The Research And Implementation Of Routing Techniques For Optical Transport Network Based On Machine Learning

With the rapid development of mobile internet,internet of things,cloud computing and other emerging services,the scale of basic transport network is increasing day by day,and the network resource data is also increasing exponentially.The increment and expansion of network services drive the communication ability,operation ability and the transformation of network design mode of the basic load bearing network,and the instrumentalization of design task has become an inevitable trend.The planning and design work of inter provincial trunk transmission network based on WDM/OTN technology has gradually transformed into instrumentalization under the background of big data.However,due to the design of many scenarios of trunk transmission network,especially for routing and wavelength selection planning work,often rely on the designer experience,the decision-making process is very difficult to be ed to a logic program,so the design and development process of this sort of software,will inevitably run into bottlenecks,in which a"information chasm" exist between developers and designers.In order to solve this problem,data mining and machine learning are used as a tool to refine the influence factor model from the historical design decision in this paper,so as to realize the intelligent routing.The main work and achievements of this paper include the following aspects.(1)According to the characteristics of OTN resource data and historical routing data of telecom operators,data preprocessing is performed to generate two types of sample data,including resource influence sample and topology influence sample.Based on the machine learning classification algorithm,a decision parameter model is established.In the parameter model construction process,based on the historical data of the current network,several popular machine classification algorithms are simulated on the simulation platform and the best algorithm is selected.(2)A routing and wavelength assignment algorithm based on the decision-making parameter model is proposed.Based on the K-shortest path algorithm,the algorithm selects a set of candidate routes by using the topological influence parameters,and selects the optimal route from the candidate routes by using resources and topology influence parameters.After the historical route regression test,the coincidence rate of the route selected from the algorithm and the historical real decision route reaches 81%and the resource hit rate reaches 84%,which meets the requirement of actual production.(3)A routing subsystem under the transport network planning and design system is designed and developed to realize the software and product of the research results in this paper.The subsystem uses the above algorithm scheme,including the link parameter model training module,routing module and user interaction module.