Survivable Traffic Grooming for Scheduled Demands Arunita Jaekel, Ying Chen and Ataul Bari School of Computer Science, University of Windsor 401 Sunset Ave., Windsor, ON N9B 3P4, Canada E-mail: {arunita, chen13r, bari1}@uwindsor.ca Abstract— There has been considerable research interest in the design of survivable grooming capable networks in recent years. For such networks, protection may take place at the lightpath level or at the connection level. The vast majority of the current work can be classified into one of two categories i) static grooming, where the demands are allocated for the entire duration of the network and ii) dynamic grooming, where the start times and durations of demands are generated randomly based on certain traffic distributions. In this paper, we propose a new technique for survivable traffic grooming under the scheduled traffic model that exploits knowledge of the connection holding times of traffic demands to lead to more efficient resource allocation. We present efficient integer linear program (ILP) formulations for the complete survivable traffic grooming problem in WDM networks. Our formulations can solve the joint problem of the topology design, traffic routing and RWA, using path protection at lightpath level. Our aim is to design a stable logical topology that can accommodate a collection of low-speed traffic demands with specified setup and teardown times. The objective function, considered in our ILP formulation, is to minimize the resource requirements. This can be easily modified to maximize the throughput under a given set of resources. We also have proposed a simplified version of our ILP formulations that can solve the problem in a way that is computationally more efficient. To the best of our knowledge, this is the first paper to address the survivable traffic grooming problem under the scheduled traffic model. I. I NTRODUCTION Traffic Grooming in wavelength division multiplexing (WDM) networks [1] can be defined as a family of techniques for combining a number of low-speed traffic streams from users so that the high capacity of each lightpath [2] (2.5 or 10 Gb/sec today, expected to be 40 Gb/sec soon) may be used as efficiently as possible [3], [4], [5], [6], [7]. Traffic grooming can be classified as either static or dynamic traffic grooming and there has been considerable research interest in both of these areas in recent years. Static traffic grooming techniques [8], [9], [10], [11] assume that each low-speed traffic demand is persistent throughout the lifetime of the network. So, there is no opportunity of sharing resources between temporally disjoint demands. In dynamic traffic grooming [12], [13], on the other hand, the arrival time of requests are not known ahead of time. Hence, it is difficult to efficiently design a stable topology that will be able to handle all traffic requests. Therefore, lightpaths are typically setup and torn down based on the current demand. In this paper, we address the problem of traffic grooming under the scheduled traffic model [15], [16], [17], which is ap- propriate for applications that require periodic use of lightpaths (e.g. once per day) at predefined times. The setup and teardown times of demands are known in advance, so resource allocation can be optimized in both space and time. Although routing and wavelength assignment [14] of scheduled lightpaths, under the scheduled traffic model have been considered in a number of recent papers [16], [17], [18], [19], these papers do not address the problem of combining low-speed scheduled demands on to high capacity lightpaths in an efficient manner. Instead, the traffic demands typically have a coarse granularity, and correspond to the number of lightpaths to be routed over the network. We propose a new approach for traffic grooming in WDM mesh networks with low-speed scheduled traffic demands. Although individual (subwavelength) demands may be short lived, it is desirable to have a logical topology that is relatively stable and not subject to frequent changes. The resulting logical topology should be capable of accommodating all scheduled traffic demands and these demands should be routed over the logical topology in a way that allows sharing of resources among non-overlapping demands. Our objective is to design a stable logical topology, which reduces the overhead required for setting up and tearing down individual lightpaths. We also provide protection at the lightpath level, such that the network can survive single link failures, using dedicated path protection (DPP) techniques [33]. In this paper, we have addressed the complete problem of survivable traffic grooming under the scheduled traffic model in WDM mesh networks and proposed integer linear program (ILP) formulations to solve this problem optimally. Our formulations optimally solve the following sub-problems: i) Design of a stable logical topology that does not change over time, ii) Routing and wavelength assignment for each primary and backup lightpath included in the logical topology, iii) Combining low-speed traffic demands on to high capacity lightpaths, such that all demands can be suc- cessfully routed, and iv) Sharing of resources among time-disjoint demands. We present an ILP formulation, ILP1, that addresses the above sub-problems. The objective of our formulation is to minimize the amount of resources required to accommodate all