Research On Spectrum Engineering And Control Mechanism In Flexible Grid Optical Networks

Recently, with the development of broadband business, new types of services such as real-time high-resolution video, distributed services and data center services are increasing, thus resulting an unprecedented huge growth of the Internet traffic. This has raised a higher requirement on the capacity of traditional optical networks. Under the service-driven scenario, it becomes an inevitable trend to develop huge-capacity, high-spectrum-efficiency, dynamic and flexible all-optical networks. This kind of flexible grid optical network is regarded as the direction of future networks, and has a wide prospect of application. Based on the control and management problem of spectrum resources in flexible grid optical networks, this paper proposes an original concept named Spectrum Engineering, and studies the rational mechanism which aims to improve the utilization of network resources. The main contributions and innovations of this paper are summarized as follows:Firstly, for the complex characteristics and status of resources in flexible grid optical networks, this paper studies the spectrum consistency and continuity constraints during the allocation of resources, and proposes two routing and spectrum allocation (RSA) heuristics. One is from the perspective of service layer, where the services are classified according to their properties, and the spectrum is divided into different segments. The same class of services are provisioned on a common spectral segments. The other one is from the perspective of spectrum layer, where the route and the spectrum allocation of services are decided according to the resources status in the network. Simulation results show that both two methods can significantly reduce the network blocking probability (up to 60%), with the one considering resource status performing better.Secondly, to address the fragmentation issues in flexible grid optical networks, this paper studies dynamic optimization and defragmentation algorithms. Specifically, based on the blocking service and network condition, this papre constructes two kinds of defragmentation algorithms. One is Maximum Spectrum Compactness Defragmentation (MSCD) algorithm, in which a local optimization is triggered when there exists a blocked services, and the path with maximum spectrum compactness is selected during the reconstruction process. The other one is Spectrum Compactness Trigger Defragmentation (SCTD) algorithm, in which an overall optimization is triggered when the network compactness is under a pre-defined threshold. Simulation results show that compared with MSCD, SCTD achieves better performance in terms of blocking probability, while with the cost of affecting much more exsiting services. In addition, the performance of SCTD has a great concern with the value of trigger threshold.Thirdly, to address the mismatches between services and spectrum resources in flexible grid optical networks, this paper studies the dynamic multi-layer traffic grooming problem. Based on the concept of spectrum engineering and auxiliary graph model, this paper proposes three traffic grooming herustics for sliceable transponders, namely Minimize Light Path-Spectrum Engineering (MLP-SE), Minimize Virtual Hop-Spectrum Engineering (MVH-SE) and Minimize Physical Hop-Spectrum Engineering (MPH-SE). All of those herustics are simulated in terms of bandwidth blocking probability, average number of transponders, average virtual hops and average physical hops. Simulation results show that different traffic grooming herustic has different advantages. For example, MLP-SE gets least energy consumption, MVH-SE achieves best QoS, and MPH-SE obtains highest spectrum efficiency. The decision on which traffic herustic should be selected depends on the major concern of network operators.Lastly, for the gradual migration from fixed grid to flexible grid in optical networks, this paper studies the inter-operation problems between fixed grid equipments and flexible grid equipments, and proposes the spectrum resource allocation schemes in this hybrid network scenario, In addition, this paper designes two kinds of gradual migration strategies. One is upgrading the entire node in the network gradually, while the other one is upgrading the WSS in the ROADMs gradually. For the former strategy, this paper has considered five heuristics, namely Highest Degree First (HDF), Highest Generated Traffic First (HGTF), Highest Carried Traffic First (HCTF), Most High-Bandwidth Traffic First (MHTF) and Most Low-Bandwidth Traffic First (MLFT). Simulation results shown that migrating to flexible grid can improve network capacity and lead to lower bandwidth blocking probability. This benefit can be achieved even if we only upgrade a small portion of node (even WSS) to support flexible grid technology.