction in the near future of colorless ROADM will also allow any wavelength to be added/dropped to any port. In this situation colorless receivers can be used in the drop ports to increase the efficiency and reduce the cost of reconfigurable optical networks. In a colorless receiver, just by tuning the local oscillator (LO), an individual wavelength-division multiplexed (WDM) channel can be selected and detected without any optical filtering device (e.g. demultiplexer or filter).
The Optical Internetworking Forum (OIF) [1] has proposed dual polarization quadrature phase-shift keying (DP-QPSK) modulation format to reach 100 Gbps per channel in the amplified C-band over the existing optical network infrastructure. Higher quadrature amplitude modulation (e.g. 16-64 QAM) is a viable alternative to further increase system transmission capacity while reducing bandwidth requirements. Thus, in the framework of the MIRTHE project [2], 16-QAM monolithically integrated transmitter and receivers for 400 Gbps are being assessed. Required coherent receivers comprise a polarization diversity network (e.g. polarization beam splitters) and two phase diversity downconverters (one per polarization). In this paper we will focus on the last part, that is, the optical downconverter. A widespread solution is the monolithic integration of the 90° optical hybrid with four photodiodes in balanced configuration on the same chip [2, 3]. In colorless reception, a measure of the suppression of the interfering direct-detection terms from coincident WDM channels is the common-mode-rejection-ratio (CMRR) [4, 5]. Therefore, balanced 90° hybrid based coherent receiver with high CMRR and high LO-to-signal power ratio can be used as colorless receiver. However, a wideband high CMRR will require stringent fabrication tolerance requirements (resulting in high cost and low fabrication yield) to reduce hardware impairments (i.e. amplitude imbalances existing in phase diversity network or photodiode responsivity mismatch) [6, 7].
A promising alternative to overcome the above problems is the 120° phase diversity receiver which, if properly calibrated, has shown to be highly tolerant to hardware impairments at microwave frequencies [8]. This is an interesting solution because, as it is known from multiport theory [9, 10], three is the minimum number of power outputs to perfectly recover IQ signals from power readings by linear means, and thus the 120° based downconverter is the simplest multiport receiver. This type of 120° downconverter has been reported several times for optical communications by using 3x3 fiber couplers [11, 12]. The authors have recently proposed a monolithic integrated downconverter, based on a 2x3 multimode interference coupler (MMI), with a simple linear calibration strategy to fully correct receiver impairments [7, 10]. Our proposal, compared to the balanced 90° downconverter (based on a 2x4 MMI), not only showed the same noise-induced penalty under ideal hardware, but exhibited a higher signal dynamic range, wider operating bandwidth and greater tolerance to fabrication errors for a single-channel reception. In this paper we compare the performance of colorless reception of 56 Gbps QAM WDM channels (enabling 112 Gbps under dual polarization), for two different types of monolithically integrated downconverters: the conventional 90°, based in a 2x4 MMI and differential transimpedance amplifiers, and the 120°, based on a 2x3 MMI [7, 10] with calibrated analog in-phase and quadrature (IQ) recovery. Simulation results show that hardware imbalances arising from typical fabrication errors reduce the CMRR and increase the interference from coincident channels in a much more limiting way for the monolithic integrated conventional 90° downconverter than for the calibrated 120° downconverter.
The paper is organized as follows: in Sections II and III monolithically integrated conventional 90° and proposed calibrated 120° downconverters are described, respectively, for colorless reception under hardware imbalances. In Section IV, the colorless performance of both schemes is numerically compared in a realistic fabrication scenario. In Section V different proposals of calibration for the 90° downconverter are briefly evaluated. Finally, Section VI provides the main conclusions.
2. Colorless operation of monolithic integrated conventional 90° hybrid downconverter
Conventional 90° hybrid integrated downconverter in Fig. 1, is based on a monolithically integrated 2x4 MMI with four photodiodes followed by differential transimpedance amplifiers (TIA) with DC offset cancellation [3]. Output electrical IQ signal components are then digitized in two analog-to-digital converters (ADC) and combined to be further processed in the digital signal processor (DSP).
Assuming a previous polarization diversity control network, the electric field of the received WDM signal (composed of N multiplexed channels) and the LO laser can be described in terms of their complex envelopes as
For homodyne detection, the LO (of power PLO) must be tuned to the angular frequency ωk of the channel to be detected. Considering an equal signal power for all multiplexed channels, individual slowly varying complex signal envelope of the nth channel can be written (neglecting fiber transmission impairments) in terms of the normalized IQ signal components, In and Qn, as
The WDM signal and LO are combined in the 2x4 MMI, with scattering parameters Skij defined between their ports at frequency ωk, and detected from the photodiodes, with responsivities Ri. Therefore, when selecting kth channel, the four output photocurrents can be calculated (from i = 3, 4, 5 or 6) as
Neglecting the high frequency beating terms at ωn-ωk, which will be completely filtered by the electronics, photocurrents iIk and iQk for IQ components can be obtained from idealized differential TIAs, and be described in matrix form as
Three terms can be identified at the right-hand side of Eq. (5): DC offset term, an interfering direct-detection term from the self-beating of adjacent channels and linear axis transformation of IQ components. Their parameters (α, γ, u, v), which were first introduced in [13], are shown again here in Table 1 for convenience. Linear terms cause a translation, rotation and imbalance of reference axes. Compensation of the linear distortion induced at each wavelength by hardware imperfections (hybrid and photodiode responsivity imbalance) will be removed in the DSP by the Gram-Schmidt orthogonalization procedure (GSOP) [14]. The second term of Eq. (5) causes a baseband interference current that cannot be removed and limits the colorless behavior of the receiver. This interfering current can therefore be expressed as
Table 1. Parameters Derived in [13] to Characterize Conventional 90° Hybrid Integrated Coherent Receiver
The interference term depends on the signal power, the number of coincident channels and the performance of the coherent receiver in terms of power imbalance. It must be noticed that the baseband interference current shows a close relation with the CMRR for a single wavelength signal [1], since it is a direct measurement of the power imbalance behavior of a coherent receiver.
Therefore, we will use the CMRR as a figure of merit of the colorless behavior of the proposed integrated colorless receivers, as it is usually done in literature [4, 5].
For multichannel reception, from Eq. (5), the self-beating interference contribution from each adjacent channel will be weighted by the signal power and the CMRR at its respective wavelength. In this way, colorless reception will require a low Ps/PLO ratio and a high CMRR over the complete received multichannel frequency band.
Figure 2(a) shows the transversal geometry of the waveguides used in this work. We will consider only two relevant cases for simulations: I. Nominal design (i.e. no fabrication errors) and II. Moderate fabrication errors. Based on our experience working with a commercial InP fabrication platform [2, 3], we have chosen as moderate fabrication errors deviations in waveguide widths |δW|<150 nm and etch depth errors |δD|<45nm. A realistic photodiode imbalance responsivity of 5% has also been included in simulations. A more detailed description of the integrated downconverter including relevant physical dimensions can be found in [7]. Figure 2(b) shows the resultant wavelength dependence of the maximum value of CMRR from Eq. (7). The OIF specifies a CMRR for the signal port greater than 20 dB in absolute value [1]. As expected, the nominal design (Case I) corresponds to an OIF-compliant 90° downconverter. However, the specification of the OIF for the CMRR will be fulfilled only in half of the C-band under reasonable fabrication errors (Case II).
Fig. 2 (a) Transversal geometry of the InP/InGaAsP rib waveguides used in this work. H = 1µm, D = 0.5µm, nInP = 3.18, nInGaAsP = 3.27. (b) CMRR for input signal port versus wavelength in the C-band for the conventional 90° downconverter as a function of the fabrication tolerance (Case I/II).
3. Colorless operation of monolithic integrated calibrated 120° coupler downconverter
Figure 3 shows the proposed colorless calibrated 120° optical downconverter based on a 2x3 MMI coupler monolithically integrated with three photodiodes followed by their respective TIAs [7, 10]. Output electrical signals are then linearly combined with a calibrated analog circuit prior to be digitized in two ADCs and digitally processed.
Following a similar analysis to that of Section 2, the WDM signal and LO waves will be combined in the 120° coupler, being described the three photocurrents when selecting kth channel as i3k, i4k and i5k.
Three parameters, showed in Table 2, describe now the three terms that express the output photocurrents [10]: a DC offset term, an interfering direct-detection term from the self-beating of adjacent channels (equivalent to the second term of Eq. (5) for the 90° downconverter) and LO-signal power-dependent linear combination from IQ signal components.
Table 2. Parameters to Characterize 120° Coherent Receiver
As it was shown previously for a single-channel reception in [10], solving the linear equations formulated in Eq. (8), undistorted IQ components of the demodulated channel can be recovered by a linear combination of the three output photocurrents, while canceling self-induced intensity interfering term.
Under ideal hardware it is easily obtained that [10, 12]
For real operation, the required coefficients (AIi, AQi) are obtained following a simple calibration method [9] applied at the central wavelength of the C-band (i.e. 1550 nm). Due to the reduced wavelength-dependence of the 2x3 MMI scattering parameters [7, 15], the calibrated coefficients calculated at 1550 nm can be used over the complete C-band to nearly cancel receiver imbalances. This allows us to electronically regenerate, as Fig. 3 shows, the IQ components of any channel in the C-band just by implementing the linear analog operation described by Eq. (9) (operation that was digitally assessed in [7, 10]). It must be highlighted that, contrary to the conventional balanced 90° hybrid approach, in this case it is not required the use of the digital GSOP algorithm, since the analog calibration almost compensates all the hardware impairments of the receiver over the complete C-band, as it will be shown later. From Eq. (8)-(9), it can be shown that the baseband interference current can be expressed as
In the ideal case (e.g. |Sij|2 = ⅓ within the working band), from Eq. (10)-(11), the interference term is zero and there is no limitation in the colorless operation of the device. When the 120° coupler is not ideal, deviations from the desired scattering parameters can be partially compensated using the calibrated coefficients (AIi, AQi), so the interference contribution from all WDM channels is highly reduced. It must be highlighted that the analog operation described by Eq. (9) has an important advantage with respect the digital approach proposed in [7, 10]. When the IQ recovery is performed in the analog domain, the interfering direct-detection term is cancelled prior to the ADC conversion, so the effective number of bits (ENoB) of the ADC in the presence of multiple adjacent channels is not seriously reduced [12]. As we stated in Section 2, the CMRR will be used as a figure of merit of the colorless behavior of the receiver. Since CMRR for a 120° downconverter is not defined in OIF, we propose to use the following expressions
It must be noticed that these expressions have a close relation with CMRR definition of the 90° downconverter from Eq. (7): the numerator coincides with the interference term defined in Eq. (11), whereas the denominator is just a normalization factor. Simulation results will show the appropriateness of Eq. (12).
Figure 4 shows the CMRR calculations for the integrated calibrated 120° optical downconverter developed in [7] under the same fabrication scenarios presented in Section 2. Figure 4(a) has been obtained for each wavelength from the maximum value of Eq. (12) using the exact calibration coefficients calculated at 1550 nm. Figure 4(b) represents a more realistic case, where deviations of a 5% from the exact values have been introduced in the calibration coefficients. In both cases, the CMRR is better than 20dB in the complete C-band, clearly improving the CMRR behavior of the conventional 90° downconverter shown in Fig. 2(b). In the next section the 120° downconverter will adopt realistic calibration coefficients, being shown that its better CMRR, see Fig. 4(b), will cause a noticeable improvement in the colorless operation with respect to the conventional 90° hybrid based receiver.
Fig. 4 CMRR versus wavelength for the calibrated 120°/conventional 90° downconverters as a function of the fabrication tolerance (Case I/II). (a) 120° downc. Exact coefficients at 1550 nm (b) 120° downc. Coefficients with a 5% deviation.
4. Colorless performance comparison of conventional 90° and calibrated 120° downconverters
In this section the colorless performance of the downconverters presented in Sections 2 and 3 will be numerically simulated and compared. An external LO power of 10 dBm and WDM channels of equal power at 56 Gbps (enabling 112 Gbps under dual polarization) centered on the C-band (50 GHz grid) have been considered. The optical fiber has been modeled as an AWGN channel with uniform spontaneous amplified emission (ASE) noise contribution to each channel. Therefore, the effect of the residual dispersion or polarization orientation of adjacent channels on the receiver performance has not been assessed here (see [4, 5] for a deeper study of the scaling factor to be introduced in the intensity interfering term). Incoming OSNR has been adjusted for incidents channel at BER = 10−4 in an ideal coherent receiver in absence of internal noise sources. TIAs have been modeled with an input referred noise current density of 20 pA/√Hz. An ADC resolution of 5 bits and 6 bits has been considered for 16-QAM and 64-QAM respectively, thus obtaining a low quantization noise penalty (≈0.5 dB) [16]. More details of the simulation scenario can be found in [10]. Figure 5 shows, for the nominal design (Case I), the OSNR penalty (for a BER = 10−4) versus the input signal power as a function of the WDM channel number under higher-order modulation (16-QAM and 64-QAM). Dashed line represents an additional 1dB OSNR penalty over the quantization noise floor (of 0.5 dB). Both receivers are limited by shot-noise in a similar way for low signal power levels. However the conventional 90° receiver performance will be particularly degraded for high signal power by the interference from the self-beating of adjacent channels. This interference contribution increases with the number of WDM channels and is due to CMRR degradation arising from non-perfect MMI coupler performance for all transmitted channels. In this way, for reasonable fabrication errors in the MMI (Case II), as Fig. 6 shows, the conventional 90° downconverter performance will be further degraded at high signal power, being limited its dynamic range. Please notice that colorless reception of 80 channels was not included in the results of Fig. 6(b) for the conventional 90° downconverter, as this type of receiver was not able to get the required BER for this scenario. The advantage of the calibrated120° downconverter is more obvious for high signal power when increasing the number of WDM channels, where the reduction of self-beating interference from adjacent channels becomes more appreciable. These results are also in close correspondence with those in Fig. 4, which showed a better value of CMRR along the complete C-band for the calibrated 120° receiver. Table 3 summarizes the dynamic range obtained for each type of downconverter as a function of the number of WDM channels, in a moderate fabrication error scenario (Case II) and for a maximum OSNR penalty of 1.5 dB.
Fig. 5 OSNR penalty (for a BER = 10−4) versus input signal power in a conventional 90° hybrid (filled circles) and calibrated 120° coupler (empty circles) downconverters, following the nominal design (Case I), as a function of the number of WDM channels (a) 16-QAM transmission (b) 64-QAM transmission.
Fig. 6 OSNR penalty (for a BER = 10−4) versus input signal power in a conventional 90° hybrid (filled circles) and calibrated120° coupler (empty circles) downconverters, following moderate fabrication errors (Case II), as a function of the number of WDM channels (a) 16-QAM (b) 64-QAM transmission.
Table 3. Dynamic Range for the Conventional 90°/ Calibrated 120° Downconverter as a Function of the Number of WDM Channels
5. Colorless operation of calibrated 90° downconverters
In the previous sections we have compared the performance of the conventional non-calibrated 90° downconverter with the calibrated 120° receiver, thus it could raise the question of whether the improved performance of the 120° downconverter may be due solely to the calibration process. To address this question the calibrated 120° downconverter should be compared with its calibrated 90° counterpart in an identical scenario.
In a first approach two different possibilities of calibrated 90° downconverters arise: A) implementing a full calibrated analog IQ recovery taking as inputs all the four photocurrents (see Fig. 7(a)), B) calibrating the weights between each pair of photocurrents (see Fig. 7(b)) prior to their amplification. Obviously Option A, besides requiring an additional TIA, is more complex than the 120° counterpart, as four independent power input signals (instead of three) with adequate weights must be used to recover IQ signal. Thus, maybe better performance could be obtained if the weights are properly adjusted. This is an interesting alternative for high precision applications whose study is beyond the scope of this paper and which should be investigated in future works. Concerning alternative B, the comparison with the proposed 120° architecture should be done on a fair basis, thus GSOP algorithm, which was used in previous sections to partially remove the hardware imbalances of the 90° downconverter but not on the 120° one, must be removed for both alternatives.
Fig. 7 Proposals of calibrated 90° downconverters: (a) 90° downc. with calibrated analog IQ recovery from the four output photocurrents, (b) 90° downc. with calibrated weights between each pair of photocurrents.
Figure 8 shows a comparison of performance for the calibrated downconverters: the 120° from Fig. 3 and the 90° Option B from Fig. 7(b). The OSNR penalty (for a BER = 10−4) versus the input signal power as a function of the WDM channel number under higher-order modulation (16-QAM and 64-QAM) are depicted in this figure for the moderate fabrication error scenario (Case II). It is observed that the calibrated 120° downconverter can still offer an improvement in the OSNR penalty of 0.8 dB for 16-QAM under 80 WDM channels and 2.8 dB for 64-QAM under 40 WDM channels (note that colorless reception of more than 40 channels was not included for 64-QAM transmission, as the Option B of the calibrated 90° downconverter was not able to get the required BER for this scenario).
Fig. 8 OSNR penalty (for a BER = 10−4) versus input signal power in a calibrated 120° downconverter (empty circles) and the calibrated Option B of the 90° downconverter (filled squares), following moderate fabrication errors (Case II), as a function of the number of WDM channels (a) 16-QAM (b) 64-QAM transmission.
Some insight on these results can be gained noting that, from the two types of distortion in Eq. (5) (second and third term) that are suffered by the 90° downconverter, the calibrated Option B nearly cancels the interference direct-detection term (second term of Eq. (5) which is closely related to CMRR), but not the linear distortion term (third term of Eq. (5) induced by hardware imperfections. Both terms can be however simultaneously compensated in the calibrated 120° downconverter [10]. This linear distortion causes a detrimental imbalance of reference axes of the received IQ constellation for the 90° approach [13] which finally leads to worse receiver performance as indicated by Fig. 8.
6. Conclusions
We have compared the colorless performance of two monolithically integrated receivers: i) the conventional 90° hybrid downconverter based on a 2x4 MMI with balanced photodetection, ii) a 120° coupler downconverter based on a 2x3 MMI with calibrated analog IQ recovery. Passive components of both devices have been designed using standard InP/InGaAsP rib waveguides [7], whereas typical fabrication errors (e.g. waveguide width and etch depth errors) have been included to define realistic simulation scenarios. Numerical results clearly show that, in a colorless multichannel high-order modulation (16-64 QAM) scenario, the calibrated 120° downconverter significantly outperforms the conventional 90° receiver. Specifically, it has been shown that, for realistic fabrication errors and 64-QAM transmission, the calibrated 120° downconverter can achieve colorless reception of 80 WDM channels within the whole C-band and over a wide dynamic range (~10 dB). In the same scenario, conventional 90° downconverter only supports up to 40 channels, with a much more reduced dynamic range (~4.5 dB). Additionally, other alternatives from the calibration of the 90° downconverter have been briefly evaluated.
Acknowledgments
The authors gratefully acknowledge the design support from Diego Pérez-Galacho. This work has been partially funded under Andalusian Regional Ministry of Science Innovation and Business project P09-TIC-5268, Spanish Ministry of Science and Innovation project TEC2009-10152 and EU 7th Framework Programme project MIRTHE ICT-2009-5 nº 257980.
References and links
1. Optical Internetworking Forum (OIF), “100G ultra long haul DWDM framework document,” document OIF-FD-100G-DWDM-01.0 (June 2009), http://www.oiforum.com/public/impagreements.html.
2. Mirthe Project, “Monolithic InP-based dual polarization QPSK integrated receiver and transmitter for coherent 100–400Gb Ethernet,” http://www.ist-mirthe.eu/.
3. R. Kunkel, H. G. Bach, D. Hoffmann, C. Weinert, I. Molina-Fernández, and R. Halir, “First monolithic InP-based 90 degrees-hybrid OEIC comprising balanced detectors for 100GE coherent frontends,” in International Conference on Indium Phosphide & Related Materials (IPRM, 2009), paper TuB2.2, pp. 167–170.
4. B. Zhang, C. Malouin, and T. J. Schmidt, “Towards full band colorless reception with coherent balanced receivers,” Opt. Express 20(9), 10339–10352 (2012). [CrossRef] [PubMed]
5. L. E. Nelson, S. L. Woodward, S. Foo, M. Moyer, D. J. S. Beckett, M. O’Sullivan, and P. D. Magill, “Detection of a single 40 Gb/s polarization-multiplexed QPSK channel with a real-time intradyne receiver in the presence of multiple coincident WDM channels,” J. Lightwave Technol. 28(20), 2933–2943 (2010). [CrossRef]
6. V. E. Houtsma, N. G. Weimann, T. Hu, R. Kopf, A. Tate, J. Frackoviak, R. Reyes, Y. K. Chen, L. Zhang, C. R. Doerr, and D. T. Neilson, “Manufacturable monolithically integrated InP dual-port coherent receiver for 100G PDM-QPSK applications,” Tech. Digest Optical Fiber Comm. (OFC) (2011), paper OML2.
7. P. J. Reyes-Iglesias, A. Ortega-Moñux, and I. Molina-Fernández, “Enhanced monolithically integrated coherent 120° downconverter with high fabrication yield,” Opt. Express 20(21), 23013–23018 (2012). [CrossRef] [PubMed]
8. P. Pérez-Lara, I. Molina-Fernández, J. G. Wangüemert-Pérez, and A. Rueda-Pérez, “Broadband five-port direct receiver based on low-pass and high-pass phase shifters,” IEEE Trans. Microw. Theory Tech. 58(4), 849–853 (2010). [CrossRef]
9. F. M. Ghannouchi and R. G. Bosisio, “An alternative explicit six-port matrix calibration formalism using five standards,” IEEE Trans. Microw. Theory Tech. 36(3), 494–498 (1988). [CrossRef]
10. P. J. Reyes-Iglesias, I. Molina-Fernández, A. Moscoso-Mártir, and A. Ortega-Moñux, “High-performance monolithically integrated 120° downconverter with relaxed hardware constraints,” Opt. Express 20(5), 5725–5741 (2012). [CrossRef] [PubMed]
11. T. Pfau, S. Hoffmann, O. Adamczyk, R. Peveling, V. Herath, M. Porrmann, and R. Noé, “Coherent optical communication: towards realtime systems at 40 Gbit/s and beyond,” Opt. Express 16(2), 866–872 (2008). [CrossRef] [PubMed]
12. C. Xie, P. J. Winzer, G. Raybon, A. H. Gnauck, B. Zhu, T. Geisler, and B. Edvold, “Colorless coherent receiver using 3x3 coupler hybrids and single-ended detection,” Opt. Express 20(2), 1164–1171 (2012). [CrossRef] [PubMed]
13. A. Moscoso-Mártir, I. Molina-Fernández, and A. Ortega-Monux, “Signal constellation distortion and BER degradation due to hardware impairments in six-port receivers with analog I/Q generation,” Prog. Electromagnetics Res. 121, 225–247 (2011). [CrossRef]
14. I. Fatadin, S. J. Savory, and D. Ives, “Compensation of quadrature imbalance in an optical QPSK coherent receiver,” IEEE Photon. Technol. Lett. 20(20), 1733–1735 (2008). [CrossRef]
15. A. Besse, M. Bachmann, H. Melchior, L. B. Soldano, and M. K. Smit, “Optical bandwidth and fabrication tolerances of multimode interference couplers,” J. Lightwave Technol. 12(6), 1004–1009 (1994). [CrossRef]
16. T. Pfau, S. Hoffmann, and R. Noé, “Hardware-efficient coherent digital receiver concept with feedforward carrier recovery for M-QAM constellations,” J. Lightwave Technol. 27(8), 989–999 (2009). [CrossRef]
