Research On Clustering-based Task Fault-tolerant Scheduling Of Ad Hoc Network

With the increasingly widespread use of mobile communication network,the mobile communication network with the infrastructure we usually use has been unable to meet people's demand for mobile communication network.It is hoped that when people have a need no matter when and where they are able to take timely access to the Internet.In these cases,we need some temporary structures which can be quickly build and put into use.Ad hoc network is such a kind of mobile network.Ad hoc network is a new mobile computing network,with no center,self-organizing,dynamic topology,etc,which is so different characteristics compared with traditional mobile network.Its own particularity to endow it with great prospects for a variety of fields,but it also brings a number of other the problems to be solved,such as network resource management,fault-tolerant task scheduling,security and QoS support,etc.These issues has caused more and more scholars of continuous attention and research.In this thesis,clustering and task fault-tolerant scheduling of Ad hoc network have been studied.In the clustering stage,this paper proposes a new PMCDA algorithm through combining Highest Degree algorithm(HCAD)and Weighted Clustering Algorithm(WCA)which is improved respectively.The algorithm takes into account both the performance and the average connection degree of the node.In task scheduling stage,we put forward a new IMP-PSO-ACO algorithm by analyzing the particle swarm optimization algorithm(PSO)and ant colony algorithm(ACO).The PSO and ACO algorithm are combined and was improved.It uses the particle swarm optimization algorithm to generate the pheromone distribution,which is also used in the ant algorithm for the exact solution.In fault detection phase,the heartbeat message mechanism is adopted to perform fault diagnosis on the task node,we present a practical calculation method to predict heartbeat margin time.In the recovery phase,to avoid the repetition of task execution and wasting a lot of time and resources,checkpoint mechanism was introduced in large-scale tasks,which can allows users who submit tasks get timely and reliable execution.Finally,simulation results prove that PMCDA clustering algorithm can improve the stability of the cluster structure to a certain extent and reduce the various aspects of the related expenses.IMP-PSO-ACO algorithm has obvious effect for task scheduling optimization,which can shorten the total time to complete the task.