Research On Reconfigurable Service Carrying Network Topology Adaption Technology

On the basis of sharing different substrate resource, Reconfigurable Flexible Network(RFNet) supports the coexistence of multiple heterogeneous network architectures byconstructing the Reconfigurable Service Carrying Networks (RSCNs), and therefore providesusers with a variety of network services. However, during the practical operation of RFNet, thesubstrate network topology will contract when an abnormal reconfiguration or a node/link failurehappens, and expand when a new node/link is added. So it’s common that the substrate networktopology is dynamically changed on different time scales. How to migrate the RSCNs to adaptthe substrate network topology change optimizedly, namely RSCN Topology Adaption Problem,is an important topic to be solved.Supported by the “Research and Development of Reconfigurable Router Components” and“Engineering Implementation of Reconfigurable Flexible Testbed Equipments” projects of theNational High-Tech Research and Development Program of China (863Program), this thesisfocuses on the aspects of the resource balance and service reliability in RSCN TopologyAdaption Problem to carry out research, and also a simulation platform and an applicationplatform are designed. The detailed research contents are outlined as follows:1. Aiming at improving the resource balance when the substrate network topology expands,a Migration-Efficiency-based resource Migration (ME-Mig) scheme is proposed. In this scheme,the load balance factor and the migration cost factor are integrated into a new concept calledmigration efficiency. Based on the principle of migration efficiency maximization, theprocedures of atomic request migration, resource fault-tolerant migration and resource balancemigration are designed. The resource unbalance problem is solved by resource balance migration,and the resource invalidation problem is solved by resource fault-tolerant migration. Simulationresults show that the ME-Mig scheme increases the request accepted ratio and the network loadbalance degree, while decreasing the migration cost of RSCNs, and make use of networkresource balancelier with less migration cost.2. Aiming at improving the service reliability of delay-insensitive RSCN when the substratenetwork topology contracts, a reliable mapping scheme based on topology impact degree andbacktracking migration is proposed. The RSCN Mapping algorithm based on Topology ImpactDegree (RNM-TID) uses topology impact degree factor, which is based on topology shape andmapping location, to evaluate and perform the RSCN mapping, and therefore reduces theinfluenced scope when a RSCN failure happens. The Migration Algorithm based onBacktracking (MA-Back) uses multi-step migration mode to increase the migration success ratio, and effectively control the migration cost. Simulation results show that this scheme increases therequest accepted ratio, migration success ratio and valid loading ratio, and therefore improve thereliability of RSCN.3. Aiming at improving the service reliability of delay-sensitive RSCN when the substratenetwork topology contracts, a reliable mapping scheme based on minimum backup topologygeneration is proposed. In this scheme, the reliable mapping is divided into two stages whichinclude backup topology generation and mapping. In the backup topology generation stage, aminimum backup topology generation algorithm is proposed to decrease the total bandwidthconstraints of backup topology. In the backup topology mapping stage, a Reconfigurable servicecarrying Network Mapping algorithm with coordinated Primary and Backup Topology(RNM-PBT) is proposed. The RNM-PBT algorithm lets the primary topology and backuptopology perform mapping by turns, so the primary and backup topology resource can beallocated effectively. Simulation results show that this scheme decreases the average resourceconsumption and execution time cost, while increasing the request accepted ratio of RSCNs.4. A simulation platform and Reconfigurable Management Platform (RMP) are designed tosupport the RSCN topology adaption technology. Based on the open source simulation tool, byadding the event sequence model and the simulation control model while implementing thealgorithm model, the simulation platform called RNM-Lab can carry out the simulation oftopology adaption algorithms in dynamic network model. By embedding the RNM-Lab functionmodels into the RMP’s core process layer, the RMP gains the topology adaption managementability.