Deep reinforcement learning for proactive spectrum defragmentation in elastic optical networks

Elastic optical network (EON) is a promising candidate for the next generation optical transport network because of its high spectrum efficiency and flexible spectrum allocation. However, to migrate from today’s dense wavelength division multiplexing (DWDM) network to EON, optical nodes such as reconfigurable optical add/drop multiplexers (ROADMs) should be replaced with the ones with expensive flexi-grid wavelength selective switches (WSSs). Thus, whether and how today’s optical network will be evolved to EON is still under debate. Meanwhile, with the technical maturity of optical super-channels that transmit bit rates at the level of 1 Tb/s, super-channels are expected to dominate the future optical transport network because of their attractiveness of high spectrum efficiency and superior cost effectiveness. In this paper, we propose a new quasi-coarse wavelength division multiplexing (Quasi-CWDM) optical network architecture that is compatible to today’s WDM network in spectrum operation, while still maintaining the flexibility of adaptation between the bit rate and the transparent reach of an optical super-channel, but not requiring expensive flexi-grid WSSs. Specifically, the Quasi-CWDM optical network employs large fixed frequency grids, whose spacing falls between the traditional CWDM and DWDM, e.g., 200 GHz or 400 GHz. To maintain the flexibility between the bit rate and the transparent reach of an optical super-channel, different modulation formats are chosen according to channel physical distances. To minimize network hardware cost and maximize carried user traffic demand, we develop a mixed integer linear programming (MILP) model and a heuristic algorithm that can efficiently groom upper-layer IP traffic onto the optical layer by optical super-channels. We compare the design based on the Quasi-CWDM network to those based on the DWDM network and the advanced EON. It is found that the proposed Quasi-CWDM network has a much lower hardware cost than the other two types of networks, while achieving spectrum efficiency close to EON.

An efficient spectrum defragmentation algorithm is proposed in elastic optical networks by properly rerouting existing connections after traffic terminated dynamically to improve spectrum efficiency, and shown to be resource efficient. Keywords-optical network; network optimization; Spectrum Defragmentation

The rapid escalation of Internet traffic and increasingly heterogeneous service demands impose stringent requirements on optical networks for dynamic resource scheduling, efficient spectrum utilization, and automated operation. Elastic optical networks (EONs) are regarded as a promising solution, yet their performance remains constrained by two critical challenges: routing, modulation format, and spectrum allocation (RMSA) and spectrum defragmentation (SD). Existing approaches predominantly focus on optimizing one of these tasks, which may lead to limited adaptability and suboptimal network efficiency. To address this gap, we propose an asynchronous cooperative multi-agent deep reinforcement learning framework, termed MADRL-JRASD, for the joint optimization of RMSA and proactive SD. The framework incorporates an RMSA agent with dynamic allocation capability and an SD agent with autonomous decision-making ability, coordinated through an asynchronous architecture that enables adaptive responses to environmental changes. Invalid-action masking and carefully designed reward functions are further integrated to enhance training stability and convergence. Comprehensive evaluations over three representative topologies demonstrate that MADRL-JRASD reduces the blocking probability by up to 81% compared with RMSA heuristics without SD and achieves an 85% reduction in overhead relative to heuristic algorithms combining RMSA and SD that attain similar blocking performance. Moreover, the sensitivity analysis shows that the SD agent improves spectrum utilization and that multi-agent cooperation enhances global decision coordination, while action masking and reward design jointly strengthen the convergence and efficiency of MADRL-JRASD.

Fragmentation-aware algorithms for multipath routing and spectrum assignment in elastic optical networks

Extending light-trail into elastic optical networks for dynamic traffic grooming

An elastic optical network (EON) plays an important role in transport technology for virtualization of networks. A key aspect of EONs is to establish lightpaths (virtual links) with exactly the amount of spectrum that is needed and with the possibility of grooming, the process of grouping many small traffic flows into larger units, creating a super-lightpath. Grooming eliminates the need for many guard bands between lightpaths and also saves transceivers; however, it often leads to the need to perform optical–electrical–optical conversions to multiple-data-rate optical signals at intermediate nodes. The aim of this paper is to provide a mixed-integer linear programming (MILP) formulation, as well as heuristic and meta-heuristic approaches, for the design of multiple virtual optical networks (VONs) in an elastic optical substrate network with bandwidth-variable lightpaths, modulation format constraints, and virtual elastic regenerator placement. Traffic grooming is allowed inside each VON, and a distance-adaptive modulation format technique is employed to guarantee efficiency in terms of bandwidth for a physical substrate, subject to several virtual topologies. A reduced MILP formulation without grooming capability is also proposed for comparison. The complete MILP formulation jointly solves the virtual topology design, regenerator placement, and grooming problems, as well as the routing, modulation, and spectrum assignment (RMSA) problem. The reduced MILP formulation, heuristics, and meta-heuristic, on the other hand, separate the virtual topology design problem from the RMSA problem. It is shown that the grooming approach can provide good results, since it solves the problem for a complete design when compared to the approach without grooming. Furthermore, heuristic solutions for large networks are proposed, which present good performance (in terms of saving spectrum) for the design with large instances.

This paper focus on the routing, modulation level and spectrum allocation (RMLSA) problem in the next generation datacentre networks, which own the advantages of both Elastic Optical Networks (EON) and Software Defined Networking (SDN). We proposes an Integer Linear Programming (ILP) model for this problem. Simulation results show the ILP model achieves an outstanding load balancing performance.

Objectives: The current technologies of wavelength routing and allocation of available for WDM networks are no longer applicable in Elastic Optical Networks (EON), since all their connections are allocated in a fixed manner, as provided for ITU. Therefore, the topic concerning routing and allocation of resources in EON is relevant given the growth projected in web traffic. This survey includes the evolution, main characteristics, and the process of routing and allocation of wavelengths required by EON. Methodology: The processes of allocation of sub-wavelengths and those of super-wavelengths are presented, as well as the multiple storage for different transmission rates and the principle of spectral contiguity. The process of routing and allocation taking OFMD into account is explained. Findings: In this study, we present a survey with routing and allocation techniques of wavelengths that improve the spectral efficiency of EON. Application/Improvements: This survey provides a wide knowledge on the methods to improve the spectral efficiency in EON. Such efficiency can promote an advance in new routing and allocation techniques for the optical transport networks. Keywords: Elastic Optical Networks, EON, Optical Networks, Routing, WDM

Elastic optical networking (EON) equipped with multiple modulation formats has recently emerged as a promising solution to flexible provisioning in optical networks. To overcome the limit of high-order modulation formats for long distance transmission, traffic can be split and carried by multiple channels, with each modulated at lower rates. However, the flexibility offered by EON and its multi-channel variant comes with a higher cost for extra hardware components (e.g., laser and modulator array). This paper studies the single- and multi- channel EON schemes from a cost effectiveness point of view, with comparison to the conventional 40/100G provisioning scheme based on the ITU-T grid. We first propose and discuss the hardware module implementation of the EON schemes. By decomposing the hardware module into mechanics and electrical, analog/digital conversion, and optics, we propose cost models for each part based on current and historical market data. Two key metrics, hardware cost and traffic-share cost, are calculated and analyzed. We show that the hardware cost based on the proposed model well matches the market trend. Also, the result indicates that the two EON schemes, though are more expensive due to hardware complexity, can effectively reduce the traffic-share cost according to our simulation of a network-wide provisioning scenario. In particular, the multi-channel EON can reduce the traffic-share cost by up to 21.1% from the single-channel EON in the investigated scenarios, achieving the highest cost effectiveness among the three schemes.

Study and Analysis of Routing and Spectrum Allocation (RSA) and Routing, Modulation and Spectrum Allocation (RMSA) Algorithms in Elastic Optical Networks (EONs)

Service differentiation-based spectrum defragmentation in mixed grid optical networks

Over the past decade, we have seen a substantial amount of research on elastic optical networks (EONs), showing great promise towards dramatically improving transmission capacity compared to traditional optical networks. The main concern is that the achievable transmission capacity for a given fiber will reach the physical limit soon in the case of single-core fibers. Multi-core fiber (MCF) has been proposed as an innovative alternative which is based on the concept of space-division multiplexing (SDM). We endeavour to work on a Crosstalk Elimination and Spectrum Defragmentation based Routing and Spectrum Allocation (CESD-RSA) algorithm has been proposed which is based on a predefined metric, which measures the spectrum status in the SDM-EONs. Past research has indicated the dependence of crosstalk effect in MCF on the assignment of signals in the context of the spectrum and the core. This specifically entails a study of the Routing, Core and Spectrum allocation in a flexi-grid multicore network. In the end we shall endeavour to evaluate the proposed method using computer simulations.KeywordsWDMSDMEONMCFRSASF

We consider network link upgrade with ultra-low loss (ULL) fibers in elastic optical backbone networks. Results demonstrate that the strategy which scans all links for ULL fiber deployment provides the best performance.

Distributed sub-light-tree based multicast provisioning with shared protection in elastic optical datacenter networks

Elastic optical networks (EONs) have been receiving extensive attention because of their inherent flexibility and the efficiency with which they allocate fiber-bandwidth. For an EON, survivability is of crucial importance because of the very high bandwidth that it carries on each optical channel. In this article, we review the current state of the art of survivable EONs which reviews the literature to summarize the following aspects: (a) spectrum resource sharing among backup lightpaths, (b) sharing of high-speed optical transponders, (c) effect of spectrum conversion, (d) bandwidth squeezed restoration (BSR), (e) joint restoration by multiple sub-band optical channels, (f) multi-layer survivability, and (g) energy efficiency. Apart from a summary on the current research status, we also discuss open research issues which are important to survivable EONs from the perspectives of (a) impact of spectrum conversion, (b) impact of elastic optical transponder configuration, (c) impact of physical layer impairments and limitations, (d) protection-path-based spectrum defragmentation, and (e) network availability.