Research On Resource Mapping Strategies In Network Virtualization

With the rapid development of Internet, the underlying networks and network service are evolving more and more heterogeneous, which makes the ossifications of the Internet highlighted. To overcome the ossifications and design flexible and extensible networks, Network Virtualization technology has been much touted as one of the most promising technologies for the next generation networks. Through virtualizing, aggregating and reallocating the physical network resource, network virtualization can decouple the service provisions from the underlying infrastructure network, and enables multiple customized Virtual Networks (VNs) to coexist on the same infrastructure independently. As one of the main research branches of the resource management in network virtualization, resource mapping can instantiate the virtual networks, and plays a crucial role to the realization of network virtualization. Considering the heterogeneity of underlying physical networks, we mainly focus on the resource mapping strategies in wired network virtualization and radio access network virtualization in this dissertation.In wired network virtualization, Infrastructure Providers (InPs) would allocate sufficient network resource to the arrived Virtual Network Request (VNR), known as Virtual Network Mapping (VNE). However, VNE is typical an online process to allocate resource to VNRs that submitted sequentially, and the embedding result of each VNR has a great impact on the oncoming ones. The existing algorithms are greedy to embed each individual VNR optimally but ignore its impacts on the future requests, which could aggregate the fragmentation of substrate network resource and ultimately lead to higher rejections of the future requests. From this perspective, all existing VNE algorithms still suffer some degree of blindness in the temporal dimension. To fill this gap, we first introduce a new link-related metric, named as link interference, for each substrate link to measure the interference caused by its bandwidth deficiency to accept VN requests, and then a heuristic Interference-based VNE (I-VNE) algorithm is proposed. Benefited from the new metric, I-VNE can jointly consider the temporal and spatial topology information of networks, and tries to embed each virtual network request with low interference to avoid rejecting the future requests. Our simulations show that, I-VNE can significantly improve the performances in terms of time averaged revenue, acceptance ratio, compared with existing VNE algorithms with only the global resource information in the spatial dimension.In radio access network virtualization, InP would allocation adequate radio resource to each Virtual Wireless Network (VWN) that coexists on the physical network. Due to the time-varying wireless channel and fluctuant traffic arrival rate, traditional static resource allocation may cause mismatch in the supply-and-demand chain, and ultimately result in inefficient resource utilization. To fill the gap, we take the traffic information into consideration and propose a throughput-maximum resource allocation in the OFDMA-based WVN, while guaranteeing the average resource requirement for each slice in a long run. As it’s difficult to guarantee the long term performance without the future traffic arrival information and wireless channel information, we employ the Lyapunov Optimization and develop an online throughput-maximum resource allocation algorithm to dynamically allocate network resource to each VWN. Theoretic analysis implies that the proposed algorithm can arbitrarily get close to the theoretic optimal throughput with the degradation in delay, and the two performances can be balanced on demand. Simulations show that the proposed resource allocation scheme significantly outperforms the existing schemes in terms of throughput and delay.