Analysis Framework of RSA Algorithms in Elastic Optical Rings

Recently proposed Elastic Optical Networks (EON) provide optimized bandwidth utilization through a better fitting match between connection bandwidth requirements and spectrum occupation. In an EON, lightpaths are established using a Routing and Spectrum Assignment (RSA) algorithm. Nonetheless, a novel problem arises in EON, known as the fragmentation problem. Indeed, as connections are dynamically established and terminated, spectrum may become fragmented and jeopardize the establishment of future connections. In this situation, the contiguity and continuity constraints in the RSA problem are harder to be guaranteed and therefore connections' blocking probability increases. In this paper we present and analyse some fragmentation-aware RSA algorithms currently found in the literature and analyse their performance when implemented alone or in tandem.

In this paper, we propose a defragmentation based load balancing technique for efficiently undertaking routing and spectrum assignment (RSA), and minimizing the fragmentation problem in elastic optical network (EON). In EON, after connections terminate, they leave behind spectrum gaps which con not be re-assigned. To counter this condition, in Defragmentation based Load Balancing Routing and Spectrum Assignment (DLBRSA), the existing connections are reconfigured. At the time of defragmentation, for route selection, among the available shortest routes, the route with the largest chunk of contiguous frequency slots (FSs) is chosen. DLBRSA has three advantages over the conventional Shortest Path RSA (SPRSA) technique: (1) defragmentation increases the network resource utilization, (2) load balancing on the basis of the largest available FS chunk along the route ensures uniform spectrum allocation, and (3) defragmentation ensures that the route length of the request is minimum. The performance of the DLBRSA is analyzed in reference to a benchmark, SPRSA strategy. The comparison is done on the metrics of bandwidth rejected, and the percentage of total network capacity allocated for the EON and NSFNET network topologies. These two parameters are evaluated under the two scenarios of increase in: network load (in Erlang) and number of FSs demand. Simulation results show lower bandwidth blocking probability of DLBRSA from 6.53% to 8.96% under the different load conditions. In addition, decrease of network resource utilization is observed, when we compare the proposed DLBRSA with SPRSA the existing strategy.

Elastic Optical Networks has drawn a lot of attention in the past few years because of its ability to transmit different data rates increasing or decreasing the optical spectrum according to the necessary demand, ensuring a high spectral efficiency. However, establishing and tearing down different connections ends up segmenting the spectrum in several small fragments, making it impossible for new requests to be attended. Besides, another recurrent problem in elastic optical networks, called routing and spectrum assignment (RSA), aims to find a path and assign to it a contiguous number of spectrum slots using the smallest possible amount of spectral resources. The main purpose of this paper is to propose an algorithm to minimize the spectrum fragmentation problem in Elastic Optical Networks by using RSA algorithms. Results obtained through simulations indicates that the proposed algorithm outperforms in terms of blocking probability and spectrum fragmentation ratio the other existing ones.

Deep reinforcement learning (DRL) has been introduced to the routing and spectrum assignment (RSA) of elastic optical networks (EONs) where the RSA policies are learnt during the interaction of a DRL agent with the EON environment. Upon each new traffic arrival, the DRL agent senses the EON state, and makes the RSA decision accordingly. Therefore, the EON state feature extraction is essential for the performance. However, current approaches, mainly based on fully connected neural networks or convolutional neural networks, are unsatisfactory in EON state feature extraction due to the following two reasons. (1) The optical network topology information is not well considered. (2) The path-level features are highly related to the RSA decision, due to the spectrum continuity constraint, while path-level feature extraction has not been well exploited. To overcome the above shortcomings, in this paper, we propose to utilize the Graph Convolutional Neural Network (GCN) and the Recurrent Neural Network (RNN) for the feature extraction of the network topology, and the aggregation of link-level features to path-level features. By doing this, critical RSA-related information can be sensed by the DRL agent to make better actions. Simulation results demonstrate that our proposed method outperforms previous approaches.

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

The emergence of Elastic Optical Networks (EON) has brought new conceptions in the operations of optical networks, improving its flexibility and efficiency. The routing and spectrum assignment (RSA) is one of the key problems in EON which deals with resource allocation. In order to reduce the bandwidth blocking ratio on the network, recently, Auxiliary-Graph-Model approaches propose spectrum reservation schemes to reduce the network exhaustion probability. However, these approaches do not take into account the energy expenditure caused by these schemes. In this work, we use an auxiliary graph model to design a new RSA heuristic algorithm to improve the network energy consumption without losses in bandwidth blocking rate. Numerical results show that our proposal may provide a reduction of up to 54% in the bandwidth blocking ratio and achieve a gain of up to 51% in energy efficiency compared to literature.

This paper presents the features of an advanced software simulator for the Elastic Optical Network (EON) - Complex Elastic Optical Networks Simulator (CEONS). The CEONS is as a test environment for solving three main issues addressed in EON: Routing and Spectrum Assignment (RSA) problems; Routing, Modulation and Spectrum Assignment (RMSA) problems, and addressing regenerator placement problem. With a universal API, users can design their own algorithms in any programming language and implement them in the simulator as plugins to solve the problems mentioned above. The ease and versatility of implementing the algorithms allows usage of the CEONS in advanced problems and for the introduction of EON technology.

De-multiplexing the required spectrum in a traffic demand into multiple non-adjacent granular spectrums for dynamic traffic grooming in EON

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.

In Elastic Optical Networks (EONs), two communications requests sharing common fiber links have to be separated in the spectrum domain by a guard band to prevent the interference or satisfy the security requirement, since there are types of physical impairments or attacks in the optical layer. For different interference or security levels, the size of guard band should be adapted, while it is supposed to be identical in most of the literature. To be more agile and realistic, we rise a new spectrum assignment model called Distance Spectrum Assignment (DSA): the size of guard band between two communications sharing a common fiber link varies according to the specific circumstance, and the goal is to minimize the maximum index of frequency slots assigned to satisfy all the guard band constraints. Since DSA is a strongly NP-hard problem, we propose an Integer Linear Program (ILP) model for computing the optimal solution. For solving the DSA problem in large-scale EONs, we develop a heuristic algorithm with time complexity O(n3Δ), where n is the number of requests and Δ is the maximum degree of the conflict graph. We prove that the proposed algorithm achieves an approximate ratio of O(log(n)) in complete conflict graphs. Numerical results demonstrate the proposed heuristic algorithm can find near-optimal spectrum assignments for solving the DSA problem in general topologies.

Efficient resource allocation in OFDM-based optical networks is directly affected by definition and formulations of Routing and Spectrum Assignment (RSA) problem. In this paper, we propose new Binary Linear Programming (BLP) formulations for the RSA problem that provide more efficient resource allocation in comparison with previous formulations. Unlike the conventional formulations that mainly aim at minimizing used spectrum slots, our formulations simultaneously aim at defragmentation and minimization of the assigned spectrum slots. We show contiguity constraint of the conventional formulations may incorrectly make a feasible solution unattainable. To solve this problem, we remove the contiguity constraint and embed its effect as a penalty term in the goal function. Two performance metrics for evaluating the amount of contiguity and defragmentation in a typical network are also defined. Simulation results show that the new formulations provide a considerable improvement in the performance of the RSA process in terms of the amount of contiguity and defragmentation with respect to the previous formulations.

This paper considers distance adaptive dynamic routing and spectrum assignment (RSA) for elastic optical networks with shared backup path protection (SBPP). Efficient heuristic algorithms based on spectrum window planes are proposed for implementing distance and modulation format adaptive RSA so as to maximize spare capacity sharing among multiple protection lightpaths. A differentiated sharable frequency slot cost was also defined for the first time to more efficiently share protection resource. Network performance is evaluated in terms of bandwidth blocking probability (BBP) through simulations. The proposed SBPP technique is effective in reducing BBP and improving spectral efficiency compared to conventional SBPP schemes and 1+1 path protection. The impact of transponder tunability on bandwidth blocking performance is also evaluated to show that a limited tuning range is sufficient to achieve a BBP performance close to that with full tunability.

Elastic optical networks (EONs) are new generation optical networks that provide flexible bandwidth and spectrum assignment characteristics to accommodate diverse demand range over traditional dense wavelength division multiplexing (DWDM)-based networks. While overcoming spectrum contiguity and continuity constraints, Routing and Spectrum Assignment (RSA) is a challenging task in EONs. In this article, we have proposed an EON network model. Using existing RSA techniques we have analyzed the performance of the proposed model on the basis of Network Blocking Probability (NBP) and Network Utilization Efficiency (NUE) under variable load conditions. It has been found to be working optimally even at a load of 200 Erlangs.

The routing and spectrum assignment (RSA) is one of the key problems in flexible optical networks. When in a gridless fashion, the shortest-path RSA algorithms have exponential computational complexity, and are thus not suitable to be employed in real networks. On the other hand, today most optical components cannot support fully gridless tunability, which also limits the application of gridless RSA schemes. In this paper, we propose a novel grid-based spectrum-scan routing (SSR) scheme in flexible optical networks. The SSR scheme achieves optimal routing with a polynomial computational complexity. Compared with other RSA schemes, SSR has shorter computation time, lower blocking probability, and higher resource utilization.

With the ever increasing demand for optical networks, many decisions about network planning are becoming increasingly challenging. One of these crucial decisions is the network provisioning, known as the Routing and Spectrum Assignment (RSA) problem. What makes this problem practically challenging to service providers is the significant difference in size between the realistic problem instances and the instances that can be solved using the existing algorithms. Motivated to solve larger problem sizes, we propose a decomposition model and a nested column generation algorithm for the RSA problem. Compared to recent studies, the proposed algorithm produces higher-quality solutions for larger data instances quite efficiently. Furthermore, it provides a measure of the solution accuracy, i.e., an upper bound on the maximum distance of the solution obtained compared to the optimal solution.