Research On Resource Allocation Algorithm

With the development of Internet, network resources as a new kind of resource are growing rapidly. Network resources may include electronic documents (e-document), digital video, network information, web application, web service and so on in broad sense. For those network resources which can be copied, such as electronic documents and video, people can download from the Internet easily. However, for those scarce network resources, such as network space allocation, web service, how to find the reasonable allocation algorithm of these scarce network resources, optimizing the overall welfare of the system has become an important research topic, and is also the main issue of this paper.Network resource allocation is defined as the efficient allocation of scarce network resources to the consumers owning the need under a certain algorithm to achieve high social welfare, that is the resource providers and consumers both achieve high welfare. The traditional resource allocation algorithm pay more attention to how to improve the allocation efficiency, but little attention to the properties of resources, and thus there is lack of a targeted, reasonable allocation algorithm to improve the allocation of scarce network resource efficiency further.This paper divides the scarce network resources into two categories - pricing based and non-pricing based network resources. And according with the characteristics of these two kinds of resources, we use modeling analysis and introduce the network flow based market clearing prices algorithm and combinatorial auction mechanism of algorithm game theory to solve the problems of these two kinds of network resource allocation.The main contributions of this paper are the following:First, network resources can be divided into pricing and non-pricing, and analyzed by modeling separately;Second, regarding the pricing of network resources can be combined with the actual space allocation of network teaching platform, using the network flow model and the algorithm of compute market clear price, we designed a resource allocation method: flexible resources quantity computing function, and network flow model based resource allocation algorithm; then we using JAVA to implement the algorithm and obtain a visual tool which can compute the equilibrium prices and flexibly adjusted resource-related parameters, and get the optimal result of resource allocation.Finally, for the non-pricing network resources, regarding the non-pricing of network resources can be combined with the actual course allocation of network teaching platform, we designed the user welfare function, introduced the combinatorial auction based resource allocation algorithm, and gained the efficient allocation of resources with high social welfare. CPLEX with JAVA combination is used to implement the algorithm, and after repeated experiments, the experimental data is obtained. Comparing with the commonly used ranking-based or reputation-based allocation algorithm, the analysis of the experimental data indicated that the allocation of non-pricing scare network resource was more efficiently and could gain higher social welfare and took into account efficiency and fairness for the resources allocation.The network resource allocation algorithms in this paper have expansibility. Network flow model based pricing network resource allocation could be use for network bandwidth allocation, space allocation and reservation in cloud computing; combinatorial auction-based non-pricing network resource allocation also suitable for the distribution of public resources. The visual tool is easy to use and could be applied in the field of other pricing resource allocation.