Research On Green Grooming Mechanisms And Optimization Algorithms In IP Over WDM Optical Networks

With the development of Generalized Multi-Protocol Label Switching (GMPLS) technology and Automatic Switching Optical Network (ASON), the IP network has been seamlessly integrated with the optical network to realize the IP over WDM (Wavelength-Division-Multiplexing) optical network. As one of core backbones, this network paradigm gives vendors enormous pressure from reducing energy consumption and equipment expenditure. More importantly, the new requirements such as green, multicasting, multi-domain, realistic, heterogeneous self-reconfiguration, and network virtualization, encourage new characteristics and applications in IP over WDM optical network. With the increasing traffic, the excessive energy consumption has become one of major obstacles for the network development. Therefore, the academic and industrial communities put extensive attentions on designing the IP over WDM optical network with high resource utilization, high energy efficiency and low port cost. Moreover, applying aforementioned designs to the new IP over WDM network, which has novel characteristics of realistic, heterogeneous self-reconfiguration, and supporting cloud computing, will become one of the hot topics with the significant prospects of academics and industrialization.Currently, most energy-saving algorithms in IP over WDM optical network deploy the traffic grooming, but energy efficiency, i.e., the trade-off between the energy consumed for establishing lightpaths and the energy saved by optical bypass, is not within the scope. Furthermore, the joint optimization of energy efficiency and port cost is not well addressed because the existing traffic grooming fails to coexist with the waveband switching in terms of complementary advantages. More importantly, the unpredictability of real-time traffic tends to be obvious, the multicast and many-to-many sessions increasingly play an important role, and the network architecture starts to present the multi-domain features. Clearly, the existing traffic grooming cannot be directly applied in such new IP over WDM optical networks. In addition, the heterogeneous self-reconfigurable and cloud computing oriented IP over WDM optical network has the urgent requirements for more flexible grooming mechanisms and optimization algorithms, in order to mitigate the gra-diversity problems and further improve the utilization of the underlying infrastructure. Therefore, the research on green grooming mechanisms and optimization algorithms for IP over WDM optical networks has the considerably academic and industrial significance.The contributions and innovations in this dissertation are summarized as follows:(1) The second chapter addresses the green unicast traffic grooming mechanism in IP over WDM optical networks.First, the author investigates the energy efficiency of the IP over WDM optical network, analyzes the energy consumption for establishing lightpaths and the energy saved by using traffic grooming, and derives a quantitative model for energy efficiency in the single-domain network. Next, the green unicast traffic grooming based on the optical bypass reconfiguration is proposed, and it separates a lightpath at every bypass node, which makes it easy to groom traffic as much as possible into the pre-established lightpaths. Then more energy is saved. Finally, the non-cooperative gaming theory is applied into the design of the green unicast traffic grooming. The double-win between vendors and users is achieved by gaming on the path link in an energy-efficient IP over WDM optical network.(2) The third chapter addresses the green traffic grooming mechanism for supporting multicasting and multi-granularity in IP over WDM optical networks.First, the green multicast traffic grooming with the different combinations of strategies is presented, and an optimal strategy combination is derived by the simulation, in order to provide the open guidelines for network designers. Next, the green routing and aggregation algorithm tailored for the emerging many-to-many sessions is devised through separating the large sessions. Finally, based on the hybrid hierarchical optical cross-connect architecture, a linear programming is designed at the node port level, with its aims to perform the joint optimization of all optical transmission port cost and the energy consumption of OEO (Optical-Electrical-Optical) matrix ports. Correspondingly, the green traffic grooming that supports waveband switching is also proposed.(3) The fourth chapter addresses the green integrated grooming mechanism in IP over WDM optical networks.First, the advantages of traffic grooming and waveband switching are effectively combined. The traffic aggregation is realized by traffic grooming, thus improving both lightpath utilization and energy efficiency. By means of waveband switching, the high-load lightpaths are then grouped into fewer waveband tunnels for reducing port costs. Next, the author extends the above green integrated grooming to the multicast application. A new statistical approach is designed to compute the blocking probability of users, and heuristics are also proposed, each of which has a different optimization objective for blocking probability. Finally, both intra-domain and network-wide energy efficiency are investigated in the multi-domain IP over WDM optical network. The author analyzes the effect of the link-cost adjustment on energy efficiency and derives the corresponding quantitative model. The integrated advantage from traffic grooming and waveband switching is also involved.(4) The fifth chapter addresses the green robust grooming mechanism in the realistic IP over WDM optical networks.First, given the limited information model in the realistic network, the corresponding power ratio is derived, and it is defined as the ratio of energy consumed for establishing lightpaths over the energy saved by traffic grooming. The green robust grooming mechanism is thus designed with the objective of optimizing this parameter. Next, the hose-model separation approach is proposed to compute power ratios for different grain levels. With this optimized parameter, the optical virtual topology is constructed followed by the principle that "higher grain level is in priority". Finally, considering the peak traffic tends to scatter over boundary nodes in the realistic multi-domain network, the author separates the hose model and derives the corresponding energy efficiency from a new perspective.(5) The sixth chapter addresses the grooming mechanism and optimization algorithm for heterogeneous self-reconfigurable IP over WDM optical networks.First, with the help of software-defined optical transmission technique, the self-shift of heterogeneous grooming approaches is realized at the software layer, so as to process the multi-grain level traffic in the various manners. Correspondingly, the traffic partition grooming is proposed to solve the gra-diversity problems. Next, the author applies the above traffic partition grooming into the multi-domain scenario. The traffic migration is also involved, and it can transfer the pre-blocked traffic into the pre-established tunnel in the plane with another grain level, which reduces the blocking probability. Finally, the author designs a novel spectrum defragmentation scheme based on virtual concatenation. The proposed approach not only mitigates the side effect of spectrum shifting on the online service performances but also reduces the blocking probability through consolidating fragments.(6) The seventh chapter addresses the resource optimization algorithm for cloud computing oriented IP over WDM optical networks.On the one hand, a provident resource defragmentation algorithm is proposed for cloud data center, where the future distribution of the arriving VM (Virtual Machine) requests can be predicted so that the maximal profit can be determined as the starting and ending point of the resource defragmentation process. This algorithm not only mitigates the excessive VM migration overhead, but also reduces the defragmentation delay. On the other hand, the author presents a collaborative resource virtualization algorithm for cloud computing oriented IP over WDM optical networks. The combination of traffic grooming, server consolidation, anycast routing and resource balancing is achieved to maximize the revenue of cloud tenants.To determine and evaluate the performance of proposed green grooming mechanisms and optimization algorithms in IP over WDM optical networks, the simulation platform is developed. Simulation results demonstrated that the proposed mechanisms and algorithms can achieve the joint optimization of energy efficiency and port cost, the highly efficient resource utilization or the maximal revenue in the multicast, multi-domain, realistic, heterogeneous self-reconfigurable or cloud computing oriented IP over WDM network, respectively. The research achievements in this dissertation are able to provide the abundant theoretical foundation and technical guarantee for the development of future optical networks.