This article discusses the routing and wavelength assignment (RWA) problem in optical networks employing wavelenength division multiplexing (WDM) technology. Two variants of the problem are studied: static RWA, whereby the tra#c requirements are known in advance, and dynamic RWA in which connection requests arrive in some random fashion. Both point-topoint and multicast tra#c demands are considered.

Abstract. In this half-day tutorial, we present the current state-of-theart in optical networks. We begin by discussing the various optical devices used in optical networks. Then, we present wavelength-routed networks, which is currently the dominant architecture for optical networks. We discuss wavelength allocation policies, calculation of call blocking probabilities, and network optimization techniques. Subsequently, we focus on the various protocols that have been proposed for wavelength-routed networks. Specifically, we present a framework for IP over optical networks, MPLS, LDP, CR-LDP, and GMPLS. Next, we discuss optical packet switching and optical burst switching, two new emerging and highly promising technologies. 1

Different integer linear programming (ILP) formulations have been proposed for the routing and wavelength assignment problem in WDM optical networks, mainly for asymmetrical systems, more than for symmetrical systems, under different objectives. We propose a synthesis of the mathematical models for symmetrical systems, with a unified and simplified notation for four widely used objectives. As for asymmetrical traffic models (Jaumard, Meyer and Thiongane, 2004), we show that all formulations, both link and path formulations, are equivalent in terms of the bound values provided by the optimal solution of their linear programming relaxation, although their number of variables and constraints differ. We propose an experimental comparison of the linear relaxation bounds with the optimal ILP solutions whenever it is possible, for several network and

ABSTRACT In an IP over WDM network architecture, IP traffic is carried over a virtual topology (VT), composed of optical transparent channels called lightpaths. We propose an energy-efficient dynamic VT adaptation method with sleep mode, which allows changing only one lightpath connectivity at a time by dynamically monitoring the IP packet traffic load on each lightpath. In order to assure the connectivity of future traffic increase a high-load threshold is used. When the packet traffic load on a specific lightpath becomes higher than a threshold, a new lightpath is added to the virtual topology. However this congestion avoidance policy increases the power consumption by activating transponders in the network. Therefore at the same time lightly loaded lightpaths are eliminated in the virtual topology by using a low-load threshold. Energy-efficient virtual topology adaptation is achieved by traffic offloading considering both lightly loaded and heavily loaded links in the network. In this regard, high-and low-load threshold values need to be carefully determined in order to both gain power and ensure load balancing with keeping the stability of virtual connectivity at a reasonable level. We have analysed the power consumption of the network during 48-hours for different values of low-and high-load thresholds together with the impact of different threshold values on the stability of the virtual topology. We have shown that there is a trade-off between number of changes in the virtual topology and the energy-efficiency. Experimental results demonstrate that a good compromise can be achieved by adjusting the high-and low-load thresholds carefully. Keywords: Energy-efficiency, IP over WDM networks, optical WDM networks, virtual topology reconfiguration, dynamic traffic. INTRODUCTION Internet has become a significant part of our daily life leading to a rapid increase in traffic rates. Power consumption per user is continuously increasing although power consumption per bit is decreasing with the evolution of power efficient devices in telecommunication networks. In order to meet the requirements of this increasing traffic demands, today's optical wavelength division multiplexing (WDM) networks as a backbone transmission system, is designed to be over-dimensioned, with extra switching capacity and excess number of deployed links, taking into account the peak rates and future growth of the traffic, resulting in waste of energy. In an IP over WDM network architecture, IP traffic is carried over a virtual topology (VT), composed of optical transparent channels called lightpaths. In transparent optical networks, lightpaths are the most power consuming elements due to the electro-optical operations during the add/drop. Generally, the intensity of traffic on lightpaths follows a daily traffic profile, decreases during the early hours of the morning, starts to increase again after 9 am, and peaks during the daytime. In different time zones, the peak traffic intensities occur at different times, relative to a global clock reference. If the lightly loaded lightpaths can be monitored and put into sleep mode by traffic engineering significant energy can be saved. Virtual topology reconfiguration with the support of standby primitives for line cards is presented in [1] and in [2] authors propose a logical topology design approach periodically solving the problem for each traffic matrix independently for different times of the day which requires to solve an NP complete problem and therefore requires complex optimization methods. In this study, we propose a simple dynamic VT reconfiguration method with sleep mode, which allows to change only one lightpath connectivity at a time by dynamically monitoring the IP packet traffic load on each lightpath. If the load of a specific lightpath is higher than a high-load threshold a new lightpath is added, if the load is lower than a low-load threshold then the lightpath is put into sleep mode. Here it is important to carefully set high-and low-load thresholds in order to save power without too many changes in VT. We show that proposed method achieves significant saving in power consumption over conventional static VT design approach by assuring the connectivity of possible future traffic increase at the same time.

Survivability approaches for multiple failures in WDM optical networks

All-optical switching or wavelength routing has the benefit of optical bypass that can eliminate expensive high-speed electronic processing at intermediate nodes and reduce significantly the cost of high-bandwidth transport. But all-optical switching has the limitations of coarse granularity, lack of multiplexing gain, and scarcity of wavelength resources, which do not mesh well with Internet traffic that has many small and diverse flows and emphasizes the importance of resource sharing. In particular, wavelength routed light paths have difficulty to seamlessly converge with multiprotocol label switching label-switched paths that have arbitrary bandwidth granularity and relatively abundant labels. In this paper, we propose a hybrid wavelength and subwavelength routing scheme that can preserve the benefits of optical bypass for large traffic flows at the same time provide multiplexing gain for small traffic flows. We first study the hybrid routing scheme using static optimization that produces an optimal path set and a partition between wavelength and subwavelength routing. We then present a dynamic heuristic that tracks the static optimization closely. During the process, we proposed a traffic arrival process called incremental arrival with sporadic random termination to more accurately model practical optical network traffic generation process.

présentée à

In Wavelength Division Multiplexing (WDM) networks, the huge capacity of wavelength channels is generally much larger than the bandwidth requirement of individual traffic streams from network users. Traffic grooming techniques aggregate low-bandwidth traffic streams onto high-bandwidth wavelength channels. In this paper, we study the optimization problem of grooming the static traffic in mesh Synchronous Optical Network (SONET) over WDM networks. The problem is formulated as a constrained integer linear programming problem and an innovative optimization objective is developed as network profit optimi-zation. The routing cost in the SONET and WDM layers as well as the revenue generated by accepting SONET traffic demands are modelled. Through the optimization process, SONET traffic demands will be selectively accepted based on the profit (i.e., the excess of revenue over network cost) they generate. Considering the complexity of the network optimization problem, a decom-position approach using Lagrangian relaxation is proposed. The overall relaxed dual problem is decomposed into routing and wavelength assignment and SONET traffic routing sub-problems. The subgradient approach is used to optimize the derived dual function by updating the Lagrange multipliers. To generate a feasible network routing scheme, a heuristic algorithm is proposed based on the dual solution. A systematic approach to obtain theoretical performance bounds is presented for an arbitrary topology mesh network. This is the first time that such theoretical performance bounds are obtained for SONET traffic grooming in mesh topology networks. The optimization results of sample networks indicate that the proposed algorithm achieves good sub-optimal