The Research Of Embedding Algorithm For Wireless Network Virtualization

With the rapid development of various wireless communication technologies and intelligent terminals, the diversified mobile services are emerging, the future network presents a densely deployed and various services coexistence state. Wireless network virtualization technology provides an effective management for the complex network environment, which is also a key technology in the fifth generation mobile communications(5G) research. Through the uniform abstract and management ignoring the differences between the underlying physical networks, wireless network virtualization provides customized services for users. The thesis mainly focuses on virtual network embedding which is a key problem for wireless network virtualization, analyzes and summaries the advantages and disadvantages of the existing embedding algorithms. Then the thesis establishes a wireless network virtualization embedding model, and put forward two virtual network embedding algorithms.Firstly, to avoid the disadvantages of existing embedding algorithms, the problem of joint allocation of bandwidth and power for wireless virtual network embedding should be considered. This thesis formulates a model to describe the problem and proposes a Wireless Virtual Network Embedding Algorithm based on Joint Bandwidth and Power(WVNEA-JBP). The algorithm considers three constraint conditions, including node resources, link resources and geographical position, and fully considers the load balancing and interference between wireless network links, using path splitting to minimize the embedding cost. The simulation results validate that WVNEA-JBP can effectively improve resource utilization, success rate of virtual network request and reduce embedding cost on the premise to meet virtual network requests.Secondly, due to dynamic virtual network requests in wireless network virtualization, it may cause load imbalance, which not only affects the performance of the existing virtual network, but also is not conducive to the subsequent request embedding successful. Therefore, the thesis rethinks embedding virtual network. At first, analyzes the key problem in the process of rethinking embedding, and introducing the corresponding solution. Then, a novel Rethinking Embedding Algorithm based on Critical Virtual Node(REA-CVN) is proposed. The algorithm considers load balancing and the influence of existing virtual network. The simulation results show that REA-CVN algorithm reduces the embedding cost, improves the resource utilization and the success rate of virtual network requests, and saves the time overhead of embedding.