Reinforcement Learning for Generalized Parameter Optimization in Elastic Optical Networks

Koch, Rebekka; Kuhl, Sebastian; Morais, Rui Manuel; Spinnler, Bernhard; Schairer, Wolfgang; Sommerkorn-Krombholz, Bernd; Pachnicke, Stephan

doi:10.1109/jlt.2021.3123271

Cited by 12 publications

(3 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It allows the agent to continuously improve its performance by self-comparison. Koch et al [27] adopted the RL algorithm for parameter optimization in EONs. In addition, a cost-efficient routing, modulation, wavelength, and port assignment algorithm based on DRL was developed in [28].…”

Section: Deep Reinforcement Learning In Rmsa Of Eonsmentioning

confidence: 99%

Deep Reinforcement Learning-Based RMSA Policy Distillation for Elastic Optical Networks

et al. 2022

View full text Add to dashboard Cite

The reinforcement learning-based routing, modulation, and spectrum assignment has been regarded as an emerging paradigm for resource allocation in the elastic optical networks. One limitation is that the learning process is highly dependent on the training environment, such as the traffic pattern or the optical network topology. Therefore, re-training is required in case of network topology or traffic pattern variations, which consumes a great amount of computation power and time. To ease the requirement of re-training, we propose a policy distillation scheme, which distills knowledge from a well-trained teacher model and then transfers the knowledge to the to-be-trained student model, so that the training of the latter can be accelerated. Specifically, the teacher model is trained for one training environment (e.g., the topology and traffic pattern) and the student model is for another training environment. The simulation results indicate that our proposed method can effectively speed up the training process of the student model, and it even leads to a lower blocking probability, compared with the case that the student model is trained without knowledge distillation.

show abstract

Section: Deep Reinforcement Learning In Rmsa Of Eonsmentioning

confidence: 99%

Deep Reinforcement Learning-Based RMSA Policy Distillation for Elastic Optical Networks

et al. 2022

View full text Add to dashboard Cite

show abstract

“…To handle the ever-increasing network traffic demands, novel technologies are desired to update optical networks to improve spectrum efficiency and transmission capacity. With the characteristic of fine-grained BW allocation, elastic optical networks (EONs) allow efficient spectrum utilization [3,4] . In addition, space division multiplexing (SDM) is a technology that exploits the core or mode as an independent data channel to further expand capacity [5,6] .…”

Section: Introductionmentioning

confidence: 99%

Experimental demonstration of a flexible-grid 1 × (2 × 3) mode- and wavelength-selective switch using silicon microring resonators and counter-tapered couplers

Kong,

Lu,

Wang

et al. 2024

中国光学快报

View full text Add to dashboard Cite

A flexible-grid 1 × 2 × 3 mode-and wavelength-selective switch which comprises counter-tapered couplers and silicon microring resonators has been proposed, optimized, and demonstrated experimentally in this work. By carefully thermally tuning phase shifters and silicon microring resonators, mode and wavelength signals can be independently and flexibly conveyed to any one of the output ports, and different bandwidths can be generated as desired. The particle swarm optimization algorithm and finite difference time-domain method are employed to optimize structural parameters of the twomode (de)multiplexer and crossing waveguide. The bandwidth-tunable wavelength-selective optical router composed of 12 microring resonators is studied by taking advantage of the transfer matrix method. Measurement results show that, for the fabricated module, cross talk less than −10.18 dB, an extinction ratio larger than 17.41 dB, an in-band ripple lower than 0.79 dB, and a 3-dB bandwidth changing from 0.38 to 1.05 nm are obtained, as the wavelength-channel spacing is 0.40 nm. The corresponding response time is measured to be 13.64 μs.

show abstract

“…However, accurate representation of EDFAs using analytical models is challenging. Machine learning (ML) [4]- [6] and data aided EDFA models [7]- [12] are alternatives that gather attention from the community. Training an ML model for an EDFA and other link components requires full access to its input and output in a lab environment.…”

Section: Introductionmentioning

confidence: 99%

Spectral Power Profile Optimization of Field-Deployed WDM Network by Remote Link Modeling

Jones¹,

Bottrill²,

Taengnoi³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

Reinforcement Learning for Generalized Parameter Optimization in Elastic Optical Networks

Cited by 12 publications

References 15 publications

Deep Reinforcement Learning-Based RMSA Policy Distillation for Elastic Optical Networks

Deep Reinforcement Learning-Based RMSA Policy Distillation for Elastic Optical Networks

Experimental demonstration of a flexible-grid 1 × (2 × 3) mode- and wavelength-selective switch using silicon microring resonators and counter-tapered couplers

Spectral Power Profile Optimization of Field-Deployed WDM Network by Remote Link Modeling

Contact Info

Product

Resources

About