Heuristic Algorithms for the Routing and Wavelength Assignment of Scheduled Lightpath Demands in Optical Networks . . 2 Nina Skorin-Kapov Efficient Multi-Layer Operational Strategies for Survivable IP-over-WDM Networks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Krishanthmohan Ratnam, Luying Zhou, and Mohan Gurusamy Fair Sharing Using Dual Service-Level Agreements to Achieve Open Access in a Passive Optical Network . . . . . . . . . . . . . . 32 Amitabha Banerjee, Glen Kramer, and Biswanath Mukherjee An Improved Algorithm for Optimal Lightpath Establishment on a Tree Topology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 Guoliang Xue, Weiyi Zhang, Jian Tang, and Krishnaiyan Thulasiraman Backup Reprovisioning to Remedy the Effect of Multiple Link Failures in WDM Mesh Networks …

We investigate benefits of backup reprovisioning after network-state updates in survivable mesh networks. Our approach achieves good capacity utilization and network robustness. We study tradeoff between capacity optimization and computation/reconfiguration overhead with different reprovisioning periods.

We investigate a novel fault-tolerant node architecture using a resilient buffer (R-buffer) in optical burst switching (OBS) networks. Our study shows that most of the lost bursts can be restored quickly using the architecture.

Survivability is a key concern in modern network design in order to achieve fast service restorability against network failures. This paper investigates the problem of survivable dynamic connection provisioning in general telecom backbone networks, which are mesh structured. These networks employ optical fibers, which may fail due to network outages such as fiber cuts, etc. Our study applies to survivability of optical wavelength-division multiplexing (WDM) as well as multi-protocol label switching (MPLS) networks. For survivability study, we assume differentiated services where connections may have different availability requirements, so they may be provisioned differently with protection (if needed) based on their availability requirements and current network state. Therefore, it may be possible that connections with the same source, destination, and availability requirement are provisioned differently (unprotected, shared-path protected, or dedicated-path protected) at different times based on current network state. Such differentiated provisioning can provide diverse levels of service performance and achieve network resource optimization flexibly. Our main contributions are as follows. First, we develop an analytical model to quantify the availabilities of connections with various protection modes, i.e., unprotected, dedicated-path protected, and shared-path protected. Particular emphasis is placed on computing a connection's availability with shared-path protection by employing the link-vector technique, because this technique can maximally explore the sharing potential among backup paths and achieve bandwidth-assignment flexibility. Based on the mathematical model, we then present a novel provisioning strategy for dynamic connection requests in which multiple levels of services are provided and different protection schemes may be applied to different connections. The strategy jointly considers both connection availability satisfaction and resource optimization. Numerical results show very good accuracy of our model and high effectiveness of our provisioning strategy

Optical burst switching (OBS) is a promising technique for supporting high-capacity, bursty data traffic over optical wavelength-division-multiplexed (WDM) networks. An optical link failure may result in a huge amount of data (and revenue) loss, and it has been an important survivability concern in optical networks. In this paper, we study the fault- management issues related with a link failure in an OBS network. We propose to use pre-planned global rerouting to balance network load and to reroute bursts after a link fails. We apply optimization techniques to pre-plan explicit backup routes for failure scenarios. Our objective is to achieve optimal load balancing both before and after a failure such that the network state can still remain stable with minimum burst-loss probability when a failure occurs. We apply the pre-planned normal and backup routing tables to an OBS network, and study the network performance after a failure occur using illustrative numerical examples. The results show that the average burst-loss probability can be significantly reduced by 60% - from an average of 0.10 to 0.04 (when the normalized link load is less than 0.5) using globally-rerouted backup routes, when compared with the scheme without global rerouting. We also observe that the burst-loss probability is reduced by 43% - from an average of 0.07 to 0.04 (when the link load is less than 0.5) if the rerouting is done using optimization techniques, when compared with shortest-path routing.for Fault Management

In an optical WDM mesh network, different protection schemes (such as dedicated or shared protection) can be used to improve the service availability against network failures. However, in order to satisfy a connections service-availability requirement in a cost-effective and resource-efficient manner, we need a systematic mechanism to select a proper protection scheme for each connection request while provisioning the connection. In this paper, we propose to use connection availability as a metric to provide differentiated protection services in a wavelength-convertible WDM mesh network. We develop a mathematical model to analyze the availabilities of connections with different protection modes (i.e., unprotected, dedicated protected, or shared protected). In the shared-protection case, we investigate how a connection's availability is affected by backup resource sharing. The sharing might cause backup resource contention between several connections when multiple simultaneous (or overlapping) failures occur in the network. Using a continuous-time Markov model, we derive the conditional probability for a connection to acquire backup resources in the presence of backup resource contention. Through this model, we show how the availability of a shared-protected connection can be quantitatively computed. Based on the analytical model, we develop provisioning strategies for a given set of connection demands in which an appropriate, possibly different, level of protection is provided to each connection according to its predefined availability requirement, e.g., 0.999, 0.997. We propose integer linear programming (ILP) and heuristic approaches to provision the connections cost effectively while satisfying the connections' availability requirements. The effectiveness of our provisioning approaches is demonstrated through numerical examples. The proposed provisioning strategies inherently facilitate the service differentiation in optical WDM mesh networks.

Survivability is a key concern in modern telecom mesh networks because of the enormous capacity of a telecom link which is usually an optical fiber employing wavelength-division multiplexing (WDM). Backup bandwidth reprovisioning has been shown to be an effective approach for improving network survivability as well as preventing existing services from unnecessary interruption. We investigate the advantages of reprovisioning new backup paths for connections when a previous failure is repaired (as well as when a network failure occurs). We consider reprovisioning of backup paths either (1) for unprotected or vulnerable connections that lose their primary or backup paths due to a previous failure and fail to be reprovisioned when the failure happens due to resource limits, or (2) for all existing connections in the network. The pros and cons of the two policies are investigated. A wavelength-convertible network and shared-path protection are assumed in this study. We compare the performance of our dynamic reprovisioning approach with a conventional scheme which reprovisions backup paths for connections only when a network failure occurs. The simulation results demonstrate that our approach achieves more network robustness and better backup capacity optimization.

As networks grow in size and complexity, both the probability and the impact of failures increase. The pre-allocated backup bandwidth, which has been widely investigated in the literature, may not be able to provide full protection guarantee when multiple failures occur in a network. In this study, we consider multiple concurrent failures where concurrent means that a new failure occurs before a previous failure is repaired. To combat the effect of multiple concurrent failures, new backups can be reprovisioned after one failure such that the next potential failure can be handled effectively and efficiently. We consider dynamic traffic where a pair of link-disjoint primary and backup paths is provisioned when a new connection request arrives. After a failure occurs, the affected connections switch traffic from their primary paths to backup paths. To protect against next potential failure, we reprovision new backups for connections that become unprotected or vulnerable because of losing their primary or their backup due to the previous failure or due to backup resource sharing. This approach is called Minimal Backup Reprovisioning (MBR). An alternative approach is to globally rearrange backups for all connections after one failure occurs, which is called Global Backup Reprovisioning (GBR). Backup reprovisioning can be performed whenever the network's state changes, e.g., (1) when a new request arrives, (2) when an existing connection terminates, (3) when a network failure occurs, (4) when a failed link/node is repaired, etc., to utilize the available resources more efficiently or to recover quickly from the next failure. In this study, we perform MBR or GBR after one network failure occurs to protect against the next potential failure in a wavelength-convertible WDM mesh network. The link-vector network model which can maximally explore the backup-sharing potential is assumed in this study. We then analyze the complexity of MBR and GBR under such a network model. A reprovisioning algorithm is proposed for MBR which can significantly reduce the connection vulnerability without the knowledge of the location of the next failure. In GBR, both integer linear program (ILP) and heuristic-based approaches are proposed. We compare capacity requirement and computational complexity of MBR to that of GBR through numerical examples. MBR demonstrates a good tradeoff between complexity and capacity efficiency to handle multiple concurrent failures

We present a reliability analysis for shared-path-protected WDM mesh networks. We develop a cost-effective provisioning approach to provide differentiated services to carry connections with both reliability guarantee and resource optimization.

Optical burst switching (OBS) and Optical packet switching (OPS) are highly dynamic transport network architectures for a future optical Internet. As they both rely on statistical multiplexing network dimensioning, routing, and efficient contention resolution are key issues in order to achieve a low burst loss probability. This paper first compares principal network dimensioning and fixed routing strategies in a US reference core network for the failure-free case and for selected single link failure cases. We show that dimensioning approaches, which take knowledge of traffic demands as well as of routing into account, combined with shortest path routing can achieve a better performance than uniform dimensioning with either shortest path or optimized explicit routing. However, for critical single link failure cases, the higher flexibility of explicit routing yields more homogeneous results. Then, we extend this comparison to alternative/deflection routing in the failure-free case. Finally, we present results for the impact of dimensioning in OBS/OPS networks which employ simple fiber-delay line buffers for contention resolution.

Ethernet has been playing an increasingly important role in wide area networks ( WANs), from both a service perspective and a transport-technology perspective. Unlike its dominant presence in local area networks ( LANs), Ethernet in WANs has been increasing its popularity in three different directions, i.e., Ethernet-based layer-2 virtual private network ( L2VPN) over layer-3 network, Ethernet over SONET ( EoS), and Ethernet directly over WDM channels or optical. fibers. In this paper, we investigate the benefits and challenges of using next-generation SONET/SDH techniques - namely SONET/SDH virtual concatenation ( VCAT) and link-capacity adjustment scheme ( LCAS) - to support Ethernet-based data services in intelligent optical WDM wide area networks. In particular, we evaluate the network performance improvement after employing VCAT. In order to fully utilize VCAT's inverse-multiplexing capability, several simple and effective heuristic algorithms are proposed and evaluated.

Recent trends in bandwidth markets show that customers are tending to ask providers for a large bandwidth but for a limited amount of time to support new short-term bandwidth-hungry applications. In order to meet these new requirements, progress in network technologies (fast and reconfigurable switching equipment) and protocols [such as generalized multiprotocol label switching (GMPLS) and automatically switched optical network (ASON)] is paving the road towards flexible optical transport networks in which leasable circuits could be set up and released on a short-term basis. As a result, for dynamic traffic, the holding time of connection requests can be known in advance. The authors propose to exploit knowledge of connection holding time to design an efficient algorithm, called PHOTO, for the dynamic provisioning of shared-path-protected connections in optical mesh networks. The core idea of the proposal consists of exploiting the knowledge of the holding time of connection requests to minimize resource overbuild (RO) due to backup capacity and hence achieve resource-usage efficiency. For a typical US nationwide network, savings on RO of up to 10% were obtained for practical scenarios compared to a holding-time-unaware, but otherwise shared-path-efficient, approach.

Progress in network technologies and protocols is paving the road towards flexible optical transport networks, in which leasable circuits could be set up and released on a short-term basis. Thus we consider it reasonable that, at least for some types of services, the holding time of connection requests can be known in advance. In this paper, we propose to exploit the knowledge of connection-holding time to improve the performance of an algorithm for shared-segment protection (SSP).For a typical US nationwide network, we compared our approach to an holding-time-unaware, but otherwise shared segmented efficient, approach, obtaining savings on resource overbuild of up to 7% for practical scenarios. Index Terms— Optical network, WDM, lightpath, dynamic traffic, holding time, shared-segment protection.

Ethernet has been playing an increasingly important role in wide area networks (WANs), from both a service perspective and a transport-technology perspective. Unlike its dominant presence in local area networks (LANs), Ethernet in WANs has been increasing its popularity in three different directions, i.e., Ethernet-based layer-2 virtual private network (L2VPN) over layer-3 network, Ethernet over SONET (EoS), and Ethernet directly over WDM channels or optical fibers. In this paper, we investigate the benefits and challenges of using next-generation SONET/SDH techniques—namely SONET/SDH virtual concatenation (VCAT) and link-capacity adjustment scheme (LCAS)—to support Ethernet-based data services in intelligent optical WDM wide area networks. In particular, we evaluate the network performance improvement after employing VCAT. In order to fully utilize VCAT’s inverse-multiplexing capability, several simple and effective heuristic algorithms are proposed and evaluated.

With the maturing of wavelength-division multiplexing (WDM) technology, one single strand of fiber can provide tremendous bandwidth (potentially a few tens of terabits per second) by multiplexing many non-overlapping wavelength channels. In a wavelength-routed WDM network, with the relatively high frequency of network failures, and the tremendous traffic loss a failure may cause, fault management becomes a critical concern in network design and its real-time operation. This dissertation investigates architectures and algorithms for survivable optical networks. The dissertation first presents a broad overview of the fault-management mechanisms involved in deploying a survivable optical mesh network. Different parameters are presented which can measure the quality of service (QoS) provided by a WDM mesh network to upper protocol layers, such as service availability, service reliability, restoration time, and service restorability. It reviews the concepts, the factors that affect them, and how to improve them. Next, the dissertation investigates the problem of using optimal path pair as the routes for the primary and backup paths in a WDM network with and without wavelength-conversion capability. It proves that the problem is NP-complete in wavelength-continuous networks. In order to satisfy a connection's service-availability requirement in a cost-effective manner, we need a systematic mechanism to select a proper protection scheme for each connection request. Connection availability is proposed to use as a metric to provide differentiated protection services. A mathematical model is developed to analyze the availabilities of connections with different protection modes. Integer linear program (ILP) and heuristic approaches are studied to provision the connections cost effectively while satisfying the connections' availability requirements. To combat the effect of multiple concurrent failures, new backups can be reprovisioned after one failure such that the next potential failure can be handled effectively and efficiently. Backup reprovisioning approaches are proposed and studied under a novel network model. Optical burst switching (OBS) is a promising technique for supporting high-capacity, bursty data traffic over optical WDM networks. The fault-management issues in an OBS network are studied. A pre-planned global rerouting scheme is proposed to balance network load and a novel fault-tolerant node architecture is proposed to quickly restore the bursts on a failed link.

This paper investigates the problem of dynamic survivable lightpath provisioning in optical mesh networks employing wavelength-division multiplexing (WDM). In particular, we focus on shared-path protection because it is resource efficient due to the fact that backup paths can share wavelength links when their corresponding working paths are mutually diverse. Our main contributions are as follows. 1) First, we prove that the problem of finding an eligible pair of working and backup paths for a new lightpath request requiring shared-path protection under the current network state is NP-complete. 2) Then, we develop a heuristic, called CAFES, to compute a feasible solution with high probability. 3) Finally, we design another heuristic, called OPT, to optimize resource consumption for a given solution. The merits of our approaches are that they capture the essence of shared-path protection and approach to optimal solutions without enumerating paths. We evaluate the effectiveness of our heuristics and the results are found to be promising.