Research On Resource Allocation Of Maximizing Multi-Class Service Utility In Peer-to-Peer Networks

In recent years,with the increasing use of Peer-to-Peer(P2P)networks in different fields,the demand for Internet resources by users has also increased rapidly.However,the transmission capacity of the current network is limited,the speed of addition and upgrade of network equipment can no longer meet the user's demand for network bandwidth growth.Therefore,how to allocate network resources for users becomes one of the hot issues in P2 P network.Based on utility theory in microeconomics,this thesis establishes a utility maximization model for resource allocation problem of multi-class services in P2 P networks from the perspective of user satisfaction,and applies nonlinear programming theory to analyze the optimal solution of the model.A subgradient-based distributed resource allocation algorithm is proposed for the case of only elastic services.For the multi-class services case where elastic services and inelastic services coexist,this thesis proposes a particle swarm optimization(PSO)algorithm to solve this complex non-convex optimization problem.The main content of the thesis is as follows:First of all,according to the different shapes of network service utility functions,the services in P2 P networks are divided into non-real-time elastic services and real-time inelastic services.Firstly,a utility-maximizing resource allocation model is established for the case where there are only elastic services in the networks.The optimal expression of elastic service resource allocation is obtained and an improved subgradient-based distributed resource allocation algorithm is proposed to solve the optimization problem.Second,this thesis considers the coexistence of elastic services and inelastic services in P2 P networks,and establishes a resource allocation model based on utility maximization for inelastic services.The model is analyzed by applying the nonlinear programming theory.This thesis extracts the threshold of upload bandwidth that the optimal resource allocation should satisfy when multi-class services coexist.A heuristic algorithm based on particle swarm optimization is proposed to solve this non-convex optimization problem with multi-class services.Finally,this thesis builds a small-scale peer-to-peer network,and analyzes effectiveness of the P2 P networks resource allocation mechanism based on utility maximization.Firstly,the improved sub-gradient algorithm is used to simulate the model with only elastic services.Then simulation results verified the effectiveness and convergence of the proposed algorithm.Secondly,for multi-class services situation in which elastic services and inelastic services coexist,this thesis uses PSO-based resource allocation algorithm,sets corresponding parameters for simulation.Simulation results verified the convergence and acceleration of the speed to seek optimal solution of the particle swarm optimization algorithm.