Helder A. Pereira

OSNR model to consider physical layer impairments in transparent optical networks

We propose a model that considers several physical impairments in all-optical networks based on optical signal-to-noise degradation. Our model considers the gain saturation effect and amplified spontaneous emission depletion in optical amplifiers, coherent crosstalk in optical switches, and four-wave mixing in transmission fibers. We apply our model to investigate the impact of different physical impairments on the performance of all-optical networks. The simulation results show the impact of each impairment on network performance in terms of blocking probability as a function of device parameters. We also apply the model as a metric for impairment-constraint routing in all-optical networks. We show that our proposed routing and wavelength assignment algorithm outperforms two common approaches.

Wavelength Assignment for Physical-Layer-Impaired Optical Networks Using Evolutionary Computation

This paper presents a wavelength assignment algorithm suitable for optical networks mainly impaired by physical layer effects, named the Intelligent Wavelength Assignment algorithm (iWA). The main idea is to determine the wavelength activation order for a first-fit algorithm that balances the impact of the physical layer effects by using a training algorithm inspired by evolutionary concepts. The iWA presents some recently proposed concepts in intelligent optimization algorithms, such as an external archive to store the best solutions. Some different physical layer effects, such as four-wave mixing and residual dispersion, were considered in the tests of our proposal. We tested our proposal for transparent optical networks. However, we believe iWA can be used in other types of network, such as opaque networks and translucent networks. The proposed wavelength assignment algorithm was compared with five other wavelength assignment algorithms for two network topologies in three different scenarios. The iWA algorithm outperformed the other WA algorithms in most cases. The robustness of our proposed algorithm to the load distribution changes was also analyzed.

Impact of physical layer impairments in all-optical networks

We investigate the impact of different physical impairments in the performance of all-optical networks. The physical impairments considered are the gain saturation and amplified spontaneous emission noise depletion in amplifiers, homodyne crosstalk in optical switches, four wave mixing and polarization mode dispersion in the transmission fibers. Our simulation results are presented in terms of blocking probability of requested calls as a function of network device parameters.

Multiobjective sparse regeneration placement algorithm in optical networks considering network performance and CAPEX

In this paper we propose the use a multiobjective evolutionary optimization algorithm to solve the regenerators placement problem. The optical network performance and the capital costs are used as optimization objectives. The blocking probability is used as the network performance indicator, which is evaluated by network simulations with an impairment aware routing and wavelength assignment algorithm, whereas the total regenerators cost is the total number of regenerators deployed in the network. Our algorithm finds which network nodes should have regeneration capability and the number of regenerators in each of these regenerating nodes. Our proposal outperformed other well known algorithms found in the literature such as NDF and SQP.

Novel strategies for sparse regenerator placement in translucent optical networks

In this paper, we propose two strategies for sparse regenerator placement (RP) in translucent optical networks, named most used regenerator placement (MU-RP), and most simultaneous used regenerator placement (MSU-RP). Our proposals are compared to well known RP algorithms presented in literature for two different network topologies for different network loads, distribution of load along the networks and number of translucent nodes. MSU-RP presented remarkable results and outperformed all previous approaches in all cases, while MSU-RP obtained a slightly superior or similar performance when compared to previous approaches presented in the literature.

Sparse regeneration placement for translucent optical networks using multiobjective evolutionary algorithms considering quality of service and capital cost

In this paper we propose the use a multiobjective evolutionary optimization algorithm to solve the regenerators placement problem (MORP). Network performance and capital costs are used as optimization objectives. We consider that optical signals are regenerated using a network element that performs 3R (re-amplifying, re-shaping and re-timing) in the electrical domain. Two algorithms for regenerator placement are proposed: multiobjective regenerator placement-not constricted (MORP-NC) and multiobjective regenerator placement-constricted (MORP-C). They are compared through the computational simulation.

Impairment aware wavelength assignment for all-optical networks based on evolutionary computation

This paper presents a wavelength assignment algorithm for all optical networks. The main idea is to determine the wavelength assignment order for a first fit algorithm taking into account the physical layer effects, such as four wave mixing and residual dispersion, by using a training algorithm inspired in evolutionary computation concepts. The intelligent algorithm uses an external archive to store the best solutions. The proposed wavelength assignment algorithm was compared with others classical algorithms such as regular first fit, best fit and random. In all the cases our proposal presented a better performance.

An adaptive path restoration algorithm based on power series routing for all-optical networks

In this paper we present a novel path restoration algorithm, named PSR-R. Our proposal is based on the power series routing (PSR) algorithm, which was proposed by Chaves et al. in 2011 for routing. The cost function for each link is given by a function expanded in a power series, considering as input variables the wavelength availability and the link length. The coefficients of the expanded function are determined by a global optimizer, which tries to minimize the rate of unsuccessful failure recovery (RUFR). We assessed our proposal by performing simulations of single link failures in a well-known network topology. The analytical model used to evaluate physical layer takes into account the following impairments: gain saturation, ASE noise depletion in the optical amplifiers, crosstalk in the optical switches, and polarization mode dispersion and residual chromatic dispersion in optical fibers. We compared our proposal with other approaches in different scenarios as a function of the network load. Our proposal outperformed all other approaches in terms of RUFR.

Applications of computational intelligence in optical networks

Computational intelligence techniques have appeared as efficient alternatives to solve complex problems, such as in multimodal and multidimensional spaces and optimization of multi-objective problems. These approaches are generally bio-inspired and many algorithms have been proposed in recent years. In this last decade, some efforts have been carried in order to demonstrate that these techniques can be useful to tackle tough problems in optical networks. This paper presents a review of some applications of computational intelligence in optical networking, such as in the impairment-aware routing and wavelength assignment, regenerator placement and physical topology design.

Noise Penalties Modeling for the Performance Evaluation of All-Optical Networks

We propose a model to consider several physical impairments in optical networks based on the OSNR. Our model considers the dependence of gain and amplified spontaneous emission depletion in optical amplifiers, homodyne crosstalk in optical switches, and four wave mixing in the transmission fibers. We present an application of our model for the evaluation of network performance in terms of blocking probability using a routing and wavelength assignment algorithm. Our results show the impact of each impairment on network performance as a function of device parameters.