Research On The Future Intelligent Network Resource Management

With the fast development of current Internet, our way of life and working is deeply changed by the new technology. However, the rapidly increasing of Internet scale and the multiple requirements leads to the fact that many problems of current Internet architecture emerge now. Firstly, it is hard to develop a new technology because of the disequilibrium between the investment and revenue. Besides, new change needs a wide range of cooperation because of the IP protocol’s end-to-end design.The network virtualization is a powerful way to overcome the mentioned problems. The network virtualization technology provides a platform which allows multiple virtual networks share a same substrate network simultaneously. Different protocols can be used between different platforms. Besides, virtual network technology separates the infrastructure providers and the service providers. The InPs mainly focus on providing and managing the substrate network while the SPs mainly focus on providing service for the end users.The network embedding technology means that multiple virtual network run on the same substrate network in an effective, parallel and fast way while satisfying the constraints such as node CPU resource and link bandwidth resource. This paper will dig into the network resource allocation and virtual network embedding algorithm to improve the system’s revenue, resource utilization and algorithm efficiency.The main innovations of our paper are as follows:1) With the investigation of the economic theory such as game theory, auction theory, our paper proposes a resource allocation approach based on auction theory. The result of resource allocation can guide the virtual network embedding.2) With the investigation of the current virtual network embedding algorithm, our paper proposes a virtual network algorithm base on the novel adjacent principle.3) Multiple simulations show that our algorithm can largely increase the revenue and R/C ratio, decrease the cost and runtime. And we can improve the load balance of the substrate network.