Minimizing maximum fiber requirement in optical networks

Recently, machine learning (ML) has attracted the attention of both researchers and practitioners to address several issues in the optical networking field. This trend has been mainly driven by the huge amount of available data (i.e., signal quality indicators, network alarms, etc.) and to the large number of optimization parameters which feature current optical networks (such as, modulation format, lightpath routes, transport wavelength, etc.). In this paper, we leverage the techniques from the ML discipline to efficiently accomplish the routing and wavelength assignment (RWA) for an input traffic matrix in an optical WDM network. Numerical results show that near-optimal RWA can be obtained with our approach, while reducing computational time up to 93% in comparison to a traditional optimization approach based on integer linear programming. Moreover, to further demonstrate the effectiveness of our approach, we deployed the ML classifier into an ONOS-based software defined optical network laboratory testbed, where we evaluate the performance of the overall RWA process in terms of computational time.

In this paper we investigate traffic-engineering issues in Wavelength Division Multiplexing (WDM) all- optical networks. In such networks, the wavelength continuity constraint along with the wavelength clash constraint, lead to poor network performances when dealing with the lightpath provisioning problem. The impact of these constraints is especially severe when traffic demands are unpredictable and characterized by random arrivals and departures. In order to alleviate the impact of these constraints, we propose to employ intentional/active rerouting. Active lightpath rerouting is to intentionally reroute already established lightpaths, during their life period, so as to achieve a better blocking performance. We here assume that due to the large geographic area an optical WDM network can cover, upgrading such a network to support the huge demand for network bandwidth can be costly. Thereby, it is extremely important for network operators to apply traffic-engineering strategies to cost-effectively optimize network throughput. Two new routing and wavelength assignment (RWA) algorithms applying intentional rerouting are proposed. Both algorithms dynamically reroute some already established lightpaths from longer paths to vacant shorter ones so as to reduce the network resources consumption and hence improve the network throughput. The first algorithm, namely, Timer-Based Active Lightpath Rerouting (TB-ALR) initiates the rerouting procedure every time a timer expires. The second algorithm, namely, Sequential Routing with Active Lightpath Rerouting (SeqRwALR) initiates the rerouting procedure when a connection leaves and its lightpaths are released. To the best of our knowledge, our global approach has not already been investigated in the literature. Simulation results show that the proposed active rerouting algorithms yield much lower connection rejection ratios than rerouting algorithms previously presented in the literature while rerouting a small number of already established lightpaths. By rerouting a small number of existing lightpaths, we hope that the disruption period incurred due to rerouting is minimized.

Extending light-trail into elastic optical networks for dynamic traffic grooming

In this paper we study on-line wavelength assignment in wavelength-routed WDM networks under both unicast and multicast traffic. We assume nodes in the networks have wavelength conversion ability. Since wavelength converters are still expensive and difficult to implement, we consider the networks that have only a limited number of converters in each node, and the converters are shared by all input channels at the node. We consider how to set up connections in such networks using as few wavelength converters as possible. For unicast traffic, we first study the problem of setting up a lightpath on a given link path with minimum number of conversions, and give a new algorithm that solves it in O(tk) time, where t is the number of links on the path and k is the number of wavelengths per fiber, as compared to the best known existing algorithm that runs in at least O(t/sup 2/k) time. We also consider the case when nodes have different conversion priorities, and give an O(tk) time algorithm for setting up a lightpath on a given link path while converting wavelength at higher priority nodes only when necessary. We then generalize this technique to WDM networks with arbitrary topologies and present an algorithm that sets up an optimal lightpath network-wide in O(Nk + Lk) time by checking the state of the entire network, where N and L are the number of nodes and links in the network, respectively. For multicast traffic, finding an optimal multicast light-tree is known to be NP-hard and is usually solved by first finding a link tree then finding a light tree on the link tree. Finding a link tree is also NP-hard and has been extensively studied. Thus, we focus on the second problem which is to set up a light tree on a given link tree with minimum number of conversions. We propose a new and more practical multicast conversion model, where the output of the wavelength converter can be split. As can be seen, the new model can save the usage of converters considerably. We first show that this problem is NP-hard and then give efficient heuristics to solve it approximately.

In this paper, we study on-line wavelength assignment in wavelength-routed WDM networks under both unicast and multicast traffic where nodes in the networks have wavelength conversion ability. Since wavelength converters are still expensive and difficult to implement, we consider the case where nodes in networks have only a limited number of converters that are shared by all input channels. We study the problem of setting up connections in such networks using minimum number of wavelength converters. For unicast traffic, we first study the problem of setting up a lightpath on a given link-path with minimum number of conversions. We give a simple algorithm that solves it in O(tk) time where t is the number of links on the path and k is the number of wavelengths per fiber, as compared to the best known existing method that needs to construct an auxiliary graph and apply the Dijkstra's algorithm. We also consider the problem of setting up a lightpath while using wavelength converters at nodes with fewer available converters only when necessary, and give an O(tk) time algorithm. We then generalize this technique to WDM networks with arbitrary topologies and give an algorithm that sets up an optimal lightpath network-wide in O(Nk+Lk) time, where N and L are the number of nodes and links in the network, respectively. We also consider multicast traffic in this paper. Finding an optimal multicast light-tree is known to be NP-hard and is usually solved by first finding a link-tree then finding a light-tree on the link-tree. Finding an optimal link-tree is also NP-hard and has been extensively studied. Thus, we focus on the second problem which is to set up a light-tree on a given link-tree with minimum number of conversions. We propose a new multicast conversion model with which the output of the wavelength converter is split-table to save the usage of converters. We show that under this model the problem of setting up an optimal light-tree is NP-hard and then give efficient heuristics to solve it approximately

The problem of dynamic routing in a multifibre time-slotted wavelength routed WDM network is considered. Requests to establish a connection are dynamic and each connection requests an integer multiple of some basic unit. Each wavelength has the capacity of carrying a multiple of these channels, where each channel is represented by a time-slot as in TDM networks. The network is represented as a layered graph model with multiple layers, where each layer represents a specific wavelength. Each link in the layered graph has one or more fibres and an associated cost. The cost of layered graph links could be a function of the loading state of the respective wavelength across the available fibres. Algorithms are proposed for fibre selection and for setting link costs, and a modified Dijkstra algorithm on the layered graph is used to select a route on the layered graph for a new connection request. The selected route on the layered graph represents the physical path and the selected wavelength. The performance of these algorithms is evaluated for a realistic mesh network topology for various combinations of the number of wavelengths, the number of fibres per link, the number of time-slots per wavelength, and the offered load for both uniform and non-uniform traffic loads. The best performing fibre selection and link cost update methods are identified. The proposed algorithm for routing and wavelength assignment in mesh optical WDM grooming networks outperforms previously proposed work in the literature.

A fundamental problem in computer and communication networks is the wavelength assignment (WA) problem: given a set of routing paths on a network, assign wavelengths (channels) to the paths such that the paths with the same wavelength are edge-disjoint. The optimization problem here is to minimize the number of wavelengths. A popular network topology is a tree of rings. It is known NP-hard to find the minimum number of wavelengths for the WA problem on a tree of rings. Let L be the maximum number of paths on any edge in the network. Then L is a lower bound on the number of wavelengths for the WA problem. We give a polynomial time algorithm which uses at most 3L wavelengths for the WA problem on a tree of rings with node degree at most eight. This improves the previous result of 4L. We also show that some instances of the WA problem require at least 3L wavelengths on a tree of rings, implying that the 3L upper bound is optimal for the worst case instances. In addition, we prove that our algorithm has approximation ratios 2 and 2.5 for a tree of rings with node degrees at most four and six, respectively.

In anycast, connection requests are routed to one of multiple destination candidates, thus helping improve the overall acceptance of service requests in the network when compared to traditional unicast. In this paper we develop a new analytical model to compute the network-wide blocking performance for anycast routing and wavelength assignment (ARWA) in wavelength division multiplexing (WDM) optical networks using the reduced-load fixed-point approximation analysis. This model is based on conventional queuing theory for loss systems combined with a joint probability analysis of the anycast service alternatives. Performance results show that our model provides good accuracy compared to simulation results.

In the past few decades, the architecture of optical networks has undergone significant evolution, from the earliest wavelength-division multiplexing (WDM) optical networks to elastic optical networks (EONs) and later to space-division multiplexing (SDM) EONs, to address the continuous growth of Internet traffic. By 2024, Pbps-level optical networks are expected, far exceeding the capacity limit of single-mode fibers. The massive SDM era is on the horizon. In this context, a newly designed architecture for optical networks called the spatial channel network (SCN) architecture, which achieves high cost efficiency by means of practical hierarchical optical cross-connects, has recently been proposed. However, the evolution of optical network architectures will simultaneously present challenges related to network optimization. For instance, with the evolution from WDM optical networks to EONs, the kernel network optimization problem was transformed from the routing and wavelength assignment (RWA) problem into the routing and spectrum assignment (RSA) problem due to the additionally introduced constraint of spectrum contiguity. Similarly, specially designed algorithms are also expected to be essential for addressing the network optimization problem in SCNs. In this paper, we define this new problem as the routing, spatial channel, and spectrum assignment (RSCSA) problem. We propose an integer linear programming (ILP) model and a heuristic algorithm to solve the RSCSA problem. We examine the performance of the proposed approaches via simulation experiments. The results show that both proposed approaches are effective in finding the optimal solutions or solutions close to the lower bounds. To the best of our knowledge, this is the first work to focus on the network optimization problem in SCNs.

The manycast communication paradigm is becoming increasingly important in today’s optical networks, wherein connection requests from a source node can be routed to a subset of a candidate destination set. This flexibility in choosing different destinations from a larger candidate destination set helps improve the network performance. In this paper, we develop a new analytical model to compute the network-wide blocking performance for generalized manycast routing and wavelength assignment (MRWA) in optical wavelength division multiplexed (WDM) networks. Specifically, we calculate the request blocking on full-wavelength convertible optical networks based on Markov chains and conditional probability analysis. Performance results show that our model is accurate and is verified by extensive simulation results.

Routing and wavelength assignment vs. EDFA reliability performance in optical backbone networks: An operational cost perspective

In this paper we consider the routing and wavelength assignment problem in a wavelength routed all optical network. Inspired by techniques from artificial intelligence, in particular the Blocking Island (BI) abstraction, we propose a simple and intelligent routing and wavelength assignment (RWA) algorithm: BI_RWA. This algorithm can be used in arbitrarily connected optical networks. In addition, it is general enough such that with some simple modifications, it can be applied to different optical networking scenarios: static or dynamic traffic, single or multiple fiber links between node pairs, with or without wavelength converters. We have conducted simulation experiments to evaluate the performance of our algorithm. The simulation is carried out in two parts: static traffic and dynamic traffic. The results will demonstrate that our RWA algorithm outperforms state-of-the-art related algorithms.

Multicast routing for wavelength division multiplexing optical network is a challenging problem in optical networks. The Multicast Routing and Wavelength Assignment (MRWA) problem has been proved to be a NP complete problem. In this paper, the MRWA problem in Wavelength Division Multiplexing (WDM) has been modeled as a Multi-Objective Multicast Routing and Wavelength Assignment (MOMRWA) optimization problem with three optimization objectives, including minimizing the resource cost, minimizing the end-to-end delay and minimizing the used channels. Some Quality of Service metrics have also been considered in MOMRWA. To tackle the MOMRWA problem, a multi-objective genetic algorithm, namely MRWA_SSNSGA-II, has been proposed, which is based on the Steady-State Non-Dominated Sorting Genetic Algorithm II combined with the Efficient Non-Domination Level Update (ENLU) strategy. The purpose of the ENLU strategy is to reduce the computing time of maintaining the Non-Domination Level structure. In order to avoid the loop generated by the gene recombination, a light tree recovery mechanism based on the Minimum Path Cost Heuristic has been proposed. We evaluate our proposed MRWA_SSNSGA-II on a large number of test instances with different sizes. Experimental results demonstrate that MRWA_SSNSGA-II outperformed some recent GA based multi-objective optimization algorithms for MOMRWA problem in terms of the solution quality.

We investigate the problem of online routing and wavelength assignment and the related throughput maximization problem in wavelength division multiplexing optical networks. It is pointed out that these problems are highly inapproximable, that is, the competitive ratio of any algorithm is at least a polynomial. We evaluate the average-case performance of several online algorithms, which have no knowledge of future arriving connection requests when processing the current connection request. Our experimental results on a wide range of optical networks demonstrate that the average-case performance of these algorithms are very close to optimal.

To construct an efficient subscriber optical fiber cable network, a highly economical and versatile distribution method is needed. This paper clarifies a technique in which by determining the installation cable size of feeder system and the ratio of common fibers to total fibers, a balance between the service requirement and cost is obtained.First, gamma distribution, practically used for metal cables, is applied to the demand forecast error distribution and the relation between the number of fibers and the ratio of common fibers to total fibers, which is required to obtain the desired equipment reliability, is clarified. Next, the relation between the cross‐connecting probability is determined and utilizing the ratio of common fibers, the cost function of optical cable system is derived. Here, both the cable installation cost and cross‐connection cost are considered. These two factors are combined and a design method of installation cable size and the ratio of common fibers to total fibers is clarified. This method ensures the desired equipment reliability and is the most economical one. Using this method and assuming all the various costs, it is clarified that for high cross‐connection cost, a cable having a large number of fibers should be installed and the cable network construction having comparatively low ratio of common fibers to total fibers is desirable. When the cross‐connection cost is less, the installation cable having a smaller number of fibers and a higher ratio of common fibers to total fibers is desirable. Moreover, with these results in mind, an economical subscriber optical fiber cable network construction is proposed.