Research On Construction Algorithms Of Service Carrying Networks

Over the years, the innovation of internet applications has been emerging boomingly. A large number of personalized and small applications spring up and gradually c hange from small-scale to large-scale. It is very difficult for single network architecture or service model to support diverse network op erations. Thus, through constructing a service carrying network(SCN) and on the basis of shared underlying physical resources, the reconfigurable flexible network(RFNet) technology system supports the coexistence of multiple heterogeneous network and provides users with a variety of network services. So the construction mechanism of SCN is bound to become one of the key issues of reconfigurable network technology research.This paper relies on the sub-topics of the national 863 Project "the development of reconfigurable routing components " and "equipment engineering implementation of reconfigurable network test". After summarizing the characteristics and request of construction, this paper studies the construction algorithms of clustered three SCN s. Specific research work is as follows:1. The normal existing public SCN is huge-scaled, easily managed and clustered. C hapter2 proposes the SCN construction algorithms based on hybrid swarm intelligence optimization(SCNC-PSO-GA). The algorithm utilizes swarm intelligence optimization to avoid the local optimum when approaching the best mapping scheme. And also, the algorithm overcomes the connectivity among nodes with the help by the balanced bandwidth multi-commodity flow algorithm. The simulation results show that the algorithm has high request rate. Compared with D-ViNE-SP、D-ViN E-LB and VNE-R-PSO, Income overhead ratio increase by 55%、23% and 6%. Revenue of the reconfigurable network raises a lot.2.The longtime leased personalized SCN is a middle-scaled, and need special function. Chapter 3 proposes a construction algorithm based on the analysis of the network centrality(SCNC-ANC) for distributed systems who ask to carry service on personalized SCN aiming at decreasing the cost of client and improving access bandwidth of user node. The algorithm quantifies the node’s connectivity with the others in the global view, the probability of the path to be mapped, and provide reference to increasing access bandwidth for outer nodes. And also, the algorithm constructs SCN according to the max- min matching principle. The simulation results show that the algorithm has a high success rate of establishment. Compared with VN E-ICC and VNE-Greedy, the access bandwidth of other nodes increase 14% and 32%,which improve the QoS of distributed systems.3.The short time existing real-time SCN is small- scaled, and require t fast construction of network. C hapter 4 proposes a SCN construction algorithm based on K-core decomposition(SCNC-KCore). In order to compress the searching space, the algorithm constructs the relay carrying subnet by a similar process of K-core decomposition. The candidate nodes connect t he relay carrying subnet and then construct SCN. As simulation results show, through reduces the search space, construction time of the algorithm is only 50% of VN E-R-PSO’s, 23% of D-ViNE-LB’s and 27% of D-ViN E-SP’s. And the algorithm improves the immediacy of the construction.