Fair QoS-Aware Adaptive Routing and Wavelength Assignment in All-Optical Networks

Compressed network monitoring for ip and all-optical networks

“…We simulate the operation of an all-optical network where BERs are computed according to the model described in Section 6.1. Physical-layer parameters for the network are described in [15]. We simulated the arrival and departure of 350 calls in a downscaled version 2 of the NSF network, depicted in Fig.…”

“…Assuming circuit-switched networks with no wavelength conversion 1 , a RWA algorithm chooses a route and a wavelength (the combination of which is called "lightpath" [4]) to accommodate each incoming call at call admission time. It is possible to increase the quality of transmission in optical networks by using appropriate RWA techniques [15,16]. In this paper, we make no assumption regarding the particular RWA used in the network.…”

“…Optical amplifiers inject amplifier spontaneous emission (ASE) noise, and the signal is also subject to node crosstalk, which refers to signal leakages caused, for instance, by imperfect filtering at the nodes [12]. We refer the reader to [23], [22], [9], [15] for more details regarding the models of ASE noise, nonlinear crosstalk, node crosstalk, and their combined effects, respectively. Note that we ignore here a number of physical impairments such as receiver noise and polarization mode dispersion, but these effects can be incorporated easily in our model as additional noise variances, as will be seen shortly.…”

Reinforcement learning based routing in all-optical networks with physical impairments

“…Indeed, crosstalk is generated by the interaction of several lightpaths. In this work, we use the physical layer model we presented in [1]. In particular, we refer the reader to [1] for efficient techniques that relate the impact of ISI, ASE noise, interchannel and node crosstalk to µ 0 , µ 1 , σ 0 , σ 1 and hence BER.…”

“…However, due to the absence of electronic regeneration, signals are subject to accumulating physical impairments during propagation and perfect physical layer is no longer a valid assumption -measuring Quality of Transmission (QoT) through bit-error rates (BER), this means that lightpaths' BERs (a lightpath is a combination of a route and a wavelength) can reach values beyond acceptable values as set by the network manager. Routing and Wavelength Assignment has emerged as a cross-layer technique to route calls in the network over lightpaths accounting for physical impairments, in order to decrease the probability a call is blocked because of the lack of available resources or because establishing the lightpath would cause this lightpath's or another lightpath's QoT beyond the acceptable threshold [1]. In this paper, we propose to use a reinforcement learning technique in a distributed fashion, to choose a tentative lightpath among a set of alternates based only on the past events seen locally.…”