Elsevier

Optics Communications

Volume 475, 15 November 2020, 126212
Optics Communications

Joint channel power and amplifier gain optimization in coherent DWDM systems

https://doi.org/10.1016/j.optcom.2020.126212Get rights and content

Highlights

  • Model presented in this paper is the first joint optimization of launch powers and amplifier gains.

  • The Physical layer impairments are taken into account using the GN model.

  • Model could be utilized in network planning problems such as power allocation, routing, and spectrum assignment.

Abstract

In this paper, the channel power and amplifier gain of coherent dense wavelength division multiplexing (DWDM) fiber-optic communication systems are jointly optimized. The objective is to maximize the minimum margin of signal to noise ratio (SNR). The presented optimization problem is non-linear and non-convex, in which the intra- and inter-channel nonlinear interference (NLI) noise of fiber is estimated by using the so-called Gaussian noise (GN) model. Furthermore, the noise and other practical aspects of optical amplifiers such as output power saturation, maximum gain limitation, gain tilt and ripple, and noise figure non-flatness are modeled. Finally, we show that this problem can be reformulated as a convex optimization. Our results reveal that joint power and amplifier gain optimization in a 5-span fiber link with span length of 80 km (100 km) leads to 1 dB (0.47 dB) higher SNR than the conventional power optimization method in which amplifier gain is fixed and equals to span loss. By considering gain ripple and tilt in a 8-span fiber link with span length of 80 km, we observed that our algorithm has 0.72 dB and 1.86 dB higher SNR than the alternative power optimization methods in which amplifier gain is equalized and unequalized, respectively.

Introduction

Optical networks have formed the backbone of the internet. Enormous growth in the demand for high data rate has led to the capacity crunch [1]. To tackle this capacity shortage, power optimization subject to maximizing the achievable rate [2], [3], space-division multiplexing [4], [5], [6], and elastic optical networks can be considered as the most promising solutions. Elastic optical networks bring some obvious benefits such as spectral- and energy-efficient traffic grooming [7]. Spectral efficiency is also improved by using advanced modulation formats, e.g., polarization multiplexed M-ary quadrature amplitude modulation (PM-M-QAM), for M=4, 8, 16, 32, 64, 128, and 256. Recently, the de-aggregation of 16-QAM to 4-level pulse amplitude modulation was experimentally performed in [8], which can be used to optimize spectral usage. The most overriding factor that restricts the capacity of the optical networks is Kerr nonlinearity [9].

Finding efficient ways to predict the performance degradation due to the Kerr nonlinearity is then of key importance for modern design of dense wavelength division multiplexing (DWDM) optical networks. Brute-force numerical approaches such as the split-step Fourier method to solve the nonlinear Schrödinger equation are not a viable option due to the high computational complexity caused by the wide transmission bandwidth considered. On the other hand, many approximated analytical models for nonlinear fiber propagation in the time and frequency domain are currently available in the literature [10], [11], [12], [13], [14], [15]. Such analytical models are key enablers of physical layer network optimization [16], [17], [18], [19], [20], [21], [22]. The authors of [17] proposed joint power allocation, routing, and wavelength assignment using the enhanced Gaussian noise model [14]. Among the others, the Gaussian noise (GN) model [12], [23] have risen to popularity due to their broad applications and relative simplicity.

Exploiting the GN models has opened up a new horizon in optimizing routing, wavelength assignment, modulation level, channel coding, and power allocation in the planning of coherent polarization-multiplexed optical networks [18], [24], [25], [26], to maximize the minimum received SNR or the network’s total achievable rate. It has been shown that the remarkable saving in capital expenditure can be obtained by power optimization for different modulation formats [27] and matching transmission equipment to network demand [28]. Using the GN model, resource allocation is optimized in elastic optical networks with the objective of minimizing the maximum bandwidth usage [29]. The same authors jointly optimize the routing, spectrum, modulation format, and power spectral density per connection in elastic optical networks [30]. Power optimization is also performed considering either a single flat (equal) launch power over all channels or individual different power values for different channels in [18], where maximizing the achievable rate and minimum margin of SNR is the objective function of the optimization. In [31], the limitation of electrical power available to the undersea optical amplifiers has been modeled in the channel powers optimization in long-haul coherent DWDM systems. The authors of [31] not only imposed the undersea power constraint and allowed gains to vary, but also accurately modeled amplifier physics including power saturation

In all the aforementioned studies (except [31]) the amplifier gains are fixed and set based on the loss of fiber spans. In addition, the power saturation of optical amplifier has not been modeled. In this paper, we propose to jointly optimize channel powers and amplifier gains, and also include the power saturation, gain tilt and ripple, and noise figure non-flatness of optical amplifiers in the optimization problem. Furthermore, in our system model a fiber link with multiple different spans (i.e., spans with different lengths or fiber types) is considered, which is more practical. We will show that the joint optimization of channel powers and amplifier gains results in higher SNR, compared with the conventional channel power optimization methods.

The rest of the paper is organized as follows. In Section 2, the system model is described. The SNR estimation based on the GN model is presented in Section 3. The proposed optimization problem is formulated in Section 4. Numerical results are presented in Section 5, and the paper is concluded in Section 6.

Section snippets

System model

We consider a long-haul fiber-optic communication system based on coherent DWDM technology, in which N Nyquist rectangular spectral shape channels with bandwidth Δf=R are transmitted over Ns fiber spans. The lengths of spans can be different, where Ls (km) indicates the length of span s, and the total length (L) of fiber link is s=1NsLs. Let Pk,s, R, and fk denote the power at the input of sth span, symbol rate, and central frequency of kth channel, respectively, and Ptot,s indicates the total

Estimating the power of NLI noise

We utilize incoherent approximation of the GN model, referred to as incoherent GN (IGN) [32], to estimate NLI noise imposed by Kerr effect. In the GN model, four-wave mixing (FWM) among all DWDM channels are categorized into three types [33]: (i) self-channel interference (SCI), (ii) cross-channel interference (XCI), and (iii) multi-channel interference (MCI), where SCI is intra-channel NLI, and XCI and MCI are inter-channel NLIs. The total power of NLI noise (PNLI) is obtained by adding the

Problem formulation

In this section, we formulate the power and gain optimization problem with objective of maximizing minimum SNR margin, where SNR margin indicates the difference of SNR and required SNR (SNRreq) in logarithmic representation.

In the former section, we showed that PNLIk is a function of input power of all channels and gains of amplifiers, and PASEk is a function of gains of amplifiers. Accordingly, SNRk is a function of input powers and gains of amplifiers. Let X indicate a N×Ns matrix which its xk

Numerical results

We solve the proposed optimization problem, referred to as joint power and gain optimization (JPGO), and compare it with its conventional alternative [3], referred to as optimized power and fixed gain (OP-FG). In OP-FG, amplifier gains are set based on the span loss, Gi=10Lsiα10, and channel powers are obtained by solving the optimization problem proposed in [3].

In all numerical results, unless specified, we assume that N=10, Ns=8, Ls=80 km, Δf=50 GHz, PsatEDFA=25 dBm, GmaxEDFA=30 dB, GBA=20

Conclusion

In this paper, we formulated the problem of joint channel powers and amplifier gains optimization in coherent DWDM systems as a convex optimization. Furthermore, we optimized a fiber link with multiple different spans and modeled the practical limitations of optical amplifiers such as output power saturation, maximum gain limitation, gain tilt and ripple, and noise figure non-flatness. We showed that in a 5-span fiber link with span length of 80 km (100 km) the proposed method leads to 1 dB

CRediT authorship contribution statement

R. Hashemi: Conceptualization, Methodology, Software, Writing - some part of original draft. H. Beyranvand: Conceptualization, Methodology, Writing - some part of original draft. H. Rabbani: Data curation, Methodology, Writing - some part of original draft.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgment

This work was supported in part by Iran National Science Foundation (INSF) .

References (40)

  • FallahpourA. et al.

    Experimental demonstration of tunable de-aggregation from 16-QAM to 4-PAM for two wavelength multiplexed channels using wave mixing in a single nonlinear element to map constellation onto axes

    Opt. Commun.

    (2019)
  • A. Chralyvy, Plenary paper: The coming capacity crunch, in: Proc. European Conf. and Exhibition on Optic. Commun.,...
  • RabbaniH. et al.

    Improving the achievable rates of optical coherent transmission with back-propagation

    (2018)
  • RobertsI. et al.

    Convex channel power optimization in nonlinear WDM systems using Gaussian noise model

    J. Lightwave Technol.

    (2016)
  • Yaghubi-NamaadM. et al.

    Adaptive modulation and flexible resource allocation in space-division- multiplexed elastic optical networks

    IEEE/OSA J. Opt. Commun. Networking

    (2018)
  • ArpanaeiF. et al.

    Three-dimensional resource allocation in space division multiplexing elastic optical networks

    J. Opt. Commun. Netw.

    (2018)
  • MoghaddamE.E. et al.

    Crosstalk-aware resource allocation in survivable space-division-multiplexed elastic optical networks supporting hybrid dedicated and shared path protection

    J. Lightwave Technol.

    (2020)
  • JinnoM.

    Elastic optical networking: Roles and benefits in beyond 100-Gb/s era

    J. Lightwave Technol.

    (2016)
  • R.-J. Essiambre, G. Kramer, P.J. Winzer, G.J. Foschini, B. Goebel, Capacity limits of optical fiber networks 28 (4)...
  • A. Carena, V. Curri, G. Bosco, P. Poggiolini, F. Forghieri, Modeling of the impact of nonlinear propagation effects in...
  • MecozziA. et al.

    Polarization scattering by intra-channel collisions

    Opt. Express

    (2012)
  • P. Johannisson, M. Karlsson, Perturbation analysis of nonlinear propagation in a strongly dispersive optical...
  • DarR. et al.

    Properties of nonlinear noise in long, dispersion-uncompensated fiber links

    Opt. Express

    (2013)
  • CarenaA. et al.

    EGN model of non-linear fiber propagation

    Opt. Express

    (2014)
  • RabbaniH. et al.

    A general analytical model of nonlinear fiber propagation in the presence of Kerr nonlinearity and stimulated Raman scattering

    (2019)
  • H. Beyranvand, J.A. Salehi, A quality-of-transmission aware dynamic routing and spectrum assignment scheme for future...
  • H. Rabbani, L. Beygi, S. Ghoshooni, H. Rabbani, E. Agrell, Quality of transmission aware optical networking using...
  • J. Ives, P. Bayvel, J. Savory, Adapting transmitter power and modulation format to improve optical network performance...
  • BeyranvandH. et al.

    Performance analysis of equal-energy two-level OCDMA system using generalized optical orthogonal codes

    J. Lightwave Technol.

    (2013)
  • MoghaddamE.E. et al.

    Routing, spectrum and modulation level assignment, and scheduling in survivable elastic optical networks supporting multi-class traffic

    J. Lightwave Technol.

    (2018)
  • Cited by (13)

    • Active learning for OPM in FMF systems

      2023, Physical Communication
    • Improving MDM–WDM optical network performance via optimized power allocation using Gaussian noise model

      2023, Optical Fiber Technology
      Citation Excerpt :

      Thereby, to have a better discussion and conclusion, we investigate both FG and NFG scenarios. The employment of a discrete data-rates set that falls below and above the highest feasible continuous communication data rates, as well as the integration of different communication hardware generations, are all practical aspects of using different SNR requirements [32–55]. M.A. Amirabadi: Conception and design of study, Writing – original draft, Writing – review & editing.

    • Research on optimal design of ultra-bandwidth multi-channel filter based on Fabry-Perot structure

      2021, Results in Optics
      Citation Excerpt :

      Wavelength Division Multiplexing (WDM) technology can meet this requirement well. As a key component of WDM technology, the function of the combiner /demultiplexer is to multiplex multiple optical carriers with closely spaced frequencies to a single fiber for transmission or vice versa (Hashemi et al., 2020; Li et al., 2017); mainly including arrayed waveguides (Zhao et al., 2020; Zheng et al., 2020; Juhari et al., 2017; Lee et al., 2020), fiber gratings (Sayed et al., 2020; Tian et al., 2020; Sultana et al., 2020), and dielectric film filters (Admassu et al., 2020; Zhang et al., 2020; Yepuri et al., 2020); etc. Singh et al. (2020) have designed a multi-channel photonic band gap for multi-channel optical filters for future optical devices.

    View all citing articles on Scopus
    View full text