3894 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 32, NO. 23, DECEMBER 1, 2014 Trading Regeneration and Spectrum Utilization in Code-Rate Adaptive Flexi-Grid Networks Isabella Cerutti, Francesca Martinelli, Nicola Sambo, Filippo Cugini, and Piero Castoldi, Member, IEEE Abstract—Introduced to improve the spectral efficiency, time- frequency packing technique allows the exploitation of different code rates, leading to different levels of spectral utilization and all-optical reach. To ensure quality of transmission of the opti- cal signal, opto-electronic regeneration must be used to propagate the transmission beyond the maximum optical reach and can in- herently offer conversion of spectrum and code rate. In this way, multiple code rates can be enabled in optical networks, leading to a flexible design in which spectrum utilization and regeneration can be properly optimized and traded. This paper addresses for the first time the joint problem of selecting the code rate, the re- generation nodes, the spectrum allocation and the route for the requested lightpaths in an optical network with a flexi-grid. A ge- netic algorithm is proposed that balances the contrasting objectives of minimizing the regeneration nodes and the spectrum utilization. Results show that when regeneration nodes are minimized, code- rate adaptation is able to reduce the regeneration nodes as well as the spectrum utilization with respect to rate-fixed optical networks. In general, a balance of the two contrasting objectives is preferred to achieve a low resource utilization. Index Terms—Design optimization, optical networks, regenera- tor placement, time frequency packing, wavelength routing. I. INTRODUCTION T HE fast and continuous increase of traffic load in opti- cal networks prompts the support of multiple line rates to enable adaptation to traffic variations [1]–[4] as well as of transmission techniques that improve the spectral efficiency of optical communications [5]–[9]. Examples of spectral efficient solutions are the use of high-order modulation formats (e.g., polarization multiplexing 16 quadrature amplitude modulation — PM-16QAM) or faster-than-Nyquist transmission techniques [5]. In both cases, more bits can be squeezed in a Hz of band- width, allowing for a flexible upgrade planning of the network. Thus, optical networks can be optimized by selecting the best rate and/or modulation format for each requested optical chan- nel or lightpath, leading to a multi rate and multi modulation format design. Manuscript received June 27, 2014; revised August 14, 2014; accepted September 12, 2014. Date of publication September 19, 2014; date of cur- rent version October 16, 2014. This work is partially supported by the EU-FP7 Idealist project and the Tuscany regional project “ARNO T3: AR-chitectures of networks and optical nodes for high-capacity transmission, the access-metro- core transport based integrated photonic technologies.” I. Cerutti, N. Sambo, and P. Castoldi are with the Scuola Superiore Sant’Anna, 56124 Pisa, Italy (e-mail: i.cerutti@sssup.it; nicola.sambo@sssup.it; piero. castoldi@sssup.it) F. Martinelli was with the CNIT, 56124 Pisa, Italy (e-mail: frenci.martinelli@ gmail.com). F. Cugini is with the CNIT, 56124 Pisa, Italy (e-mail: filippo.cugini@cnit.it). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/JLT.2014.2359179 The downside of the improved spectral efficiency (e.g., by using a PM-16QAM instead of a polarization multiplexing quadrature phase shift keying—PM-QPSK) is that a stronger degradation of the optical signal occurs during propagation [10], requiring more frequent signal regeneration to guarantee an ad- equate quality of transmission or QoT (e.g., bit error rate level). More specifically, regeneration is necessary to avoid that the signal propagates farther than the the maximum optical reach, i.e., the maximum propagation distance of the optical signal over which the QoT can be ensured. Thus, it is necessary to prop- erly place the regenerators and select the transmission rate and modulation format for the specific physical characteristics of the propagation path [10]–[12]. However, the support of multi- ple modulation formats requires complex transponders enabled for different formats and has a rather limited variance of the maximum optical reach, reducing the design flexibility [10]. Another technique that has been recently proposed for im- proving the spectral efficiency is the time frequency packing (TFP) [5], [13], [14] of the optical signals. Similar to faster- than-Nyquist, TFP is a transmission technique in which optical pulses can overlap in time or frequency or both to achieve high spectral efficiency, at the cost of intersymbol and inter-carrier interference. Error correction code and the coherent receiver are properly designed to recover for the introduced interference [5]. Thanks to this, TFP achieves very high spectral efficiency (e.g., a super channel at 1 Tb/s in 200 GHz) and can be used for all-optical transmissions over long distances (e.g., as long as 5000 km [14]). Moreover, by enabling multiple code rates (through software), it is possible to provide different levels of robustness, leading to different maximum optical reaches. As in the other spectral-efficient techniques, a trade-off between spec- tral efficiency and the maximum optical reach exists and can be exploited for a flexible and optimized planning of the network. When high spectral efficiency is sought, low code redundancy (achieving relative short reach) may be used requiring a cer- tain number of regenerations. Alternatively, regeneration points (thus, costs related to regeneration) could be reduced by increas- ing the code redundancy (i.e., lower code rate), at the expense of a larger bandwidth. Notice that when considering 3R (opto- electronic) regeneration, conversion of spectrum and code rate can be offered for free. When applied to flexi-grid optical networks [10], [11], [15], planning optimization with code-rate adaptation (RA) requires strategies and algorithms for selecting the most suitable code rate as well as the most suitable regeneration nodes. In addition, routing and spectrum assignment should be solved jointly. The regenerator placement problem has been well studied in the past in conventional wavelength-switched optical networks: linear 0733-8724 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications standards/publications/rights/index.html for more information.