Resource-aware provisioning strategies in translucent elastic optical networks
Introduction
Wavelength division multiplexing (WDM) and dense wavelength division multiplexing (DWDM) networks base their operation on a fixed 100/50 GHz grid [1]. Even if these networks can transport large bandwidths they become inefficient in presence of heterogeneous and variable traffic demands. The elastic optical network (EON) architecture [2], [3], proposed almost a decade ago and extensively studied since then [4], enables a more efficient use of spectrum resources by means of a flexible grid structure. In this architecture, demands are provisioned using multiple adjacent 12.5 GHz frequency slots. The amount of slots required by a single demand depends on the demanded bitrate, the transmission length, the modulation format, the baud rate, the number of carriers and the error correction overhead. Bandwidth-variable transponders (BVTs) can adjust the baud rate, the modulation format, the number of carriers and the error correction overhead to transport a demand with a specified bitrate minimizing the required frequency slots. This flexibility enables EONs to establish optical connections with “arbitrary” bandwidths (frequency slots) [4], [5].
On the other hand, regeneration has been widely used in optical networks to reduce the blocking probability when provisioning traffic demands. In traditional WDM networks, regeneration has been mainly proposed to overcome physical impairments that limit the maximum transmission distance [6]. Besides, regeneration can also be used to reduce blocking by providing wavelength conversion capabilities that can tackle spectrum fragmentation issues. However, regeneration in traditional WDM networks has no impact on the spectrum utilization because each lightpath exploits either a 50 GHz or a 100 GHz channel regardless of the path length or assigned wavelength. Instead, regeneration in EONs can also be used to compress the amount of spectrum required for provisioning a demand. To this end, a flexible regenerator composed of back-to-back bandwidth-variable transponders [7], [8], [9] can be used, where each transponder handles the same rate but not necessarily requires the same amount of frequency slots. These regenerators can be configured to support, independently on each carrier, a specific modulation format and baud-rate. The shorter the distance between regeneration points, the higher the modulation format that can be used, leading to an increase in spectral efficiency and spectrum saving. Flexible regeneration enables a trade-off between spectrum and transponder costs, as described in [10], [11].
The use of regeneration and its impact on blocking ratio has been widely studied in the past. Nevertheless, recently it has gained relevance mainly due to the trade-off between regeneration and spectrum use in EONs that we described in the previous paragraph. In recent years, the effort have focused on obtaining provisioning algorithms that take advantage of this trade-off [8], [12], [13], [14], [15], [16]. In this context, regenerators can be considered either as a cost to be minimized or as a resource to be smartly used. In the first case, spectrum is considered as the limiting resource and the use of regeneration to reduce the blocking probability translates to node cost increase. In the second scenario, regenerators are bounded to the amount of transponders that each node is equipped with, at network design phase. The goal becomes to efficiently use available transponders. If spectrum slot lack becomes the main bottleneck when accepting traffic demands, transponders can be used to save spectrum to reduce the demand blocking probability. In operational networks, regeneration capabilities can be increased much faster and smoothly by equipping nodes with more transponders, while increasing spectrum resources is typically slower and may require a huge investment due to fiber cable deployments or renting costs.
In this paper we evaluate different provisioning strategies that tackle both unbounded and bounded regeneration scenarios to illustrate the value of regeneration in EONs. We name these strategies as resource-aware strategies, because they aim at selecting the best combination of available spectrum and transponder resources for establishing a translucent lightpath. Based on our previous contribution [14], originally designed for the full regeneration capacity scenario, we propose a novel strategy, suited to both scenarios. The strategy differs from others proposed in the literature, which are typically agnostic to available resources. We demonstrate that resource awareness helps in decreasing the blocking probability and increases the actual network capacity. thanks to a more efficient assignment of both spectrum and transponder resources. The paper is organized as follows: in Section 2 previous research activities are summarized and compared. In Section 3 we analyze the problem of selecting an optimal provisioning candidate in Elastic Optical Networks. We present the used system model in Section 4. Then we introduce the proposed resource-aware algorithms in Section 5. Finally, in Section 6 we show simulation results and in Section 7 we derive the conclusions.
Section snippets
Related work
Regeneration in optical networks has been widely studied in terms of Regeneration Placement (RP), where regeneration is mainly used to compensate for transmission length and/or spectrum fragmentation issues that preclude lightpath establishment on a given path. [17], [18]. Thus, the use of regeneration focus on reach extension and/or wavelength conversion as a mean to reduce the lightpath blocking probability. This problem has also been studied for EONs in [19], [20], [21] focusing on selecting
Optimal provisioning candidates
Multiple candidate solutions exist for provisioning a traffic demand over a path . These solutions range from the transparent case (no regeneration in any node) to the opaque one (regeneration at each intermediate node), including all translucent solutions which implement regeneration at some intermediate nodes.
Each candidate solution can have different requirements in terms of network resources. The two main resources of an EON are spectrum and transponders. In this work, we focus on the
System model
To determine all possible solutions in for provisioning a traffic demand over a path , we introduce the main assumptions and the necessary background. In particular, we discuss how to determine for each the associated cost in terms of spectrum when additional transponders are used for regeneration purposes. We briefly describe the general network model as well as the ROADM one, and then provide a more detailed analysis of the transponder model which has a direct impact on
Candidate selection
Given an optimal set of candidate solutions for provisioning a traffic demand over path , one of them needs to be selected. As earlier discussed, we argue that this decision needs to be aware of the available resources along path . In this section, we describe two different strategies that can be used for this purpose.
Results
In this section we analyze simulation results to evaluate the proposed strategies and compare them with existing ones. Our main goal is to demonstrate the value of strategies that are aware of available resources. First, we consider those strategies that assume full regeneration capacity, where nodes are equipped with a large amount of transponders. Hence, these strategies aim at minimizing the required spectrum with the minimum regeneration cost. Next, we consider those strategies that are
Conclusions
In this work we analyzed the trade-off between regeneration and spectrum costs when provisioning lightpaths in translucent EONs. Existing strategies aimed at minimizing either one of these costs, instead we focused on a joint optimization. To this end, we modeled the problem as a two-dimensional resource assignment problem and proposed two different provisioning strategies. The first one, named threshold-aware (TA), is well suited for scenarios with full regeneration capacity. The second one,
CRediT authorship contribution statement
Nehuen Gonzalez-Montoro: Software, Simulation, Validation, Investigation, Data curation, Writing - original draft, Visualization. Jorge M. Finochietto: Conceptualization, Resources, Supervision, Visualization, Writing - original draft, Writing - review & editing. Andrea Bianco: Methodology, Supervision, Visualization, Writing - review & editing.
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.
References (30)
- et al.
A survey on physical layer impairments aware routing and wavelength assignment algorithms in optical networks
Comput. Netw.
(2009) - et al.
Manycast routing, modulation level and spectrum assignment over elastic optical networks
Opt. Fiber Technol., Mater. Devices Syst.
(2017) - ITU-T Recommendation “G.694.1 : Spectral grids for WDM applications: DWDM frequency grid”,...
- et al.
Spectrum-efficient and scalable elastic optical path network: architecture, benefits, and enabling technologies
IEEE Commun. Mag.
(2009) - et al.
Elastic optical networking: A new dawn for the optical layer?
IEEE Commun. Mag.
(2012) - et al.
Routing and spectrum allocation in elastic optical networks: A tutorial
IEEE Commun. Surv. Tutor.
(2015) - et al.
A tutorial on the flexible optical networking paradigm: State of the art, trends, and research challenges
Proc. IEEE
(2014) Optical Network Design and Planning
(2014)- et al.
On performance gains of flexible regeneration and modulation conversion in translucent elastic optical networks with superchannel transmission
J. Lightwave Technol.
(2016) - et al.
Performance analysis of flexible regeneration and modulation conversion in elastic optical networks