Research On Intelligent Routing And Resource Optimization In Software-Defined Multi-Dimension Optical Networks

The Generalized Multi-Protocol Label Switching(GMPLS)protocol based optical networks can provide a large-capacity,high-speed network connection.However,with the rapid development of network technology,network application services such as 5 G,Internet of Things,big data,and cloud computing are emerging one after another,which leads to the gradually becoming rigid control plane of optical networks.For example,it is both difficult to configure the optical network according to predefined policies,and to reconfigure it to respond to faults,load and application changes.Therefore,complex and ever-changing business demands pose new challenges to the flexible management and control of optical networks.Meanwhile,they also push the development of optical networks toward a more intelligent direction.As a new type of network architecture,software-defined networking(SDN)breaks the vertical rigid situation by separating the network's control logic from the underlying devices,which promotes the centralization of the network's control logic and introduces the ability to program the network.Thanks to the deep integration of SDN ideas and optical networks,the birth of software-defined optical networks(SDON)has been promoted.SDONs make it easier for optical control planes to create and introduce new network abstractions,simplifying network management,and facilitating network evolution.However,based on SDONs,there are still many issues that need to address resiliency communication,delay-sensitive traffic transmission,multi-granularity network control,and open and flexible network management.At present,the failure recovery scheme in the optical network adopts a protection strategy,and a protection path-which is parallel to the working path-is pre-configured by the controller.When a network failure is detected,it allows the data forwarding plane to play the role of failure recovery.However,protection solutions cannot handle situations where both working and protected paths break down.At the same time,for delay-sensitive services,there is no effective delay monitoring mechanism.Note that,both the delay measurement and the traffic engineering are important.If operators want to provide a delay guaranteed service for various users in SDONs,it is necessary to design a monitoring mechanism and develop a traffic scheduling algorithm.In addition,in the hybrid optical network with fixed-grid wavelength division multiplexing and flexi-grid orthogonal frequency division multiplexing technologies,a more flexible routing and resource allocation scheme is needed to further improve the utilization efficiency of spectral resources.Finally,in order to further open up the network flexibility based on architecture on demand nodes and multi-cores fiber links,it is imperative to design a versatile optical control plane to manage and control the underlying complex infrastructure.Therefore,the research on intelligent routing and resource optimization in software-defined multi-dimension optical networks has important academic significance and application value.The contributions in this dissertation are sunmmarized as follows:(1)The second chapter studies the routing scheme of failure recovery in software-defined optical networks.First,the author focuses on failure recovery solutions for data planes in software-defined optical networks.The out-of-band control mechanism is utilized for the communication between the controller and the data forwarding elements.Next,a failure detection scheme,a dynamic all pairs shortest paths algorithm,and a failure recovery application are proposed.Then,the overall feasibility of dynamic end-to-end path recovery is verified by demonstrating the experiment.Finally,the performance is quantitatively evaluated in terms of the end-to-end delay,CPU utilization of controller,and traffic blocking probability,which shows the effectiveness of the overall solution.The results indicate that,in designed three tested scenarios,the whole processes including the connection setup and the failure restoration can be completed in 50 ms,which satisfies the carrier-grade requirements.(2)The third chapter studies a delay-sensitive service guaranteed routing scheme in software-defined optical networks.First,the author designs a monitoring mechanism of link delay by using both the link layer discovery protocol(LLDP)monitoring module and the Echo monitoring module.Then,a dynamic shortest delay routing algorithm is proposed,i.e.,SDRA,in order to enhance the security of data transmission and meet delay-sensitive traffic needs.Next,the author develops a routing application assorted with both the monitoring mechanism and the SDRA algorithm by extending the RYU controller.In addition,the overall feasibility of the proposed solution is experimentally demonstrated and verified in a semi-practical SDON testbed.Finally,based on the NSFNET network topology,the author quantitatively evaluates the performance of the SDRA scheme in terms of the average path delay,the processing latency of the controller,and the end-to-end delay.The SDRA-based delay solution is more advantageous than the shortest hop routing algorithm.(3)The fourth chapter studies software-defined routing and resource allocation schemes in multi-granularity optical networks.First,the author elaborates on the overall network architecture of software-defined multigranularity optical networks,and extend the OpenFlow protocol to implement the seamless operation between control and data planes.Next,the mathematical problem of allocating multigranularity network resources is formulated.For transparent flexi-grid orthogonal frequency division multiplexing(OFDM)optical networks,a novel routing,modulation level and spectrum allocation algorithm(RMLSA)based on spectral efficiency and connectivity(SEC)is designed,i.e.,SEC-RMLSA.For opaque fixed-grid wavelength division multiplexing(WDM)optical networks,the algorithm is proposed to select the lightpath with the maximized capacityover-hops first(MCHF)from the large-capacity virtual network.Then,well-designed weights and maximum flow algorithms are utilized to build virtual networks between source and destination nodes,which helps algorithms to find the optimal traffic scheduling scheme in real time according to dynamic link load.Moreover,the author develops multi-granularity routing applications by extending the RYU controller.Finally,the designed solution is implemented in a semi-practical system testbed.The overall feasibility and efficiency of the proposed solution are experimentally verified and evaluated based on the NSFNET topology.The results show that the SEC-RMLSA and MCHF algorithms can improve spectral efficiency and thus reduce the traffic blocking probability.(4)The fifth chapter studies the routing and resource allocation scheme in softwaredefined space division multiplexing optical networks.First,the software-defined space division multiplexing optical network architecture is described,and the routing and resource allocation optimization models are formulated based on architecture on demand(AoD)reconfigurable nodes in the software-defined space division multiplexing optical network.Next,the fiber core classification algorithm is designed according to the service type to avoid the crosstalk between adjacent cores in multi-cores fiber.Based on the core classification algorithm,the required AoD optical nodes can be synthesized.In addition,in the optical networks with both the multi-cores fiber and the AoD nodes,the routing and resource allocation algorithms for multiple types of services are proposed,and a related routing application is developed.Finally,under different traffic loads,the performance based on the maximum capacity first(MCF),the shortest distance first(SDF),and the minimum hop first(MHF)algorithms are compared.For time division multiplexing,wavelength division multiplexing,and orthogonal frequency division multiplexing(three types of service requests),the three algorithms exhibit different network performance in terms of the blocking probability,the average number of core conversions,the average number of wavelength conversions,and the average number of spectrum slot conversions.Therefore,under different network loads,flexible switching of different routing and resource allocation algorithms is recommended to adaptively control the network.