The Research And Implementation Of Packet Transport Network End–to–End Routing Optimization

The modern communication technology is developing fast and changing rapidly.The growing consumer demand makes the traditional circuit-switched networks are difficult to meet the needs of the new market.PTN is the newly invited technology which is applied in order to solve the above-described problem.The core of PTN is MPLS-TP,it makes some difference of traditional MPLS network,the new features of OAM was added on and some features of connectionless like IP was cut off.The most important new added feature is end-toend of the end-customer demand,It requires only need to set the basic information and QoS traffic from source to sink o that network management software itself can selects an optimal path to meet the needs of services.Especially in the case of large-scale of network deployment,the PTN Network Management System need to create services route automatically.In order to protect the stability of the services.PTN network routing algorithm must be accurate and smart.The thesis have analyzed the PTN NMS routing algorithm and routing selection strategy,and found two ways to improve the routing algorithm in PTN,the major research of the thesis is as follws:1)The thesis has analyzed the mathematical modeling of history alarm of PTN.After sampling form the history alarm database and weighting operation,the time series of alarm was established.The time series must go through the stationary processing.Next step is being fitting with the ARMA model.After finding the fitted ARMA model,It can be used to forecast the reliability of the routes in PTN networks.2)The thesis has analyzed SPF algorithm used to routing in the original network,and introduced the new another algorithm-PSO.discussed that the particle swarm optimization for solving the shortest path problem,transform the original particle swarm optimization formula into required form for solving the PTN routing Then solve the low efficiency of particle searching,solved the problem of fitness function value in coonvergence,while the velocity and position update formula particles were improved by increasing the disturbance,reduced particle algorithm precocious leaving Solutions possibility.3)The thesis has researched the requirements of NMS LSP management module,designed the object model of LSP with covering the requirements.Implement the module of management of LSP trail service with combining forecast of link alarm data and improved routing algorithm,The module has increased the stability of services and reduced the resource utilization of system,risk of end-toend services failure and cost of maintenance.Innovation point of the thesis:Using ARMA model to forecast the stability of service link based on NMS history alarm data;Implement the optimization of NMS LSP management routing module by using improved PSO algorithm.